Model save
Browse files- README.md +12 -12
- adapter_model.safetensors +1 -1
- all_results.json +17 -17
- eval_results.json +14 -14
- train_results.json +3 -3
- trainer_state.json +1026 -1026
README.md
CHANGED
@@ -15,17 +15,17 @@ should probably proofread and complete it, then remove this comment. -->
|
|
15 |
|
16 |
This model is a fine-tuned version of [alignment-handbook/zephyr-7b-sft-full](https://huggingface.co/alignment-handbook/zephyr-7b-sft-full) on the None dataset.
|
17 |
It achieves the following results on the evaluation set:
|
18 |
-
- Loss: 0.
|
19 |
-
- Rewards/chosen: -
|
20 |
-
- Rewards/rejected: -
|
21 |
-
- Rewards/accuracies: 0.
|
22 |
-
- Rewards/margins:
|
23 |
-
- Logps/rejected: -
|
24 |
-
- Logps/chosen: -
|
25 |
-
- Logits/rejected: -2.
|
26 |
-
- Logits/chosen: -2.
|
27 |
-
- Use Label:
|
28 |
-
- Pred Label:
|
29 |
|
30 |
## Model description
|
31 |
|
@@ -62,7 +62,7 @@ The following hyperparameters were used during training:
|
|
62 |
|
63 |
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen | Use Label | Pred Label |
|
64 |
|:-------------:|:-----:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|:---------:|:----------:|
|
65 |
-
| 0.
|
66 |
|
67 |
|
68 |
### Framework versions
|
|
|
15 |
|
16 |
This model is a fine-tuned version of [alignment-handbook/zephyr-7b-sft-full](https://huggingface.co/alignment-handbook/zephyr-7b-sft-full) on the None dataset.
|
17 |
It achieves the following results on the evaluation set:
|
18 |
+
- Loss: 0.2409
|
19 |
+
- Rewards/chosen: -2.7432
|
20 |
+
- Rewards/rejected: -6.3660
|
21 |
+
- Rewards/accuracies: 0.7340
|
22 |
+
- Rewards/margins: 3.6228
|
23 |
+
- Logps/rejected: -322.9595
|
24 |
+
- Logps/chosen: -311.6890
|
25 |
+
- Logits/rejected: -2.6650
|
26 |
+
- Logits/chosen: -2.6975
|
27 |
+
- Use Label: 6842.4238
|
28 |
+
- Pred Label: 9189.5762
|
29 |
|
30 |
## Model description
|
31 |
|
|
|
62 |
|
63 |
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen | Use Label | Pred Label |
|
64 |
|:-------------:|:-----:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|:---------:|:----------:|
|
65 |
+
| 0.2135 | 1.0 | 955 | 0.2409 | -2.7432 | -6.3660 | 0.7340 | 3.6228 | -322.9595 | -311.6890 | -2.6650 | -2.6975 | 6698.4238 | 8833.5762 |
|
66 |
|
67 |
|
68 |
### Framework versions
|
adapter_model.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 218138576
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:ca4e6feb10113e9de9b08081514996934a6b42d53e8760d60528af986c57ddb4
|
3 |
size 218138576
|
all_results.json
CHANGED
@@ -1,23 +1,23 @@
|
|
1 |
{
|
2 |
"epoch": 1.0,
|
3 |
-
"eval_logits/chosen": -2.
|
4 |
-
"eval_logits/rejected": -2.
|
5 |
-
"eval_logps/chosen": -
|
6 |
-
"eval_logps/rejected": -
|
7 |
-
"eval_loss": 0.
|
8 |
-
"eval_pred_label":
|
9 |
-
"eval_rewards/accuracies": 0.
|
10 |
-
"eval_rewards/chosen": -
|
11 |
-
"eval_rewards/margins":
|
12 |
-
"eval_rewards/rejected": -
|
13 |
-
"eval_runtime":
|
14 |
"eval_samples": 2000,
|
15 |
-
"eval_samples_per_second": 4.
|
16 |
-
"eval_steps_per_second": 0.
|
17 |
-
"eval_use_label":
|
18 |
-
"train_loss": 0.
|
19 |
-
"train_runtime":
|
20 |
"train_samples": 61135,
|
21 |
-
"train_samples_per_second": 2.
|
22 |
"train_steps_per_second": 0.038
|
23 |
}
|
|
|
1 |
{
|
2 |
"epoch": 1.0,
|
3 |
+
"eval_logits/chosen": -2.6974706649780273,
|
4 |
+
"eval_logits/rejected": -2.665019989013672,
|
5 |
+
"eval_logps/chosen": -311.6889953613281,
|
6 |
+
"eval_logps/rejected": -322.95947265625,
|
7 |
+
"eval_loss": 0.2408759742975235,
|
8 |
+
"eval_pred_label": 9189.576171875,
|
9 |
+
"eval_rewards/accuracies": 0.734000027179718,
|
10 |
+
"eval_rewards/chosen": -2.7431609630584717,
|
11 |
+
"eval_rewards/margins": 3.6228184700012207,
|
12 |
+
"eval_rewards/rejected": -6.36598014831543,
|
13 |
+
"eval_runtime": 452.5604,
|
14 |
"eval_samples": 2000,
|
15 |
+
"eval_samples_per_second": 4.419,
|
16 |
+
"eval_steps_per_second": 0.276,
|
17 |
+
"eval_use_label": 6842.423828125,
|
18 |
+
"train_loss": 0.31699458866219243,
|
19 |
+
"train_runtime": 25218.7851,
|
20 |
"train_samples": 61135,
|
21 |
+
"train_samples_per_second": 2.424,
|
22 |
"train_steps_per_second": 0.038
|
23 |
}
|
eval_results.json
CHANGED
@@ -1,18 +1,18 @@
|
|
1 |
{
|
2 |
"epoch": 1.0,
|
3 |
-
"eval_logits/chosen": -2.
|
4 |
-
"eval_logits/rejected": -2.
|
5 |
-
"eval_logps/chosen": -
|
6 |
-
"eval_logps/rejected": -
|
7 |
-
"eval_loss": 0.
|
8 |
-
"eval_pred_label":
|
9 |
-
"eval_rewards/accuracies": 0.
|
10 |
-
"eval_rewards/chosen": -
|
11 |
-
"eval_rewards/margins":
|
12 |
-
"eval_rewards/rejected": -
|
13 |
-
"eval_runtime":
|
14 |
"eval_samples": 2000,
|
15 |
-
"eval_samples_per_second": 4.
|
16 |
-
"eval_steps_per_second": 0.
|
17 |
-
"eval_use_label":
|
18 |
}
|
|
|
1 |
{
|
2 |
"epoch": 1.0,
|
3 |
+
"eval_logits/chosen": -2.6974706649780273,
|
4 |
+
"eval_logits/rejected": -2.665019989013672,
|
5 |
+
"eval_logps/chosen": -311.6889953613281,
|
6 |
+
"eval_logps/rejected": -322.95947265625,
|
7 |
+
"eval_loss": 0.2408759742975235,
|
8 |
+
"eval_pred_label": 9189.576171875,
|
9 |
+
"eval_rewards/accuracies": 0.734000027179718,
|
10 |
+
"eval_rewards/chosen": -2.7431609630584717,
|
11 |
+
"eval_rewards/margins": 3.6228184700012207,
|
12 |
+
"eval_rewards/rejected": -6.36598014831543,
|
13 |
+
"eval_runtime": 452.5604,
|
14 |
"eval_samples": 2000,
|
15 |
+
"eval_samples_per_second": 4.419,
|
16 |
+
"eval_steps_per_second": 0.276,
|
17 |
+
"eval_use_label": 6842.423828125
|
18 |
}
|
train_results.json
CHANGED
@@ -1,8 +1,8 @@
|
|
1 |
{
|
2 |
"epoch": 1.0,
|
3 |
-
"train_loss": 0.
|
4 |
-
"train_runtime":
|
5 |
"train_samples": 61135,
|
6 |
-
"train_samples_per_second": 2.
|
7 |
"train_steps_per_second": 0.038
|
8 |
}
|
|
|
1 |
{
|
2 |
"epoch": 1.0,
|
3 |
+
"train_loss": 0.31699458866219243,
|
4 |
+
"train_runtime": 25218.7851,
|
5 |
"train_samples": 61135,
|
6 |
+
"train_samples_per_second": 2.424,
|
7 |
"train_steps_per_second": 0.038
|
8 |
}
|
trainer_state.json
CHANGED
@@ -75,1500 +75,1500 @@
|
|
75 |
{
|
76 |
"epoch": 0.04,
|
77 |
"learning_rate": 2.0833333333333336e-05,
|
78 |
-
"logits/chosen": -2.
|
79 |
-
"logits/rejected": -2.
|
80 |
-
"logps/chosen": -281.
|
81 |
-
"logps/rejected": -277.
|
82 |
-
"loss": 0.
|
83 |
-
"pred_label": 0.
|
84 |
-
"rewards/accuracies": 0.
|
85 |
-
"rewards/chosen": 0.
|
86 |
-
"rewards/margins": 0.
|
87 |
-
"rewards/rejected": -0.
|
88 |
"step": 40,
|
89 |
-
"use_label": 561.
|
90 |
},
|
91 |
{
|
92 |
"epoch": 0.05,
|
93 |
"learning_rate": 2.604166666666667e-05,
|
94 |
-
"logits/chosen": -2.
|
95 |
-
"logits/rejected": -2.
|
96 |
-
"logps/chosen": -
|
97 |
-
"logps/rejected": -
|
98 |
-
"loss": 0.
|
99 |
-
"pred_label":
|
100 |
-
"rewards/accuracies": 0.
|
101 |
-
"rewards/chosen": -0.
|
102 |
-
"rewards/margins": 0.
|
103 |
-
"rewards/rejected": -0.
|
104 |
"step": 50,
|
105 |
-
"use_label":
|
106 |
},
|
107 |
{
|
108 |
"epoch": 0.06,
|
109 |
"learning_rate": 3.125e-05,
|
110 |
-
"logits/chosen": -2.
|
111 |
-
"logits/rejected": -2.
|
112 |
-
"logps/chosen": -
|
113 |
-
"logps/rejected": -
|
114 |
-
"loss": 0.
|
115 |
-
"pred_label":
|
116 |
-
"rewards/accuracies": 0.
|
117 |
-
"rewards/chosen": -0.
|
118 |
-
"rewards/margins": 0.
|
119 |
-
"rewards/rejected": -0.
|
120 |
"step": 60,
|
121 |
-
"use_label":
|
122 |
},
|
123 |
{
|
124 |
"epoch": 0.07,
|
125 |
"learning_rate": 3.6458333333333336e-05,
|
126 |
-
"logits/chosen": -2.
|
127 |
-
"logits/rejected": -2.
|
128 |
-
"logps/chosen": -
|
129 |
-
"logps/rejected": -
|
130 |
-
"loss": 0.
|
131 |
-
"pred_label":
|
132 |
-
"rewards/accuracies": 0.
|
133 |
-
"rewards/chosen": -0.
|
134 |
-
"rewards/margins": 0.
|
135 |
-
"rewards/rejected": -
|
136 |
"step": 70,
|
137 |
-
"use_label":
|
138 |
},
|
139 |
{
|
140 |
"epoch": 0.08,
|
141 |
"learning_rate": 4.166666666666667e-05,
|
142 |
-
"logits/chosen": -2.
|
143 |
-
"logits/rejected": -2.
|
144 |
-
"logps/chosen": -
|
145 |
-
"logps/rejected": -
|
146 |
-
"loss": 0.
|
147 |
-
"pred_label":
|
148 |
-
"rewards/accuracies": 0.
|
149 |
-
"rewards/chosen": -
|
150 |
-
"rewards/margins":
|
151 |
-
"rewards/rejected": -
|
152 |
"step": 80,
|
153 |
-
"use_label":
|
154 |
},
|
155 |
{
|
156 |
"epoch": 0.09,
|
157 |
"learning_rate": 4.6875e-05,
|
158 |
-
"logits/chosen": -2.
|
159 |
-
"logits/rejected": -2.
|
160 |
-
"logps/chosen": -
|
161 |
-
"logps/rejected": -
|
162 |
-
"loss": 0.
|
163 |
-
"pred_label":
|
164 |
-
"rewards/accuracies": 0.
|
165 |
-
"rewards/chosen":
|
166 |
-
"rewards/margins":
|
167 |
-
"rewards/rejected": -
|
168 |
"step": 90,
|
169 |
-
"use_label":
|
170 |
},
|
171 |
{
|
172 |
"epoch": 0.1,
|
173 |
"learning_rate": 4.976717112922003e-05,
|
174 |
-
"logits/chosen": -2.
|
175 |
-
"logits/rejected": -2.
|
176 |
-
"logps/chosen": -
|
177 |
-
"logps/rejected": -
|
178 |
-
"loss": 0.
|
179 |
-
"pred_label":
|
180 |
-
"rewards/accuracies": 0.
|
181 |
-
"rewards/chosen": -
|
182 |
-
"rewards/margins":
|
183 |
-
"rewards/rejected": -
|
184 |
"step": 100,
|
185 |
-
"use_label":
|
186 |
},
|
187 |
{
|
188 |
"epoch": 0.12,
|
189 |
"learning_rate": 4.918509895227008e-05,
|
190 |
-
"logits/chosen": -2.
|
191 |
-
"logits/rejected": -2.
|
192 |
-
"logps/chosen": -
|
193 |
-
"logps/rejected": -
|
194 |
-
"loss": 0.
|
195 |
-
"pred_label":
|
196 |
-
"rewards/accuracies": 0.
|
197 |
-
"rewards/chosen": -
|
198 |
-
"rewards/margins":
|
199 |
-
"rewards/rejected": -
|
200 |
"step": 110,
|
201 |
-
"use_label":
|
202 |
},
|
203 |
{
|
204 |
"epoch": 0.13,
|
205 |
"learning_rate": 4.860302677532014e-05,
|
206 |
-
"logits/chosen": -2.
|
207 |
-
"logits/rejected": -2.
|
208 |
-
"logps/chosen": -
|
209 |
-
"logps/rejected": -
|
210 |
-
"loss": 0.
|
211 |
-
"pred_label":
|
212 |
-
"rewards/accuracies": 0.
|
213 |
-
"rewards/chosen": -
|
214 |
-
"rewards/margins":
|
215 |
-
"rewards/rejected": -
|
216 |
"step": 120,
|
217 |
-
"use_label":
|
218 |
},
|
219 |
{
|
220 |
"epoch": 0.14,
|
221 |
"learning_rate": 4.80209545983702e-05,
|
222 |
-
"logits/chosen": -2.
|
223 |
-
"logits/rejected": -2.
|
224 |
-
"logps/chosen": -
|
225 |
-
"logps/rejected": -
|
226 |
-
"loss": 0.
|
227 |
-
"pred_label":
|
228 |
-
"rewards/accuracies": 0.
|
229 |
-
"rewards/chosen": -
|
230 |
-
"rewards/margins":
|
231 |
-
"rewards/rejected": -
|
232 |
"step": 130,
|
233 |
-
"use_label":
|
234 |
},
|
235 |
{
|
236 |
"epoch": 0.15,
|
237 |
"learning_rate": 4.743888242142026e-05,
|
238 |
-
"logits/chosen": -2.
|
239 |
-
"logits/rejected": -2.
|
240 |
-
"logps/chosen": -
|
241 |
-
"logps/rejected": -
|
242 |
-
"loss": 0.
|
243 |
-
"pred_label":
|
244 |
-
"rewards/accuracies": 0.
|
245 |
-
"rewards/chosen": -
|
246 |
-
"rewards/margins":
|
247 |
-
"rewards/rejected": -
|
248 |
"step": 140,
|
249 |
-
"use_label":
|
250 |
},
|
251 |
{
|
252 |
"epoch": 0.16,
|
253 |
"learning_rate": 4.685681024447032e-05,
|
254 |
-
"logits/chosen": -2.
|
255 |
-
"logits/rejected": -2.
|
256 |
-
"logps/chosen": -
|
257 |
-
"logps/rejected": -
|
258 |
-
"loss": 0.
|
259 |
-
"pred_label":
|
260 |
"rewards/accuracies": 0.612500011920929,
|
261 |
-
"rewards/chosen": -
|
262 |
-
"rewards/margins":
|
263 |
-
"rewards/rejected": -
|
264 |
"step": 150,
|
265 |
-
"use_label":
|
266 |
},
|
267 |
{
|
268 |
"epoch": 0.17,
|
269 |
"learning_rate": 4.6274738067520374e-05,
|
270 |
-
"logits/chosen": -2.
|
271 |
-
"logits/rejected": -2.
|
272 |
-
"logps/chosen": -
|
273 |
-
"logps/rejected": -
|
274 |
-
"loss": 0.
|
275 |
-
"pred_label":
|
276 |
-
"rewards/accuracies": 0.
|
277 |
-
"rewards/chosen": -
|
278 |
-
"rewards/margins":
|
279 |
-
"rewards/rejected": -
|
280 |
"step": 160,
|
281 |
-
"use_label":
|
282 |
},
|
283 |
{
|
284 |
"epoch": 0.18,
|
285 |
"learning_rate": 4.5692665890570435e-05,
|
286 |
-
"logits/chosen": -2.
|
287 |
-
"logits/rejected": -2.
|
288 |
-
"logps/chosen": -
|
289 |
-
"logps/rejected": -
|
290 |
-
"loss": 0.
|
291 |
-
"pred_label":
|
292 |
-
"rewards/accuracies": 0.
|
293 |
-
"rewards/chosen": -
|
294 |
-
"rewards/margins":
|
295 |
-
"rewards/rejected": -
|
296 |
"step": 170,
|
297 |
-
"use_label":
|
298 |
},
|
299 |
{
|
300 |
"epoch": 0.19,
|
301 |
"learning_rate": 4.511059371362049e-05,
|
302 |
-
"logits/chosen": -2.
|
303 |
-
"logits/rejected": -2.
|
304 |
-
"logps/chosen": -
|
305 |
-
"logps/rejected": -
|
306 |
-
"loss": 0.
|
307 |
-
"pred_label":
|
308 |
-
"rewards/accuracies": 0.
|
309 |
-
"rewards/chosen": -
|
310 |
-
"rewards/margins":
|
311 |
-
"rewards/rejected": -
|
312 |
"step": 180,
|
313 |
-
"use_label":
|
314 |
},
|
315 |
{
|
316 |
"epoch": 0.2,
|
317 |
"learning_rate": 4.452852153667055e-05,
|
318 |
-
"logits/chosen": -2.
|
319 |
-
"logits/rejected": -2.
|
320 |
-
"logps/chosen": -
|
321 |
-
"logps/rejected": -
|
322 |
-
"loss": 0.
|
323 |
-
"pred_label":
|
324 |
-
"rewards/accuracies": 0.
|
325 |
-
"rewards/chosen": -
|
326 |
-
"rewards/margins":
|
327 |
-
"rewards/rejected": -
|
328 |
"step": 190,
|
329 |
-
"use_label":
|
330 |
},
|
331 |
{
|
332 |
"epoch": 0.21,
|
333 |
"learning_rate": 4.394644935972061e-05,
|
334 |
-
"logits/chosen": -2.
