yiran-wang3 commited on
Commit
8847d1a
1 Parent(s): 096e11e

End of training

Browse files
README.md ADDED
@@ -0,0 +1,64 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: transformers
3
+ license: other
4
+ base_model: yiran-wang3/ds_coder6.7b_reflct_adamw_iter5
5
+ tags:
6
+ - alignment-handbook
7
+ - generated_from_trainer
8
+ - trl
9
+ - dpo
10
+ datasets:
11
+ - self-generate/ds_coder6.7b_reflct_sppo_hard_new_cn_mining_oj_iter5-binarized-reflection-scored
12
+ model-index:
13
+ - name: ds_coder6.7b_reflct_adamw_iter6
14
+ results: []
15
+ ---
16
+
17
+ <!-- This model card has been generated automatically according to the information the Trainer had access to. You
18
+ should probably proofread and complete it, then remove this comment. -->
19
+
20
+ # ds_coder6.7b_reflct_adamw_iter6
21
+
22
+ This model is a fine-tuned version of [yiran-wang3/ds_coder6.7b_reflct_adamw_iter5](https://huggingface.co/yiran-wang3/ds_coder6.7b_reflct_adamw_iter5) on the self-generate/ds_coder6.7b_reflct_sppo_hard_new_cn_mining_oj_iter5-binarized-reflection-scored dataset.
23
+
24
+ ## Model description
25
+
26
+ More information needed
27
+
28
+ ## Intended uses & limitations
29
+
30
+ More information needed
31
+
32
+ ## Training and evaluation data
33
+
34
+ More information needed
35
+
36
+ ## Training procedure
37
+
38
+ ### Training hyperparameters
39
+
40
+ The following hyperparameters were used during training:
41
+ - learning_rate: 1e-06
42
+ - train_batch_size: 8
43
+ - eval_batch_size: 4
44
+ - seed: 42
45
+ - distributed_type: multi-GPU
46
+ - num_devices: 8
47
+ - total_train_batch_size: 64
48
+ - total_eval_batch_size: 32
49
+ - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
50
+ - lr_scheduler_type: constant
51
+ - lr_scheduler_warmup_ratio: 0.1
52
+ - lr_scheduler_warmup_steps: 100
53
+ - num_epochs: 1.0
54
+
55
+ ### Training results
56
+
57
+
58
+
59
+ ### Framework versions
60
+
61
+ - Transformers 4.45.0
62
+ - Pytorch 2.4.0+cu121
63
+ - Datasets 2.14.6
64
+ - Tokenizers 0.20.3
all_results.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 1.0,
3
+ "total_flos": 0.0,
4
+ "train_loss": 0.47894636502391413,
5
+ "train_runtime": 421.3681,
6
+ "train_samples": 2378,
7
+ "train_samples_per_second": 5.644,
8
+ "train_steps_per_second": 0.09
9
+ }
config.json CHANGED
@@ -29,6 +29,6 @@
29
  "tie_word_embeddings": false,
30
  "torch_dtype": "bfloat16",
31
  "transformers_version": "4.45.0",
32
- "use_cache": false,
33
  "vocab_size": 32256
34
  }
 
29
  "tie_word_embeddings": false,
30
  "torch_dtype": "bfloat16",
31
  "transformers_version": "4.45.0",
32
+ "use_cache": true,
33
  "vocab_size": 32256
34
  }
generation_config.json ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 32013,
4
+ "eos_token_id": 32021,
5
+ "transformers_version": "4.45.0"
6
+ }
train_results.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 1.0,
3
+ "total_flos": 0.0,
4
+ "train_loss": 0.47894636502391413,
5
+ "train_runtime": 421.3681,
6
+ "train_samples": 2378,
7
+ "train_samples_per_second": 5.644,
8
+ "train_steps_per_second": 0.09
9
+ }
trainer_state.json ADDED
@@ -0,0 +1,840 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 1.0,
5
+ "eval_steps": 100,
