Model save
Browse files- README.md +2 -2
- all_results.json +6 -5
- config.json +1 -1
- generation_config.json +1 -1
- model-00001-of-00003.safetensors +1 -1
- model-00002-of-00003.safetensors +1 -1
- model-00003-of-00003.safetensors +1 -1
- runs/Jul08_05-29-38_n136-082-130/events.out.tfevents.1720387900.n136-082-130.2711574.0 +2 -2
- tokenizer.json +1 -0
- train_results.json +6 -5
- trainer_state.json +579 -567
- training_args.bin +2 -2
README.md
CHANGED
@@ -52,7 +52,7 @@ The following hyperparameters were used during training:
|
|
52 |
|
53 |
### Framework versions
|
54 |
|
55 |
-
- Transformers 4.
|
56 |
- Pytorch 2.1.2+cu118
|
57 |
- Datasets 2.16.1
|
58 |
-
- Tokenizers 0.
|
|
|
52 |
|
53 |
### Framework versions
|
54 |
|
55 |
+
- Transformers 4.41.1
|
56 |
- Pytorch 2.1.2+cu118
|
57 |
- Datasets 2.16.1
|
58 |
+
- Tokenizers 0.19.1
|
all_results.json
CHANGED
@@ -1,8 +1,9 @@
|
|
1 |
{
|
2 |
-
"epoch":
|
3 |
-
"
|
4 |
-
"
|
|
|
5 |
"train_samples": 66084,
|
6 |
-
"train_samples_per_second":
|
7 |
-
"train_steps_per_second": 0.
|
8 |
}
|
|
|
1 |
{
|
2 |
+
"epoch": 0.9990319457889641,
|
3 |
+
"total_flos": 0.0,
|
4 |
+
"train_loss": 1.2622329578843228,
|
5 |
+
"train_runtime": 8815.3772,
|
6 |
"train_samples": 66084,
|
7 |
+
"train_samples_per_second": 7.496,
|
8 |
+
"train_steps_per_second": 0.059
|
9 |
}
|
config.json
CHANGED
@@ -20,7 +20,7 @@
|
|
20 |
"sliding_window": 4096,
|
21 |
"tie_word_embeddings": false,
|
22 |
"torch_dtype": "bfloat16",
|
23 |
-
"transformers_version": "4.
|
24 |
"use_cache": false,
|
25 |
"vocab_size": 32000
|
26 |
}
|
|
|
20 |
"sliding_window": 4096,
|
21 |
"tie_word_embeddings": false,
|
22 |
"torch_dtype": "bfloat16",
|
23 |
+
"transformers_version": "4.41.1",
|
24 |
"use_cache": false,
|
25 |
"vocab_size": 32000
|
26 |
}
|
generation_config.json
CHANGED
@@ -2,5 +2,5 @@
|
|
2 |
"_from_model_config": true,
|
3 |
"bos_token_id": 1,
|
4 |
"eos_token_id": 2,
|
5 |
-
"transformers_version": "4.
|
6 |
}
|
|
|
2 |
"_from_model_config": true,
|
3 |
"bos_token_id": 1,
|
4 |
"eos_token_id": 2,
|
5 |
+
"transformers_version": "4.41.1"
|
6 |
}
|
model-00001-of-00003.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 4943162336
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:ea42776f0193c66b4e372e22097d2fc65400f7bdb2f3ca46f982c79196cf49f7
|
3 |
size 4943162336
|
model-00002-of-00003.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 4999819336
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:6eb3a99a80f98514c49768d558a81159ceee2e5cf406379207b21f1de70e3733
|
3 |
size 4999819336
|
model-00003-of-00003.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 4540516344
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:4c08bb1fbdcc4816d468af7fec45a822a9da800bb4f19eff2e0946970a0e8a2a
|
3 |
size 4540516344
|
runs/Jul08_05-29-38_n136-082-130/events.out.tfevents.1720387900.n136-082-130.2711574.0
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:8233040679669edc5de74dbf2c369f093b9a8fb17949255be5d238d6a8a3326b
|
3 |
+
size 41168
|
tokenizer.json
CHANGED
@@ -134,6 +134,7 @@
|
|
134 |
"end_of_word_suffix": null,
|
135 |
"fuse_unk": true,
|
136 |
"byte_fallback": true,
|
|
|
137 |
"vocab": {
|
138 |
"<unk>": 0,
|
139 |
"<s>": 1,
|
|
|
134 |
"end_of_word_suffix": null,
|
135 |
"fuse_unk": true,
|
136 |
"byte_fallback": true,
|
137 |
+
"ignore_merges": false,
|
138 |
"vocab": {
|
139 |
"<unk>": 0,
|
140 |
"<s>": 1,
|
train_results.json
CHANGED
@@ -1,8 +1,9 @@
|
|
1 |
{
|
2 |
-
"epoch":
|
3 |
-
"
|
4 |
-
"
|
|
|
5 |
"train_samples": 66084,
|
6 |
-
"train_samples_per_second":
|
7 |
-
"train_steps_per_second": 0.
|
8 |
}
|
|
|
1 |
{
|
2 |
+
"epoch": 0.9990319457889641,
|
3 |
+
"total_flos": 0.0,
|
4 |
+
"train_loss": 1.2622329578843228,
|
5 |
+
"train_runtime": 8815.3772,
|
6 |
"train_samples": 66084,
|
7 |
+
"train_samples_per_second": 7.496,
|
8 |
+
"train_steps_per_second": 0.059
|
9 |
}
|
trainer_state.json
CHANGED
@@ -9,8 +9,8 @@
|
|
9 |
"is_world_process_zero": true,
|
10 |
"log_history": [
|
11 |
{
|
12 |
-
"epoch": 0.
|
13 |
-
"grad_norm":
|
14 |
"learning_rate": 1.9230769230769234e-11,
|
15 |
"logits/chosen": -1.8683955669403076,
|
16 |
"logits/rejected": -1.7658718824386597,
|
@@ -24,778 +24,778 @@
|
|
24 |
"step": 1
|
25 |
},
|
26 |
{
|
27 |
-
"epoch": 0.
|
28 |
-
"grad_norm":
|
29 |
"learning_rate": 1.9230769230769234e-10,
|
30 |
-
"logits/chosen": -1.
|
31 |
-
"logits/rejected": -1.
|
32 |
-
"logps/chosen": -0.
|
33 |
-
"logps/rejected": -0.
|
34 |
-
"loss": 1.
|
35 |
-
"rewards/accuracies": 0.
|
36 |
-
"rewards/chosen": -0.
|
37 |
-
"rewards/margins": -0.
|
38 |
-
"rewards/rejected": 0.
|
39 |
"step": 10
|
40 |
},
|
41 |
{
|
42 |
-
"epoch": 0.
|
43 |
-
"grad_norm":
|
44 |
"learning_rate": 3.8461538461538467e-10,
|
45 |
-
"logits/chosen": -1.
|
46 |
-
"logits/rejected": -1.
|
47 |
-
"logps/chosen": -1.
|
48 |
-
"logps/rejected": -0.
|
49 |
-
"loss": 1.
|
50 |
-
"rewards/accuracies": 0.
|
51 |
-
"rewards/chosen":
|
52 |
-
"rewards/margins": 0.
|
53 |
-
"rewards/rejected": -0.
|
54 |
"step": 20
|
55 |
},
|
56 |
{
|
57 |
-
"epoch": 0.
|
58 |
-
"grad_norm":
|
59 |
"learning_rate": 5.769230769230769e-10,
|
60 |
-
"logits/chosen": -1.
|
61 |
-
"logits/rejected": -1.
|
62 |
-
"logps/chosen": -1.
|
63 |
-
"logps/rejected": -0.
|
64 |
-
"loss": 1.
|
65 |
-
"rewards/accuracies": 0.
|
66 |
-
"rewards/chosen": -0.
|
67 |
-
"rewards/margins": 0.
|
68 |
-
"rewards/rejected": -0.
|
69 |
"step": 30
|
70 |
},
|
71 |
{
|
72 |
-
"epoch": 0.
|
73 |
-
"grad_norm":
|
74 |
"learning_rate": 7.692307692307693e-10,
|
75 |
-
"logits/chosen": -1.
|
76 |
-
"logits/rejected": -1.
|
77 |
-
"logps/chosen": -1.
|
78 |
-
"logps/rejected": -0.
|
79 |
-
"loss": 1.
|
80 |
-
"rewards/accuracies": 0.
|
81 |
-
"rewards/chosen": -0.
|
82 |
-
"rewards/margins": -0.
|
83 |
-
"rewards/rejected": 0.
|
84 |
"step": 40
|
85 |
},
|
86 |
{
|
87 |
-
"epoch": 0.
|
88 |
-
"grad_norm":
|
89 |
"learning_rate": 9.615384615384616e-10,
|
90 |
-
"logits/chosen": -1.
|
91 |
-
"logits/rejected": -1.
|
92 |
-
"logps/chosen": -0.
|
93 |
-
"logps/rejected": -0.
|
94 |
-
"loss": 1.
|
95 |
-
"rewards/accuracies": 0.
|
96 |
-
"rewards/chosen": -0.
|
97 |
-
"rewards/margins": 0.
|
98 |
-
"rewards/rejected":
|
99 |
"step": 50
|
100 |
},
|
101 |
{
|
102 |
-
"epoch": 0.
|
103 |
-
"grad_norm":
|
104 |
"learning_rate": 9.99266706925562e-10,
|
105 |
-
"logits/chosen": -1.
|
106 |
-
"logits/rejected": -1.
|
107 |
-
"logps/chosen": -0.
|
108 |
-
"logps/rejected": -0.
|
109 |
-
"loss": 1.
|
110 |
-
"rewards/accuracies": 0.
|
111 |
-
"rewards/chosen": 0.
|
112 |
-
"rewards/margins":
|
113 |
-
"rewards/rejected": 0.
|
114 |
"step": 60
|
115 |
},
|
116 |
{
|
117 |
-
"epoch": 0.
|
118 |
-
"grad_norm":
|
119 |
"learning_rate": 9.96291389741603e-10,
|
120 |
-
"logits/chosen": -1.
|
121 |
-
"logits/rejected": -1.
|
122 |
-
"logps/chosen": -0.
|
123 |
-
"logps/rejected": -0.
|
124 |
-
"loss": 1.
|
125 |
-
"rewards/accuracies": 0.
|
126 |
-
"rewards/chosen": -0.
|
127 |
-
"rewards/margins": 0.
|
128 |
-
"rewards/rejected": -0.
|
129 |
"step": 70
|
130 |
},
|
131 |
{
|
132 |
-
"epoch": 0.