|
335 |
-
"logits/rejected": -2.
|
336 |
-
"logps/chosen": -
|
337 |
-
"logps/rejected": -
|
338 |
-
"loss": 0.
|
339 |
-
"pred_label":
|
340 |
-
"rewards/accuracies": 0.
|
341 |
-
"rewards/chosen": -
|
342 |
-
"rewards/margins":
|
343 |
-
"rewards/rejected": -
|
344 |
"step": 200,
|
345 |
-
"use_label":
|
346 |
},
|
347 |
{
|
348 |
"epoch": 0.22,
|
349 |
"learning_rate": 4.336437718277067e-05,
|
350 |
-
"logits/chosen": -2.
|
351 |
-
"logits/rejected": -2.
|
352 |
-
"logps/chosen": -
|
353 |
-
"logps/rejected": -
|
354 |
-
"loss": 0.
|
355 |
-
"pred_label":
|
356 |
-
"rewards/accuracies": 0.
|
357 |
-
"rewards/chosen": -
|
358 |
-
"rewards/margins":
|
359 |
-
"rewards/rejected": -
|
360 |
"step": 210,
|
361 |
-
"use_label":
|
362 |
},
|
363 |
{
|
364 |
"epoch": 0.23,
|
365 |
"learning_rate": 4.278230500582072e-05,
|
366 |
-
"logits/chosen": -2.
|
367 |
-
"logits/rejected": -2.
|
368 |
-
"logps/chosen": -
|
369 |
-
"logps/rejected": -
|
370 |
-
"loss": 0.
|
371 |
-
"pred_label":
|
372 |
-
"rewards/accuracies": 0.
|
373 |
-
"rewards/chosen": -
|
374 |
-
"rewards/margins":
|
375 |
-
"rewards/rejected": -
|
376 |
"step": 220,
|
377 |
-
"use_label":
|
378 |
},
|
379 |
{
|
380 |
"epoch": 0.24,
|
381 |
"learning_rate": 4.220023282887078e-05,
|
382 |
-
"logits/chosen": -2.
|
383 |
-
"logits/rejected": -2.
|
384 |
-
"logps/chosen": -
|
385 |
-
"logps/rejected": -
|
386 |
-
"loss": 0.
|
387 |
-
"pred_label":
|
388 |
-
"rewards/accuracies": 0.
|
389 |
-
"rewards/chosen": -
|
390 |
-
"rewards/margins":
|
391 |
-
"rewards/rejected": -
|
392 |
"step": 230,
|
393 |
-
"use_label":
|
394 |
},
|
395 |
{
|
396 |
"epoch": 0.25,
|
397 |
"learning_rate": 4.161816065192084e-05,
|
398 |
-
"logits/chosen": -2.
|
399 |
-
"logits/rejected": -2.
|
400 |
-
"logps/chosen": -
|
401 |
-
"logps/rejected": -
|
402 |
-
"loss": 0.
|
403 |
-
"pred_label":
|
404 |
-
"rewards/accuracies": 0.
|
405 |
-
"rewards/chosen": -
|
406 |
-
"rewards/margins":
|
407 |
-
"rewards/rejected": -
|
408 |
"step": 240,
|
409 |
-
"use_label":
|
410 |
},
|
411 |
{
|
412 |
"epoch": 0.26,
|
413 |
"learning_rate": 4.10360884749709e-05,
|
414 |
-
"logits/chosen": -2.
|
415 |
-
"logits/rejected": -2.
|
416 |
-
"logps/chosen": -
|
417 |
-
"logps/rejected": -
|
418 |
-
"loss": 0.
|
419 |
-
"pred_label":
|
420 |
-
"rewards/accuracies": 0.
|
421 |
-
"rewards/chosen": -
|
422 |
-
"rewards/margins":
|
423 |
-
"rewards/rejected": -
|
424 |
"step": 250,
|
425 |
-
"use_label":
|
426 |
},
|
427 |
{
|
428 |
"epoch": 0.27,
|
429 |
"learning_rate": 4.045401629802096e-05,
|
430 |
-
"logits/chosen": -2.
|
431 |
-
"logits/rejected": -2.
|
432 |
-
"logps/chosen": -
|
433 |
-
"logps/rejected": -
|
434 |
-
"loss": 0.
|
435 |
-
"pred_label":
|
436 |
-
"rewards/accuracies": 0.
|
437 |
-
"rewards/chosen": -
|
438 |
-
"rewards/margins":
|
439 |
-
"rewards/rejected": -
|
440 |
"step": 260,
|
441 |
-
"use_label":
|
442 |
},
|
443 |
{
|
444 |
"epoch": 0.28,
|
445 |
"learning_rate": 3.9871944121071014e-05,
|
446 |
-
"logits/chosen": -2.
|
447 |
-
"logits/rejected": -2.
|
448 |
-
"logps/chosen": -
|
449 |
-
"logps/rejected": -
|
450 |
-
"loss": 0.
|
451 |
-
"pred_label":
|
452 |
-
"rewards/accuracies": 0.
|
453 |
-
"rewards/chosen": -
|
454 |
-
"rewards/margins":
|
455 |
-
"rewards/rejected": -
|
456 |
"step": 270,
|
457 |
-
"use_label":
|
458 |
},
|
459 |
{
|
460 |
"epoch": 0.29,
|
461 |
"learning_rate": 3.928987194412107e-05,
|
462 |
-
"logits/chosen": -2.
|
463 |
-
"logits/rejected": -2.
|
464 |
-
"logps/chosen": -
|
465 |
-
"logps/rejected": -
|
466 |
-
"loss": 0.
|
467 |
-
"pred_label":
|
468 |
-
"rewards/accuracies": 0.
|
469 |
-
"rewards/chosen": -
|
470 |
-
"rewards/margins":
|
471 |
-
"rewards/rejected": -
|
472 |
"step": 280,
|
473 |
-
"use_label":
|
474 |
},
|
475 |
{
|
476 |
"epoch": 0.3,
|
477 |
"learning_rate": 3.870779976717113e-05,
|
478 |
-
"logits/chosen": -2.
|
479 |
-
"logits/rejected": -2.
|
480 |
-
"logps/chosen": -
|
481 |
-
"logps/rejected": -
|
482 |
-
"loss": 0.
|
483 |
-
"pred_label":
|
484 |
-
"rewards/accuracies": 0.
|
485 |
-
"rewards/chosen": -
|
486 |
-
"rewards/margins":
|
487 |
-
"rewards/rejected": -
|
488 |
"step": 290,
|
489 |
-
"use_label":
|
490 |
},
|
491 |
{
|
492 |
"epoch": 0.31,
|
493 |
"learning_rate": 3.812572759022119e-05,
|
494 |
-
"logits/chosen": -2.
|
495 |
-
"logits/rejected": -2.
|
496 |
-
"logps/chosen": -
|
497 |
-
"logps/rejected": -
|
498 |
-
"loss": 0.
|
499 |
-
"pred_label":
|
500 |
-
"rewards/accuracies": 0.
|
501 |
-
"rewards/chosen": -
|
502 |
-
"rewards/margins":
|
503 |
-
"rewards/rejected": -
|
504 |
"step": 300,
|
505 |
-
"use_label":
|
506 |
},
|
507 |
{
|
508 |
"epoch": 0.32,
|
509 |
"learning_rate": 3.7543655413271246e-05,
|
510 |
-
"logits/chosen": -2.
|
511 |
-
"logits/rejected": -2.
|
512 |
-
"logps/chosen": -
|
513 |
-
"logps/rejected": -
|
514 |
-
"loss": 0.
|
515 |
-
"pred_label":
|
516 |
-
"rewards/accuracies": 0.
|
517 |
-
"rewards/chosen": -
|
518 |
-
"rewards/margins":
|
519 |
-
"rewards/rejected": -
|
520 |
"step": 310,
|
521 |
-
"use_label":
|
522 |
},
|
523 |
{
|
524 |
"epoch": 0.33,
|
525 |
"learning_rate": 3.696158323632131e-05,
|
526 |
-
"logits/chosen": -2.
|
527 |
-
"logits/rejected": -2.
|
528 |
-
"logps/chosen": -
|
529 |
-
"logps/rejected": -
|
530 |
-
"loss": 0.
|
531 |
-
"pred_label":
|
532 |
-
"rewards/accuracies": 0.
|
533 |
-
"rewards/chosen": -
|
534 |
-
"rewards/margins":
|
535 |
-
"rewards/rejected": -
|
536 |
"step": 320,
|
537 |
-
"use_label":
|
538 |
},
|
539 |
{
|
540 |
"epoch": 0.35,
|
541 |
"learning_rate": 3.637951105937136e-05,
|
542 |
-
"logits/chosen": -2.
|
543 |
-
"logits/rejected": -2.
|
544 |
-
"logps/chosen": -
|
545 |
-
"logps/rejected": -
|
546 |
-
"loss": 0.
|
547 |
-
"pred_label":
|
548 |
-
"rewards/accuracies": 0.
|
549 |
-
"rewards/chosen": -
|
550 |
-
"rewards/margins":
|
551 |
-
"rewards/rejected": -
|
552 |
"step": 330,
|
553 |
-
"use_label":
|
554 |
},
|
555 |
{
|
556 |
"epoch": 0.36,
|
557 |
"learning_rate": 3.579743888242142e-05,
|
558 |
-
"logits/chosen": -2.
|
559 |
-
"logits/rejected": -2.
|
560 |
-
"logps/chosen": -
|
561 |
-
"logps/rejected": -
|
562 |
-
"loss": 0.
|
563 |
-
"pred_label":
|
564 |
-
"rewards/accuracies": 0.
|
565 |
-
"rewards/chosen": -
|
566 |
-
"rewards/margins":
|
567 |
-
"rewards/rejected": -
|
568 |
"step": 340,
|
569 |
-
"use_label":
|
570 |
},
|
571 |
{
|
572 |
"epoch": 0.37,
|
573 |
"learning_rate": 3.5215366705471484e-05,
|
574 |
-
"logits/chosen": -2.
|
575 |
-
"logits/rejected": -2.
|
576 |
-
"logps/chosen": -
|
577 |
-
"logps/rejected": -
|
578 |
-
"loss": 0.
|
579 |
-
"pred_label":
|
580 |
-
"rewards/accuracies": 0.
|
581 |
-
"rewards/chosen": -
|
582 |
-
"rewards/margins":
|
583 |
-
"rewards/rejected": -
|
584 |
"step": 350,
|
585 |
-
"use_label":
|
586 |
},
|
587 |
{
|
588 |
"epoch": 0.38,
|
589 |
"learning_rate": 3.463329452852154e-05,
|
590 |
-
"logits/chosen": -2.
|
591 |
-
"logits/rejected": -2.
|
592 |
-
"logps/chosen": -
|
593 |
-
"logps/rejected": -
|
594 |
-
"loss": 0.
|
595 |
-
"pred_label":
|
596 |
-
"rewards/accuracies": 0.
|
597 |
-
"rewards/chosen": -
|
598 |
-
"rewards/margins":
|
599 |
-
"rewards/rejected": -
|
600 |
"step": 360,
|
601 |
-
"use_label":
|
602 |
},
|
603 |
{
|
604 |
"epoch": 0.39,
|
605 |
"learning_rate": 3.40512223515716e-05,
|
606 |
-
"logits/chosen": -2.
|
607 |
-
"logits/rejected": -2.
|
608 |
-
"logps/chosen": -
|
609 |
-
"logps/rejected": -
|
610 |
-
"loss": 0.
|
611 |
-
"pred_label":
|
612 |
-
"rewards/accuracies": 0.
|
613 |
-
"rewards/chosen": -
|
614 |
-
"rewards/margins":
|
615 |
-
"rewards/rejected": -
|
616 |
"step": 370,
|
617 |
-
"use_label":
|
618 |
},
|
619 |
{
|
620 |
"epoch": 0.4,
|
621 |
"learning_rate": 3.3469150174621654e-05,
|
622 |
-
"logits/chosen": -2.
|
623 |
-
"logits/rejected": -2.
|
624 |
-
"logps/chosen": -
|
625 |
-
"logps/rejected": -
|
626 |
-
"loss": 0.
|
627 |
-
"pred_label":
|
628 |
-
"rewards/accuracies": 0.
|
629 |
-
"rewards/chosen": -
|
630 |
-
"rewards/margins":
|
631 |
-
"rewards/rejected": -
|
632 |
"step": 380,
|
633 |
-
"use_label":
|
634 |
},
|
635 |
{
|
636 |
"epoch": 0.41,
|
637 |
"learning_rate": 3.288707799767171e-05,
|
638 |
-
"logits/chosen": -2.
|
639 |
-
"logits/rejected": -2.
|
640 |
-
"logps/chosen": -
|
641 |
-
"logps/rejected": -
|
642 |
-
"loss": 0.
|
643 |
-
"pred_label":
|
644 |
-
"rewards/accuracies": 0.
|
645 |
-
"rewards/chosen": -
|
646 |
-
"rewards/margins":
|
647 |
-
"rewards/rejected": -
|
648 |
"step": 390,
|
649 |
-
"use_label":
|
650 |
},
|
651 |
{
|
652 |
"epoch": 0.42,
|
653 |
"learning_rate": 3.2305005820721776e-05,
|
654 |
-
"logits/chosen": -2.
|
655 |
-
"logits/rejected": -2.
|
656 |
-
"logps/chosen": -
|
657 |
-
"logps/rejected": -
|
658 |
-
"loss": 0.
|
659 |
-
"pred_label":
|
660 |
-
"rewards/accuracies": 0.
|
661 |
-
"rewards/chosen": -
|
662 |
-
"rewards/margins":
|
663 |
-
"rewards/rejected": -
|
664 |
"step": 400,
|
665 |
-
"use_label":
|
666 |
},
|
667 |
{
|
668 |
"epoch": 0.43,
|
669 |
"learning_rate": 3.172293364377183e-05,
|
670 |
-
"logits/chosen": -2.
|
671 |
-
"logits/rejected": -2.
|
672 |
-
"logps/chosen": -
|
673 |
-
"logps/rejected": -
|
674 |
-
"loss": 0.
|
675 |
-
"pred_label":
|
676 |
-
"rewards/accuracies": 0.
|
677 |
-
"rewards/chosen": -
|
678 |
-
"rewards/margins":
|
679 |
-
"rewards/rejected": -
|
680 |
"step": 410,
|
681 |
-
"use_label":
|
682 |
},
|
683 |
{
|
684 |
"epoch": 0.44,
|
685 |
"learning_rate": 3.1140861466821885e-05,
|
686 |
-
"logits/chosen": -2.
|
687 |
-
"logits/rejected": -2.
|
688 |
-
"logps/chosen": -
|
689 |
-
"logps/rejected": -
|
690 |
-
"loss": 0.
|
691 |
-
"pred_label":
|
692 |
-
"rewards/accuracies": 0.
|
693 |
-
"rewards/chosen": -
|
694 |
-
"rewards/margins":
|
695 |
-
"rewards/rejected": -
|
696 |
"step": 420,
|
697 |
-
"use_label":
|
698 |
},
|
699 |
{
|
700 |
"epoch": 0.45,
|
701 |
"learning_rate": 3.055878928987195e-05,
|
702 |
-
"logits/chosen": -2.
|
703 |
-
"logits/rejected": -2.
|
704 |
-
"logps/chosen": -
|
705 |
-
"logps/rejected": -
|
706 |
-
"loss": 0.
|
707 |
-
"pred_label":
|
708 |
-
"rewards/accuracies": 0.
|
709 |
-
"rewards/chosen": -
|
710 |
-
"rewards/margins":
|
711 |
-
"rewards/rejected": -
|
712 |
"step": 430,
|
713 |
-
"use_label":
|
714 |
},
|
715 |
{
|
716 |
"epoch": 0.46,
|
717 |
"learning_rate": 2.9976717112922005e-05,
|
718 |
-
"logits/chosen": -2.
|
719 |
-
"logits/rejected": -2.
|
720 |
-
"logps/chosen": -
|
721 |
-
"logps/rejected": -
|
722 |
-
"loss": 0.
|
723 |
-
"pred_label":
|
724 |
-
"rewards/accuracies": 0.
|
725 |
-
"rewards/chosen": -
|
726 |
-
"rewards/margins":
|
727 |
-
"rewards/rejected": -
|
728 |
"step": 440,
|
729 |
-
"use_label":
|
730 |
},
|
731 |
{
|
732 |
"epoch": 0.47,
|
733 |
"learning_rate": 2.939464493597206e-05,
|
734 |
-
"logits/chosen": -2.
|
735 |
-
"logits/rejected": -2.
|
736 |
-
"logps/chosen": -
|
737 |
-
"logps/rejected": -
|
738 |
-
"loss": 0.
|
739 |
-
"pred_label":
|
740 |
-
"rewards/accuracies": 0.
|
741 |
-
"rewards/chosen": -
|
742 |
-
"rewards/margins":
|
743 |
-
"rewards/rejected": -
|
744 |
"step": 450,
|
745 |
-
"use_label":
|
746 |
},
|
747 |
{
|
748 |
"epoch": 0.48,
|
749 |
"learning_rate": 2.881257275902212e-05,
|
750 |
-
"logits/chosen": -2.
|
751 |
-
"logits/rejected": -2.
|
752 |
-
"logps/chosen": -
|
753 |
-
"logps/rejected": -
|
754 |
-
"loss": 0.
|
755 |
-
"pred_label":
|
756 |
-
"rewards/accuracies": 0.
|
757 |
-
"rewards/chosen": -
|
758 |
-
"rewards/margins":
|
759 |
-
"rewards/rejected": -
|
760 |
"step": 460,
|
761 |
-
"use_label":
|
762 |
},
|
763 |
{
|
764 |
"epoch": 0.49,
|
765 |
"learning_rate": 2.8230500582072178e-05,
|
766 |
-
"logits/chosen": -2.
|
767 |
-
"logits/rejected": -2.
|
768 |
-
"logps/chosen": -
|
769 |
-
"logps/rejected": -
|
770 |
-
"loss": 0.
|
771 |
-
"pred_label":
|
772 |
-
"rewards/accuracies": 0.
|
773 |
-
"rewards/chosen": -
|
774 |
-
"rewards/margins":
|
775 |
-
"rewards/rejected": -
|
776 |
"step": 470,
|
777 |
-
"use_label":
|
778 |
},
|
779 |
{
|
780 |
"epoch": 0.5,
|
781 |
"learning_rate": 2.7648428405122233e-05,
|
782 |
-
"logits/chosen": -2.
|
783 |
-
"logits/rejected": -2.
|
784 |
-
"logps/chosen": -
|
785 |
-
"logps/rejected": -
|
786 |
-
"loss": 0.
|
787 |
-
"pred_label":
|
788 |
-
"rewards/accuracies": 0.
|
789 |
-
"rewards/chosen": -
|
790 |
-
"rewards/margins":
|
791 |
-
"rewards/rejected": -
|
792 |
"step": 480,
|
793 |
-
"use_label":
|
794 |
},
|
795 |
{
|
796 |
"epoch": 0.51,
|
797 |
"learning_rate": 2.7066356228172297e-05,
|
798 |
-
"logits/chosen": -2.