6
+ "global_step": 38,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "debug/policy_chosen_logits": 1.8110721111297607,
13
+ "debug/policy_chosen_logps": -242.31497192382812,
14
+ "debug/policy_rejected_logits": 1.8570362329483032,
15
+ "debug/policy_rejected_logps": -306.15191650390625,
16
+ "debug/reference_chosen_logps": -242.31497192382812,
17
+ "debug/reference_rejected_logps": -306.15191650390625,
18
+ "epoch": 0.02631578947368421,
19
+ "grad_norm": 5.1162553610364885,
20
+ "learning_rate": 1e-06,
21
+ "logits/chosen": 1.8110721111297607,
22
+ "logits/rejected": 1.8570362329483032,
23
+ "logps/chosen": -242.31497192382812,
24
+ "logps/rejected": -306.15191650390625,
25
+ "loss": 0.5,
26
+ "rewards/accuracies": 0.0,
27
+ "rewards/chosen": 0.0,
28
+ "rewards/margins": 0.0,
29
+ "rewards/rejected": 0.0,
30
+ "step": 1
31
+ },
32
+ {
33
+ "debug/policy_chosen_logits": 1.535241723060608,
34
+ "debug/policy_chosen_logps": -233.50314331054688,
35
+ "debug/policy_rejected_logits": 1.5806331634521484,
36
+ "debug/policy_rejected_logps": -236.88345336914062,
37
+ "debug/reference_chosen_logps": -233.77830505371094,
38
+ "debug/reference_rejected_logps": -236.92105102539062,
39
+ "epoch": 0.05263157894736842,
40
+ "grad_norm": 4.391641485844322,
41
+ "learning_rate": 1e-06,
42
+ "logits/chosen": 1.535241723060608,
43
+ "logits/rejected": 1.5806331634521484,
44
+ "logps/chosen": -233.50314331054688,
45
+ "logps/rejected": -236.88345336914062,
46
+ "loss": 0.4982,
47
+ "rewards/accuracies": 0.5,
48
+ "rewards/chosen": 0.0027516556438058615,
49
+ "rewards/margins": 0.002375659765675664,
50
+ "rewards/rejected": 0.0003759957617148757,
51
+ "step": 2
52
+ },
53
+ {
54
+ "debug/policy_chosen_logits": 1.6230586767196655,
55
+ "debug/policy_chosen_logps": -261.1310119628906,
56
+ "debug/policy_rejected_logits": 1.8340940475463867,
57
+ "debug/policy_rejected_logps": -378.02337646484375,
58
+ "debug/reference_chosen_logps": -261.3377685546875,
59
+ "debug/reference_rejected_logps": -378.5300598144531,
60
+ "epoch": 0.07894736842105263,
61
+ "grad_norm": 5.337649724861765,
62
+ "learning_rate": 1e-06,
63
+ "logits/chosen": 1.6230586767196655,
64
+ "logits/rejected": 1.8340940475463867,
65
+ "logps/chosen": -261.1310119628906,
66
+ "logps/rejected": -378.02337646484375,
67
+ "loss": 0.4995,
68
+ "rewards/accuracies": 0.5,
69
+ "rewards/chosen": 0.0020675850100815296,
70
+ "rewards/margins": -0.002999210264533758,
71
+ "rewards/rejected": 0.0050667948089540005,
72
+ "step": 3
73
+ },
74
+ {
75
+ "debug/policy_chosen_logits": 1.801438570022583,
76
+ "debug/policy_chosen_logps": -233.70458984375,
77
+ "debug/policy_rejected_logits": 1.6214240789413452,
78
+ "debug/policy_rejected_logps": -303.5933837890625,
79
+ "debug/reference_chosen_logps": -233.42352294921875,
80
+ "debug/reference_rejected_logps": -302.8963623046875,
81
+ "epoch": 0.10526315789473684,
82
+ "grad_norm": 5.416209205950235,
83
+ "learning_rate": 1e-06,
84
+ "logits/chosen": 1.801438570022583,
85
+ "logits/rejected": 1.6214240789413452,
86
+ "logps/chosen": -233.70458984375,
87
+ "logps/rejected": -303.5933837890625,
88
+ "loss": 0.5002,
89
+ "rewards/accuracies": 0.625,
90
+ "rewards/chosen": -0.0028107455000281334,