|
133 |
-
"grad_norm":
|
134 |
"learning_rate": 9.91041841371078e-10,
|
135 |
-
"logits/chosen": -1.
|
136 |
-
"logits/rejected": -1.
|
137 |
-
"logps/chosen": -1.
|
138 |
-
"logps/rejected": -0.
|
139 |
-
"loss": 1.
|
140 |
-
"rewards/accuracies": 0.
|
141 |
-
"rewards/chosen":
|
142 |
-
"rewards/margins":
|
143 |
-
"rewards/rejected": -0.
|
144 |
"step": 80
|
145 |
},
|
146 |
{
|
147 |
-
"epoch": 0.
|
148 |
-
"grad_norm":
|
149 |
"learning_rate": 9.835421176144035e-10,
|
150 |
-
"logits/chosen": -1.
|
151 |
-
"logits/rejected": -1.
|
152 |
-
"logps/chosen": -1.
|
153 |
-
"logps/rejected": -0.
|
154 |
-
"loss": 1.
|
155 |
-
"rewards/accuracies": 0.
|
156 |
-
"rewards/chosen": 0.
|
157 |
-
"rewards/margins": 0.
|
158 |
-
"rewards/rejected": 0.
|
159 |
"step": 90
|
160 |
},
|
161 |
{
|
162 |
-
"epoch": 0.
|
163 |
-
"grad_norm":
|
164 |
"learning_rate": 9.738265855914014e-10,
|
165 |
-
"logits/chosen": -1.
|
166 |
-
"logits/rejected": -1.
|
167 |
-
"logps/chosen": -0.
|
168 |
-
"logps/rejected": -0.
|
169 |
-
"loss": 1.
|
170 |
-
"rewards/accuracies": 0.
|
171 |
-
"rewards/chosen":
|
172 |
-
"rewards/margins": -0.
|
173 |
-
"rewards/rejected": 0.
|
174 |
"step": 100
|
175 |
},
|
176 |
{
|
177 |
-
"epoch": 0.
|
178 |
-
"grad_norm":
|
179 |
"learning_rate": 9.619397662556434e-10,
|
180 |
-
"logits/chosen": -1.
|
181 |
-
"logits/rejected": -1.
|
182 |
-
"logps/chosen": -0.
|
183 |
-
"logps/rejected": -0.
|
184 |
-
"loss": 1.
|
185 |
-
"rewards/accuracies": 0.
|
186 |
-
"rewards/chosen": 0.
|
187 |
-
"rewards/margins": -0.
|
188 |
-
"rewards/rejected": 0.
|
189 |
"step": 110
|
190 |
},
|
191 |
{
|
192 |
-
"epoch": 0.
|
193 |
-
"grad_norm":
|
194 |
"learning_rate": 9.47936130379344e-10,
|
195 |
-
"logits/chosen": -1.
|
196 |
-
"logits/rejected": -1.
|
197 |
-
"logps/chosen": -0.
|
198 |
-
"logps/rejected": -0.
|
199 |
-
"loss": 1.
|
200 |
-
"rewards/accuracies": 0.
|
201 |
-
"rewards/chosen": 0.
|
202 |
-
"rewards/margins": 0.
|
203 |
-
"rewards/rejected": 0.
|
204 |
"step": 120
|
205 |
},
|
206 |
{
|
207 |
-
"epoch": 0.
|
208 |
-
"grad_norm":
|
209 |
"learning_rate": 9.318798489436919e-10,
|
210 |
-
"logits/chosen": -1.
|
211 |
-
"logits/rejected": -1.
|
212 |
-
"logps/chosen": -0.
|
213 |
-
"logps/rejected": -0.
|
214 |
-
"loss": 1.
|
215 |
-
"rewards/accuracies": 0.
|
216 |
-
"rewards/chosen": 0.
|
217 |
-
"rewards/margins": 0.
|
218 |
-
"rewards/rejected":
|
219 |
"step": 130
|
220 |
},
|
221 |
{
|
222 |
-
"epoch": 0.
|
223 |
-
"grad_norm":
|
224 |
"learning_rate": 9.138444990784454e-10,
|
225 |
-
"logits/chosen": -1.
|
226 |
-
"logits/rejected": -1.
|
227 |
-
"logps/chosen": -0.
|
228 |
-
"logps/rejected": -0.
|
229 |
-
"loss": 1.
|
230 |
-
"rewards/accuracies": 0.
|
231 |
-
"rewards/chosen": -0.
|
232 |
-
"rewards/margins": -0.
|
233 |
-
"rewards/rejected": 0.
|
234 |
"step": 140
|
235 |
},
|
236 |
{
|
237 |
-
"epoch": 0.
|
238 |
-
"grad_norm":
|
239 |
"learning_rate": 8.939127268983109e-10,
|
240 |
-
"logits/chosen": -1.
|
241 |
-
"logits/rejected": -1.
|
242 |
-
"logps/chosen": -1.
|
243 |
-
"logps/rejected": -0.
|
244 |
-
"loss": 1.
|
245 |
-
"rewards/accuracies": 0.
|
246 |
-
"rewards/chosen": 0.
|
247 |
-
"rewards/margins": 0.
|
248 |
-
"rewards/rejected": 0.
|
249 |
"step": 150
|
250 |
},
|
251 |
{
|
252 |
-
"epoch": 0.
|
253 |
-
"grad_norm":
|
254 |
"learning_rate": 8.721758687811352e-10,
|
255 |
-
"logits/chosen": -1.
|
256 |
-
"logits/rejected": -1.
|
257 |
-
"logps/chosen": -0.
|
258 |
-
"logps/rejected": -0.
|
259 |
-
"loss": 1.
|
260 |
-
"rewards/accuracies": 0.
|
261 |
-
"rewards/chosen": -0.
|
262 |
-
"rewards/margins": -0.
|
263 |
-
"rewards/rejected":
|
264 |
"step": 160
|
265 |
},
|
266 |
{
|
267 |
-
"epoch": 0.
|
268 |
-
"grad_norm":
|
269 |
"learning_rate": 8.487335328233912e-10,
|
270 |
-
"logits/chosen": -1.
|
271 |
-
"logits/rejected": -1.
|
272 |
-
"logps/chosen": -0.
|
273 |
-
"logps/rejected": -0.
|
274 |
-
"loss": 1.
|
275 |
-
"rewards/accuracies": 0.
|
276 |
-
"rewards/chosen": 0.
|
277 |
-
"rewards/margins": 0.
|
278 |
-
"rewards/rejected": 0.
|
279 |
"step": 170
|
280 |
},
|
281 |
{
|
282 |
-
"epoch": 0.
|
283 |
-
"grad_norm":
|
284 |
"learning_rate": 8.236931423909139e-10,
|
285 |
-
"logits/chosen": -1.
|
286 |
-
"logits/rejected": -1.
|
287 |
-
"logps/chosen": -0.
|
288 |
-
"logps/rejected": -0.
|
289 |
-
"loss": 1.
|
290 |
-
"rewards/accuracies": 0.
|
291 |
-
"rewards/chosen": 0.
|
292 |
-
"rewards/margins": 0.
|
293 |
-
"rewards/rejected": -0.
|
294 |
"step": 180
|
295 |
},
|
296 |
{
|
297 |
-
"epoch": 0.
|
298 |
-
"grad_norm":
|
299 |
"learning_rate": 7.971694438565449e-10,
|
300 |
-
"logits/chosen": -1.
|
301 |
-
"logits/rejected": -1.
|
302 |
-
"logps/chosen": -0.
|
303 |
-
"logps/rejected": -0.
|
304 |
-
"loss": 1.
|
305 |
-
"rewards/accuracies": 0.
|
306 |
-
"rewards/chosen":
|
307 |
-
"rewards/margins": 0.
|
308 |
-
"rewards/rejected": -0.
|
309 |
"step": 190
|
310 |
},
|
311 |
{
|
312 |
-
"epoch": 0.
|
313 |
-
"grad_norm":
|
314 |
"learning_rate": 7.692839807804521e-10,
|
315 |
-
"logits/chosen": -1.
|
316 |
-
"logits/rejected": -1.
|
317 |
-
"logps/chosen": -0.
|
318 |
-
"logps/rejected": -0.
|
319 |
-
"loss": 1.
|
320 |
-
"rewards/accuracies": 0.
|
321 |
-
"rewards/chosen": -0.
|
322 |
-
"rewards/margins": -0.
|
323 |
-
"rewards/rejected": 0.
|
324 |
"step": 200
|
325 |
},
|
326 |
{
|
327 |
-
"epoch": 0.
|
328 |
-
"grad_norm":
|
329 |
"learning_rate": 7.401645369426697e-10,
|
330 |
-
"logits/chosen": -1.
|
331 |
-
"logits/rejected": -1.
|
332 |
-
"logps/chosen": -0.
|
333 |
-
"logps/rejected": -0.
|
334 |
-
"loss": 1.
|
335 |
-
"rewards/accuracies": 0.
|
336 |
-
"rewards/chosen": 0.
|
337 |
-
"rewards/margins": -0.
|
338 |
-
"rewards/rejected": 0.
|
339 |
"step": 210
|
340 |
},
|
341 |
{
|
342 |
-
"epoch": 0.
|
343 |
-
"grad_norm":
|
344 |
"learning_rate": 7.099445507801324e-10,
|
345 |
-
"logits/chosen": -1.
|
346 |
-
"logits/rejected": -1.
|
347 |
-
"logps/chosen": -0.
|
348 |
-
"logps/rejected": -0.
|
349 |
-
"loss": 1.
|
350 |
-
"rewards/accuracies": 0.
|
351 |
-
"rewards/chosen": 0.
|
352 |
-
"rewards/margins": 0.
|
353 |
-
"rewards/rejected": 0.
|
354 |
"step": 220
|
355 |
},
|
356 |
{
|
357 |
-
"epoch": 0.
|
358 |
-
"grad_norm":
|
359 |
"learning_rate": 6.7876250391152e-10,
|
360 |
-
"logits/chosen": -1.
|
361 |
-
"logits/rejected": -1.
|
362 |
-
"logps/chosen": -0.
|
363 |
-
"logps/rejected": -0.
|
364 |
-
"loss": 1.
|
365 |
-
"rewards/accuracies": 0.
|
366 |
-
"rewards/chosen":
|
367 |
-
"rewards/margins": 0.
|
368 |
-
"rewards/rejected": -0.
|
369 |
"step": 230
|
370 |
},
|
371 |
{
|
372 |
-
"epoch": 0.
|
373 |
-
"grad_norm":
|
374 |
"learning_rate": 6.467612865519674e-10,
|
375 |
-
"logits/chosen": -1.