|
799 |
-
"logits/rejected": -2.
|
800 |
-
"logps/chosen": -
|
801 |
-
"logps/rejected": -
|
802 |
-
"loss": 0.
|
803 |
-
"pred_label":
|
804 |
-
"rewards/accuracies": 0.
|
805 |
-
"rewards/chosen": -
|
806 |
-
"rewards/margins":
|
807 |
-
"rewards/rejected": -
|
808 |
"step": 490,
|
809 |
-
"use_label":
|
810 |
},
|
811 |
{
|
812 |
"epoch": 0.52,
|
813 |
"learning_rate": 2.6484284051222352e-05,
|
814 |
-
"logits/chosen": -2.
|
815 |
-
"logits/rejected": -2.
|
816 |
-
"logps/chosen": -
|
817 |
-
"logps/rejected": -
|
818 |
-
"loss": 0.
|
819 |
-
"pred_label":
|
820 |
-
"rewards/accuracies": 0.
|
821 |
-
"rewards/chosen": -
|
822 |
-
"rewards/margins":
|
823 |
-
"rewards/rejected": -
|
824 |
"step": 500,
|
825 |
-
"use_label":
|
826 |
},
|
827 |
{
|
828 |
"epoch": 0.53,
|
829 |
"learning_rate": 2.590221187427241e-05,
|
830 |
-
"logits/chosen": -2.
|
831 |
-
"logits/rejected": -2.
|
832 |
-
"logps/chosen": -
|
833 |
-
"logps/rejected": -
|
834 |
-
"loss": 0.
|
835 |
-
"pred_label":
|
836 |
-
"rewards/accuracies": 0.
|
837 |
-
"rewards/chosen": -
|
838 |
-
"rewards/margins":
|
839 |
-
"rewards/rejected": -
|
840 |
"step": 510,
|
841 |
-
"use_label":
|
842 |
},
|
843 |
{
|
844 |
"epoch": 0.54,
|
845 |
"learning_rate": 2.532013969732247e-05,
|
846 |
-
"logits/chosen": -2.
|
847 |
-
"logits/rejected": -2.
|
848 |
-
"logps/chosen": -
|
849 |
-
"logps/rejected": -
|
850 |
-
"loss": 0.
|
851 |
-
"pred_label":
|
852 |
-
"rewards/accuracies": 0.
|
853 |
-
"rewards/chosen": -
|
854 |
-
"rewards/margins":
|
855 |
-
"rewards/rejected": -
|
856 |
"step": 520,
|
857 |
-
"use_label":
|
858 |
},
|
859 |
{
|
860 |
"epoch": 0.55,
|
861 |
"learning_rate": 2.4738067520372525e-05,
|
862 |
-
"logits/chosen": -2.
|
863 |
-
"logits/rejected": -2.
|
864 |
-
"logps/chosen": -
|
865 |
-
"logps/rejected": -
|
866 |
-
"loss": 0.
|
867 |
-
"pred_label":
|
868 |
-
"rewards/accuracies": 0.
|
869 |
-
"rewards/chosen": -
|
870 |
-
"rewards/margins":
|
871 |
-
"rewards/rejected": -
|
872 |
"step": 530,
|
873 |
-
"use_label":
|
874 |
},
|
875 |
{
|
876 |
"epoch": 0.57,
|
877 |
"learning_rate": 2.4155995343422587e-05,
|
878 |
-
"logits/chosen": -2.
|
879 |
-
"logits/rejected": -2.
|
880 |
-
"logps/chosen": -
|
881 |
-
"logps/rejected": -
|
882 |
-
"loss": 0.
|
883 |
-
"pred_label":
|
884 |
-
"rewards/accuracies": 0.
|
885 |
-
"rewards/chosen": -
|
886 |
-
"rewards/margins":
|
887 |
-
"rewards/rejected": -
|
888 |
"step": 540,
|
889 |
-
"use_label":
|
890 |
},
|
891 |
{
|
892 |
"epoch": 0.58,
|
893 |
"learning_rate": 2.3573923166472644e-05,
|
894 |
-
"logits/chosen": -2.
|
895 |
-
"logits/rejected": -
|
896 |
-
"logps/chosen": -
|
897 |
-
"logps/rejected": -
|
898 |
-
"loss": 0.
|
899 |
-
"pred_label":
|
900 |
-
"rewards/accuracies": 0.
|
901 |
-
"rewards/chosen": -
|
902 |
-
"rewards/margins":
|
903 |
-
"rewards/rejected": -
|
904 |
"step": 550,
|
905 |
-
"use_label":
|
906 |
},
|
907 |
{
|
908 |
"epoch": 0.59,
|
909 |
"learning_rate": 2.2991850989522702e-05,
|
910 |
-
"logits/chosen": -2.
|
911 |
-
"logits/rejected": -2.
|
912 |
-
"logps/chosen": -
|
913 |
-
"logps/rejected": -
|
914 |
-
"loss": 0.
|
915 |
-
"pred_label":
|
916 |
-
"rewards/accuracies": 0.
|
917 |
-
"rewards/chosen": -
|
918 |
-
"rewards/margins":
|
919 |
-
"rewards/rejected": -
|
920 |
"step": 560,
|
921 |
-
"use_label":
|
922 |
},
|
923 |
{
|
924 |
"epoch": 0.6,
|
925 |
"learning_rate": 2.240977881257276e-05,
|
926 |
-
"logits/chosen": -2.
|
927 |
-
"logits/rejected": -2.
|
928 |
-
"logps/chosen": -
|
929 |
-
"logps/rejected": -
|
930 |
-
"loss": 0.
|
931 |
-
"pred_label":
|
932 |
"rewards/accuracies": 0.706250011920929,
|
933 |
-
"rewards/chosen": -
|
934 |
-
"rewards/margins":
|
935 |
-
"rewards/rejected": -
|
936 |
"step": 570,
|
937 |
-
"use_label":
|
938 |
},
|
939 |
{
|
940 |
"epoch": 0.61,
|
941 |
"learning_rate": 2.1827706635622818e-05,
|
942 |
-
"logits/chosen": -2.
|
943 |
-
"logits/rejected": -2.
|
944 |
-
"logps/chosen": -
|
945 |
-
"logps/rejected": -
|
946 |
-
"loss": 0.
|
947 |
-
"pred_label":
|
948 |
-
"rewards/accuracies": 0.
|
949 |
-
"rewards/chosen": -
|
950 |
-
"rewards/margins":
|
951 |
-
"rewards/rejected": -
|
952 |
"step": 580,
|
953 |
-
"use_label":
|
954 |
},
|
955 |
{
|
956 |
"epoch": 0.62,
|
957 |
"learning_rate": 2.124563445867288e-05,
|
958 |
-
"logits/chosen": -2.
|
959 |
-
"logits/rejected": -2.
|
960 |
-
"logps/chosen": -
|
961 |
-
"logps/rejected": -
|
962 |
-
"loss": 0.
|
963 |
-
"pred_label":
|
964 |
-
"rewards/accuracies": 0.
|
965 |
-
"rewards/chosen": -
|
966 |
-
"rewards/margins":
|
967 |
-
"rewards/rejected": -
|
968 |
"step": 590,
|
969 |
-
"use_label":
|
970 |
},
|
971 |
{
|
972 |
"epoch": 0.63,
|
973 |
"learning_rate": 2.0663562281722934e-05,
|
974 |
-
"logits/chosen": -2.
|
975 |
-
"logits/rejected": -2.
|
976 |
-
"logps/chosen": -
|
977 |
-
"logps/rejected": -
|
978 |
-
"loss": 0.
|
979 |
-
"pred_label":
|
980 |
-
"rewards/accuracies": 0.
|
981 |
-
"rewards/chosen": -
|
982 |
-
"rewards/margins":
|
983 |
-
"rewards/rejected": -
|
984 |
"step": 600,
|
985 |
-
"use_label":
|
986 |
},
|
987 |
{
|
988 |
"epoch": 0.64,
|
989 |
"learning_rate": 2.0081490104772992e-05,
|
990 |
-
"logits/chosen": -2.
|
991 |
-
"logits/rejected": -2.
|
992 |
-
"logps/chosen": -
|
993 |
-
"logps/rejected": -
|
994 |
-
"loss": 0.
|
995 |
-
"pred_label":
|
996 |
-
"rewards/accuracies": 0.
|
997 |
-
"rewards/chosen": -
|
998 |
-
"rewards/margins":
|
999 |
-
"rewards/rejected": -
|
1000 |
"step": 610,
|
1001 |
-
"use_label":
|
1002 |
},
|
1003 |
{
|
1004 |
"epoch": 0.65,
|
1005 |
"learning_rate": 1.9499417927823053e-05,
|
1006 |
-
"logits/chosen": -2.
|
1007 |
-
"logits/rejected": -2.
|
1008 |
-
"logps/chosen": -
|
1009 |
-
"logps/rejected": -
|
1010 |
-
"loss": 0.
|
1011 |
-
"pred_label":
|
1012 |
-
"rewards/accuracies": 0.
|
1013 |
-
"rewards/chosen": -
|
1014 |
-
"rewards/margins":
|
1015 |
-
"rewards/rejected": -
|
1016 |
"step": 620,
|
1017 |
-
"use_label":
|
1018 |
},
|
1019 |
{
|
1020 |
"epoch": 0.66,
|
1021 |
"learning_rate": 1.8917345750873107e-05,
|
1022 |
-
"logits/chosen": -2.
|
1023 |
-
"logits/rejected": -2.
|
1024 |
-
"logps/chosen": -
|
1025 |
-
"logps/rejected": -
|
1026 |
-
"loss": 0.
|
1027 |
-
"pred_label":
|
1028 |
-
"rewards/accuracies": 0.
|
1029 |
-
"rewards/chosen": -
|
1030 |
-
"rewards/margins":
|
1031 |
-
"rewards/rejected": -
|
1032 |
"step": 630,
|
1033 |
-
"use_label":
|
1034 |
},
|
1035 |
{
|
1036 |
"epoch": 0.67,
|
1037 |
"learning_rate": 1.833527357392317e-05,
|
1038 |
-
"logits/chosen": -2.
|
1039 |
-
"logits/rejected": -
|
1040 |
-
"logps/chosen": -
|
1041 |
-
"logps/rejected": -
|
1042 |
-
"loss": 0.
|
1043 |
-
"pred_label":
|
1044 |
-
"rewards/accuracies": 0.
|
1045 |
-
"rewards/chosen": -
|
1046 |
-
"rewards/margins":
|
1047 |
-
"rewards/rejected": -
|
1048 |
"step": 640,
|
1049 |
-
"use_label":
|
1050 |
},
|
1051 |
{
|
1052 |
"epoch": 0.68,
|
1053 |
"learning_rate": 1.7753201396973227e-05,
|
1054 |
-
"logits/chosen": -2.
|
1055 |
-
"logits/rejected": -2.
|
1056 |
-
"logps/chosen": -
|
1057 |
-
"logps/rejected": -
|
1058 |
-
"loss": 0.
|
1059 |
-
"pred_label":
|
1060 |
-
"rewards/accuracies": 0.
|
1061 |
-
"rewards/chosen": -
|
1062 |
-
"rewards/margins":
|
1063 |
-
"rewards/rejected": -
|
1064 |
"step": 650,
|
1065 |
-
"use_label":
|
1066 |
},
|
1067 |
{
|
1068 |
"epoch": 0.69,
|
1069 |
"learning_rate": 1.717112922002328e-05,
|
1070 |
-
"logits/chosen": -2.
|
1071 |
-
"logits/rejected": -2.
|
1072 |
-
"logps/chosen": -
|
1073 |
-
"logps/rejected": -
|
1074 |
-
"loss": 0.
|
1075 |
-
"pred_label":
|
1076 |
-
"rewards/accuracies": 0.
|
1077 |
-
"rewards/chosen": -
|
1078 |
-
"rewards/margins":
|
1079 |
-
"rewards/rejected": -
|
1080 |
"step": 660,
|
1081 |
-
"use_label":
|
1082 |
},
|
1083 |
{
|
1084 |
"epoch": 0.7,
|
1085 |
"learning_rate": 1.6589057043073342e-05,
|
1086 |
-
"logits/chosen": -2.
|
1087 |
-
"logits/rejected": -2.
|
1088 |
-
"logps/chosen": -
|
1089 |
-
"logps/rejected": -
|
1090 |
-
"loss": 0.
|
1091 |
-
"pred_label":
|
1092 |
-
"rewards/accuracies": 0.
|
1093 |
-
"rewards/chosen": -
|
1094 |
-
"rewards/margins":
|
1095 |
-
"rewards/rejected": -
|
1096 |
"step": 670,
|
1097 |
-
"use_label":
|
1098 |
},
|
1099 |
{
|
1100 |
"epoch": 0.71,
|
1101 |
"learning_rate": 1.60069848661234e-05,
|
1102 |
-
"logits/chosen": -2.
|
1103 |
-
"logits/rejected": -2.
|
1104 |
-
"logps/chosen": -
|
1105 |
-
"logps/rejected": -
|
1106 |
-
"loss": 0.
|
1107 |
-
"pred_label":
|
1108 |
-
"rewards/accuracies": 0.
|
1109 |
-
"rewards/chosen": -
|
1110 |
-
"rewards/margins":
|
1111 |
-
"rewards/rejected": -
|
1112 |
"step": 680,
|
1113 |
-
"use_label":
|
1114 |
},
|
1115 |
{
|
1116 |
"epoch": 0.72,
|
1117 |
"learning_rate": 1.5424912689173458e-05,
|
1118 |
-
"logits/chosen": -2.
|
1119 |
-
"logits/rejected": -2.
|
1120 |
-
"logps/chosen": -
|
1121 |
-
"logps/rejected": -
|
1122 |
-
"loss": 0.
|
1123 |
-
"pred_label":
|
1124 |
-
"rewards/accuracies": 0.
|
1125 |
-
"rewards/chosen": -
|
1126 |
-
"rewards/margins":
|
1127 |
-
"rewards/rejected": -
|
1128 |
"step": 690,
|
1129 |
-
"use_label":
|
1130 |
},
|
1131 |
{
|
1132 |
"epoch": 0.73,
|
1133 |
"learning_rate": 1.4842840512223516e-05,
|
1134 |
-
"logits/chosen": -2.
|
1135 |
-
"logits/rejected": -2.
|
1136 |
-
"logps/chosen": -
|
1137 |
-
"logps/rejected": -
|
1138 |
-
"loss": 0.
|
1139 |
-
"pred_label":
|
1140 |
-
"rewards/accuracies": 0.
|
1141 |
-
"rewards/chosen": -
|
1142 |
-
"rewards/margins":
|
1143 |
-
"rewards/rejected": -
|
1144 |
"step": 700,
|
1145 |
-
"use_label":
|
1146 |
},
|
1147 |
{
|
1148 |
"epoch": 0.74,
|
1149 |
"learning_rate": 1.4260768335273575e-05,
|
1150 |
-
"logits/chosen": -2.
|
1151 |
-
"logits/rejected": -2.
|
1152 |
-
"logps/chosen": -
|
1153 |
-
"logps/rejected": -
|
1154 |
-
"loss": 0.
|
1155 |
-
"pred_label":
|
1156 |
-
"rewards/accuracies": 0.
|
1157 |
-
"rewards/chosen": -
|
1158 |
-
"rewards/margins":
|
1159 |
-
"rewards/rejected": -
|
1160 |
"step": 710,
|
1161 |
-
"use_label":
|
1162 |
},
|
1163 |
{
|
1164 |
"epoch": 0.75,
|
1165 |
"learning_rate": 1.3678696158323633e-05,
|
1166 |
-
"logits/chosen": -2.
|
1167 |
-
"logits/rejected": -2.
|
1168 |
-
"logps/chosen": -
|
1169 |
-
"logps/rejected": -
|
1170 |
-
"loss": 0.
|
1171 |
-
"pred_label":
|
1172 |
-
"rewards/accuracies": 0.
|
1173 |
-
"rewards/chosen": -
|
1174 |
-
"rewards/margins":
|
1175 |
-
"rewards/rejected": -
|
1176 |
"step": 720,
|
1177 |
-
"use_label":
|
1178 |
},
|
1179 |
{
|
1180 |
"epoch": 0.76,
|
1181 |
"learning_rate": 1.309662398137369e-05,
|
1182 |
-
"logits/chosen": -2.
|
1183 |
-
"logits/rejected": -2.
|
1184 |
-
"logps/chosen": -
|
1185 |
-
"logps/rejected": -
|
1186 |
-
"loss": 0.
|
1187 |
-
"pred_label":
|
1188 |
-
"rewards/accuracies": 0.
|
1189 |
-
"rewards/chosen": -
|
1190 |
-
"rewards/margins":
|
1191 |
-
"rewards/rejected": -
|
1192 |
"step": 730,
|
1193 |
-
"use_label":
|
1194 |
},
|
1195 |
{
|
1196 |
"epoch": 0.77,
|
1197 |
"learning_rate": 1.2514551804423749e-05,
|
1198 |
-
"logits/chosen": -2.
|
1199 |
-
"logits/rejected": -2.
|
1200 |
-
"logps/chosen": -
|
1201 |
-
"logps/rejected": -
|
1202 |
-
"loss": 0.
|
1203 |
-
"pred_label":
|
1204 |
-
"rewards/accuracies": 0.
|
1205 |
-
"rewards/chosen": -
|
1206 |
-
"rewards/margins":
|
1207 |
-
"rewards/rejected": -
|
1208 |
"step": 740,
|
1209 |
-
"use_label":
|
1210 |
},
|
1211 |
{
|
1212 |
"epoch": 0.79,
|
1213 |
"learning_rate": 1.1932479627473807e-05,
|
1214 |
-
"logits/chosen": -2.
|
1215 |
-
"logits/rejected": -2.
|
1216 |
-
"logps/chosen": -
|
1217 |
-
"logps/rejected": -
|
1218 |
-
"loss": 0.
|
1219 |
-
"pred_label":
|
1220 |
-
"rewards/accuracies": 0.
|
1221 |
-
"rewards/chosen": -
|
1222 |
-
"rewards/margins":
|
1223 |
-
"rewards/rejected": -
|
1224 |
"step": 750,
|
1225 |
-
"use_label":
|
1226 |
},
|
1227 |
{
|
1228 |
"epoch": 0.8,
|
1229 |
"learning_rate": 1.1350407450523866e-05,
|
1230 |
-
"logits/chosen": -2.
|
1231 |
-
"logits/rejected": -2.
|
1232 |
-
"logps/chosen": -
|
1233 |
-
"logps/rejected": -
|
1234 |
-
"loss": 0.
|
1235 |
-
"pred_label":
|
1236 |
-
"rewards/accuracies": 0.
|
1237 |
-
"rewards/chosen": -
|
1238 |
-
"rewards/margins":
|
1239 |
-
"rewards/rejected": -
|
1240 |
"step": 760,
|
1241 |
-
"use_label":
|
1242 |
},
|
1243 |
{
|
1244 |
"epoch": 0.81,
|
1245 |
"learning_rate": 1.0768335273573923e-05,
|
1246 |
-
"logits/chosen": -2.
|
1247 |
-
"logits/rejected": -2.