91
+ "rewards/margins": 0.0041594123467803,
92
+ "rewards/rejected": -0.0069701578468084335,
93
+ "step": 4
94
+ },
95
+ {
96
+ "debug/policy_chosen_logits": 1.62641179561615,
97
+ "debug/policy_chosen_logps": -220.23741149902344,
98
+ "debug/policy_rejected_logits": 1.7156729698181152,
99
+ "debug/policy_rejected_logps": -306.0166015625,
100
+ "debug/reference_chosen_logps": -219.86209106445312,
101
+ "debug/reference_rejected_logps": -305.0866394042969,
102
+ "epoch": 0.13157894736842105,
103
+ "grad_norm": 4.132996885138251,
104
+ "learning_rate": 1e-06,
105
+ "logits/chosen": 1.62641179561615,
106
+ "logits/rejected": 1.7156729698181152,
107
+ "logps/chosen": -220.23741149902344,
108
+ "logps/rejected": -306.0166015625,
109
+ "loss": 0.4991,
110
+ "rewards/accuracies": 0.625,
111
+ "rewards/chosen": -0.0037531470879912376,
112
+ "rewards/margins": 0.005546645727008581,
113
+ "rewards/rejected": -0.009299792349338531,
114
+ "step": 5
115
+ },
116
+ {
117
+ "debug/policy_chosen_logits": 1.862275242805481,
118
+ "debug/policy_chosen_logps": -232.3478240966797,
119
+ "debug/policy_rejected_logits": 1.7659251689910889,
120
+ "debug/policy_rejected_logps": -222.363037109375,
121
+ "debug/reference_chosen_logps": -231.81625366210938,
122
+ "debug/reference_rejected_logps": -221.9506072998047,
123
+ "epoch": 0.15789473684210525,
124
+ "grad_norm": 4.973887804037034,
125
+ "learning_rate": 1e-06,
126
+ "logits/chosen": 1.862275242805481,
127
+ "logits/rejected": 1.7659251689910889,
128
+ "logps/chosen": -232.3478240966797,
129
+ "logps/rejected": -222.363037109375,
130
+ "loss": 0.4961,
131
+ "rewards/accuracies": 0.375,
132
+ "rewards/chosen": -0.005315742455422878,
133
+ "rewards/margins": -0.0011914825299754739,
134
+ "rewards/rejected": -0.004124259576201439,
135
+ "step": 6
136
+ },
137
+ {
138
+ "debug/policy_chosen_logits": 1.8720442056655884,
139
+ "debug/policy_chosen_logps": -241.5899658203125,
140
+ "debug/policy_rejected_logits": 1.792407751083374,
141
+ "debug/policy_rejected_logps": -224.8570556640625,
142
+ "debug/reference_chosen_logps": -241.32894897460938,
143
+ "debug/reference_rejected_logps": -224.58128356933594,
144
+ "epoch": 0.18421052631578946,
145
+ "grad_norm": 4.979808676717409,
146
+ "learning_rate": 1e-06,
147
+ "logits/chosen": 1.8720442056655884,
148
+ "logits/rejected": 1.792407751083374,
149
+ "logps/chosen": -241.5899658203125,
150
+ "logps/rejected": -224.8570556640625,
151
+ "loss": 0.4938,
152
+ "rewards/accuracies": 0.5,
153
+ "rewards/chosen": -0.0026101875118911266,
154
+ "rewards/margins": 0.000147590646520257,
155
+ "rewards/rejected": -0.0027577781584113836,
156
+ "step": 7
157
+ },
158
+ {
159
+ "debug/policy_chosen_logits": 1.626907467842102,
160
+ "debug/policy_chosen_logps": -217.16574096679688,
161
+ "debug/policy_rejected_logits": 1.626301646232605,
162
+ "debug/policy_rejected_logps": -300.71978759765625,
163
+ "debug/reference_chosen_logps": -216.7672119140625,
164
+ "debug/reference_rejected_logps": -298.59197998046875,
165
+ "epoch": 0.21052631578947367,
166
+ "grad_norm": 4.585134841741792,
167
+ "learning_rate": 1e-06,
168
+ "logits/chosen": 1.626907467842102,
169