|
376 |
-
"logits/rejected": -1.
|
377 |
-
"logps/chosen": -0.
|
378 |
-
"logps/rejected": -0.
|
379 |
-
"loss": 1.
|
380 |
-
"rewards/accuracies": 0.
|
381 |
-
"rewards/chosen":
|
382 |
-
"rewards/margins": -0.
|
383 |
-
"rewards/rejected": 0.
|
384 |
"step": 240
|
385 |
},
|
386 |
{
|
387 |
-
"epoch": 0.
|
388 |
-
"grad_norm":
|
389 |
"learning_rate": 6.14087542725593e-10,
|
390 |
-
"logits/chosen": -1.
|
391 |
-
"logits/rejected": -1.
|
392 |
-
"logps/chosen": -1.
|
393 |
-
"logps/rejected": -0.
|
394 |
-
"loss": 1.
|
395 |
-
"rewards/accuracies": 0.
|
396 |
-
"rewards/chosen": -0.
|
397 |
-
"rewards/margins": -0.
|
398 |
-
"rewards/rejected": 0.
|
399 |
"step": 250
|
400 |
},
|
401 |
{
|
402 |
-
"epoch": 0.
|
403 |
-
"grad_norm":
|
404 |
"learning_rate": 5.808909982763825e-10,
|
405 |
-
"logits/chosen": -1.
|
406 |
-
"logits/rejected": -1.
|
407 |
-
"logps/chosen": -0.
|
408 |
-
"logps/rejected": -0.
|
409 |
-
"loss": 1.
|
410 |
-
"rewards/accuracies": 0.
|
411 |
-
"rewards/chosen": 0.
|
412 |
-
"rewards/margins": 0.
|
413 |
-
"rewards/rejected":
|
414 |
"step": 260
|
415 |
},
|
416 |
{
|
417 |
-
"epoch": 0.
|
418 |
-
"grad_norm":
|
419 |
"learning_rate": 5.473237747567806e-10,
|
420 |
-
"logits/chosen": -1.
|
421 |
-
"logits/rejected": -1.
|
422 |
-
"logps/chosen": -0.
|
423 |
-
"logps/rejected": -0.
|
424 |
-
"loss": 1.
|
425 |
-
"rewards/accuracies": 0.
|
426 |
-
"rewards/chosen": -0.
|
427 |
-
"rewards/margins": -0.
|
428 |
-
"rewards/rejected": 0.
|
429 |
"step": 270
|
430 |
},
|
431 |
{
|
432 |
-
"epoch": 0.
|
433 |
-
"grad_norm":
|
434 |
"learning_rate": 5.135396923380673e-10,
|
435 |
-
"logits/chosen": -1.
|
436 |
-
"logits/rejected": -1.
|
437 |
-
"logps/chosen": -0.
|
438 |
-
"logps/rejected": -0.
|
439 |
-
"loss": 1.
|
440 |
"rewards/accuracies": 0.4749999940395355,
|
441 |
-
"rewards/chosen": 0.
|
442 |
-
"rewards/margins":
|
443 |
-
"rewards/rejected": 0.
|
444 |
"step": 280
|
445 |
},
|
446 |
{
|
447 |
-
"epoch": 0.
|
448 |
-
"grad_norm":
|
449 |
"learning_rate": 4.796935649368935e-10,
|
450 |
-
"logits/chosen": -1.
|
451 |
-
"logits/rejected": -1.
|
452 |
-
"logps/chosen": -1.
|
453 |
-
"logps/rejected": -0.
|
454 |
-
"loss": 1.
|
455 |
-
"rewards/accuracies": 0.
|
456 |
-
"rewards/chosen": 0.
|
457 |
-
"rewards/margins": 0.
|
458 |
-
"rewards/rejected": -0.
|
459 |
"step": 290
|
460 |
},
|
461 |
{
|
462 |
-
"epoch": 0.
|
463 |
-
"grad_norm":
|
464 |
"learning_rate": 4.4594049078802925e-10,
|
465 |
-
"logits/chosen": -1.
|
466 |
-
"logits/rejected": -1.
|
467 |
-
"logps/chosen": -0.
|
468 |
-
"logps/rejected": -0.
|
469 |
-
"loss": 1.
|
470 |
-
"rewards/accuracies": 0.
|
471 |
-
"rewards/chosen": 0.
|
472 |
-
"rewards/margins": 0.
|
473 |
-
"rewards/rejected": 0.
|
474 |
"step": 300
|
475 |
},
|
476 |
{
|
477 |
-
"epoch": 0.
|
478 |
-
"grad_norm":
|
479 |
"learning_rate": 4.1243514171423466e-10,
|
480 |
-
"logits/chosen": -1.
|
481 |
-
"logits/rejected": -1.
|
482 |
-
"logps/chosen": -0.
|
483 |
-
"logps/rejected": -0.
|
484 |
-
"loss": 1.
|
485 |
-
"rewards/accuracies": 0.
|
486 |
-
"rewards/chosen":
|
487 |
-
"rewards/margins": -0.
|
488 |
-
"rewards/rejected": 0.
|
489 |
"step": 310
|
490 |
},
|
491 |
{
|
492 |
-
"epoch": 0.
|
493 |
-
"grad_norm":
|
494 |
"learning_rate": 3.793310543501473e-10,
|
495 |
-
"logits/chosen": -1.
|
496 |
-
"logits/rejected": -1.
|
497 |
-
"logps/chosen": -0.
|
498 |
-
"logps/rejected": -0.
|
499 |
-
"loss": 1.
|
500 |
-
"rewards/accuracies": 0.
|
501 |
-
"rewards/chosen":
|
502 |
-
"rewards/margins": 0.
|
503 |
-
"rewards/rejected":
|
504 |
"step": 320
|
505 |
},
|
506 |
{
|
507 |
-
"epoch": 0.
|
508 |
-
"grad_norm":
|
509 |
"learning_rate": 3.4677992656811053e-10,
|
510 |
-
"logits/chosen": -1.
|
511 |
-
"logits/rejected": -1.
|
512 |
-
"logps/chosen": -1.
|
513 |
-
"logps/rejected": -0.
|
514 |
-
"loss": 1.
|
515 |
-
"rewards/accuracies": 0.
|
516 |
-
"rewards/chosen": -0.
|
517 |
-
"rewards/margins": 0.
|
518 |
-
"rewards/rejected": -0.
|
519 |
"step": 330
|
520 |
},
|
521 |
{
|
522 |
-
"epoch": 0.
|
523 |
-
"grad_norm":
|
524 |
"learning_rate": 3.149309223300428e-10,
|
525 |
-
"logits/chosen": -1.
|
526 |
-
"logits/rejected": -1.
|
527 |
-
"logps/chosen": -1.
|
528 |
-
"logps/rejected": -0.
|
529 |
-
"loss": 1.
|
530 |
-
"rewards/accuracies": 0.
|
531 |
-
"rewards/chosen": 0.
|
532 |
-
"rewards/margins": 0.
|
533 |
-
"rewards/rejected": 0.
|
534 |
"step": 340
|
535 |
},
|
536 |
{
|
537 |
-
"epoch": 0.
|
538 |
-
"grad_norm":
|
539 |
"learning_rate": 2.8392998815082717e-10,
|
540 |
-
"logits/chosen": -1.
|
541 |
-
"logits/rejected": -1.
|
542 |
-
"logps/chosen": -1.
|
543 |
-
"logps/rejected": -1.
|
544 |
-
"loss": 1.
|
545 |
-
"rewards/accuracies": 0.
|
546 |
-
"rewards/chosen": 0.
|
547 |
-
"rewards/margins": 0.
|
548 |
-
"rewards/rejected": 0.
|
549 |
"step": 350
|
550 |
},
|
551 |
{
|
552 |
-
"epoch": 0.
|
553 |
-
"grad_norm":
|
554 |
"learning_rate": 2.5391918430549634e-10,
|
555 |
-
"logits/chosen": -1.
|
556 |
-
"logits/rejected": -1.
|
557 |
-
"logps/chosen": -1.
|
558 |
-
"logps/rejected": -0.
|
559 |
-
"loss": 1.
|
560 |
-
"rewards/accuracies": 0.
|
561 |
-
"rewards/chosen": 0.
|
562 |
-
"rewards/margins": 0.
|
563 |
-
"rewards/rejected": -0.
|
564 |
"step": 360
|
565 |
},
|
566 |
{
|
567 |
-
"epoch": 0.
|
568 |
-
"grad_norm":
|
569 |
"learning_rate": 2.250360338449226e-10,
|
570 |
-
"logits/chosen": -1.
|
571 |
-
"logits/rejected": -1.
|
572 |
-
"logps/chosen": -0.
|
573 |
-
"logps/rejected": -0.
|
574 |
-
"loss": 1.
|
575 |
-
"rewards/accuracies": 0.
|
576 |
-
"rewards/chosen": 0.
|
577 |
-
"rewards/margins": 0.
|
578 |
-
"rewards/rejected": -0.
|
579 |
"step": 370
|
580 |
},
|
581 |
{
|
582 |
-
"epoch": 0.
|
583 |
-
"grad_norm":
|
584 |
"learning_rate": 1.9741289240311756e-10,
|
585 |
-
"logits/chosen": -1.
|
586 |
-
"logits/rejected": -1.
|
587 |
-
"logps/chosen": -0.
|
588 |
-
"logps/rejected": -0.
|
589 |
-
"loss": 1.
|
590 |
-
"rewards/accuracies": 0.
|
591 |
-
"rewards/chosen": 0.
|
592 |
-
"rewards/margins": 0.
|
593 |
-
"rewards/rejected":
|
594 |
"step": 380
|
595 |
},
|
596 |
{
|
597 |
-
"epoch": 0.
|
598 |
-
"grad_norm":
|
599 |
"learning_rate": 1.7117634168396773e-10,
|
600 |
-
"logits/chosen": -1.
|
601 |
-
"logits/rejected": -1.
|
602 |
-
"logps/chosen": -1.
|
603 |
-
"logps/rejected": -0.
|
604 |
-
"loss": 1.
|
605 |
-
"rewards/accuracies": 0.
|
606 |
-
"rewards/chosen": 0.
|
607 |
-
"rewards/margins": 0.
|
608 |
-
"rewards/rejected": 0.
|
609 |
"step": 390
|
610 |
},
|
611 |
{
|
612 |
-
"epoch": 0.
|
613 |
-
"grad_norm":
|
614 |
"learning_rate": 1.4644660940672628e-10,
|
615 |
-
"logits/chosen": -1.
|
616 |
-
"logits/rejected": -1.