|
1248 |
-
"logps/chosen": -
|
1249 |
-
"logps/rejected": -
|
1250 |
-
"loss": 0.
|
1251 |
-
"pred_label":
|
1252 |
-
"rewards/accuracies": 0.
|
1253 |
-
"rewards/chosen": -
|
1254 |
-
"rewards/margins":
|
1255 |
-
"rewards/rejected": -
|
1256 |
"step": 770,
|
1257 |
-
"use_label":
|
1258 |
},
|
1259 |
{
|
1260 |
"epoch": 0.82,
|
1261 |
"learning_rate": 1.0186263096623982e-05,
|
1262 |
-
"logits/chosen": -2.
|
1263 |
-
"logits/rejected": -2.
|
1264 |
-
"logps/chosen": -
|
1265 |
-
"logps/rejected": -
|
1266 |
-
"loss": 0.
|
1267 |
-
"pred_label":
|
1268 |
-
"rewards/accuracies": 0.
|
1269 |
-
"rewards/chosen": -
|
1270 |
-
"rewards/margins":
|
1271 |
-
"rewards/rejected": -
|
1272 |
"step": 780,
|
1273 |
-
"use_label":
|
1274 |
},
|
1275 |
{
|
1276 |
"epoch": 0.83,
|
1277 |
"learning_rate": 9.60419091967404e-06,
|
1278 |
-
"logits/chosen": -2.
|
1279 |
-
"logits/rejected": -2.
|
1280 |
-
"logps/chosen": -
|
1281 |
-
"logps/rejected": -
|
1282 |
-
"loss": 0.
|
1283 |
-
"pred_label":
|
1284 |
-
"rewards/accuracies": 0.
|
1285 |
-
"rewards/chosen": -
|
1286 |
-
"rewards/margins":
|
1287 |
-
"rewards/rejected": -
|
1288 |
"step": 790,
|
1289 |
-
"use_label":
|
1290 |
},
|
1291 |
{
|
1292 |
"epoch": 0.84,
|
1293 |
"learning_rate": 9.022118742724098e-06,
|
1294 |
-
"logits/chosen": -2.
|
1295 |
-
"logits/rejected": -2.
|
1296 |
-
"logps/chosen": -
|
1297 |
-
"logps/rejected": -
|
1298 |
-
"loss": 0.
|
1299 |
-
"pred_label":
|
1300 |
-
"rewards/accuracies": 0.
|
1301 |
-
"rewards/chosen": -
|
1302 |
-
"rewards/margins":
|
1303 |
-
"rewards/rejected": -
|
1304 |
"step": 800,
|
1305 |
-
"use_label":
|
1306 |
},
|
1307 |
{
|
1308 |
"epoch": 0.85,
|
1309 |
"learning_rate": 8.440046565774158e-06,
|
1310 |
-
"logits/chosen": -2.
|
1311 |
-
"logits/rejected": -2.
|
1312 |
-
"logps/chosen": -
|
1313 |
-
"logps/rejected": -
|
1314 |
-
"loss": 0.
|
1315 |
-
"pred_label":
|
1316 |
-
"rewards/accuracies": 0.
|
1317 |
-
"rewards/chosen": -
|
1318 |
-
"rewards/margins":
|
1319 |
-
"rewards/rejected": -
|
1320 |
"step": 810,
|
1321 |
-
"use_label":
|
1322 |
},
|
1323 |
{
|
1324 |
"epoch": 0.86,
|
1325 |
"learning_rate": 7.857974388824214e-06,
|
1326 |
-
"logits/chosen": -2.
|
1327 |
-
"logits/rejected": -2.
|
1328 |
-
"logps/chosen": -
|
1329 |
-
"logps/rejected": -
|
1330 |
-
"loss": 0.
|
1331 |
-
"pred_label":
|
1332 |
-
"rewards/accuracies": 0.
|
1333 |
-
"rewards/chosen": -
|
1334 |
-
"rewards/margins":
|
1335 |
-
"rewards/rejected": -
|
1336 |
"step": 820,
|
1337 |
-
"use_label":
|
1338 |
},
|
1339 |
{
|
1340 |
"epoch": 0.87,
|
1341 |
"learning_rate": 7.275902211874273e-06,
|
1342 |
-
"logits/chosen": -2.
|
1343 |
-
"logits/rejected": -2.
|
1344 |
-
"logps/chosen": -
|
1345 |
-
"logps/rejected": -
|
1346 |
-
"loss": 0.
|
1347 |
-
"pred_label":
|
1348 |
-
"rewards/accuracies": 0.
|
1349 |
-
"rewards/chosen": -
|
1350 |
-
"rewards/margins":
|
1351 |
-
"rewards/rejected": -
|
1352 |
"step": 830,
|
1353 |
-
"use_label":
|
1354 |
},
|
1355 |
{
|
1356 |
"epoch": 0.88,
|
1357 |
"learning_rate": 6.693830034924331e-06,
|
1358 |
-
"logits/chosen": -2.
|
1359 |
-
"logits/rejected": -2.
|
1360 |
-
"logps/chosen": -
|
1361 |
-
"logps/rejected": -
|
1362 |
-
"loss": 0.
|
1363 |
-
"pred_label":
|
1364 |
-
"rewards/accuracies": 0.
|
1365 |
-
"rewards/chosen": -
|
1366 |
-
"rewards/margins":
|
1367 |
-
"rewards/rejected": -
|
1368 |
"step": 840,
|
1369 |
-
"use_label":
|
1370 |
},
|
1371 |
{
|
1372 |
"epoch": 0.89,
|
1373 |
"learning_rate": 6.111757857974389e-06,
|
1374 |
-
"logits/chosen": -2.
|
1375 |
-
"logits/rejected": -2.
|
1376 |
-
"logps/chosen": -
|
1377 |
-
"logps/rejected": -
|
1378 |
-
"loss": 0.
|
1379 |
-
"pred_label":
|
1380 |
-
"rewards/accuracies": 0.
|
1381 |
-
"rewards/chosen": -
|
1382 |
-
"rewards/margins":
|
1383 |
-
"rewards/rejected": -
|
1384 |
"step": 850,
|
1385 |
-
"use_label":
|
1386 |
},
|
1387 |
{
|
1388 |
"epoch": 0.9,
|
1389 |
"learning_rate": 5.529685681024447e-06,
|
1390 |
-
"logits/chosen": -2.
|
1391 |
-
"logits/rejected": -2.
|
1392 |
-
"logps/chosen": -
|
1393 |
-
"logps/rejected": -
|
1394 |
-
"loss": 0.
|
1395 |
-
"pred_label":
|
1396 |
-
"rewards/accuracies": 0.
|
1397 |
-
"rewards/chosen": -
|
1398 |
-
"rewards/margins":
|
1399 |
-
"rewards/rejected": -
|
1400 |
"step": 860,
|
1401 |
-
"use_label":
|
1402 |
},
|
1403 |
{
|
1404 |
"epoch": 0.91,
|
1405 |
"learning_rate": 4.947613504074506e-06,
|
1406 |
-
"logits/chosen": -2.
|
1407 |
-
"logits/rejected": -2.
|
1408 |
-
"logps/chosen": -
|
1409 |
-
"logps/rejected": -
|
1410 |
-
"loss": 0.
|
1411 |
-
"pred_label":
|
1412 |
-
"rewards/accuracies": 0.
|
1413 |
-
"rewards/chosen": -
|
1414 |
-
"rewards/margins":
|
1415 |
-
"rewards/rejected": -
|
1416 |
"step": 870,
|
1417 |
-
"use_label":
|
1418 |
},
|
1419 |
{
|
1420 |
"epoch": 0.92,
|
1421 |
"learning_rate": 4.3655413271245635e-06,
|
1422 |
-
"logits/chosen": -2.
|
1423 |
-
"logits/rejected": -2.
|
1424 |
-
"logps/chosen": -
|
1425 |
-
"logps/rejected": -
|
1426 |
-
"loss": 0.
|
1427 |
-
"pred_label":
|
1428 |
-
"rewards/accuracies": 0.
|
1429 |
-
"rewards/chosen": -
|
1430 |
-
"rewards/margins":
|
1431 |
-
"rewards/rejected": -
|
1432 |
"step": 880,
|
1433 |
-
"use_label":
|
1434 |
},
|
1435 |
{
|
1436 |
"epoch": 0.93,
|
1437 |
"learning_rate": 3.7834691501746217e-06,
|
1438 |
-
"logits/chosen": -2.
|
1439 |
-
"logits/rejected": -2.
|
1440 |
-
"logps/chosen": -
|
1441 |
-
"logps/rejected": -
|
1442 |
-
"loss": 0.
|
1443 |
-
"pred_label":
|
1444 |
-
"rewards/accuracies": 0.
|
1445 |
-
"rewards/chosen": -
|
1446 |
-
"rewards/margins":
|
1447 |
-
"rewards/rejected": -
|
1448 |
"step": 890,
|
1449 |
-
"use_label":
|
1450 |
},
|
1451 |
{
|
1452 |
"epoch": 0.94,
|
1453 |
"learning_rate": 3.2013969732246805e-06,
|
1454 |
-
"logits/chosen": -2.
|
1455 |
-
"logits/rejected": -2.
|
1456 |
-
"logps/chosen": -
|
1457 |
-
"logps/rejected": -
|
1458 |
-
"loss": 0.
|
1459 |
-
"pred_label":
|
1460 |
-
"rewards/accuracies": 0.
|
1461 |
-
"rewards/chosen": -
|
1462 |
-
"rewards/margins":
|
1463 |
-
"rewards/rejected": -
|
1464 |
"step": 900,
|
1465 |
-
"use_label":
|
1466 |
},
|
1467 |
{
|
1468 |
"epoch": 0.95,
|
1469 |
"learning_rate": 2.6193247962747383e-06,
|
1470 |
-
"logits/chosen": -2.
|
1471 |
-
"logits/rejected": -2.
|
1472 |
-
"logps/chosen": -
|
1473 |
-
"logps/rejected": -
|
1474 |
-
"loss": 0.
|
1475 |
-
"pred_label":
|
1476 |
-
"rewards/accuracies": 0.
|
1477 |
-
"rewards/chosen": -
|
1478 |
-
"rewards/margins":
|
1479 |
-
"rewards/rejected": -
|
1480 |
"step": 910,
|
1481 |
-
"use_label":
|
1482 |
},
|
1483 |
{
|
1484 |
"epoch": 0.96,
|
1485 |
"learning_rate": 2.037252619324796e-06,
|
1486 |
-
"logits/chosen": -2.
|
1487 |
-
"logits/rejected": -2.
|
1488 |
-
"logps/chosen": -
|
1489 |
-
"logps/rejected": -
|
1490 |
-
"loss": 0.
|
1491 |
-
"pred_label":
|
1492 |
-
"rewards/accuracies": 0.
|
1493 |
-
"rewards/chosen": -
|
1494 |
-
"rewards/margins":
|
1495 |
-
"rewards/rejected": -
|
1496 |
"step": 920,
|
1497 |
-
"use_label":
|
1498 |
},
|
1499 |
{
|
1500 |
"epoch": 0.97,
|
1501 |
"learning_rate": 1.4551804423748545e-06,
|
1502 |
-
"logits/chosen": -2.
|
1503 |
-
"logits/rejected": -2.
|
1504 |
-
"logps/chosen": -
|
1505 |
-
"logps/rejected": -
|
1506 |
-
"loss": 0.
|
1507 |
-
"pred_label":
|
1508 |
"rewards/accuracies": 0.8125,
|
1509 |
-
"rewards/chosen": -
|
1510 |
-
"rewards/margins":
|
1511 |
-
"rewards/rejected": -
|
1512 |
"step": 930,
|
1513 |
-
"use_label":
|
1514 |
},
|
1515 |
{
|
1516 |
"epoch": 0.98,
|
1517 |
"learning_rate": 8.731082654249127e-07,
|
1518 |
-
"logits/chosen": -2.
|
1519 |
-
"logits/rejected": -2.
|
1520 |
-
"logps/chosen": -
|
1521 |
-
"logps/rejected": -
|
1522 |
-
"loss": 0.
|
1523 |
-
"pred_label":
|
1524 |
-
"rewards/accuracies": 0.
|
1525 |
-
"rewards/chosen": -
|
1526 |
-
"rewards/margins":
|
1527 |
-
"rewards/rejected": -
|
1528 |
"step": 940,
|
1529 |
-
"use_label":
|
1530 |
},
|
1531 |
{
|
1532 |
"epoch": 0.99,
|
1533 |
"learning_rate": 2.910360884749709e-07,
|
1534 |
-
"logits/chosen": -2.
|
1535 |
-
"logits/rejected": -2.
|
1536 |
-
"logps/chosen": -
|
1537 |
-
"logps/rejected": -
|
1538 |
-
"loss": 0.
|
1539 |
-
"pred_label":
|
1540 |
-
"rewards/accuracies": 0.
|
1541 |
-
"rewards/chosen": -
|
1542 |
-
"rewards/margins":
|
1543 |
-
"rewards/rejected": -
|
1544 |
"step": 950,
|
1545 |
-
"use_label":
|
1546 |
},
|
1547 |
{
|
1548 |
"epoch": 1.0,
|
1549 |
-
"eval_logits/chosen": -2.
|
1550 |
-
"eval_logits/rejected": -2.
|
1551 |
-
"eval_logps/chosen": -
|
1552 |
-
"eval_logps/rejected": -
|
1553 |
-
"eval_loss": 0.
|
1554 |
-
"eval_pred_label":
|
1555 |
-
"eval_rewards/accuracies": 0.
|
1556 |
-
"eval_rewards/chosen": -
|
1557 |
-
"eval_rewards/margins":
|
1558 |
-
"eval_rewards/rejected": -
|
1559 |
-
"eval_runtime":
|
1560 |
-
"eval_samples_per_second": 4.
|
1561 |
-
"eval_steps_per_second": 0.
|
1562 |
-
"eval_use_label":
|
1563 |
"step": 955
|
1564 |
},
|
1565 |
{
|
1566 |
"epoch": 1.0,
|
1567 |
"step": 955,
|
1568 |
"total_flos": 0.0,
|
1569 |
-
"train_loss": 0.
|
1570 |
-
"train_runtime":
|
1571 |
-
"train_samples_per_second": 2.