+ "logits/rejected": 1.626301646232605,
170
+ "logps/chosen": -217.16574096679688,
171
+ "logps/rejected": -300.71978759765625,
172
+ "loss": 0.4937,
173
+ "rewards/accuracies": 0.75,
174
+ "rewards/chosen": -0.0039853667840361595,
175
+ "rewards/margins": 0.01729259453713894,
176
+ "rewards/rejected": -0.021277960389852524,
177
+ "step": 8
178
+ },
179
+ {
180
+ "debug/policy_chosen_logits": 2.008640766143799,
181
+ "debug/policy_chosen_logps": -246.12429809570312,
182
+ "debug/policy_rejected_logits": 1.8793882131576538,
183
+ "debug/policy_rejected_logps": -265.8034973144531,
184
+ "debug/reference_chosen_logps": -245.7059783935547,
185
+ "debug/reference_rejected_logps": -264.12457275390625,
186
+ "epoch": 0.23684210526315788,
187
+ "grad_norm": 4.595004508027087,
188
+ "learning_rate": 1e-06,
189
+ "logits/chosen": 2.008640766143799,
190
+ "logits/rejected": 1.8793882131576538,
191
+ "logps/chosen": -246.12429809570312,
192
+ "logps/rejected": -265.8034973144531,
193
+ "loss": 0.4933,
194
+ "rewards/accuracies": 0.875,
195
+ "rewards/chosen": -0.004182987846434116,
196
+ "rewards/margins": 0.012606277130544186,
197
+ "rewards/rejected": -0.016789264976978302,
198
+ "step": 9
199
+ },
200
+ {
201
+ "debug/policy_chosen_logits": 1.7032949924468994,
202
+ "debug/policy_chosen_logps": -232.8529815673828,
203
+ "debug/policy_rejected_logits": 1.5618568658828735,
204
+ "debug/policy_rejected_logps": -253.05532836914062,
205
+ "debug/reference_chosen_logps": -233.14126586914062,
206
+ "debug/reference_rejected_logps": -252.65480041503906,
207
+ "epoch": 0.2631578947368421,
208
+ "grad_norm": 4.0613383659805065,
209
+ "learning_rate": 1e-06,
210
+ "logits/chosen": 1.7032949924468994,
211
+ "logits/rejected": 1.5618568658828735,
212
+ "logps/chosen": -232.8529815673828,
213
+ "logps/rejected": -253.05532836914062,
214
+ "loss": 0.4957,
215
+ "rewards/accuracies": 0.75,
216
+ "rewards/chosen": 0.0028829958755522966,
217
+ "rewards/margins": 0.006888084579259157,
218
+ "rewards/rejected": -0.004005088936537504,
219
+ "step": 10
220
+ },
221
+ {
222
+ "debug/policy_chosen_logits": 1.684201717376709,
223
+ "debug/policy_chosen_logps": -232.94818115234375,
224
+ "debug/policy_rejected_logits": 1.8217120170593262,
225
+ "debug/policy_rejected_logps": -266.5169982910156,
226
+ "debug/reference_chosen_logps": -231.85365295410156,
227
+ "debug/reference_rejected_logps": -266.5830993652344,
228
+ "epoch": 0.2894736842105263,
229
+ "grad_norm": 4.384299997547265,
230
+ "learning_rate": 1e-06,
231
+ "logits/chosen": 1.684201717376709,
232
+ "logits/rejected": 1.8217120170593262,
233
+ "logps/chosen": -232.94818115234375,
234
+ "logps/rejected": -266.5169982910156,
235
+ "loss": 0.497,
236
+ "rewards/accuracies": 0.5,
237
+ "rewards/chosen": -0.010945243760943413,
238
+ "rewards/margins": -0.011606387794017792,
239
+ "rewards/rejected": 0.0006611444987356663,
240
+ "step": 11
241
+ },
242
+ {
243
+ "debug/policy_chosen_logits": 1.6037061214447021,
244
+ "debug/policy_chosen_logps": -271.25543212890625,
245
+ "debug/policy_rejected_logits": 1.5078108310699463,
246
+ "debug/policy_rejected_logps": -325.62957763671875,
247
+ "debug/reference_chosen_logps": -270.1516418457031,
248