|
617 |
-
"logps/chosen": -0.
|
618 |
-
"logps/rejected": -0.
|
619 |
-
"loss": 1.
|
620 |
-
"rewards/accuracies": 0.
|
621 |
-
"rewards/chosen": 0.
|
622 |
-
"rewards/margins": -0.
|
623 |
-
"rewards/rejected": 0.
|
624 |
"step": 400
|
625 |
},
|
626 |
{
|
627 |
-
"epoch": 0.
|
628 |
-
"grad_norm":
|
629 |
"learning_rate": 1.2333701836832813e-10,
|
630 |
-
"logits/chosen": -1.
|
631 |
-
"logits/rejected": -1.
|
632 |
-
"logps/chosen": -0.
|
633 |
-
"logps/rejected": -0.
|
634 |
-
"loss": 1.
|
635 |
-
"rewards/accuracies": 0.
|
636 |
-
"rewards/chosen": 0.
|
637 |
-
"rewards/margins": 0.
|
638 |
-
"rewards/rejected": -0.
|
639 |
"step": 410
|
640 |
},
|
641 |
{
|
642 |
-
"epoch": 0.
|
643 |
-
"grad_norm":
|
644 |
"learning_rate": 1.0195346714717813e-10,
|
645 |
-
"logits/chosen": -1.
|
646 |
-
"logits/rejected": -1.
|
647 |
-
"logps/chosen": -0.
|
648 |
-
"logps/rejected": -0.
|
649 |
-
"loss": 1.
|
650 |
-
"rewards/accuracies": 0.
|
651 |
-
"rewards/chosen": -0.
|
652 |
-
"rewards/margins": -0.
|
653 |
-
"rewards/rejected": 0.
|
654 |
"step": 420
|
655 |
},
|
656 |
{
|
657 |
-
"epoch": 0.
|
658 |
-
"grad_norm":
|
659 |
"learning_rate": 8.239394482805996e-11,
|
660 |
-
"logits/chosen": -1.
|
661 |
-
"logits/rejected": -1.
|
662 |
-
"logps/chosen": -1.
|
663 |
-
"logps/rejected": -0.
|
664 |
-
"loss": 1.
|
665 |
-
"rewards/accuracies": 0.
|
666 |
-
"rewards/chosen": 0.
|
667 |
-
"rewards/margins": 0.
|
668 |
-
"rewards/rejected": -0.
|
669 |
"step": 430
|
670 |
},
|
671 |
{
|
672 |
-
"epoch": 0.
|
673 |
-
"grad_norm":
|
674 |
"learning_rate": 6.474808197191401e-11,
|
675 |
-
"logits/chosen": -1.
|
676 |
-
"logits/rejected": -1.
|
677 |
-
"logps/chosen": -1.
|
678 |
-
"logps/rejected": -0.
|
679 |
-
"loss": 1.
|
680 |
-
"rewards/accuracies": 0.
|
681 |
-
"rewards/chosen": 0.
|
682 |
-
"rewards/margins": 0.
|
683 |
-
"rewards/rejected": -0.
|
684 |
"step": 440
|
685 |
},
|
686 |
{
|
687 |
-
"epoch": 0.
|
688 |
-
"grad_norm":
|
689 |
"learning_rate": 4.9096739888146e-11,
|
690 |
-
"logits/chosen": -1.
|
691 |
-
"logits/rejected": -1.
|
692 |
-
"logps/chosen": -1.
|
693 |
-
"logps/rejected": -0.
|
694 |
-
"loss": 1.
|
695 |
-
"rewards/accuracies": 0.
|
696 |
-
"rewards/chosen": 0.
|
697 |
-
"rewards/margins": 0.
|
698 |
-
"rewards/rejected": 0.
|
699 |
"step": 450
|
700 |
},
|
701 |
{
|
702 |
-
"epoch": 0.
|
703 |
-
"grad_norm":
|
704 |
"learning_rate": 3.5511640091604293e-11,
|
705 |
-
"logits/chosen": -1.
|
706 |
-
"logits/rejected": -1.
|
707 |
-
"logps/chosen": -1.
|
708 |
-
"logps/rejected": -0.
|
709 |
-
"loss": 1.
|
710 |
-
"rewards/accuracies": 0.
|
711 |
-
"rewards/chosen":
|
712 |
-
"rewards/margins": 0.
|
713 |
-
"rewards/rejected": -0.
|
714 |
"step": 460
|
715 |
},
|
716 |
{
|
717 |
-
"epoch": 0.
|
718 |
-
"grad_norm":
|
719 |
"learning_rate": 2.4055035642222225e-11,
|
720 |
-
"logits/chosen": -1.
|
721 |
-
"logits/rejected": -1.
|
722 |
-
"logps/chosen": -0.
|
723 |
-
"logps/rejected": -0.
|
724 |
-
"loss": 1.
|
725 |
-
"rewards/accuracies": 0.
|
726 |
-
"rewards/chosen": 0.
|
727 |
-
"rewards/margins":
|
728 |
-
"rewards/rejected": 0.
|
729 |
"step": 470
|
730 |
},
|
731 |
{
|
732 |
-
"epoch": 0.
|
733 |
-
"grad_norm":
|
734 |
"learning_rate": 1.477942587339426e-11,
|
735 |
-
"logits/chosen": -1.
|
736 |
-
"logits/rejected": -1.
|
737 |
-
"logps/chosen": -0.
|
738 |
-
"logps/rejected": -0.
|
739 |
-
"loss": 1.
|
740 |
-
"rewards/accuracies": 0.
|
741 |
-
"rewards/chosen": -0.
|
742 |
-
"rewards/margins": -0.
|
743 |
-
"rewards/rejected": 0.
|
744 |
"step": 480
|
745 |
},
|
746 |
{
|
747 |
-
"epoch": 0.
|
748 |
-
"grad_norm":
|
749 |
"learning_rate": 7.727315816331515e-12,
|
750 |
-
"logits/chosen": -1.
|
751 |
-
"logits/rejected": -1.
|
752 |
-
"logps/chosen": -1.
|
753 |
-
"logps/rejected": -1.
|
754 |
-
"loss": 1.
|
755 |
-
"rewards/accuracies": 0.
|
756 |
-
"rewards/chosen": 0.
|
757 |
-
"rewards/margins": 0.
|
758 |
-
"rewards/rejected": -0.
|
759 |
"step": 490
|
760 |
},
|
761 |
{
|
762 |
-
"epoch": 0.
|
763 |
-
"grad_norm":
|
764 |
"learning_rate": 2.9310214228202016e-12,
|
765 |
-
"logits/chosen": -1.
|
766 |
-
"logits/rejected": -1.
|
767 |
-
"logps/chosen": -0.
|
768 |
-
"logps/rejected": -0.
|
769 |
-
"loss": 1.
|
770 |
-
"rewards/accuracies": 0.
|
771 |
-
"rewards/chosen": -0.
|
772 |
-
"rewards/margins": -0.
|
773 |
-
"rewards/rejected": 0.
|
774 |
"step": 500
|
775 |
},
|
776 |
{
|
777 |
-
"epoch": 0.
|
778 |
-
"grad_norm":
|
779 |
"learning_rate": 4.125214789427734e-13,
|
780 |
-
"logits/chosen": -1.
|
781 |
-
"logits/rejected": -1.
|
782 |
-
"logps/chosen": -0.
|
783 |
-
"logps/rejected": -0.
|
784 |
-
"loss": 1.
|
785 |
-
"rewards/accuracies": 0.
|
786 |
-
"rewards/chosen": 0.
|
787 |
-
"rewards/margins": 0.
|
788 |
-
"rewards/rejected": 0.
|
789 |
"step": 510
|
790 |
},
|
791 |
{
|
792 |
-
"epoch":
|
793 |
"step": 516,
|
794 |
"total_flos": 0.0,
|
795 |
-
"train_loss": 1.
|
796 |
-
"train_runtime":
|
797 |
-
"train_samples_per_second":
|
798 |
-
"train_steps_per_second": 0.