|
1572 |
"train_steps_per_second": 0.038
|
1573 |
}
|
1574 |
],
|
|
|
75 |
{
|
76 |
"epoch": 0.04,
|
77 |
"learning_rate": 2.0833333333333336e-05,
|
78 |
+
"logits/chosen": -2.840946674346924,
|
79 |
+
"logits/rejected": -2.8493659496307373,
|
80 |
+
"logps/chosen": -281.32928466796875,
|
81 |
+
"logps/rejected": -277.8607482910156,
|
82 |
+
"loss": 0.6339,
|
83 |
+
"pred_label": 0.10000000149011612,
|
84 |
+
"rewards/accuracies": 0.6875,
|
85 |
+
"rewards/chosen": 0.02641097828745842,
|
86 |
+
"rewards/margins": 0.2079576551914215,
|
87 |
+
"rewards/rejected": -0.1815466731786728,
|
88 |
"step": 40,
|
89 |
+
"use_label": 561.9000244140625
|
90 |
},
|
91 |
{
|
92 |
"epoch": 0.05,
|
93 |
"learning_rate": 2.604166666666667e-05,
|
94 |
+
"logits/chosen": -2.854478120803833,
|
95 |
+
"logits/rejected": -2.839433193206787,
|
96 |
+
"logps/chosen": -266.80517578125,
|
97 |
+
"logps/rejected": -261.9176025390625,
|
98 |
+
"loss": 0.5914,
|
99 |
+
"pred_label": 2.950000047683716,
|
100 |
+
"rewards/accuracies": 0.643750011920929,
|
101 |
+
"rewards/chosen": -0.06968289613723755,
|
102 |
+
"rewards/margins": 0.33043327927589417,
|
103 |
+
"rewards/rejected": -0.4001162648200989,
|
104 |
"step": 50,
|
105 |
+
"use_label": 719.0499877929688
|
106 |
},
|
107 |
{
|
108 |
"epoch": 0.06,
|
109 |
"learning_rate": 3.125e-05,
|
110 |
+
"logits/chosen": -2.817157506942749,
|
111 |
+
"logits/rejected": -2.8072521686553955,
|
112 |
+
"logps/chosen": -300.76080322265625,
|
113 |
+
"logps/rejected": -290.3218994140625,
|
114 |
+
"loss": 0.5988,
|
115 |
+
"pred_label": 11.574999809265137,
|
116 |
+
"rewards/accuracies": 0.699999988079071,
|
117 |
+
"rewards/chosen": -0.03455673158168793,
|
118 |
+
"rewards/margins": 0.3870925307273865,
|
119 |
+
"rewards/rejected": -0.4216492772102356,
|
120 |
"step": 60,
|
121 |
+
"use_label": 870.4249877929688
|
122 |
},
|
123 |
{
|
124 |
"epoch": 0.07,
|
125 |
"learning_rate": 3.6458333333333336e-05,
|
126 |
+
"logits/chosen": -2.820021152496338,
|
127 |
+
"logits/rejected": -2.813854217529297,
|
128 |
+
"logps/chosen": -294.513671875,
|
129 |
+
"logps/rejected": -278.9858093261719,
|
130 |
+
"loss": 0.5227,
|
131 |
+
"pred_label": 25.399999618530273,
|
132 |
+
"rewards/accuracies": 0.7437499761581421,
|
133 |
+
"rewards/chosen": -0.020013216882944107,
|
134 |
+
"rewards/margins": 0.5820196866989136,
|
135 |
+
"rewards/rejected": -0.6020328998565674,
|
136 |
"step": 70,
|
137 |
+
"use_label": 1016.5999755859375
|
138 |
},
|
139 |
{
|
140 |
"epoch": 0.08,
|
141 |
"learning_rate": 4.166666666666667e-05,
|
142 |
+
"logits/chosen": -2.7645280361175537,
|
143 |
+
"logits/rejected": -2.7516016960144043,
|
144 |
+
"logps/chosen": -306.9416809082031,
|
145 |
+
"logps/rejected": -288.9856262207031,
|
146 |
+
"loss": 0.4536,
|
147 |
+
"pred_label": 58.599998474121094,
|
148 |
+
"rewards/accuracies": 0.75,
|
149 |
+
"rewards/chosen": -0.16363248229026794,
|
150 |
+
"rewards/margins": 0.8336677551269531,
|
151 |
+
"rewards/rejected": -0.9973002672195435,
|
152 |
"step": 80,
|
153 |
+
"use_label": 1143.4000244140625
|
154 |
},
|
155 |
{
|
156 |
"epoch": 0.09,
|
157 |
"learning_rate": 4.6875e-05,
|
158 |
+
"logits/chosen": -2.6945321559906006,
|
159 |
+
"logits/rejected": -2.747893810272217,
|
160 |
+
"logps/chosen": -295.63934326171875,
|
161 |
+
"logps/rejected": -275.24127197265625,
|
162 |
+
"loss": 0.4735,
|
163 |
+
"pred_label": 106.5,
|
164 |
+
"rewards/accuracies": 0.731249988079071,
|
165 |
+
"rewards/chosen": 0.014383295550942421,
|
166 |
+
"rewards/margins": 0.911063551902771,
|
167 |
+
"rewards/rejected": -0.8966802358627319,
|
168 |
"step": 90,
|
169 |
+
"use_label": 1255.5
|
170 |
},
|
171 |
{
|
172 |
"epoch": 0.1,
|
173 |
"learning_rate": 4.976717112922003e-05,
|
174 |
+
"logits/chosen": -2.719611406326294,
|
175 |
+
"logits/rejected": -2.718784809112549,
|
176 |
+
"logps/chosen": -276.9577941894531,
|
177 |
+
"logps/rejected": -306.4429626464844,
|
178 |
+
"loss": 0.484,
|
179 |
+
"pred_label": 153.3249969482422,
|
180 |
+
"rewards/accuracies": 0.706250011920929,
|
181 |
+
"rewards/chosen": -0.28260549902915955,
|
182 |
+
"rewards/margins": 1.008049726486206,
|
183 |
+
"rewards/rejected": -1.290655255317688,
|
184 |
"step": 100,
|
185 |
+
"use_label": 1368.675048828125
|
186 |
},
|
187 |
{
|
188 |
"epoch": 0.12,
|
189 |
"learning_rate": 4.918509895227008e-05,
|
190 |
+
"logits/chosen": -2.7281861305236816,
|
191 |
+
"logits/rejected": -2.687361478805542,
|
192 |
+
"logps/chosen": -269.6018981933594,
|
193 |
+
"logps/rejected": -263.89166259765625,
|
194 |
+
"loss": 0.4361,
|
195 |
+
"pred_label": 213.60000610351562,
|
196 |
+
"rewards/accuracies": 0.6312500238418579,
|
197 |
+
"rewards/chosen": -0.4927287697792053,
|
198 |
+
"rewards/margins": 0.9790979623794556,
|
199 |
+
"rewards/rejected": -1.4718266725540161,
|
200 |
"step": 110,
|
201 |
+
"use_label": 1468.4000244140625
|
202 |
},
|
203 |
{
|
204 |
"epoch": 0.13,
|
205 |
"learning_rate": 4.860302677532014e-05,
|
206 |
+
"logits/chosen": -2.7501158714294434,
|
207 |
+
"logits/rejected": -2.742772340774536,
|
208 |
+
"logps/chosen": -281.5338439941406,
|
209 |
+
"logps/rejected": -259.219970703125,
|
210 |
+
"loss": 0.4674,
|
211 |
+
"pred_label": 270.0249938964844,
|
212 |
+
"rewards/accuracies": 0.6625000238418579,
|
213 |
+
"rewards/chosen": -0.20209825038909912,
|
214 |
+
"rewards/margins": 0.9515292048454285,
|
215 |
+
"rewards/rejected": -1.1536273956298828,
|
216 |
"step": 120,
|
217 |
+
"use_label": 1571.9749755859375
|
218 |
},
|
219 |
{
|
220 |
"epoch": 0.14,
|
221 |
"learning_rate": 4.80209545983702e-05,
|
222 |
+
"logits/chosen": -2.8077096939086914,
|
223 |
+
"logits/rejected": -2.7647416591644287,
|
224 |
+
"logps/chosen": -308.7791748046875,
|
225 |
+
"logps/rejected": -282.69976806640625,
|
226 |
+
"loss": 0.4803,
|
227 |
+
"pred_label": 317.4750061035156,
|
228 |
+
"rewards/accuracies": 0.6812499761581421,
|
229 |
+
"rewards/chosen": -0.15117435157299042,
|
230 |
+
"rewards/margins": 0.8076679110527039,
|
231 |
+
"rewards/rejected": -0.958842396736145,
|
232 |
"step": 130,
|
233 |
+
"use_label": 1684.5250244140625
|
234 |
},
|
235 |
{
|
236 |
"epoch": 0.15,
|
237 |
"learning_rate": 4.743888242142026e-05,
|
238 |
+
"logits/chosen": -2.814915657043457,
|
239 |
+
"logits/rejected": -2.8093409538269043,
|
240 |
+
"logps/chosen": -261.60418701171875,
|
241 |
+
"logps/rejected": -249.7925567626953,
|
242 |
+
"loss": 0.3944,
|
243 |
+
"pred_label": 370.67498779296875,
|
244 |
+
"rewards/accuracies": 0.706250011920929,
|
245 |
+
"rewards/chosen": -0.34973591566085815,
|
246 |
+
"rewards/margins": 1.1276283264160156,
|
247 |
+
"rewards/rejected": -1.4773643016815186,
|
248 |
"step": 140,
|
249 |
+
"use_label": 1791.324951171875
|
250 |
},
|
251 |
{
|
252 |
"epoch": 0.16,
|
253 |
"learning_rate": 4.685681024447032e-05,
|
254 |
+
"logits/chosen": -2.71696138381958,
|
255 |
+
"logits/rejected": -2.740691661834717,
|
256 |
+
"logps/chosen": -302.32843017578125,
|
257 |
+
"logps/rejected": -298.6430358886719,
|
258 |
+
"loss": 0.3948,
|
259 |
+
"pred_label": 430.0,
|
260 |
"rewards/accuracies": 0.612500011920929,
|
261 |
+
"rewards/chosen": -0.7442375421524048,
|
262 |
+
"rewards/margins": 0.9453207850456238,
|
263 |
+
"rewards/rejected": -1.6895582675933838,
|
264 |
"step": 150,
|
265 |
+
"use_label": 1892.0
|
266 |
},
|
267 |
{
|
268 |
"epoch": 0.17,
|
269 |
"learning_rate": 4.6274738067520374e-05,
|
270 |
+
"logits/chosen": -2.7621922492980957,
|
271 |
+
"logits/rejected": -2.707430601119995,
|
272 |
+
"logps/chosen": -264.4729919433594,
|
273 |
+
"logps/rejected": -283.64837646484375,
|
274 |
+
"loss": 0.4037,
|
275 |
+
"pred_label": 494.9750061035156,
|
276 |
+
"rewards/accuracies": 0.71875,
|
277 |
+
"rewards/chosen": -0.20909884572029114,
|
278 |
+
"rewards/margins": 1.2514889240264893,
|
279 |
+
"rewards/rejected": -1.4605878591537476,
|
280 |
"step": 160,
|
281 |
+
"use_label": 1987.0250244140625
|
282 |
},
|
283 |
{
|
284 |
"epoch": 0.18,
|
285 |
"learning_rate": 4.5692665890570435e-05,
|
286 |
+
"logits/chosen": -2.800288438796997,
|
287 |
+
"logits/rejected": -2.8005611896514893,
|
288 |
+
"logps/chosen": -298.9527282714844,
|
289 |
+
"logps/rejected": -264.09014892578125,
|
290 |
+
"loss": 0.3923,
|
291 |
+
"pred_label": 555.2249755859375,
|
292 |
+
"rewards/accuracies": 0.6625000238418579,
|
293 |
+
"rewards/chosen": -0.24765756726264954,
|
294 |
+
"rewards/margins": 1.1968435049057007,
|
295 |
+
"rewards/rejected": -1.4445011615753174,
|
296 |
"step": 170,
|
297 |
+
"use_label": 2086.77490234375
|
298 |
},
|
299 |
{
|
300 |
"epoch": 0.19,
|
301 |
"learning_rate": 4.511059371362049e-05,
|
302 |
+
"logits/chosen": -2.7035715579986572,
|
303 |
+
"logits/rejected": -2.6763572692871094,
|
304 |
+
"logps/chosen": -261.8144836425781,
|
305 |
+
"logps/rejected": -294.1943054199219,
|
306 |
+
"loss": 0.3547,
|
307 |
+
"pred_label": 639.875,
|
308 |
+
"rewards/accuracies": 0.731249988079071,
|
309 |
+
"rewards/chosen": -0.588528573513031,
|
310 |
+
"rewards/margins": 1.7626692056655884,
|
311 |
+
"rewards/rejected": -2.3511977195739746,
|
312 |
"step": 180,
|
313 |
+
"use_label": 2162.125
|
314 |
},
|
315 |
{
|
316 |
"epoch": 0.2,
|
317 |
"learning_rate": 4.452852153667055e-05,
|
318 |
+
"logits/chosen": -2.788900375366211,
|
319 |
+
"logits/rejected": -2.745344638824463,
|
320 |
+
"logps/chosen": -311.311767578125,
|
321 |
+
"logps/rejected": -287.326416015625,
|
322 |
+
"loss": 0.3357,
|
323 |
+
"pred_label": 725.2999877929688,
|
324 |
+
"rewards/accuracies": 0.793749988079071,
|
325 |
+
"rewards/chosen": -0.4055810868740082,
|
326 |
+
"rewards/margins": 2.22284197807312,
|
327 |
+
"rewards/rejected": -2.6284232139587402,
|
328 |
"step": 190,
|
329 |
+
"use_label": 2236.699951171875
|
330 |
},
|
331 |
{
|
332 |
"epoch": 0.21,
|
333 |
"learning_rate": 4.394644935972061e-05,
|
334 |
+
"logits/chosen": -2.789668560028076,
|
335 |
+
"logits/rejected": -2.7655420303344727,
|
336 |
+
"logps/chosen": -274.904052734375,
|
337 |
+
"logps/rejected": -299.71356201171875,
|
338 |
+
"loss": 0.3242,
|
339 |
+
"pred_label": 815.0,
|
340 |
+
"rewards/accuracies": 0.71875,
|
341 |
+
"rewards/chosen": -1.3575278520584106,
|
342 |
+
"rewards/margins": 1.7897872924804688,
|
343 |
+
"rewards/rejected": -3.147315263748169,
|
344 |
"step": 200,
|
345 |
+
"use_label": 2307.0
|
346 |
},
|
347 |
{
|
348 |
"epoch": 0.22,
|
349 |
"learning_rate": 4.336437718277067e-05,
|
350 |
+
"logits/chosen": -2.7797892093658447,
|
351 |
+
"logits/rejected": -2.796938419342041,
|
352 |
+
"logps/chosen": -310.7392883300781,
|
353 |
+
"logps/rejected": -278.1353454589844,
|
354 |
+
"loss": 0.3789,
|
355 |
+
"pred_label": 897.2750244140625,
|
356 |
+
"rewards/accuracies": 0.75,
|
357 |
+
"rewards/chosen": -0.9491372108459473,
|
358 |
+
"rewards/margins": 1.4454370737075806,
|
359 |
+
"rewards/rejected": -2.394573926925659,
|
360 |
"step": 210,
|
361 |
+
"use_label": 2384.72509765625
|
362 |
},
|
363 |
{
|
364 |
"epoch": 0.23,
|
365 |
"learning_rate": 4.278230500582072e-05,
|
366 |
+
"logits/chosen": -2.729732036590576,
|
367 |
+
"logits/rejected": -2.7413415908813477,
|
368 |
+
"logps/chosen": -283.9934387207031,
|
369 |
+
"logps/rejected": -297.1932678222656,
|
370 |
+
"loss": 0.3046,
|
371 |
+
"pred_label": 983.3499755859375,
|
372 |
+
"rewards/accuracies": 0.75,
|
373 |
+
"rewards/chosen": -0.9001734852790833,
|
374 |
+
"rewards/margins": 1.7534692287445068,
|
375 |
+
"rewards/rejected": -2.6536426544189453,
|
376 |
"step": 220,
|
377 |
+
"use_label": 2458.64990234375
|
378 |
},
|
379 |
{
|
380 |
"epoch": 0.24,
|
381 |
"learning_rate": 4.220023282887078e-05,
|
382 |
+
"logits/chosen": -2.7093918323516846,
|
383 |
+
"logits/rejected": -2.724216938018799,
|
384 |
+
"logps/chosen": -320.80218505859375,
|
385 |
+
"logps/rejected": -295.5621643066406,
|
386 |
+
"loss": 0.2963,
|
387 |
+
"pred_label": 1077.550048828125,
|
388 |
+
"rewards/accuracies": 0.7437499761581421,
|
389 |
+
"rewards/chosen": -1.2629055976867676,
|
390 |
+
"rewards/margins": 1.8823902606964111,
|
391 |
+
"rewards/rejected": -3.1452958583831787,
|
392 |
"step": 230,
|
393 |
+
"use_label": 2524.449951171875
|
394 |
},
|
395 |
{
|
396 |
"epoch": 0.25,
|
397 |
"learning_rate": 4.161816065192084e-05,
|
398 |
+
"logits/chosen": -2.707616090774536,
|
399 |
+
"logits/rejected": -2.6788878440856934,
|
400 |
+
"logps/chosen": -300.3539123535156,
|
401 |
+
"logps/rejected": -273.60333251953125,
|
402 |
+
"loss": 0.2668,
|
403 |
+
"pred_label": 1176.5250244140625,
|
404 |
+
"rewards/accuracies": 0.6875,
|
405 |
+
"rewards/chosen": -2.5776500701904297,
|
406 |
+
"rewards/margins": 2.4807069301605225,
|
407 |
+
"rewards/rejected": -5.058356761932373,
|
408 |
"step": 240,
|
409 |
+
"use_label": 2585.47509765625
|
410 |
},
|
411 |
{
|
412 |
"epoch": 0.26,
|
413 |
"learning_rate": 4.10360884749709e-05,
|
414 |
+
"logits/chosen": -2.7368381023406982,
|
415 |
+
"logits/rejected": -2.722391366958618,
|
416 |
+
"logps/chosen": -316.1106262207031,
|
417 |
+
"logps/rejected": -321.246337890625,
|
418 |
+
"loss": 0.3166,
|
419 |
+
"pred_label": 1270.574951171875,
|
420 |
+
"rewards/accuracies": 0.706250011920929,
|
421 |
+
"rewards/chosen": -2.5470499992370605,
|
422 |
+
"rewards/margins": 2.635700225830078,
|
423 |
+
"rewards/rejected": -5.182750225067139,
|
424 |
"step": 250,
|
425 |
+
"use_label": 2651.425048828125
|
426 |
},
|
427 |
{
|
428 |
"epoch": 0.27,
|
429 |
"learning_rate": 4.045401629802096e-05,
|
430 |
+
"logits/chosen": -2.7675890922546387,
|
431 |
+
"logits/rejected": -2.7713561058044434,
|
432 |
+
"logps/chosen": -266.34136962890625,
|
433 |
+
"logps/rejected": -286.3191833496094,
|
434 |
+
"loss": 0.3408,
|
435 |
+
"pred_label": 1354.6500244140625,
|
436 |
+
"rewards/accuracies": 0.7250000238418579,
|
437 |
+
"rewards/chosen": -1.7731956243515015,
|
438 |
+
"rewards/margins": 1.8298133611679077,
|
439 |
+
"rewards/rejected": -3.603008985519409,
|
440 |
"step": 260,
|
441 |
+
"use_label": 2727.35009765625
|
442 |
},
|
443 |
{
|
444 |
"epoch": 0.28,
|
445 |
"learning_rate": 3.9871944121071014e-05,
|
446 |
+
"logits/chosen": -2.764836311340332,
|
447 |
+
"logits/rejected": -2.7679736614227295,
|
448 |
+
"logps/chosen": -306.4355773925781,
|
449 |
+
"logps/rejected": -316.5330505371094,
|
450 |
+
"loss": 0.3,
|
451 |
+
"pred_label": 1444.324951171875,