+ "debug/reference_rejected_logps": -321.0948486328125,
249
+ "epoch": 0.3157894736842105,
250
+ "grad_norm": 4.955955505999082,
251
+ "learning_rate": 1e-06,
252
+ "logits/chosen": 1.6037061214447021,
253
+ "logits/rejected": 1.5078108310699463,
254
+ "logps/chosen": -271.25543212890625,
255
+ "logps/rejected": -325.62957763671875,
256
+ "loss": 0.4883,
257
+ "rewards/accuracies": 0.625,
258
+ "rewards/chosen": -0.011037863790988922,
259
+ "rewards/margins": 0.0343095026910305,
260
+ "rewards/rejected": -0.045347362756729126,
261
+ "step": 12
262
+ },
263
+ {
264
+ "debug/policy_chosen_logits": 1.9024295806884766,
265
+ "debug/policy_chosen_logps": -230.404541015625,
266
+ "debug/policy_rejected_logits": 1.754141092300415,
267
+ "debug/policy_rejected_logps": -230.44754028320312,
268
+ "debug/reference_chosen_logps": -230.56103515625,
269
+ "debug/reference_rejected_logps": -230.1070556640625,
270
+ "epoch": 0.34210526315789475,
271
+ "grad_norm": 4.738162826032535,
272
+ "learning_rate": 1e-06,
273
+ "logits/chosen": 1.9024295806884766,
274
+ "logits/rejected": 1.754141092300415,
275
+ "logps/chosen": -230.404541015625,
276
+ "logps/rejected": -230.44754028320312,
277
+ "loss": 0.4882,
278
+ "rewards/accuracies": 0.625,
279
+ "rewards/chosen": 0.0015649220440536737,
280
+ "rewards/margins": 0.004969825968146324,
281
+ "rewards/rejected": -0.0034049032256007195,
282
+ "step": 13
283
+ },
284
+ {
285
+ "debug/policy_chosen_logits": 1.6058262586593628,
286
+ "debug/policy_chosen_logps": -256.1600341796875,
287
+ "debug/policy_rejected_logits": 1.8204787969589233,
288
+ "debug/policy_rejected_logps": -283.00579833984375,
289
+ "debug/reference_chosen_logps": -255.710205078125,
290
+ "debug/reference_rejected_logps": -283.3298034667969,
291
+ "epoch": 0.3684210526315789,
292
+ "grad_norm": 5.473371289645614,
293
+ "learning_rate": 1e-06,
294
+ "logits/chosen": 1.6058262586593628,
295
+ "logits/rejected": 1.8204787969589233,
296
+ "logps/chosen": -256.1600341796875,
297
+ "logps/rejected": -283.00579833984375,
298
+ "loss": 0.4909,
299
+ "rewards/accuracies": 0.375,
300
+ "rewards/chosen": -0.004498443566262722,
301
+ "rewards/margins": -0.007738151587545872,
302
+ "rewards/rejected": 0.003239707788452506,
303
+ "step": 14
304
+ },
305
+ {
306
+ "debug/policy_chosen_logits": 1.6739165782928467,
307
+ "debug/policy_chosen_logps": -260.7919616699219,
308
+ "debug/policy_rejected_logits": 1.597163200378418,
309
+ "debug/policy_rejected_logps": -275.9894714355469,
310
+ "debug/reference_chosen_logps": -262.1143798828125,
311
+ "debug/reference_rejected_logps": -275.02569580078125,
312
+ "epoch": 0.39473684210526316,
313
+ "grad_norm": 4.973250917361795,
314
+ "learning_rate": 1e-06,
315
+ "logits/chosen": 1.6739165782928467,
316
+ "logits/rejected": 1.597163200378418,
317
+ "logps/chosen": -260.7919616699219,
318
+ "logps/rejected": -275.9894714355469,
319
+ "loss": 0.4818,
320
+ "rewards/accuracies": 0.75,
321
+ "rewards/chosen": 0.013224296271800995,
322
+ "rewards/margins": 0.022862261161208153,
323
+ "rewards/rejected": -0.009637965820729733,
324
+ "step": 15
325
+ },
326
+ {
327
+ "debug/policy_chosen_logits": 1.7346552610397339,
328
+ "debug/policy_chosen_logps": -235.43736267089844,