|
799 |
}
|
800 |
],
|
801 |
"logging_steps": 10,
|
@@ -803,6 +803,18 @@
|
|
803 |
"num_input_tokens_seen": 0,
|
804 |
"num_train_epochs": 1,
|
805 |
"save_steps": 100,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
806 |
"total_flos": 0.0,
|
807 |
"train_batch_size": 4,
|
808 |
"trial_name": null,
|
|
|
9 |
"is_world_process_zero": true,
|
10 |
"log_history": [
|
11 |
{
|
12 |
+
"epoch": 0.001936108422071636,
|
13 |
+
"grad_norm": 4289.771574316242,
|
14 |
"learning_rate": 1.9230769230769234e-11,
|
15 |
"logits/chosen": -1.8683955669403076,
|
16 |
"logits/rejected": -1.7658718824386597,
|
|
|
24 |
"step": 1
|
25 |
},
|
26 |
{
|
27 |
+
"epoch": 0.01936108422071636,
|
28 |
+
"grad_norm": 4190.035270648744,
|
29 |
"learning_rate": 1.9230769230769234e-10,
|
30 |
+
"logits/chosen": -1.6616647243499756,
|
31 |
+
"logits/rejected": -1.6193790435791016,
|
32 |
+
"logps/chosen": -0.9486603140830994,
|
33 |
+
"logps/rejected": -0.9298955202102661,
|
34 |
+
"loss": 1.2029,
|
35 |
+
"rewards/accuracies": 0.3333333432674408,
|
36 |
+
"rewards/chosen": -0.0510055311024189,
|
37 |
+
"rewards/margins": -0.1265178769826889,
|
38 |
+
"rewards/rejected": 0.0755123570561409,
|
39 |
"step": 10
|
40 |
},
|
41 |
{
|
42 |
+
"epoch": 0.03872216844143272,
|
43 |
+
"grad_norm": 3083.054373722184,
|
44 |
"learning_rate": 3.8461538461538467e-10,
|
45 |
+
"logits/chosen": -1.5834848880767822,
|
46 |
+
"logits/rejected": -1.5355803966522217,
|
47 |
+
"logps/chosen": -1.0245015621185303,
|
48 |
+
"logps/rejected": -0.9704240560531616,
|
49 |
+
"loss": 1.2623,
|
50 |
+
"rewards/accuracies": 0.5687500238418579,
|
51 |
+
"rewards/chosen": 0.0036691308487206697,
|
52 |
+
"rewards/margins": 0.06183544546365738,
|
53 |
+
"rewards/rejected": -0.05816630274057388,
|
54 |
"step": 20
|
55 |
},
|
56 |
{
|
57 |
+
"epoch": 0.05808325266214908,
|
58 |
+
"grad_norm": 4908.628774557542,
|
59 |
"learning_rate": 5.769230769230769e-10,
|
60 |
+
"logits/chosen": -1.5294702053070068,
|
61 |
+
"logits/rejected": -1.4707310199737549,
|
62 |
+
"logps/chosen": -1.0035126209259033,
|
63 |
+
"logps/rejected": -0.9810468554496765,
|
64 |
+
"loss": 1.2911,
|
65 |
+
"rewards/accuracies": 0.512499988079071,
|
66 |
+
"rewards/chosen": -0.10282345861196518,
|
67 |
+
"rewards/margins": -0.00801654439419508,
|
68 |
+
"rewards/rejected": -0.09480690211057663,
|
69 |
"step": 30
|
70 |
},
|
71 |
{
|
72 |
+
"epoch": 0.07744433688286544,
|
73 |
+
"grad_norm": 3634.399853197314,
|
74 |
"learning_rate": 7.692307692307693e-10,
|
75 |
+
"logits/chosen": -1.5545880794525146,
|
76 |
+
"logits/rejected": -1.4971392154693604,
|
77 |
+
"logps/chosen": -1.0002901554107666,
|
78 |
+
"logps/rejected": -0.9374505281448364,
|
79 |
+
"loss": 1.2544,
|
80 |
+
"rewards/accuracies": 0.4375,
|
81 |
+
"rewards/chosen": -0.056221622973680496,
|
82 |
+
"rewards/margins": -0.12876693904399872,
|
83 |
+
"rewards/rejected": 0.07254532724618912,
|
84 |
"step": 40
|
85 |
},
|
86 |
{
|
87 |
+
"epoch": 0.0968054211035818,
|
88 |
+
"grad_norm": 3743.017714256134,
|
89 |
"learning_rate": 9.615384615384616e-10,
|
90 |
+
"logits/chosen": -1.6496717929840088,
|
91 |
+
"logits/rejected": -1.6047435998916626,
|
92 |
+
"logps/chosen": -0.990554928779602,
|
93 |
+
"logps/rejected": -0.968321681022644,
|
94 |
+
"loss": 1.2784,
|
95 |
+
"rewards/accuracies": 0.512499988079071,
|
96 |
+
"rewards/chosen": -0.044189296662807465,
|
97 |
+
"rewards/margins": -0.09925868362188339,
|
98 |
+
"rewards/rejected": 0.05506938695907593,
|
99 |
"step": 50
|
100 |
},
|
101 |
{
|
102 |
+
"epoch": 0.11616650532429816,
|
103 |
+
"grad_norm": 3989.4789916346926,
|
104 |
"learning_rate": 9.99266706925562e-10,
|
105 |
+
"logits/chosen": -1.6014108657836914,
|
106 |
+
"logits/rejected": -1.5485652685165405,
|
107 |
+
"logps/chosen": -0.9952412843704224,
|
108 |
+
"logps/rejected": -0.9312202334403992,
|
109 |
+
"loss": 1.2834,
|
110 |
+
"rewards/accuracies": 0.53125,
|
111 |
+
"rewards/chosen": 0.05288747698068619,
|
112 |
+
"rewards/margins": 0.01693376898765564,
|
113 |
+
"rewards/rejected": 0.03595370799303055,
|
114 |
"step": 60
|
115 |
},
|
116 |
{
|
117 |
+
"epoch": 0.1355275895450145,
|
118 |
+
"grad_norm": 4801.740697750488,
|
119 |
"learning_rate": 9.96291389741603e-10,
|
120 |
+
"logits/chosen": -1.5917268991470337,
|
121 |
+
"logits/rejected": -1.505897045135498,
|
122 |
+
"logps/chosen": -0.9961442947387695,
|
123 |
+
"logps/rejected": -0.946877658367157,
|
124 |
+
"loss": 1.2574,
|
125 |
+
"rewards/accuracies": 0.550000011920929,
|
126 |
+
"rewards/chosen": -0.024737417697906494,
|
127 |
+
"rewards/margins": 0.13343551754951477,
|
128 |
+
"rewards/rejected": -0.15817293524742126,
|
129 |
"step": 70
|
130 |
},
|
131 |
{
|
132 |
+
"epoch": 0.15488867376573087,
|
133 |
+
"grad_norm": 4317.0356676312595,
|
134 |
"learning_rate": 9.91041841371078e-10,
|
135 |
+
"logits/chosen": -1.5460880994796753,
|
136 |
+
"logits/rejected": -1.5285828113555908,
|
137 |
+
"logps/chosen": -1.0357868671417236,
|
138 |
+
"logps/rejected": -0.985648512840271,
|
139 |
+
"loss": 1.2938,
|
140 |
+
"rewards/accuracies": 0.512499988079071,
|
141 |
+
"rewards/chosen": 0.03525885194540024,
|
142 |
+
"rewards/margins": 0.09618574380874634,
|
143 |
+
"rewards/rejected": -0.0609268844127655,
|
144 |
"step": 80
|
145 |
},
|
146 |
{
|
147 |
+
"epoch": 0.17424975798644723,
|
148 |
+
"grad_norm": 4079.0129090665787,
|
149 |
"learning_rate": 9.835421176144035e-10,
|
150 |
+
"logits/chosen": -1.678056001663208,
|
151 |
+
"logits/rejected": -1.6193243265151978,
|
152 |
+
"logps/chosen": -1.0110713243484497,
|
153 |
+
"logps/rejected": -0.9249661564826965,
|
154 |
+
"loss": 1.2586,
|
155 |
+
"rewards/accuracies": 0.512499988079071,
|
156 |
+
"rewards/chosen": 0.03245333582162857,
|
157 |
+
"rewards/margins": 0.027137070894241333,
|
158 |
+
"rewards/rejected": 0.005316261202096939,
|
159 |
"step": 90
|
160 |
},
|
161 |
{
|
162 |
+
"epoch": 0.1936108422071636,
|
163 |
+
"grad_norm": 4056.2821430675544,
|
164 |
"learning_rate": 9.738265855914014e-10,
|
165 |
+
"logits/chosen": -1.6319992542266846,
|
166 |
+
"logits/rejected": -1.5686506032943726,
|
167 |
+
"logps/chosen": -0.978125274181366,
|
168 |
+
"logps/rejected": -0.9383082389831543,
|
169 |
+
"loss": 1.2723,
|
170 |
+
"rewards/accuracies": 0.42500001192092896,
|
171 |
+
"rewards/chosen": 0.002950614783912897,
|
172 |
+
"rewards/margins": -0.10372404754161835,
|
173 |
+
"rewards/rejected": 0.10667465627193451,
|
174 |
"step": 100
|
175 |
},
|
176 |
{
|
177 |
+
"epoch": 0.21297192642787996,
|
178 |
+
"grad_norm": 5228.540901226267,
|
179 |
"learning_rate": 9.619397662556434e-10,
|
180 |
+
"logits/chosen": -1.6645421981811523,
|
181 |
+
"logits/rejected": -1.588181495666504,
|
182 |
+
"logps/chosen": -0.8982473611831665,
|
183 |
+
"logps/rejected": -0.8726890683174133,
|
184 |
+
"loss": 1.2749,
|
185 |
+
"rewards/accuracies": 0.5062500238418579,
|
186 |
+
"rewards/chosen": 0.06572394073009491,
|
187 |
+
"rewards/margins": -0.014699941501021385,
|
188 |
+
"rewards/rejected": 0.08042389899492264,
|
189 |
"step": 110
|
190 |
},
|
191 |
{
|
192 |
+
"epoch": 0.23233301064859632,
|
193 |
+
"grad_norm": 3837.5381027526573,
|
194 |
"learning_rate": 9.47936130379344e-10,
|
195 |
+
"logits/chosen": -1.5425523519515991,
|
196 |
+
"logits/rejected": -1.5234339237213135,
|
197 |
+
"logps/chosen": -0.9705616235733032,
|
198 |
+
"logps/rejected": -0.9401592016220093,
|
199 |
+
"loss": 1.2288,
|
200 |
+
"rewards/accuracies": 0.543749988079071,
|
201 |
+
"rewards/chosen": 0.0972176343202591,
|
202 |
+
"rewards/margins": 0.09945462644100189,
|
203 |
+
"rewards/rejected": -0.0022369951475411654,
|
204 |
"step": 120
|
205 |
},
|
206 |
{
|
207 |
+
"epoch": 0.25169409486931266,
|
208 |
+
"grad_norm": 3814.5542415994746,
|
209 |
"learning_rate": 9.318798489436919e-10,
|
210 |
+
"logits/chosen": -1.579369306564331,
|
211 |
+
"logits/rejected": -1.4889588356018066,
|
212 |
+
"logps/chosen": -0.9654415845870972,