|
452 |
+
"rewards/accuracies": 0.793749988079071,
|
453 |
+
"rewards/chosen": -1.4564238786697388,
|
454 |
+
"rewards/margins": 2.7216858863830566,
|
455 |
+
"rewards/rejected": -4.178110122680664,
|
456 |
"step": 270,
|
457 |
+
"use_label": 2797.675048828125
|
458 |
},
|
459 |
{
|
460 |
"epoch": 0.29,
|
461 |
"learning_rate": 3.928987194412107e-05,
|
462 |
+
"logits/chosen": -2.81805682182312,
|
463 |
+
"logits/rejected": -2.8056905269622803,
|
464 |
+
"logps/chosen": -309.92047119140625,
|
465 |
+
"logps/rejected": -300.94268798828125,
|
466 |
+
"loss": 0.3011,
|
467 |
+
"pred_label": 1543.4749755859375,
|
468 |
+
"rewards/accuracies": 0.75,
|
469 |
+
"rewards/chosen": -1.595649003982544,
|
470 |
+
"rewards/margins": 2.275697708129883,
|
471 |
+
"rewards/rejected": -3.8713467121124268,
|
472 |
"step": 280,
|
473 |
+
"use_label": 2858.52490234375
|
474 |
},
|
475 |
{
|
476 |
"epoch": 0.3,
|
477 |
"learning_rate": 3.870779976717113e-05,
|
478 |
+
"logits/chosen": -2.786515235900879,
|
479 |
+
"logits/rejected": -2.7539620399475098,
|
480 |
+
"logps/chosen": -312.62060546875,
|
481 |
+
"logps/rejected": -268.4746398925781,
|
482 |
+
"loss": 0.3045,
|
483 |
+
"pred_label": 1641.2249755859375,
|
484 |
+
"rewards/accuracies": 0.7124999761581421,
|
485 |
+
"rewards/chosen": -1.9698221683502197,
|
486 |
+
"rewards/margins": 1.8559925556182861,
|
487 |
+
"rewards/rejected": -3.825814723968506,
|
488 |
"step": 290,
|
489 |
+
"use_label": 2920.77490234375
|
490 |
},
|
491 |
{
|
492 |
"epoch": 0.31,
|
493 |
"learning_rate": 3.812572759022119e-05,
|
494 |
+
"logits/chosen": -2.79560923576355,
|
495 |
+
"logits/rejected": -2.744600296020508,
|
496 |
+
"logps/chosen": -279.0300598144531,
|
497 |
+
"logps/rejected": -275.5624084472656,
|
498 |
+
"loss": 0.3083,
|
499 |
+
"pred_label": 1731.300048828125,
|
500 |
+
"rewards/accuracies": 0.706250011920929,
|
501 |
+
"rewards/chosen": -1.5351210832595825,
|
502 |
+
"rewards/margins": 2.036700963973999,
|
503 |
+
"rewards/rejected": -3.571821928024292,
|
504 |
"step": 300,
|
505 |
+
"use_label": 2990.699951171875
|
506 |
},
|
507 |
{
|
508 |
"epoch": 0.32,
|
509 |
"learning_rate": 3.7543655413271246e-05,
|
510 |
+
"logits/chosen": -2.7574381828308105,
|
511 |
+
"logits/rejected": -2.6999497413635254,
|
512 |
+
"logps/chosen": -285.3556213378906,
|
513 |
+
"logps/rejected": -344.51849365234375,
|
514 |
+
"loss": 0.2794,
|
515 |
+
"pred_label": 1830.0,
|
516 |
+
"rewards/accuracies": 0.78125,
|
517 |
+
"rewards/chosen": -1.8423988819122314,
|
518 |
+
"rewards/margins": 3.7551066875457764,
|
519 |
+
"rewards/rejected": -5.597506046295166,
|
520 |
"step": 310,
|
521 |
+
"use_label": 3052.0
|
522 |
},
|
523 |
{
|
524 |
"epoch": 0.33,
|
525 |
"learning_rate": 3.696158323632131e-05,
|
526 |
+
"logits/chosen": -2.7264797687530518,
|
527 |
+
"logits/rejected": -2.687101125717163,
|
528 |
+
"logps/chosen": -318.2635192871094,
|
529 |
+
"logps/rejected": -331.156005859375,
|
530 |
+
"loss": 0.2657,
|
531 |
+
"pred_label": 1930.050048828125,
|
532 |
+
"rewards/accuracies": 0.699999988079071,
|
533 |
+
"rewards/chosen": -2.228361129760742,
|
534 |
+
"rewards/margins": 2.806436777114868,
|
535 |
+
"rewards/rejected": -5.034797668457031,
|
536 |
"step": 320,
|
537 |
+
"use_label": 3111.949951171875
|
538 |
},
|
539 |
{
|
540 |
"epoch": 0.35,
|
541 |
"learning_rate": 3.637951105937136e-05,
|
542 |
+
"logits/chosen": -2.7709312438964844,
|
543 |
+
"logits/rejected": -2.7176883220672607,
|
544 |
+
"logps/chosen": -289.0472717285156,
|
545 |
+
"logps/rejected": -277.4143371582031,
|
546 |
+
"loss": 0.2868,
|
547 |
+
"pred_label": 2034.1500244140625,
|
548 |
+
"rewards/accuracies": 0.6875,
|
549 |
+
"rewards/chosen": -1.8098747730255127,
|
550 |
+
"rewards/margins": 2.50335955619812,
|
551 |
+
"rewards/rejected": -4.313233852386475,
|
552 |
"step": 330,
|
553 |
+
"use_label": 3167.85009765625
|
554 |
},
|
555 |
{
|
556 |
"epoch": 0.36,
|
557 |
"learning_rate": 3.579743888242142e-05,
|
558 |
+
"logits/chosen": -2.69682240486145,
|
559 |
+
"logits/rejected": -2.7000985145568848,
|
560 |
+
"logps/chosen": -308.01141357421875,
|
561 |
+
"logps/rejected": -279.4844665527344,
|
562 |
+
"loss": 0.3048,
|
563 |
+
"pred_label": 2128.949951171875,
|
564 |
+
"rewards/accuracies": 0.675000011920929,
|
565 |
+
"rewards/chosen": -1.883141279220581,
|
566 |
+
"rewards/margins": 1.98026442527771,
|
567 |
+
"rewards/rejected": -3.863405704498291,
|
568 |
"step": 340,
|
569 |
+
"use_label": 3233.050048828125
|
570 |
},
|
571 |
{
|
572 |
"epoch": 0.37,
|
573 |
"learning_rate": 3.5215366705471484e-05,
|
574 |
+
"logits/chosen": -2.663508176803589,
|
575 |
+
"logits/rejected": -2.6349689960479736,
|
576 |
+
"logps/chosen": -314.67962646484375,
|
577 |
+
"logps/rejected": -331.1473693847656,
|
578 |
+
"loss": 0.2869,
|
579 |
+
"pred_label": 2234.77490234375,
|
580 |
+
"rewards/accuracies": 0.8187500238418579,
|
581 |
+
"rewards/chosen": -1.7484877109527588,
|
582 |
+
"rewards/margins": 3.6000218391418457,
|
583 |
+
"rewards/rejected": -5.348509788513184,
|
584 |
"step": 350,
|
585 |
+
"use_label": 3287.22509765625
|
586 |
},
|
587 |
{
|
588 |
"epoch": 0.38,
|
589 |
"learning_rate": 3.463329452852154e-05,
|
590 |
+
"logits/chosen": -2.7377424240112305,
|
591 |
+
"logits/rejected": -2.729678153991699,
|
592 |
+
"logps/chosen": -297.26873779296875,
|
593 |
+
"logps/rejected": -313.16888427734375,
|
594 |
+
"loss": 0.2666,
|
595 |
+
"pred_label": 2342.22509765625,
|
596 |
+
"rewards/accuracies": 0.762499988079071,
|
597 |
+
"rewards/chosen": -2.4474658966064453,
|
598 |
+
"rewards/margins": 2.349332094192505,
|
599 |
+
"rewards/rejected": -4.796797752380371,
|
600 |
"step": 360,
|
601 |
+
"use_label": 3339.77490234375
|
602 |
},
|
603 |
{
|
604 |
"epoch": 0.39,
|
605 |
"learning_rate": 3.40512223515716e-05,
|
606 |
+
"logits/chosen": -2.6696877479553223,
|
607 |
+
"logits/rejected": -2.5973799228668213,
|
608 |
+
"logps/chosen": -310.88897705078125,
|
609 |
+
"logps/rejected": -354.97222900390625,
|
610 |
+
"loss": 0.2556,
|
611 |
+
"pred_label": 2450.97509765625,
|
612 |
+
"rewards/accuracies": 0.7875000238418579,
|
613 |
+
"rewards/chosen": -3.468627452850342,
|
614 |
+
"rewards/margins": 4.456360816955566,
|
615 |
+
"rewards/rejected": -7.924988746643066,
|
616 |
"step": 370,
|
617 |
+
"use_label": 3391.02490234375
|
618 |
},
|
619 |
{
|
620 |
"epoch": 0.4,
|
621 |
"learning_rate": 3.3469150174621654e-05,
|
622 |
+
"logits/chosen": -2.7021727561950684,
|
623 |
+
"logits/rejected": -2.679241418838501,
|
624 |
+
"logps/chosen": -320.2778015136719,
|
625 |
+
"logps/rejected": -327.36431884765625,
|
626 |
+
"loss": 0.2394,
|
627 |
+
"pred_label": 2560.375,
|
628 |
+
"rewards/accuracies": 0.71875,
|
629 |
+
"rewards/chosen": -3.5047569274902344,
|
630 |
+
"rewards/margins": 3.173809289932251,
|
631 |
+
"rewards/rejected": -6.678565979003906,
|
632 |
"step": 380,
|
633 |
+
"use_label": 3441.625
|
634 |
},
|
635 |
{
|
636 |
"epoch": 0.41,
|
637 |
"learning_rate": 3.288707799767171e-05,
|
638 |
+
"logits/chosen": -2.6417362689971924,
|
639 |
+
"logits/rejected": -2.6089835166931152,
|
640 |
+
"logps/chosen": -295.4016418457031,
|
641 |
+
"logps/rejected": -288.3475341796875,
|
642 |
+
"loss": 0.2799,
|
643 |
+
"pred_label": 2666.675048828125,
|
644 |
+
"rewards/accuracies": 0.737500011920929,
|
645 |
+
"rewards/chosen": -1.2609713077545166,
|
646 |
+
"rewards/margins": 3.1455700397491455,
|
647 |
+
"rewards/rejected": -4.406540870666504,
|
648 |
"step": 390,
|
649 |
+
"use_label": 3495.324951171875
|
650 |
},
|
651 |
{
|
652 |
"epoch": 0.42,
|
653 |
"learning_rate": 3.2305005820721776e-05,
|
654 |
+
"logits/chosen": -2.774970054626465,
|
655 |
+
"logits/rejected": -2.757179021835327,
|
656 |
+
"logps/chosen": -317.8477478027344,
|
657 |
+
"logps/rejected": -290.2752685546875,
|
658 |
+
"loss": 0.2669,
|
659 |
+
"pred_label": 2758.64990234375,
|
660 |
+
"rewards/accuracies": 0.71875,
|
661 |
+
"rewards/chosen": -1.0796750783920288,
|
662 |
+
"rewards/margins": 2.0139389038085938,
|
663 |
+
"rewards/rejected": -3.093614101409912,
|
664 |
"step": 400,
|
665 |
+
"use_label": 3563.35009765625
|
666 |
},
|
667 |
{
|
668 |
"epoch": 0.43,
|
669 |
"learning_rate": 3.172293364377183e-05,
|
670 |
+
"logits/chosen": -2.777430295944214,
|
671 |
+
"logits/rejected": -2.7574374675750732,
|
672 |
+
"logps/chosen": -312.574951171875,
|
673 |
+
"logps/rejected": -291.488037109375,
|
674 |
+
"loss": 0.296,
|
675 |
+
"pred_label": 2843.10009765625,
|
676 |
+
"rewards/accuracies": 0.668749988079071,
|
677 |
+
"rewards/chosen": -1.4477269649505615,
|
678 |
+
"rewards/margins": 1.8462398052215576,
|
679 |
+
"rewards/rejected": -3.2939670085906982,
|
680 |
"step": 410,
|
681 |
+
"use_label": 3638.89990234375
|
682 |
},
|
683 |
{
|
684 |
"epoch": 0.44,
|
685 |
"learning_rate": 3.1140861466821885e-05,
|
686 |
+
"logits/chosen": -2.718291759490967,
|
687 |
+
"logits/rejected": -2.688828706741333,
|
688 |
+
"logps/chosen": -254.98593139648438,
|
689 |
+
"logps/rejected": -283.18560791015625,
|
690 |
+
"loss": 0.3046,
|
691 |
+
"pred_label": 2935.14990234375,
|
692 |
+
"rewards/accuracies": 0.6937500238418579,
|
693 |
+
"rewards/chosen": -1.440553069114685,
|
694 |
+
"rewards/margins": 1.9665199518203735,
|
695 |
+
"rewards/rejected": -3.4070727825164795,
|
696 |
"step": 420,
|
697 |
+
"use_label": 3706.85009765625
|
698 |
},
|
699 |
{
|
700 |
"epoch": 0.45,
|
701 |
"learning_rate": 3.055878928987195e-05,
|
702 |
+
"logits/chosen": -2.7530155181884766,
|
703 |
+
"logits/rejected": -2.7446746826171875,
|
704 |
+
"logps/chosen": -288.42120361328125,
|
705 |
+
"logps/rejected": -287.4293518066406,
|
706 |
+
"loss": 0.3446,
|
707 |
+
"pred_label": 3031.39990234375,
|
708 |
+
"rewards/accuracies": 0.6875,
|
709 |
+
"rewards/chosen": -1.0520861148834229,
|
710 |
+
"rewards/margins": 1.7391481399536133,
|
711 |
+
"rewards/rejected": -2.791234254837036,
|
712 |
"step": 430,
|
713 |
+
"use_label": 3770.60009765625
|
714 |
},
|
715 |
{
|
716 |
"epoch": 0.46,
|
717 |
"learning_rate": 2.9976717112922005e-05,
|
718 |
+
"logits/chosen": -2.8176846504211426,
|
719 |
+
"logits/rejected": -2.8231449127197266,
|
720 |
+
"logps/chosen": -296.26837158203125,
|
721 |
+
"logps/rejected": -284.00335693359375,
|
722 |
+
"loss": 0.2945,
|
723 |
+
"pred_label": 3121.39990234375,
|
724 |
+
"rewards/accuracies": 0.71875,
|
725 |
+
"rewards/chosen": -1.2998032569885254,
|
726 |
+
"rewards/margins": 1.8008592128753662,
|
727 |
+
"rewards/rejected": -3.1006627082824707,
|
728 |
"step": 440,
|
729 |
+
"use_label": 3840.60009765625
|
730 |
},
|
731 |
{
|
732 |
"epoch": 0.47,
|
733 |
"learning_rate": 2.939464493597206e-05,
|
734 |
+
"logits/chosen": -2.6171724796295166,
|
735 |
+
"logits/rejected": -2.6079916954040527,
|
736 |
+
"logps/chosen": -248.51025390625,
|
737 |
+
"logps/rejected": -273.4137268066406,
|
738 |
+
"loss": 0.287,
|
739 |
+
"pred_label": 3211.925048828125,
|
740 |
+
"rewards/accuracies": 0.7250000238418579,
|
741 |
+
"rewards/chosen": -1.8791544437408447,
|
742 |
+
"rewards/margins": 2.497044086456299,
|
743 |
+
"rewards/rejected": -4.3761982917785645,
|
744 |
"step": 450,
|
745 |
+
"use_label": 3910.074951171875
|
746 |
},
|
747 |
{
|
748 |
"epoch": 0.48,
|
749 |
"learning_rate": 2.881257275902212e-05,
|
750 |
+
"logits/chosen": -2.815080165863037,
|
751 |
+
"logits/rejected": -2.8055264949798584,
|
752 |
+
"logps/chosen": -318.76666259765625,
|
753 |
+
"logps/rejected": -305.4552917480469,
|
754 |
+
"loss": 0.3106,
|
755 |
+
"pred_label": 3311.449951171875,
|
756 |
+
"rewards/accuracies": 0.6937500238418579,
|
757 |
+
"rewards/chosen": -1.6964454650878906,
|
758 |
+
"rewards/margins": 1.949159026145935,
|
759 |
+
"rewards/rejected": -3.645604372024536,
|
760 |
"step": 460,
|
761 |
+
"use_label": 3970.550048828125
|
762 |
},
|
763 |
{
|
764 |
"epoch": 0.49,
|
765 |
"learning_rate": 2.8230500582072178e-05,
|
766 |
+
"logits/chosen": -2.736795425415039,
|
767 |
+
"logits/rejected": -2.70324444770813,
|
768 |
+
"logps/chosen": -289.38824462890625,
|
769 |
+
"logps/rejected": -302.501708984375,
|
770 |
+
"loss": 0.2652,
|
771 |
+
"pred_label": 3412.550048828125,
|
772 |
+
"rewards/accuracies": 0.6875,
|
773 |
+
"rewards/chosen": -2.2287776470184326,
|
774 |
+
"rewards/margins": 2.337141513824463,
|
775 |
+
"rewards/rejected": -4.565918922424316,
|
776 |
"step": 470,
|
777 |
+
"use_label": 4029.449951171875
|
778 |
},
|
779 |
{
|
780 |
"epoch": 0.5,
|
781 |
"learning_rate": 2.7648428405122233e-05,
|
782 |
+
"logits/chosen": -2.6519317626953125,
|
783 |
+
"logits/rejected": -2.616342067718506,
|
784 |
+
"logps/chosen": -274.76422119140625,
|
785 |
+
"logps/rejected": -303.0647277832031,
|
786 |
+
"loss": 0.3052,
|
787 |
+
"pred_label": 3516.85009765625,
|
788 |
+
"rewards/accuracies": 0.6812499761581421,
|
789 |
+
"rewards/chosen": -2.918997287750244,
|
790 |
+
"rewards/margins": 2.5355329513549805,
|
791 |
+
"rewards/rejected": -5.454530239105225,
|
792 |
"step": 480,
|
793 |
+
"use_label": 4085.14990234375
|
794 |
},
|
795 |
{
|
796 |
"epoch": 0.51,
|
797 |
"learning_rate": 2.7066356228172297e-05,
|
798 |
+
"logits/chosen": -2.6925861835479736,
|
799 |
+
"logits/rejected": -2.67038631439209,
|
800 |
+
"logps/chosen": -289.875,
|
801 |
+
"logps/rejected": -282.19903564453125,
|
802 |
+
"loss": 0.2758,
|
803 |
+
"pred_label": 3613.125,
|
804 |
+
"rewards/accuracies": 0.7250000238418579,
|
805 |
+
"rewards/chosen": -2.284597873687744,
|
806 |
+
"rewards/margins": 2.195889711380005,
|
807 |
+
"rewards/rejected": -4.48048734664917,
|
808 |
"step": 490,
|
809 |
+
"use_label": 4148.875
|
810 |
},
|
811 |
{
|
812 |
"epoch": 0.52,
|
813 |
"learning_rate": 2.6484284051222352e-05,
|
814 |
+
"logits/chosen": -2.7040975093841553,
|
815 |
+
"logits/rejected": -2.6474618911743164,
|
816 |
+
"logps/chosen": -308.2983703613281,
|
817 |
+
"logps/rejected": -305.19525146484375,
|
818 |
+
"loss": 0.2732,
|
819 |
+
"pred_label": 3721.64990234375,
|
820 |
+
"rewards/accuracies": 0.7437499761581421,
|
821 |
+
"rewards/chosen": -1.8841865062713623,
|
822 |
+
"rewards/margins": 2.809114456176758,
|
823 |
+
"rewards/rejected": -4.693300724029541,
|
824 |
"step": 500,
|
825 |
+
"use_label": 4200.35009765625
|
826 |
},
|
827 |
{
|
828 |
"epoch": 0.53,
|
829 |
"learning_rate": 2.590221187427241e-05,
|
830 |
+
"logits/chosen": -2.7423815727233887,