329
+ "debug/policy_rejected_logits": 1.7165241241455078,
330
+ "debug/policy_rejected_logps": -291.98858642578125,
331
+ "debug/reference_chosen_logps": -238.7733154296875,
332
+ "debug/reference_rejected_logps": -289.6514587402344,
333
+ "epoch": 0.42105263157894735,
334
+ "grad_norm": 4.492419940892433,
335
+ "learning_rate": 1e-06,
336
+ "logits/chosen": 1.7346552610397339,
337
+ "logits/rejected": 1.7165241241455078,
338
+ "logps/chosen": -235.43736267089844,
339
+ "logps/rejected": -291.98858642578125,
340
+ "loss": 0.4743,
341
+ "rewards/accuracies": 0.875,
342
+ "rewards/chosen": 0.03335927799344063,
343
+ "rewards/margins": 0.05673057585954666,
344
+ "rewards/rejected": -0.023371294140815735,
345
+ "step": 16
346
+ },
347
+ {
348
+ "debug/policy_chosen_logits": 1.7080655097961426,
349
+ "debug/policy_chosen_logps": -244.19293212890625,
350
+ "debug/policy_rejected_logits": 1.6485531330108643,
351
+ "debug/policy_rejected_logps": -293.0354309082031,
352
+ "debug/reference_chosen_logps": -245.80865478515625,
353
+ "debug/reference_rejected_logps": -287.3146667480469,
354
+ "epoch": 0.4473684210526316,
355
+ "grad_norm": 4.423794468647003,
356
+ "learning_rate": 1e-06,
357
+ "logits/chosen": 1.7080655097961426,
358
+ "logits/rejected": 1.6485531330108643,
359
+ "logps/chosen": -244.19293212890625,
360
+ "logps/rejected": -293.0354309082031,
361
+ "loss": 0.4811,
362
+ "rewards/accuracies": 1.0,
363
+ "rewards/chosen": 0.01615716889500618,
364
+ "rewards/margins": 0.07336486876010895,
365
+ "rewards/rejected": -0.05720769986510277,
366
+ "step": 17
367
+ },
368
+ {
369
+ "debug/policy_chosen_logits": 1.799321174621582,
370
+ "debug/policy_chosen_logps": -238.0970458984375,
371
+ "debug/policy_rejected_logits": 1.5929076671600342,
372
+ "debug/policy_rejected_logps": -263.6419677734375,
373
+ "debug/reference_chosen_logps": -240.28492736816406,
374
+ "debug/reference_rejected_logps": -263.82159423828125,
375
+ "epoch": 0.47368421052631576,
376
+ "grad_norm": 6.480796436023613,
377
+ "learning_rate": 1e-06,
378
+ "logits/chosen": 1.799321174621582,
379
+ "logits/rejected": 1.5929076671600342,
380
+ "logps/chosen": -238.0970458984375,
381
+ "logps/rejected": -263.6419677734375,
382
+ "loss": 0.4842,
383
+ "rewards/accuracies": 0.625,
384
+ "rewards/chosen": 0.021878758445382118,
385
+ "rewards/margins": 0.020082702860236168,
386
+ "rewards/rejected": 0.0017960541881620884,
387
+ "step": 18
388
+ },
389
+ {
390
+ "debug/policy_chosen_logits": 1.5554420948028564,
391
+ "debug/policy_chosen_logps": -259.22515869140625,
392
+ "debug/policy_rejected_logits": 1.9916545152664185,
393
+ "debug/policy_rejected_logps": -345.74395751953125,
394
+ "debug/reference_chosen_logps": -258.97943115234375,
395
+ "debug/reference_rejected_logps": -340.2289123535156,
396
+ "epoch": 0.5,
397
+ "grad_norm": 4.996200485758474,
398
+ "learning_rate": 1e-06,
399
+ "logits/chosen": 1.5554420948028564,
400
+ "logits/rejected": 1.9916545152664185,
401
+ "logps/chosen": -259.22515869140625,
402
+ "logps/rejected": -345.74395751953125,
403
+ "loss": 0.4614,
404
+ "rewards/accuracies": 0.75,
405
+ "rewards/chosen": -0.002457238733768463,
406