|
213 |
+
"logps/rejected": -0.9286600351333618,
|
214 |
+
"loss": 1.2977,
|
215 |
+
"rewards/accuracies": 0.53125,
|
216 |
+
"rewards/chosen": 0.14246916770935059,
|
217 |
+
"rewards/margins": 0.08274303376674652,
|
218 |
+
"rewards/rejected": 0.05972614139318466,
|
219 |
"step": 130
|
220 |
},
|
221 |
{
|
222 |
+
"epoch": 0.271055179090029,
|
223 |
+
"grad_norm": 3751.478373618925,
|
224 |
"learning_rate": 9.138444990784454e-10,
|
225 |
+
"logits/chosen": -1.5760021209716797,
|
226 |
+
"logits/rejected": -1.5278505086898804,
|
227 |
+
"logps/chosen": -0.9930068850517273,
|
228 |
+
"logps/rejected": -0.9939811825752258,
|
229 |
+
"loss": 1.2583,
|
230 |
+
"rewards/accuracies": 0.4625000059604645,
|
231 |
+
"rewards/chosen": -0.029570287093520164,
|
232 |
+
"rewards/margins": -0.044873449951410294,
|
233 |
+
"rewards/rejected": 0.015303166583180428,
|
234 |
"step": 140
|
235 |
},
|
236 |
{
|
237 |
+
"epoch": 0.2904162633107454,
|
238 |
+
"grad_norm": 3907.5968331867316,
|
239 |
"learning_rate": 8.939127268983109e-10,
|
240 |
+
"logits/chosen": -1.5609729290008545,
|
241 |
+
"logits/rejected": -1.5375111103057861,
|
242 |
+
"logps/chosen": -1.0732953548431396,
|
243 |
+
"logps/rejected": -0.9959889650344849,
|
244 |
+
"loss": 1.2631,
|
245 |
+
"rewards/accuracies": 0.53125,
|
246 |
+
"rewards/chosen": 0.12664008140563965,
|
247 |
+
"rewards/margins": 0.0977608785033226,
|
248 |
+
"rewards/rejected": 0.02887919172644615,
|
249 |
"step": 150
|
250 |
},
|
251 |
{
|
252 |
+
"epoch": 0.30977734753146174,
|
253 |
+
"grad_norm": 4306.071797965924,
|
254 |
"learning_rate": 8.721758687811352e-10,
|
255 |
+
"logits/chosen": -1.6766704320907593,
|
256 |
+
"logits/rejected": -1.6015819311141968,
|
257 |
+
"logps/chosen": -0.9719392657279968,
|
258 |
+
"logps/rejected": -0.9425565600395203,
|
259 |
+
"loss": 1.3024,
|
260 |
+
"rewards/accuracies": 0.44999998807907104,
|
261 |
+
"rewards/chosen": -0.02740299701690674,
|
262 |
+
"rewards/margins": -0.19174733757972717,
|
263 |
+
"rewards/rejected": 0.16434435546398163,
|
264 |
"step": 160
|
265 |
},
|
266 |
{
|
267 |
+
"epoch": 0.3291384317521781,
|
268 |
+
"grad_norm": 4085.902013165705,
|
269 |
"learning_rate": 8.487335328233912e-10,
|
270 |
+
"logits/chosen": -1.5358489751815796,
|
271 |
+
"logits/rejected": -1.4411264657974243,
|
272 |
+
"logps/chosen": -0.9960495829582214,
|
273 |
+
"logps/rejected": -0.9736671447753906,
|
274 |
+
"loss": 1.2834,
|
275 |
+
"rewards/accuracies": 0.5,
|
276 |
+
"rewards/chosen": -0.07462203502655029,
|
277 |
+
"rewards/margins": -0.08827298879623413,
|
278 |
+
"rewards/rejected": 0.013650953769683838,
|
279 |
"step": 170
|
280 |
},
|
281 |
{
|
282 |
+
"epoch": 0.34849951597289447,
|
283 |
+
"grad_norm": 4830.939949744352,
|
284 |
"learning_rate": 8.236931423909139e-10,
|
285 |
+
"logits/chosen": -1.6721004247665405,
|
286 |
+
"logits/rejected": -1.5736545324325562,
|
287 |
+
"logps/chosen": -0.9753511548042297,
|
288 |
+
"logps/rejected": -0.9517275094985962,
|
289 |
+
"loss": 1.2611,
|
290 |
+
"rewards/accuracies": 0.5562499761581421,
|
291 |
+
"rewards/chosen": 0.11362428963184357,
|
292 |
+
"rewards/margins": 0.17257475852966309,
|
293 |
+
"rewards/rejected": -0.05895046144723892,
|
294 |
"step": 180
|
295 |
},
|
296 |
{
|
297 |
+
"epoch": 0.36786060019361083,
|
298 |
+
"grad_norm": 4117.362535725707,
|
299 |
"learning_rate": 7.971694438565449e-10,
|
300 |
+
"logits/chosen": -1.6258817911148071,
|
301 |
+
"logits/rejected": -1.563320517539978,
|
302 |
+
"logps/chosen": -0.9729933738708496,
|
303 |
+
"logps/rejected": -0.9748104810714722,
|
304 |
+
"loss": 1.2509,
|
305 |
+
"rewards/accuracies": 0.5562499761581421,
|
306 |
+
"rewards/chosen": 0.025180306285619736,
|
307 |
+
"rewards/margins": 0.146215558052063,
|
308 |
+
"rewards/rejected": -0.12103524059057236,
|
309 |
"step": 190
|
310 |
},
|
311 |
{
|
312 |
+
"epoch": 0.3872216844143272,
|
313 |
+
"grad_norm": 4321.444634590991,
|
314 |
"learning_rate": 7.692839807804521e-10,
|
315 |
+
"logits/chosen": -1.6311115026474,
|
316 |
+
"logits/rejected": -1.5964945554733276,
|
317 |
+
"logps/chosen": -0.9733870625495911,
|
318 |
+
"logps/rejected": -0.9268864393234253,
|
319 |
+
"loss": 1.2991,
|
320 |
+
"rewards/accuracies": 0.4937500059604645,
|
321 |
+
"rewards/chosen": -0.10145823657512665,
|
322 |
+
"rewards/margins": -0.11137993633747101,
|
323 |
+
"rewards/rejected": 0.009921704418957233,
|
324 |
"step": 200
|
325 |
},
|
326 |
{
|
327 |
+
"epoch": 0.40658276863504356,
|
328 |
+
"grad_norm": 4028.0205254352395,
|
329 |
"learning_rate": 7.401645369426697e-10,
|
330 |
+
"logits/chosen": -1.6063648462295532,
|
331 |
+
"logits/rejected": -1.5398415327072144,
|
332 |
+
"logps/chosen": -0.9638331532478333,
|
333 |
+
"logps/rejected": -0.9147375822067261,
|
334 |
+
"loss": 1.2499,
|
335 |
+
"rewards/accuracies": 0.512499988079071,
|
336 |
+
"rewards/chosen": -0.012107854709029198,
|
337 |
+
"rewards/margins": -0.00303336838260293,
|
338 |
+
"rewards/rejected": -0.009074489586055279,
|
339 |
"step": 210
|
340 |
},
|
341 |
{
|
342 |
+
"epoch": 0.4259438528557599,
|
343 |
+
"grad_norm": 4531.775441431939,
|
344 |
"learning_rate": 7.099445507801324e-10,
|
345 |
+
"logits/chosen": -1.6380093097686768,
|
346 |
+
"logits/rejected": -1.589648962020874,
|
347 |
+
"logps/chosen": -0.999239444732666,
|
348 |
+
"logps/rejected": -0.9409993886947632,
|
349 |
+
"loss": 1.2661,
|
350 |
+
"rewards/accuracies": 0.543749988079071,
|
351 |
+
"rewards/chosen": 0.08351825177669525,
|
352 |
+
"rewards/margins": 0.06452666223049164,
|
353 |
+
"rewards/rejected": 0.018991602584719658,
|
354 |
"step": 220
|
355 |
},
|
356 |
{
|
357 |
+
"epoch": 0.4453049370764763,
|
358 |
+
"grad_norm": 4190.546129288723,
|
359 |
"learning_rate": 6.7876250391152e-10,
|
360 |
+
"logits/chosen": -1.5896263122558594,
|
361 |
+
"logits/rejected": -1.5483357906341553,
|
362 |
+
"logps/chosen": -0.9513187408447266,
|
363 |
+
"logps/rejected": -0.9711192846298218,
|
364 |
+
"loss": 1.2465,
|
365 |
+
"rewards/accuracies": 0.512499988079071,
|
366 |
+
"rewards/chosen": 0.011254754848778248,
|
367 |
+
"rewards/margins": 0.014234659262001514,
|
368 |
+
"rewards/rejected": -0.002979907440021634,
|
369 |
"step": 230
|
370 |
},
|
371 |
{
|
372 |
+
"epoch": 0.46466602129719264,
|
373 |
+
"grad_norm": 4241.882144515242,
|
374 |
"learning_rate": 6.467612865519674e-10,
|
375 |
+
"logits/chosen": -1.6262489557266235,
|
376 |
+
"logits/rejected": -1.6008937358856201,
|
377 |
+
"logps/chosen": -0.9875959157943726,
|
378 |
+
"logps/rejected": -0.9101887941360474,
|
379 |
+
"loss": 1.2863,
|
380 |
+
"rewards/accuracies": 0.48124998807907104,
|
381 |
+
"rewards/chosen": 0.02070781961083412,
|
382 |
+
"rewards/margins": -0.0031043351627886295,
|
383 |
+
"rewards/rejected": 0.023812144994735718,
|
384 |
"step": 240
|
385 |
},
|
386 |
{
|
387 |
+
"epoch": 0.484027105517909,
|
388 |
+
"grad_norm": 3918.096198448725,
|
389 |
"learning_rate": 6.14087542725593e-10,
|
390 |
+
"logits/chosen": -1.6483112573623657,
|
391 |
+
"logits/rejected": -1.6147336959838867,
|
392 |
+
"logps/chosen": -1.020185112953186,
|
393 |
+
"logps/rejected": -0.9349339604377747,
|
394 |
+
"loss": 1.2945,
|
395 |
+
"rewards/accuracies": 0.42500001192092896,
|
396 |
+
"rewards/chosen": -0.05541396886110306,
|
397 |
+
"rewards/margins": -0.13103903830051422,
|
398 |
+
"rewards/rejected": 0.07562507688999176,
|
399 |
"step": 250
|
400 |
},
|
401 |
{
|
402 |
+
"epoch": 0.5033881897386253,
|
403 |
+
"grad_norm": 3626.964808344265,
|
404 |
"learning_rate": 5.808909982763825e-10,
|
405 |
+
"logits/chosen": -1.6457252502441406,
|
406 |
+
"logits/rejected": -1.5552377700805664,
|
407 |
+
"logps/chosen": -0.9940276145935059,
|
408 |
+
"logps/rejected": -0.9484678506851196,
|
409 |
+
"loss": 1.274,
|
410 |
+
"rewards/accuracies": 0.45625001192092896,
|
411 |
+
"rewards/chosen": -0.017910266295075417,
|
412 |
+
"rewards/margins": -0.03398443013429642,