|
831 |
+
"logits/rejected": -2.699392795562744,
|
832 |
+
"logps/chosen": -296.5220642089844,
|
833 |
+
"logps/rejected": -285.8718566894531,
|
834 |
+
"loss": 0.2433,
|
835 |
+
"pred_label": 3824.625,
|
836 |
+
"rewards/accuracies": 0.7749999761581421,
|
837 |
+
"rewards/chosen": -2.1184868812561035,
|
838 |
+
"rewards/margins": 3.1191751956939697,
|
839 |
+
"rewards/rejected": -5.237661361694336,
|
840 |
"step": 510,
|
841 |
+
"use_label": 4257.375
|
842 |
},
|
843 |
{
|
844 |
"epoch": 0.54,
|
845 |
"learning_rate": 2.532013969732247e-05,
|
846 |
+
"logits/chosen": -2.7525784969329834,
|
847 |
+
"logits/rejected": -2.727067708969116,
|
848 |
+
"logps/chosen": -304.08343505859375,
|
849 |
+
"logps/rejected": -308.287841796875,
|
850 |
+
"loss": 0.2989,
|
851 |
+
"pred_label": 3925.175048828125,
|
852 |
+
"rewards/accuracies": 0.706250011920929,
|
853 |
+
"rewards/chosen": -1.2129416465759277,
|
854 |
+
"rewards/margins": 2.644366502761841,
|
855 |
+
"rewards/rejected": -3.8573079109191895,
|
856 |
"step": 520,
|
857 |
+
"use_label": 4316.8251953125
|
858 |
},
|
859 |
{
|
860 |
"epoch": 0.55,
|
861 |
"learning_rate": 2.4738067520372525e-05,
|
862 |
+
"logits/chosen": -2.759995937347412,
|
863 |
+
"logits/rejected": -2.6951136589050293,
|
864 |
+
"logps/chosen": -314.9224548339844,
|
865 |
+
"logps/rejected": -335.85101318359375,
|
866 |
+
"loss": 0.2862,
|
867 |
+
"pred_label": 4020.449951171875,
|
868 |
+
"rewards/accuracies": 0.706250011920929,
|
869 |
+
"rewards/chosen": -1.4606602191925049,
|
870 |
+
"rewards/margins": 2.726986885070801,
|
871 |
+
"rewards/rejected": -4.187647342681885,
|
872 |
"step": 530,
|
873 |
+
"use_label": 4381.5498046875
|
874 |
},
|
875 |
{
|
876 |
"epoch": 0.57,
|
877 |
"learning_rate": 2.4155995343422587e-05,
|
878 |
+
"logits/chosen": -2.7623682022094727,
|
879 |
+
"logits/rejected": -2.743201732635498,
|
880 |
+
"logps/chosen": -307.76995849609375,
|
881 |
+
"logps/rejected": -319.3795471191406,
|
882 |
+
"loss": 0.2676,
|
883 |
+
"pred_label": 4129.97509765625,
|
884 |
+
"rewards/accuracies": 0.78125,
|
885 |
+
"rewards/chosen": -1.9233381748199463,
|
886 |
+
"rewards/margins": 3.4177041053771973,
|
887 |
+
"rewards/rejected": -5.3410420417785645,
|
888 |
"step": 540,
|
889 |
+
"use_label": 4432.02490234375
|
890 |
},
|
891 |
{
|
892 |
"epoch": 0.58,
|
893 |
"learning_rate": 2.3573923166472644e-05,
|
894 |
+
"logits/chosen": -2.766664981842041,
|
895 |
+
"logits/rejected": -2.727250576019287,
|
896 |
+
"logps/chosen": -307.920166015625,
|
897 |
+
"logps/rejected": -313.8031311035156,
|
898 |
+
"loss": 0.2619,
|
899 |
+
"pred_label": 4236.1748046875,
|
900 |
+
"rewards/accuracies": 0.71875,
|
901 |
+
"rewards/chosen": -1.871700644493103,
|
902 |
+
"rewards/margins": 3.314380645751953,
|
903 |
+
"rewards/rejected": -5.186081886291504,
|
904 |
"step": 550,
|
905 |
+
"use_label": 4485.8251953125
|
906 |
},
|
907 |
{
|
908 |
"epoch": 0.59,
|
909 |
"learning_rate": 2.2991850989522702e-05,
|
910 |
+
"logits/chosen": -2.747933864593506,
|
911 |
+
"logits/rejected": -2.743129253387451,
|
912 |
+
"logps/chosen": -296.66705322265625,
|
913 |
+
"logps/rejected": -261.8858947753906,
|
914 |
+
"loss": 0.281,
|
915 |
+
"pred_label": 4334.72509765625,
|
916 |
+
"rewards/accuracies": 0.75,
|
917 |
+
"rewards/chosen": -2.2262706756591797,
|
918 |
+
"rewards/margins": 2.213707685470581,
|
919 |
+
"rewards/rejected": -4.43997859954834,
|
920 |
"step": 560,
|
921 |
+
"use_label": 4547.27490234375
|
922 |
},
|
923 |
{
|
924 |
"epoch": 0.6,
|
925 |
"learning_rate": 2.240977881257276e-05,
|
926 |
+
"logits/chosen": -2.7517714500427246,
|
927 |
+
"logits/rejected": -2.7313544750213623,
|
928 |
+
"logps/chosen": -288.4844970703125,
|
929 |
+
"logps/rejected": -302.5806579589844,
|
930 |
+
"loss": 0.2905,
|
931 |
+
"pred_label": 4442.875,
|
932 |
"rewards/accuracies": 0.706250011920929,
|
933 |
+
"rewards/chosen": -1.8995840549468994,
|
934 |
+
"rewards/margins": 2.1551976203918457,
|
935 |
+
"rewards/rejected": -4.054781913757324,
|
936 |
"step": 570,
|
937 |
+
"use_label": 4599.125
|
938 |
},
|
939 |
{
|
940 |
"epoch": 0.61,
|
941 |
"learning_rate": 2.1827706635622818e-05,
|
942 |
+
"logits/chosen": -2.7617430686950684,
|
943 |
+
"logits/rejected": -2.73646879196167,
|
944 |
+
"logps/chosen": -295.9949645996094,
|
945 |
+
"logps/rejected": -310.08331298828125,
|
946 |
+
"loss": 0.2821,
|
947 |
+
"pred_label": 4553.9501953125,
|
948 |
+
"rewards/accuracies": 0.768750011920929,
|
949 |
+
"rewards/chosen": -1.7886556386947632,
|
950 |
+
"rewards/margins": 2.772958993911743,
|
951 |
+
"rewards/rejected": -4.561614513397217,
|
952 |
"step": 580,
|
953 |
+
"use_label": 4648.0498046875
|
954 |
},
|
955 |
{
|
956 |
"epoch": 0.62,
|
957 |
"learning_rate": 2.124563445867288e-05,
|
958 |
+
"logits/chosen": -2.6858975887298584,
|
959 |
+
"logits/rejected": -2.6842644214630127,
|
960 |
+
"logps/chosen": -292.4759216308594,
|
961 |
+
"logps/rejected": -323.04754638671875,
|
962 |
+
"loss": 0.2683,
|
963 |
+
"pred_label": 4648.1748046875,
|
964 |
+
"rewards/accuracies": 0.6875,
|
965 |
+
"rewards/chosen": -2.0878047943115234,
|
966 |
+
"rewards/margins": 2.138942241668701,
|
967 |
+
"rewards/rejected": -4.226747035980225,
|
968 |
"step": 590,
|
969 |
+
"use_label": 4713.8251953125
|
970 |
},
|
971 |
{
|
972 |
"epoch": 0.63,
|
973 |
"learning_rate": 2.0663562281722934e-05,
|
974 |
+
"logits/chosen": -2.755965232849121,
|
975 |
+
"logits/rejected": -2.7383854389190674,
|
976 |
+
"logps/chosen": -327.86962890625,
|
977 |
+
"logps/rejected": -351.33074951171875,
|
978 |
+
"loss": 0.2585,
|
979 |
+
"pred_label": 4749.9501953125,
|
980 |
+
"rewards/accuracies": 0.7562500238418579,
|
981 |
+
"rewards/chosen": -1.4506783485412598,
|
982 |
+
"rewards/margins": 3.4063594341278076,
|
983 |
+
"rewards/rejected": -4.8570380210876465,
|
984 |
"step": 600,
|
985 |
+
"use_label": 4772.0498046875
|
986 |
},
|
987 |
{
|
988 |
"epoch": 0.64,
|
989 |
"learning_rate": 2.0081490104772992e-05,
|
990 |
+
"logits/chosen": -2.7139110565185547,
|
991 |
+
"logits/rejected": -2.7096176147460938,
|
992 |
+
"logps/chosen": -322.1976013183594,
|
993 |
+
"logps/rejected": -281.54266357421875,
|
994 |
+
"loss": 0.267,
|
995 |
+
"pred_label": 4857.875,
|
996 |
+
"rewards/accuracies": 0.706250011920929,
|
997 |
+
"rewards/chosen": -1.6288321018218994,
|
998 |
+
"rewards/margins": 2.5627994537353516,
|
999 |
+
"rewards/rejected": -4.19163179397583,
|
1000 |
"step": 610,
|
1001 |
+
"use_label": 4824.125
|
1002 |
},
|
1003 |
{
|
1004 |
"epoch": 0.65,
|
1005 |
"learning_rate": 1.9499417927823053e-05,
|
1006 |
+
"logits/chosen": -2.733909845352173,
|
1007 |
+
"logits/rejected": -2.6813321113586426,
|
1008 |
+
"logps/chosen": -296.90032958984375,
|
1009 |
+
"logps/rejected": -298.9272766113281,
|
1010 |
+
"loss": 0.2764,
|
1011 |
+
"pred_label": 4962.77490234375,
|
1012 |
+
"rewards/accuracies": 0.768750011920929,
|
1013 |
+
"rewards/chosen": -1.691881775856018,
|
1014 |
+
"rewards/margins": 2.9819297790527344,
|
1015 |
+
"rewards/rejected": -4.673811912536621,
|
1016 |
"step": 620,
|
1017 |
+
"use_label": 4879.22509765625
|
1018 |
},
|
1019 |
{
|
1020 |
"epoch": 0.66,
|
1021 |
"learning_rate": 1.8917345750873107e-05,
|
1022 |
+
"logits/chosen": -2.7139275074005127,
|
1023 |
+
"logits/rejected": -2.7132036685943604,
|
1024 |
+
"logps/chosen": -297.9289855957031,
|
1025 |
+
"logps/rejected": -306.44940185546875,
|
1026 |
+
"loss": 0.2715,
|
1027 |
+
"pred_label": 5066.4501953125,
|
1028 |
+
"rewards/accuracies": 0.7749999761581421,
|
1029 |
+
"rewards/chosen": -1.2809231281280518,
|
1030 |
+
"rewards/margins": 3.2467029094696045,
|
1031 |
+
"rewards/rejected": -4.527626991271973,
|
1032 |
"step": 630,
|
1033 |
+
"use_label": 4935.5498046875
|
1034 |
},
|
1035 |
{
|
1036 |
"epoch": 0.67,
|
1037 |
"learning_rate": 1.833527357392317e-05,
|
1038 |
+
"logits/chosen": -2.7025070190429688,
|
1039 |
+
"logits/rejected": -2.669739246368408,
|
1040 |
+
"logps/chosen": -266.36798095703125,
|
1041 |
+
"logps/rejected": -293.29669189453125,
|
1042 |
+
"loss": 0.2907,
|
1043 |
+
"pred_label": 5163.5,
|
1044 |
+
"rewards/accuracies": 0.768750011920929,
|
1045 |
+
"rewards/chosen": -1.3176727294921875,
|
1046 |
+
"rewards/margins": 3.1546790599823,
|
1047 |
+
"rewards/rejected": -4.472352027893066,
|
1048 |
"step": 640,
|
1049 |
+
"use_label": 4998.5
|
1050 |
},
|
1051 |
{
|
1052 |
"epoch": 0.68,
|
1053 |
"learning_rate": 1.7753201396973227e-05,
|
1054 |
+
"logits/chosen": -2.721750497817993,
|
1055 |
+
"logits/rejected": -2.6910438537597656,
|
1056 |
+
"logps/chosen": -316.4061279296875,
|
1057 |
+
"logps/rejected": -322.0241394042969,
|
1058 |
+
"loss": 0.2773,
|
1059 |
+
"pred_label": 5267.2001953125,
|
1060 |
+
"rewards/accuracies": 0.78125,
|
1061 |
+
"rewards/chosen": -1.4457476139068604,
|
1062 |
+
"rewards/margins": 2.958343982696533,
|
1063 |
+
"rewards/rejected": -4.404091835021973,
|
1064 |
"step": 650,
|
1065 |
+
"use_label": 5054.7998046875
|
1066 |
},
|
1067 |
{
|
1068 |
"epoch": 0.69,
|
1069 |
"learning_rate": 1.717112922002328e-05,
|
1070 |
+
"logits/chosen": -2.7757656574249268,
|
1071 |
+
"logits/rejected": -2.7493858337402344,
|
1072 |
+
"logps/chosen": -311.73321533203125,
|
1073 |
+
"logps/rejected": -285.9647216796875,
|
1074 |
+
"loss": 0.2968,
|
1075 |
+
"pred_label": 5371.77490234375,
|
1076 |
+
"rewards/accuracies": 0.6625000238418579,
|
1077 |
+
"rewards/chosen": -1.636950135231018,
|
1078 |
+
"rewards/margins": 1.971967339515686,
|
1079 |
+
"rewards/rejected": -3.6089179515838623,
|
1080 |
"step": 660,
|
1081 |
+
"use_label": 5110.22509765625
|
1082 |
},
|
1083 |
{
|
1084 |
"epoch": 0.7,
|
1085 |
"learning_rate": 1.6589057043073342e-05,
|
1086 |
+
"logits/chosen": -2.7224373817443848,
|
1087 |
+
"logits/rejected": -2.686645030975342,
|
1088 |
+
"logps/chosen": -299.3620300292969,
|
1089 |
+
"logps/rejected": -292.6378173828125,
|
1090 |
+
"loss": 0.2702,
|
1091 |
+
"pred_label": 5477.2001953125,
|
1092 |
+
"rewards/accuracies": 0.7875000238418579,
|
1093 |
+
"rewards/chosen": -1.828905701637268,
|
1094 |
+
"rewards/margins": 3.295133113861084,
|
1095 |
+
"rewards/rejected": -5.124039173126221,
|
1096 |
"step": 670,
|
1097 |
+
"use_label": 5164.7998046875
|
1098 |
},
|
1099 |
{
|
1100 |
"epoch": 0.71,
|
1101 |
"learning_rate": 1.60069848661234e-05,
|
1102 |
+
"logits/chosen": -2.7378199100494385,
|
1103 |
+
"logits/rejected": -2.7310595512390137,
|
1104 |
+
"logps/chosen": -304.2184753417969,
|
1105 |
+
"logps/rejected": -292.1669921875,
|
1106 |
+
"loss": 0.3204,
|
1107 |
+
"pred_label": 5576.47509765625,
|
1108 |
+
"rewards/accuracies": 0.668749988079071,
|
1109 |
+
"rewards/chosen": -1.7283070087432861,
|
1110 |
+
"rewards/margins": 1.6653932332992554,
|
1111 |
+
"rewards/rejected": -3.393700361251831,
|
1112 |
"step": 680,
|
1113 |
+
"use_label": 5225.52490234375
|
1114 |
},
|
1115 |
{
|
1116 |
"epoch": 0.72,
|
1117 |
"learning_rate": 1.5424912689173458e-05,
|
1118 |
+
"logits/chosen": -2.7175381183624268,
|
1119 |
+
"logits/rejected": -2.7000772953033447,
|
1120 |
+
"logps/chosen": -266.01263427734375,
|
1121 |
+
"logps/rejected": -258.3471374511719,
|
1122 |
+
"loss": 0.258,
|
1123 |
+
"pred_label": 5677.625,
|
1124 |
+
"rewards/accuracies": 0.7749999761581421,
|
1125 |
+
"rewards/chosen": -1.56284499168396,
|
1126 |
+
"rewards/margins": 3.3434224128723145,
|
1127 |
+
"rewards/rejected": -4.9062676429748535,
|
1128 |
"step": 690,
|
1129 |
+
"use_label": 5284.375
|
1130 |
},
|
1131 |
{
|
1132 |
"epoch": 0.73,
|
1133 |
"learning_rate": 1.4842840512223516e-05,
|
1134 |
+
"logits/chosen": -2.735502243041992,
|
1135 |
+
"logits/rejected": -2.690195322036743,
|
1136 |
+
"logps/chosen": -293.5663146972656,
|
1137 |
+
"logps/rejected": -299.68121337890625,
|
1138 |
+
"loss": 0.2598,
|
1139 |
+
"pred_label": 5784.4248046875,
|
1140 |
+
"rewards/accuracies": 0.7250000238418579,
|
1141 |
+
"rewards/chosen": -1.7876224517822266,
|
1142 |
+
"rewards/margins": 3.3526370525360107,
|
1143 |
+
"rewards/rejected": -5.140259742736816,
|
1144 |
"step": 700,
|
1145 |
+
"use_label": 5337.5751953125
|
1146 |
},
|
1147 |
{
|
1148 |
"epoch": 0.74,
|
1149 |
"learning_rate": 1.4260768335273575e-05,
|
1150 |
+
"logits/chosen": -2.7820751667022705,
|
1151 |
+
"logits/rejected": -2.74092173576355,
|
1152 |
+
"logps/chosen": -305.2626953125,
|
1153 |
+
"logps/rejected": -284.5972900390625,
|
1154 |
+
"loss": 0.2792,
|
1155 |
+
"pred_label": 5887.0498046875,
|
1156 |
+
"rewards/accuracies": 0.7250000238418579,
|
1157 |
+
"rewards/chosen": -1.5688129663467407,
|
1158 |
+
"rewards/margins": 2.764045000076294,
|
1159 |
+
"rewards/rejected": -4.332858085632324,
|
1160 |
"step": 710,
|
1161 |
+
"use_label": 5394.9501953125
|
1162 |
},
|
1163 |
{
|
1164 |
"epoch": 0.75,
|
1165 |
"learning_rate": 1.3678696158323633e-05,
|
1166 |
+
"logits/chosen": -2.741055727005005,
|
1167 |
+
"logits/rejected": -2.739257335662842,
|
1168 |
+
"logps/chosen": -288.95611572265625,
|
1169 |
+
"logps/rejected": -306.3753967285156,
|
1170 |
+
"loss": 0.2461,
|
1171 |
+
"pred_label": 5995.5498046875,
|
1172 |
+
"rewards/accuracies": 0.78125,
|
1173 |
+
"rewards/chosen": -1.7503833770751953,
|
1174 |
+
"rewards/margins": 3.5468757152557373,
|
1175 |
+
"rewards/rejected": -5.297258377075195,
|
1176 |
"step": 720,
|
1177 |
+
"use_label": 5446.4501953125
|
1178 |
},
|
1179 |
{
|
1180 |
"epoch": 0.76,
|
1181 |
"learning_rate": 1.309662398137369e-05,
|
1182 |
+
"logits/chosen": -2.7259597778320312,
|
1183 |
+
"logits/rejected": -2.719268321990967,
|
1184 |
+
"logps/chosen": -315.80181884765625,
|
1185 |
+
"logps/rejected": -282.5977783203125,
|
1186 |
+
"loss": 0.2265,
|
1187 |
+
"pred_label": 6100.625,
|
1188 |
+
"rewards/accuracies": 0.7124999761581421,
|
1189 |
+
"rewards/chosen": -1.8351987600326538,
|
1190 |
+
"rewards/margins": 2.257732629776001,
|
1191 |
+
"rewards/rejected": -4.092931270599365,
|
1192 |
"step": 730,
|
1193 |
+
"use_label": 5501.375
|
1194 |
},
|
1195 |
{
|
1196 |
"epoch": 0.77,
|
1197 |
"learning_rate": 1.2514551804423749e-05,
|
1198 |
+
"logits/chosen": -2.704035758972168,
|
1199 |
+
"logits/rejected": -2.694396495819092,
|
1200 |
+
"logps/chosen": -332.0672912597656,
|
1201 |
+
"logps/rejected": -348.5420837402344,
|
1202 |
+
"loss": 0.254,
|
1203 |
+
"pred_label": 6200.5751953125,
|
1204 |
+