|
413 |
+
"rewards/rejected": 0.016074160113930702,
|
414 |
"step": 260
|
415 |
},
|
416 |
{
|
417 |
+
"epoch": 0.5227492739593417,
|
418 |
+
"grad_norm": 3993.5474645065597,
|
419 |
"learning_rate": 5.473237747567806e-10,
|
420 |
+
"logits/chosen": -1.6354029178619385,
|
421 |
+
"logits/rejected": -1.5715049505233765,
|
422 |
+
"logps/chosen": -0.9696807861328125,
|
423 |
+
"logps/rejected": -0.9575467109680176,
|
424 |
+
"loss": 1.2403,
|
425 |
+
"rewards/accuracies": 0.5249999761581421,
|
426 |
+
"rewards/chosen": -0.052422285079956055,
|
427 |
+
"rewards/margins": -0.0962902158498764,
|
428 |
+
"rewards/rejected": 0.04386794939637184,
|
429 |
"step": 270
|
430 |
},
|
431 |
{
|
432 |
+
"epoch": 0.542110358180058,
|
433 |
+
"grad_norm": 4125.713570850562,
|
434 |
"learning_rate": 5.135396923380673e-10,
|
435 |
+
"logits/chosen": -1.5687922239303589,
|
436 |
+
"logits/rejected": -1.495790719985962,
|
437 |
+
"logps/chosen": -0.9840999841690063,
|
438 |
+
"logps/rejected": -0.9467118978500366,
|
439 |
+
"loss": 1.274,
|
440 |
"rewards/accuracies": 0.4749999940395355,
|
441 |
+
"rewards/chosen": 0.07109468430280685,
|
442 |
+
"rewards/margins": 0.008125528693199158,
|
443 |
+
"rewards/rejected": 0.0629691407084465,
|
444 |
"step": 280
|
445 |
},
|
446 |
{
|
447 |
+
"epoch": 0.5614714424007744,
|
448 |
+
"grad_norm": 4172.164367513521,
|
449 |
"learning_rate": 4.796935649368935e-10,
|
450 |
+
"logits/chosen": -1.5752254724502563,
|
451 |
+
"logits/rejected": -1.4961917400360107,
|
452 |
+
"logps/chosen": -1.047147512435913,
|
453 |
+
"logps/rejected": -0.9933904409408569,
|
454 |
+
"loss": 1.2562,
|
455 |
+
"rewards/accuracies": 0.6000000238418579,
|
456 |
+
"rewards/chosen": 0.0501960925757885,
|
457 |
+
"rewards/margins": 0.15902431309223175,
|
458 |
+
"rewards/rejected": -0.10882820934057236,
|
459 |
"step": 290
|
460 |
},
|
461 |
{
|
462 |
+
"epoch": 0.5808325266214908,
|
463 |
+
"grad_norm": 4075.4312369744443,
|
464 |
"learning_rate": 4.4594049078802925e-10,
|
465 |
+
"logits/chosen": -1.5983613729476929,
|
466 |
+
"logits/rejected": -1.501022458076477,
|
467 |
+
"logps/chosen": -0.9565431475639343,
|
468 |
+
"logps/rejected": -0.9199585914611816,
|
469 |
+
"loss": 1.251,
|
470 |
+
"rewards/accuracies": 0.518750011920929,
|
471 |
+
"rewards/chosen": 0.11743637174367905,
|
472 |
+
"rewards/margins": 0.11259187757968903,
|
473 |
+
"rewards/rejected": 0.004844509996473789,
|
474 |
"step": 300
|
475 |
},
|
476 |
{
|
477 |
+
"epoch": 0.6001936108422071,
|
478 |
+
"grad_norm": 4415.771404967329,
|
479 |
"learning_rate": 4.1243514171423466e-10,
|
480 |
+
"logits/chosen": -1.5818376541137695,
|
481 |
+
"logits/rejected": -1.5412604808807373,
|
482 |
+
"logps/chosen": -0.9797090291976929,
|
483 |
+
"logps/rejected": -0.9499006271362305,
|
484 |
+
"loss": 1.2733,
|
485 |
+
"rewards/accuracies": 0.46875,
|
486 |
+
"rewards/chosen": 0.04855315759778023,
|
487 |
+
"rewards/margins": -0.004827280528843403,
|
488 |
+
"rewards/rejected": 0.053380437195301056,
|
489 |
"step": 310
|
490 |
},
|
491 |
{
|
492 |
+
"epoch": 0.6195546950629235,
|
493 |
+
"grad_norm": 4020.4622713375675,
|
494 |
"learning_rate": 3.793310543501473e-10,
|
495 |
+
"logits/chosen": -1.661425232887268,
|
496 |
+
"logits/rejected": -1.589224100112915,
|
497 |
+
"logps/chosen": -0.9785275459289551,
|
498 |
+
"logps/rejected": -0.9554710388183594,
|
499 |
+
"loss": 1.281,
|
500 |
+
"rewards/accuracies": 0.48124998807907104,
|
501 |
+
"rewards/chosen": 0.017457595095038414,
|
502 |
+
"rewards/margins": -0.04888144135475159,
|
503 |
+
"rewards/rejected": 0.06633903831243515,
|
504 |
"step": 320
|
505 |
},
|
506 |
{
|
507 |
+
"epoch": 0.6389157792836399,
|
508 |
+
"grad_norm": 4011.0955669502655,
|
509 |
"learning_rate": 3.4677992656811053e-10,
|
510 |
+
"logits/chosen": -1.6328061819076538,
|
511 |
+
"logits/rejected": -1.6010538339614868,
|
512 |
+
"logps/chosen": -1.0184727907180786,
|
513 |
+
"logps/rejected": -0.9700371623039246,
|
514 |
+
"loss": 1.2624,
|
515 |
+
"rewards/accuracies": 0.46875,
|
516 |
+
"rewards/chosen": -0.015060502104461193,
|
517 |
+
"rewards/margins": -0.0113009512424469,
|
518 |
+
"rewards/rejected": -0.0037595562171190977,
|
519 |
"step": 330
|
520 |
},
|
521 |
{
|
522 |
+
"epoch": 0.6582768635043562,
|
523 |
+
"grad_norm": 4234.390545263189,
|
524 |
"learning_rate": 3.149309223300428e-10,
|
525 |
+
"logits/chosen": -1.532965064048767,
|
526 |
+
"logits/rejected": -1.5039539337158203,
|
527 |
+
"logps/chosen": -1.0615012645721436,
|
528 |
+
"logps/rejected": -0.975587010383606,
|
529 |
+
"loss": 1.2536,
|
530 |
+
"rewards/accuracies": 0.6187499761581421,
|
531 |
+
"rewards/chosen": 0.23547323048114777,
|
532 |
+
"rewards/margins": 0.2764241397380829,
|
533 |
+
"rewards/rejected": -0.04095090553164482,
|
534 |
"step": 340
|
535 |
},
|
536 |
{
|
537 |
+
"epoch": 0.6776379477250726,
|
538 |
+
"grad_norm": 4310.225886680266,
|
539 |
"learning_rate": 2.8392998815082717e-10,
|
540 |
+
"logits/chosen": -1.6574161052703857,
|
541 |
+
"logits/rejected": -1.5642915964126587,
|
542 |
+
"logps/chosen": -1.0545518398284912,
|
543 |
+
"logps/rejected": -1.055479645729065,
|
544 |
+
"loss": 1.2634,
|
545 |
+
"rewards/accuracies": 0.518750011920929,
|
546 |
+
"rewards/chosen": 0.025425389409065247,
|
547 |
+
"rewards/margins": 0.06826835870742798,
|
548 |
+
"rewards/rejected": -0.04284298047423363,
|
549 |
"step": 350
|
550 |
},
|
551 |
{
|
552 |
+
"epoch": 0.6969990319457889,
|
553 |
+
"grad_norm": 3627.3654619519734,
|
554 |
"learning_rate": 2.5391918430549634e-10,
|
555 |
+
"logits/chosen": -1.6917625665664673,
|
556 |
+
"logits/rejected": -1.6304069757461548,
|
557 |
+
"logps/chosen": -1.0242336988449097,
|
558 |
+
"logps/rejected": -0.9425589442253113,
|
559 |
+
"loss": 1.2358,
|
560 |
+
"rewards/accuracies": 0.550000011920929,
|
561 |
+
"rewards/chosen": 0.0516384020447731,
|
562 |
+
"rewards/margins": 0.07235859334468842,
|
563 |
+
"rewards/rejected": -0.020720209926366806,
|
564 |
"step": 360
|
565 |
},
|
566 |
{
|
567 |
+
"epoch": 0.7163601161665053,
|
568 |
+
"grad_norm": 3729.350909018597,
|
569 |
"learning_rate": 2.250360338449226e-10,
|
570 |
+
"logits/chosen": -1.7181346416473389,
|
571 |
+
"logits/rejected": -1.7049201726913452,
|
572 |
+
"logps/chosen": -0.9749780893325806,
|
573 |
+
"logps/rejected": -0.9269036054611206,
|
574 |
+
"loss": 1.2295,
|
575 |
+
"rewards/accuracies": 0.5249999761581421,
|
576 |
+
"rewards/chosen": -0.060578178614377975,
|
577 |
+
"rewards/margins": 0.004380314145237207,
|
578 |
+
"rewards/rejected": -0.06495849788188934,
|
579 |
"step": 370
|
580 |
},
|
581 |
{
|
582 |
+
"epoch": 0.7357212003872217,
|
583 |
+
"grad_norm": 3635.364845851746,
|
584 |
"learning_rate": 1.9741289240311756e-10,
|
585 |
+
"logits/chosen": -1.6250957250595093,
|
586 |
+
"logits/rejected": -1.5766202211380005,
|
587 |
+
"logps/chosen": -0.9923038482666016,
|
588 |
+
"logps/rejected": -0.95228511095047,
|
589 |
+
"loss": 1.279,
|
590 |
+
"rewards/accuracies": 0.5062500238418579,
|
591 |
+
"rewards/chosen": 0.03353533893823624,
|
592 |
+
"rewards/margins": -0.006679633166640997,
|
593 |
+
"rewards/rejected": 0.04021497443318367,
|
594 |
"step": 380
|
595 |
},
|
596 |
{
|
597 |
+
"epoch": 0.755082284607938,
|
598 |
+
"grad_norm": 3489.267034208794,
|
599 |
"learning_rate": 1.7117634168396773e-10,
|
600 |
+
"logits/chosen": -1.6205106973648071,
|
601 |
+
"logits/rejected": -1.5565264225006104,
|
602 |
+
"logps/chosen": -1.0025845766067505,
|
603 |
+
"logps/rejected": -0.9802389144897461,
|
604 |
+
"loss": 1.2588,
|
605 |
+
"rewards/accuracies": 0.53125,
|
606 |
+
"rewards/chosen": 0.1880512535572052,
|
607 |
+
"rewards/margins": 0.2229468822479248,
|
608 |
+
"rewards/rejected": -0.034895628690719604,
|
609 |
"step": 390
|
610 |
},
|
611 |
{
|
612 |
+
"epoch": 0.7744433688286544,
|
613 |
+
"grad_norm": 3578.1741607281892,
|
614 |
"learning_rate": 1.4644660940672628e-10,
|
615 |
+
"logits/chosen": -1.6632875204086304,
|
616 |
+
"logits/rejected": -1.5778895616531372,
|
617 |
+
"logps/chosen": -0.9968886375427246,
|
618 |
+
"logps/rejected": -0.9527746438980103,
|
619 |
+
"loss": 1.2605,
|
620 |
+
"rewards/accuracies": 0.48750001192092896,
|
621 |
+
"rewards/chosen": 0.01907891035079956,
|
622 |
+
"rewards/margins": -0.09163277596235275,
|
623 |
+
"rewards/rejected": 0.11071167141199112,
|
624 |
"step": 400
|
625 |
},
|
626 |
{
|
627 |
+
"epoch": 0.7938044530493708,
|
628 |
+
"grad_norm": 4697.104208557154,
|
629 |
"learning_rate": 1.2333701836832813e-10,
|
630 |
+
"logits/chosen": -1.6258773803710938,
|
631 |
+
"logits/rejected": -1.5613044500350952,
|
632 |
+
"logps/chosen": -0.9721547365188599,
|
633 |
+
"logps/rejected": -0.9434949159622192,
|
634 |
+
"loss": 1.2668,
|
635 |
+
"rewards/accuracies": 0.5249999761581421,
|
636 |
+
"rewards/chosen": 0.04442809149622917,
|
637 |
+
"rewards/margins": 0.06200051307678223,
|
638 |
+
"rewards/rejected": -0.01757242903113365,
|
639 |
"step": 410
|
640 |
},
|
641 |
{
|
642 |
+
"epoch": 0.8131655372700871,
|
643 |
+
"grad_norm": 4633.7276813164935,
|
644 |
"learning_rate": 1.0195346714717813e-10,
|
645 |
+
"logits/chosen": -1.537536859512329,
|
646 |
+
"logits/rejected": -1.5266730785369873,
|
647 |
+
"logps/chosen": -0.965029239654541,
|
648 |
+
"logps/rejected": -0.9463868141174316,
|
649 |
+
"loss": 1.2845,
|
650 |
+
"rewards/accuracies": 0.4437499940395355,
|
651 |
+
"rewards/chosen": -0.0605890154838562,
|
652 |
+
"rewards/margins": -0.1169745922088623,
|
653 |
+
"rewards/rejected": 0.05638556554913521,
|
654 |
"step": 420
|
655 |
},
|
656 |
{
|
657 |
+
"epoch": 0.8325266214908035,
|
658 |
+
"grad_norm": 4035.95185171689,
|
659 |
"learning_rate": 8.239394482805996e-11,
|
660 |
+
"logits/chosen": -1.5938284397125244,
|
661 |
+
"logits/rejected": -1.5492713451385498,
|
662 |
+
"logps/chosen": -1.0192432403564453,
|
663 |
+
"logps/rejected": -0.9626362919807434,
|
664 |
+
"loss": 1.2589,
|
665 |
+
"rewards/accuracies": 0.518750011920929,
|
666 |
+
"rewards/chosen": 0.001451274729333818,
|
667 |
+
"rewards/margins": 0.09064897149801254,
|
668 |
+
"rewards/rejected": -0.08919770270586014,
|
669 |
"step": 430
|
670 |
},
|
671 |
{
|
672 |
+
"epoch": 0.8518877057115198,
|
673 |
+
"grad_norm": 4317.7906947065,
|
674 |
"learning_rate": 6.474808197191401e-11,
|
675 |
+
"logits/chosen": -1.631870985031128,
|
676 |
+
"logits/rejected": -1.5819487571716309,
|
677 |
+
"logps/chosen": -1.0460567474365234,
|
678 |
+
"logps/rejected": -0.9752202033996582,
|
679 |
+
"loss": 1.2476,
|
680 |
+
"rewards/accuracies": 0.53125,
|
681 |
+
"rewards/chosen": 0.15858839452266693,
|
682 |
+
"rewards/margins": 0.21004195511341095,
|
683 |
+
"rewards/rejected": -0.051453519612550735,
|
684 |
"step": 440
|
685 |
},
|
686 |
{
|
687 |
+
"epoch": 0.8712487899322362,
|
688 |
+
"grad_norm": 4211.087790819232,
|
689 |
"learning_rate": 4.9096739888146e-11,
|
690 |
+
"logits/chosen": -1.6219565868377686,
|
691 |
+
"logits/rejected": -1.5289008617401123,
|
692 |
+
"logps/chosen": -1.013887643814087,
|
693 |
+
"logps/rejected": -0.9734300374984741,
|
694 |
+
"loss": 1.2654,
|
695 |
+
"rewards/accuracies": 0.518750011920929,
|
696 |
+
"rewards/chosen": 0.10689739137887955,
|
697 |
+
"rewards/margins": 0.05857623741030693,
|
698 |
+
"rewards/rejected": 0.04832116514444351,
|
699 |
"step": 450
|
700 |
},
|
701 |
{
|
702 |
+
"epoch": 0.8906098741529526,
|
703 |
+
"grad_norm": 3769.26467222043,
|
704 |
"learning_rate": 3.5511640091604293e-11,
|
705 |
+
"logits/chosen": -1.5646103620529175,
|
706 |
+
"logits/rejected": -1.5271342992782593,
|
707 |
+
"logps/chosen": -1.0559688806533813,
|
708 |
+
"logps/rejected": -0.9385896921157837,
|
709 |
+
"loss": 1.2187,
|
710 |
+
"rewards/accuracies": 0.543749988079071,
|
711 |
+
"rewards/chosen": 0.07790811359882355,
|
712 |
+
"rewards/margins": 0.21173524856567383,
|
713 |
+
"rewards/rejected": -0.13382713496685028,
|
714 |
"step": 460
|
715 |
},
|
716 |
{
|
717 |
+
"epoch": 0.9099709583736689,
|
718 |
+
"grad_norm": 4757.142899979977,
|
719 |
"learning_rate": 2.4055035642222225e-11,
|
720 |
+
"logits/chosen": -1.6382324695587158,
|
721 |
+
"logits/rejected": -1.5700418949127197,
|
722 |
+
"logps/chosen": -0.9839082956314087,
|
723 |
+
"logps/rejected": -0.9138771295547485,
|
724 |
+
"loss": 1.2532,
|
725 |
+
"rewards/accuracies": 0.4749999940395355,
|
726 |
+
"rewards/chosen": 0.08895201981067657,
|
727 |
+
"rewards/margins": 0.04646927863359451,
|
728 |
+
"rewards/rejected": 0.04248274117708206,
|
729 |
"step": 470
|
730 |
},
|
731 |
{
|
732 |
+
"epoch": 0.9293320425943853,
|
733 |
+
"grad_norm": 3836.5764604419264,
|
734 |
"learning_rate": 1.477942587339426e-11,
|
735 |
+
"logits/chosen": -1.6090694665908813,
|
736 |
+
"logits/rejected": -1.5662223100662231,
|
737 |
+
"logps/chosen": -0.9489457011222839,
|
738 |
+
"logps/rejected": -0.90757817029953,
|
739 |
+
"loss": 1.2642,
|
740 |
+
"rewards/accuracies": 0.48750001192092896,
|
741 |
+
"rewards/chosen": -0.05137147754430771,
|
742 |
+
"rewards/margins": -0.01941109262406826,
|
743 |
+
"rewards/rejected": -0.0319603867828846,
|
744 |
"step": 480
|
745 |
},
|
746 |
{
|
747 |
+
"epoch": 0.9486931268151017,
|
748 |
+
"grad_norm": 4250.965954777766,
|
749 |
"learning_rate": 7.727315816331515e-12,
|
750 |
+
"logits/chosen": -1.6554877758026123,
|
751 |
+
"logits/rejected": -1.6055545806884766,
|
752 |
+
"logps/chosen": -1.0777404308319092,
|
753 |
+
"logps/rejected": -1.0004560947418213,
|
754 |
+
"loss": 1.2376,
|
755 |
+
"rewards/accuracies": 0.543749988079071,
|
756 |
+
"rewards/chosen": 0.07162754982709885,
|
757 |
+
"rewards/margins": 0.12462921440601349,
|
758 |
+
"rewards/rejected": -0.05300166457891464,
|
759 |
"step": 490
|
760 |
},
|
761 |
{
|
762 |
+
"epoch": 0.968054211035818,
|
763 |
+
"grad_norm": 3914.5437511702235,
|
764 |
"learning_rate": 2.9310214228202016e-12,
|
765 |
+
"logits/chosen": -1.6229422092437744,
|
766 |
+
"logits/rejected": -1.5482664108276367,
|
767 |
+
"logps/chosen": -0.9587628245353699,
|
768 |
+
"logps/rejected": -0.9172463417053223,
|
769 |
+
"loss": 1.2265,
|
770 |
+
"rewards/accuracies": 0.44999998807907104,
|
771 |
+
"rewards/chosen": -0.032034747302532196,
|
772 |
+
"rewards/margins": -0.1518467366695404,
|
773 |
+
"rewards/rejected": 0.1198119968175888,
|
774 |
"step": 500
|
775 |
},
|
776 |
{
|
777 |
+
"epoch": 0.9874152952565344,
|
778 |
+
"grad_norm": 4236.531801010516,
|
779 |
"learning_rate": 4.125214789427734e-13,
|
780 |
+
"logits/chosen": -1.5640289783477783,
|
781 |
+
"logits/rejected": -1.5025882720947266,
|
782 |
+
"logps/chosen": -0.9930515289306641,
|
783 |
+
"logps/rejected": -0.9385469555854797,
|
784 |
+
"loss": 1.254,
|
785 |
+
"rewards/accuracies": 0.48750001192092896,
|
786 |
+
"rewards/chosen": 0.09590280055999756,
|
787 |
+
"rewards/margins": 0.0512530580163002,
|
788 |
+
"rewards/rejected": 0.04464975371956825,
|
789 |
"step": 510
|
790 |
},
|
791 |
{
|
792 |
+
"epoch": 0.9990319457889641,
|
793 |
"step": 516,
|
794 |
"total_flos": 0.0,
|
795 |
+
"train_loss": 1.2622329578843228,
|
796 |
+
"train_runtime": 8815.3772,
|
797 |
+
"train_samples_per_second": 7.496,
|
798 |
+
"train_steps_per_second": 0.059
|
799 |
}
|
800 |
],
|
801 |
"logging_steps": 10,
|
|
|
803 |
"num_input_tokens_seen": 0,
|
804 |
"num_train_epochs": 1,
|
805 |
"save_steps": 100,
|
806 |
+
"stateful_callbacks": {
|
807 |
+
"TrainerControl": {
|
808 |
+
"args": {
|
809 |
+
"should_epoch_stop": false,
|
810 |
+
"should_evaluate": false,
|
811 |
+
"should_log": false,
|
812 |
+
"should_save": true,
|
813 |
+
"should_training_stop": false
|
814 |
+
},
|
815 |
+
"attributes": {}
|
816 |
+
}
|
817 |
+
},
|
818 |
"total_flos": 0.0,
|
819 |
"train_batch_size": 4,
|
820 |
"trial_name": null,
|
training_args.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:cd72d8e0cd286217c9a7fddf58a057686f97c0f94432feaca0ca148ca8dbc12e
|
3 |
+
size 6520
|