"rewards/accuracies": 0.7749999761581421,
|
1205 |
+
"rewards/chosen": -2.469543933868408,
|
1206 |
+
"rewards/margins": 3.7357654571533203,
|
1207 |
+
"rewards/rejected": -6.2053093910217285,
|
1208 |
"step": 740,
|
1209 |
+
"use_label": 5561.4248046875
|
1210 |
},
|
1211 |
{
|
1212 |
"epoch": 0.79,
|
1213 |
"learning_rate": 1.1932479627473807e-05,
|
1214 |
+
"logits/chosen": -2.717568874359131,
|
1215 |
+
"logits/rejected": -2.6717875003814697,
|
1216 |
+
"logps/chosen": -285.0014343261719,
|
1217 |
+
"logps/rejected": -322.31671142578125,
|
1218 |
+
"loss": 0.2139,
|
1219 |
+
"pred_label": 6312.14990234375,
|
1220 |
+
"rewards/accuracies": 0.762499988079071,
|
1221 |
+
"rewards/chosen": -2.451322078704834,
|
1222 |
+
"rewards/margins": 3.727611541748047,
|
1223 |
+
"rewards/rejected": -6.178933620452881,
|
1224 |
"step": 750,
|
1225 |
+
"use_label": 5609.85009765625
|
1226 |
},
|
1227 |
{
|
1228 |
"epoch": 0.8,
|
1229 |
"learning_rate": 1.1350407450523866e-05,
|
1230 |
+
"logits/chosen": -2.719574451446533,
|
1231 |
+
"logits/rejected": -2.7501304149627686,
|
1232 |
+
"logps/chosen": -310.88677978515625,
|
1233 |
+
"logps/rejected": -308.3591003417969,
|
1234 |
+
"loss": 0.2668,
|
1235 |
+
"pred_label": 6422.97509765625,
|
1236 |
+
"rewards/accuracies": 0.768750011920929,
|
1237 |
+
"rewards/chosen": -1.5423939228057861,
|
1238 |
+
"rewards/margins": 3.0822057723999023,
|
1239 |
+
"rewards/rejected": -4.624599456787109,
|
1240 |
"step": 760,
|
1241 |
+
"use_label": 5659.02490234375
|
1242 |
},
|
1243 |
{
|
1244 |
"epoch": 0.81,
|
1245 |
"learning_rate": 1.0768335273573923e-05,
|
1246 |
+
"logits/chosen": -2.746647357940674,
|
1247 |
+
"logits/rejected": -2.733283519744873,
|
1248 |
+
"logps/chosen": -320.37213134765625,
|
1249 |
+
"logps/rejected": -333.1225280761719,
|
1250 |
+
"loss": 0.248,
|
1251 |
+
"pred_label": 6535.22509765625,
|
1252 |
+
"rewards/accuracies": 0.737500011920929,
|
1253 |
+
"rewards/chosen": -1.807631492614746,
|
1254 |
+
"rewards/margins": 3.0322647094726562,
|
1255 |
+
"rewards/rejected": -4.839896202087402,
|
1256 |
"step": 770,
|
1257 |
+
"use_label": 5706.77490234375
|
1258 |
},
|
1259 |
{
|
1260 |
"epoch": 0.82,
|
1261 |
"learning_rate": 1.0186263096623982e-05,
|
1262 |
+
"logits/chosen": -2.7445006370544434,
|
1263 |
+
"logits/rejected": -2.7238264083862305,
|
1264 |
+
"logps/chosen": -303.70404052734375,
|
1265 |
+
"logps/rejected": -301.3880615234375,
|
1266 |
+
"loss": 0.2467,
|
1267 |
+
"pred_label": 6641.10009765625,
|
1268 |
+
"rewards/accuracies": 0.737500011920929,
|
1269 |
+
"rewards/chosen": -2.2510251998901367,
|
1270 |
+
"rewards/margins": 3.1742377281188965,
|
1271 |
+
"rewards/rejected": -5.425262928009033,
|
1272 |
"step": 780,
|
1273 |
+
"use_label": 5760.89990234375
|
1274 |
},
|
1275 |
{
|
1276 |
"epoch": 0.83,
|
1277 |
"learning_rate": 9.60419091967404e-06,
|
1278 |
+
"logits/chosen": -2.7392661571502686,
|
1279 |
+
"logits/rejected": -2.7158615589141846,
|
1280 |
+
"logps/chosen": -294.07037353515625,
|
1281 |
+
"logps/rejected": -322.58758544921875,
|
1282 |
+
"loss": 0.2523,
|
1283 |
+
"pred_label": 6745.9501953125,
|
1284 |
+
"rewards/accuracies": 0.7124999761581421,
|
1285 |
+
"rewards/chosen": -2.626100778579712,
|
1286 |
+
"rewards/margins": 3.5596566200256348,
|
1287 |
+
"rewards/rejected": -6.185757637023926,
|
1288 |
"step": 790,
|
1289 |
+
"use_label": 5816.0498046875
|
1290 |
},
|
1291 |
{
|
1292 |
"epoch": 0.84,
|
1293 |
"learning_rate": 9.022118742724098e-06,
|
1294 |
+
"logits/chosen": -2.6276087760925293,
|
1295 |
+
"logits/rejected": -2.5911715030670166,
|
1296 |
+
"logps/chosen": -343.8062438964844,
|
1297 |
+
"logps/rejected": -343.41070556640625,
|
1298 |
+
"loss": 0.2253,
|
1299 |
+
"pred_label": 6867.2998046875,
|
1300 |
+
"rewards/accuracies": 0.737500011920929,
|
1301 |
+
"rewards/chosen": -2.9334328174591064,
|
1302 |
+
"rewards/margins": 3.8014297485351562,
|
1303 |
+
"rewards/rejected": -6.73486328125,
|
1304 |
"step": 800,
|
1305 |
+
"use_label": 5854.7001953125
|
1306 |
},
|
1307 |
{
|
1308 |
"epoch": 0.85,
|
1309 |
"learning_rate": 8.440046565774158e-06,
|
1310 |
+
"logits/chosen": -2.6645331382751465,
|
1311 |
+
"logits/rejected": -2.640127658843994,
|
1312 |
+
"logps/chosen": -313.86480712890625,
|
1313 |
+
"logps/rejected": -287.8795471191406,
|
1314 |
+
"loss": 0.2183,
|
1315 |
+
"pred_label": 6983.14990234375,
|
1316 |
+
"rewards/accuracies": 0.6937500238418579,
|
1317 |
+
"rewards/chosen": -3.2840819358825684,
|
1318 |
+
"rewards/margins": 2.622573137283325,
|
1319 |
+
"rewards/rejected": -5.9066548347473145,
|
1320 |
"step": 810,
|
1321 |
+
"use_label": 5898.85009765625
|
1322 |
},
|
1323 |
{
|
1324 |
"epoch": 0.86,
|
1325 |
"learning_rate": 7.857974388824214e-06,
|
1326 |
+
"logits/chosen": -2.684762477874756,
|
1327 |
+
"logits/rejected": -2.6731903553009033,
|
1328 |
+
"logps/chosen": -316.68157958984375,
|
1329 |
+
"logps/rejected": -307.9013671875,
|
1330 |
+
"loss": 0.2752,
|
1331 |
+
"pred_label": 7098.10009765625,
|
1332 |
+
"rewards/accuracies": 0.768750011920929,
|
1333 |
+
"rewards/chosen": -2.3363418579101562,
|
1334 |
+
"rewards/margins": 4.297635078430176,
|
1335 |
+
"rewards/rejected": -6.63397741317749,
|
1336 |
"step": 820,
|
1337 |
+
"use_label": 5943.89990234375
|
1338 |
},
|
1339 |
{
|
1340 |
"epoch": 0.87,
|
1341 |
"learning_rate": 7.275902211874273e-06,
|
1342 |
+
"logits/chosen": -2.7039098739624023,
|
1343 |
+
"logits/rejected": -2.6936678886413574,
|
1344 |
+
"logps/chosen": -289.11761474609375,
|
1345 |
+
"logps/rejected": -341.92388916015625,
|
1346 |
+
"loss": 0.2433,
|
1347 |
+
"pred_label": 7206.47509765625,
|
1348 |
+
"rewards/accuracies": 0.71875,
|
1349 |
+
"rewards/chosen": -2.6304259300231934,
|
1350 |
+
"rewards/margins": 3.5417304039001465,
|
1351 |
+
"rewards/rejected": -6.172156810760498,
|
1352 |
"step": 830,
|
1353 |
+
"use_label": 5995.52490234375
|
1354 |
},
|
1355 |
{
|
1356 |
"epoch": 0.88,
|
1357 |
"learning_rate": 6.693830034924331e-06,
|
1358 |
+
"logits/chosen": -2.6950812339782715,
|
1359 |
+
"logits/rejected": -2.695385456085205,
|
1360 |
+
"logps/chosen": -295.2992858886719,
|
1361 |
+
"logps/rejected": -315.1665954589844,
|
1362 |
+
"loss": 0.2626,
|
1363 |
+
"pred_label": 7316.375,
|
1364 |
+
"rewards/accuracies": 0.7124999761581421,
|
1365 |
+
"rewards/chosen": -2.7282135486602783,
|
1366 |
+
"rewards/margins": 2.3400490283966064,
|
1367 |
+
"rewards/rejected": -5.068262577056885,
|
1368 |
"step": 840,
|
1369 |
+
"use_label": 6045.625
|
1370 |
},
|
1371 |
{
|
1372 |
"epoch": 0.89,
|
1373 |
"learning_rate": 6.111757857974389e-06,
|
1374 |
+
"logits/chosen": -2.691633701324463,
|
1375 |
+
"logits/rejected": -2.7306087017059326,
|
1376 |
+
"logps/chosen": -319.0354919433594,
|
1377 |
+
"logps/rejected": -299.2270812988281,
|
1378 |
+
"loss": 0.2547,
|
1379 |
+
"pred_label": 7422.2001953125,
|
1380 |
+
"rewards/accuracies": 0.7749999761581421,
|
1381 |
+
"rewards/chosen": -2.400865316390991,
|
1382 |
+
"rewards/margins": 3.1857247352600098,
|
1383 |
+
"rewards/rejected": -5.586589813232422,
|
1384 |
"step": 850,
|
1385 |
+
"use_label": 6099.7998046875
|
1386 |
},
|
1387 |
{
|
1388 |
"epoch": 0.9,
|
1389 |
"learning_rate": 5.529685681024447e-06,
|
1390 |
+
"logits/chosen": -2.6894938945770264,
|
1391 |
+
"logits/rejected": -2.6826186180114746,
|
1392 |
+
"logps/chosen": -314.03753662109375,
|
1393 |
+
"logps/rejected": -334.26690673828125,
|
1394 |
+
"loss": 0.2358,
|
1395 |
+
"pred_label": 7535.10009765625,
|
1396 |
+
"rewards/accuracies": 0.75,
|
1397 |
+
"rewards/chosen": -2.9154160022735596,
|
1398 |
+
"rewards/margins": 3.2621803283691406,
|
1399 |
+
"rewards/rejected": -6.177596092224121,
|
1400 |
"step": 860,
|
1401 |
+
"use_label": 6146.89990234375
|
1402 |
},
|
1403 |
{
|
1404 |
"epoch": 0.91,
|
1405 |
"learning_rate": 4.947613504074506e-06,
|
1406 |
+
"logits/chosen": -2.6914477348327637,
|
1407 |
+
"logits/rejected": -2.6861438751220703,
|
1408 |
+
"logps/chosen": -267.033203125,
|
1409 |
+
"logps/rejected": -298.70428466796875,
|
1410 |
+
"loss": 0.2361,
|
1411 |
+
"pred_label": 7641.60009765625,
|
1412 |
+
"rewards/accuracies": 0.737500011920929,
|
1413 |
+
"rewards/chosen": -2.404568910598755,
|
1414 |
+
"rewards/margins": 3.4409728050231934,
|
1415 |
+
"rewards/rejected": -5.845541477203369,
|
1416 |
"step": 870,
|
1417 |
+
"use_label": 6200.39990234375
|
1418 |
},
|
1419 |
{
|
1420 |
"epoch": 0.92,
|
1421 |
"learning_rate": 4.3655413271245635e-06,
|
1422 |
+
"logits/chosen": -2.7052195072174072,
|
1423 |
+
"logits/rejected": -2.6577229499816895,
|
1424 |
+
"logps/chosen": -311.93719482421875,
|
1425 |
+
"logps/rejected": -314.8600769042969,
|
1426 |
+
"loss": 0.2224,
|
1427 |
+
"pred_label": 7745.8251953125,
|
1428 |
+
"rewards/accuracies": 0.7749999761581421,
|
1429 |
+
"rewards/chosen": -2.5347952842712402,
|
1430 |
+
"rewards/margins": 2.786163330078125,
|
1431 |
+
"rewards/rejected": -5.320958614349365,
|
1432 |
"step": 880,
|
1433 |
+
"use_label": 6256.1748046875
|
1434 |
},
|
1435 |
{
|
1436 |
"epoch": 0.93,
|
1437 |
"learning_rate": 3.7834691501746217e-06,
|
1438 |
+
"logits/chosen": -2.718468189239502,
|
1439 |
+
"logits/rejected": -2.710252523422241,
|
1440 |
+
"logps/chosen": -304.1473693847656,
|
1441 |
+
"logps/rejected": -332.4493103027344,
|
1442 |
+
"loss": 0.2246,
|
1443 |
+
"pred_label": 7864.375,
|
1444 |
+
"rewards/accuracies": 0.793749988079071,
|
1445 |
+
"rewards/chosen": -2.2862813472747803,
|
1446 |
+
"rewards/margins": 3.7558963298797607,
|
1447 |
+
"rewards/rejected": -6.042177677154541,
|
1448 |
"step": 890,
|
1449 |
+
"use_label": 6297.625
|
1450 |
},
|
1451 |
{
|
1452 |
"epoch": 0.94,
|
1453 |
"learning_rate": 3.2013969732246805e-06,
|
1454 |
+
"logits/chosen": -2.6227777004241943,
|
1455 |
+
"logits/rejected": -2.571324110031128,
|
1456 |
+
"logps/chosen": -307.7371826171875,
|
1457 |
+
"logps/rejected": -324.693359375,
|
1458 |
+
"loss": 0.2493,
|
1459 |
+
"pred_label": 7978.625,
|
1460 |
+
"rewards/accuracies": 0.7562500238418579,
|
1461 |
+
"rewards/chosen": -2.859910488128662,
|
1462 |
+
"rewards/margins": 3.757611036300659,
|
1463 |
+
"rewards/rejected": -6.617520809173584,
|
1464 |
"step": 900,
|
1465 |
+
"use_label": 6343.375
|
1466 |
},
|
1467 |
{
|
1468 |
"epoch": 0.95,
|
1469 |
"learning_rate": 2.6193247962747383e-06,
|
1470 |
+
"logits/chosen": -2.701064109802246,
|
1471 |
+
"logits/rejected": -2.665743350982666,
|
1472 |
+
"logps/chosen": -323.77337646484375,
|
1473 |
+
"logps/rejected": -322.24432373046875,
|
1474 |
+
"loss": 0.238,
|
1475 |
+
"pred_label": 8087.5,
|
1476 |
+
"rewards/accuracies": 0.75,
|
1477 |
+
"rewards/chosen": -2.7357375621795654,
|
1478 |
+
"rewards/margins": 3.1752612590789795,
|
1479 |
+
"rewards/rejected": -5.910999298095703,
|
1480 |
"step": 910,
|
1481 |
+
"use_label": 6394.5
|
1482 |
},
|
1483 |
{
|
1484 |
"epoch": 0.96,
|
1485 |
"learning_rate": 2.037252619324796e-06,
|
1486 |
+
"logits/chosen": -2.639965772628784,
|
1487 |
+
"logits/rejected": -2.6299610137939453,
|
1488 |
+
"logps/chosen": -327.05975341796875,
|
1489 |
+
"logps/rejected": -325.63909912109375,
|
1490 |
+
"loss": 0.2258,
|
1491 |
+
"pred_label": 8202.6748046875,
|
1492 |
+
"rewards/accuracies": 0.737500011920929,
|
1493 |
+
"rewards/chosen": -3.017204761505127,
|
1494 |
+
"rewards/margins": 3.5882301330566406,
|
1495 |
+
"rewards/rejected": -6.605435371398926,
|
1496 |
"step": 920,
|
1497 |
+
"use_label": 6439.3251953125
|
1498 |
},
|
1499 |
{
|
1500 |
"epoch": 0.97,
|
1501 |
"learning_rate": 1.4551804423748545e-06,
|
1502 |
+
"logits/chosen": -2.69065260887146,
|
1503 |
+
"logits/rejected": -2.6345713138580322,
|
1504 |
+
"logps/chosen": -286.16278076171875,
|
1505 |
+
"logps/rejected": -308.4196472167969,
|
1506 |
+
"loss": 0.2207,
|
1507 |
+
"pred_label": 8314.349609375,
|
1508 |
"rewards/accuracies": 0.8125,
|
1509 |
+
"rewards/chosen": -1.7949655055999756,
|
1510 |
+
"rewards/margins": 4.867483139038086,
|
1511 |
+
"rewards/rejected": -6.662448883056641,
|
1512 |
"step": 930,
|
1513 |
+
"use_label": 6487.64990234375
|
1514 |
},
|
1515 |
{
|
1516 |
"epoch": 0.98,
|
1517 |
"learning_rate": 8.731082654249127e-07,
|
1518 |
+
"logits/chosen": -2.71934175491333,
|
1519 |
+
"logits/rejected": -2.67936372756958,
|
1520 |
+
"logps/chosen": -318.2463073730469,
|
1521 |
+
"logps/rejected": -326.56634521484375,
|
1522 |
+
"loss": 0.234,
|
1523 |
+
"pred_label": 8430.5751953125,
|
1524 |
+
"rewards/accuracies": 0.7250000238418579,
|
1525 |
+
"rewards/chosen": -3.0853824615478516,
|
1526 |
+
"rewards/margins": 2.8827648162841797,
|
1527 |
+
"rewards/rejected": -5.968147277832031,
|
1528 |
"step": 940,
|
1529 |
+
"use_label": 6531.4248046875
|
1530 |
},
|
1531 |
{
|
1532 |
"epoch": 0.99,
|
1533 |
"learning_rate": 2.910360884749709e-07,
|
1534 |
+
"logits/chosen": -2.6776463985443115,
|
1535 |
+
"logits/rejected": -2.6152024269104004,
|
1536 |
+
"logps/chosen": -300.53662109375,
|
1537 |
+
"logps/rejected": -322.4374694824219,
|
1538 |
+
"loss": 0.2135,
|
1539 |
+
"pred_label": 8549.625,
|
1540 |
+
"rewards/accuracies": 0.7562500238418579,
|
1541 |
+
"rewards/chosen": -3.2081990242004395,
|
1542 |
+
"rewards/margins": 3.740739107131958,
|
1543 |
+
"rewards/rejected": -6.94893741607666,
|
1544 |
"step": 950,
|
1545 |
+
"use_label": 6572.375
|
1546 |
},
|
1547 |
{
|
1548 |
"epoch": 1.0,
|
1549 |
+
"eval_logits/chosen": -2.6974706649780273,
|
1550 |
+
"eval_logits/rejected": -2.665019989013672,
|
1551 |
+
"eval_logps/chosen": -311.6889953613281,
|
1552 |
+
"eval_logps/rejected": -322.95947265625,
|
1553 |
+
"eval_loss": 0.2408759742975235,
|
1554 |
+
"eval_pred_label": 8833.576171875,
|
1555 |
+
"eval_rewards/accuracies": 0.734000027179718,
|
1556 |
+
"eval_rewards/chosen": -2.7431609630584717,
|
1557 |
+
"eval_rewards/margins": 3.6228184700012207,
|
1558 |
+
"eval_rewards/rejected": -6.36598014831543,
|
1559 |
+
"eval_runtime": 452.5439,
|
1560 |
+
"eval_samples_per_second": 4.419,
|
1561 |
+
"eval_steps_per_second": 0.276,
|
1562 |
+
"eval_use_label": 6698.423828125,
|
1563 |
"step": 955
|
1564 |
},
|
1565 |
{
|
1566 |
"epoch": 1.0,
|
1567 |
"step": 955,
|
1568 |
"total_flos": 0.0,
|
1569 |
+
"train_loss": 0.31699458866219243,
|
1570 |
+
"train_runtime": 25218.7851,
|
1571 |
+
"train_samples_per_second": 2.424,
|
1572 |
"train_steps_per_second": 0.038
|
1573 |
}
|
1574 |
],
|