+ "rewards/margins": 0.05269308015704155,
407
+ "rewards/rejected": -0.05515032261610031,
408
+ "step": 19
409
+ },
410
+ {
411
+ "debug/policy_chosen_logits": 1.8484898805618286,
412
+ "debug/policy_chosen_logps": -220.50599670410156,
413
+ "debug/policy_rejected_logits": 1.7760157585144043,
414
+ "debug/policy_rejected_logps": -262.2606201171875,
415
+ "debug/reference_chosen_logps": -222.98545837402344,
416
+ "debug/reference_rejected_logps": -261.75836181640625,
417
+ "epoch": 0.5263157894736842,
418
+ "grad_norm": 4.774581974562177,
419
+ "learning_rate": 1e-06,
420
+ "logits/chosen": 1.8484898805618286,
421
+ "logits/rejected": 1.7760157585144043,
422
+ "logps/chosen": -220.50599670410156,
423
+ "logps/rejected": -262.2606201171875,
424
+ "loss": 0.4799,
425
+ "rewards/accuracies": 0.625,
426
+ "rewards/chosen": 0.024794597178697586,
427
+ "rewards/margins": 0.029817257076501846,
428
+ "rewards/rejected": -0.005022658966481686,
429
+ "step": 20
430
+ },
431
+ {
432
+ "debug/policy_chosen_logits": 1.560202717781067,
433
+ "debug/policy_chosen_logps": -221.38250732421875,
434
+ "debug/policy_rejected_logits": 1.6546127796173096,
435
+ "debug/policy_rejected_logps": -251.31719970703125,
436
+ "debug/reference_chosen_logps": -223.98187255859375,
437
+ "debug/reference_rejected_logps": -250.66644287109375,
438
+ "epoch": 0.5526315789473685,
439
+ "grad_norm": 4.351924103595798,
440
+ "learning_rate": 1e-06,
441
+ "logits/chosen": 1.560202717781067,
442
+ "logits/rejected": 1.6546127796173096,
443
+ "logps/chosen": -221.38250732421875,
444
+ "logps/rejected": -251.31719970703125,
445
+ "loss": 0.473,
446
+ "rewards/accuracies": 0.875,
447
+ "rewards/chosen": 0.025993727147579193,
448
+ "rewards/margins": 0.032501161098480225,
449
+ "rewards/rejected": -0.006507434882223606,
450
+ "step": 21
451
+ },
452
+ {
453
+ "debug/policy_chosen_logits": 1.7773959636688232,
454
+ "debug/policy_chosen_logps": -270.9798889160156,
455
+ "debug/policy_rejected_logits": 1.8666154146194458,
456
+ "debug/policy_rejected_logps": -288.00830078125,
457
+ "debug/reference_chosen_logps": -270.544189453125,
458
+ "debug/reference_rejected_logps": -285.48175048828125,
459
+ "epoch": 0.5789473684210527,
460
+ "grad_norm": 4.442244693551726,
461
+ "learning_rate": 1e-06,
462
+ "logits/chosen": 1.7773959636688232,
463
+ "logits/rejected": 1.8666154146194458,
464
+ "logps/chosen": -270.9798889160156,
465
+ "logps/rejected": -288.00830078125,
466
+ "loss": 0.484,
467
+ "rewards/accuracies": 0.75,
468
+ "rewards/chosen": -0.004356803838163614,
469
+ "rewards/margins": 0.020908568054437637,
470
+ "rewards/rejected": -0.025265369564294815,
471
+ "step": 22
472
+ },
473
+ {
474
+ "debug/policy_chosen_logits": 1.6907881498336792,
475
+ "debug/policy_chosen_logps": -232.88339233398438,
476
+ "debug/policy_rejected_logits": 1.4464173316955566,
477
+ "debug/policy_rejected_logps": -248.91647338867188,
478
+ "debug/reference_chosen_logps": -235.3972625732422,
479
+ "debug/reference_rejected_logps": -248.42919921875,
480
+ "epoch": 0.6052631578947368,
481
+ "grad_norm": 4.8897128390396265,
482
+ "learning_rate": 1e-06,
483
+ "logits/chosen": 1.6907881498336792,
484
+ "logits/rejected": 1.4464173316955566,
485
+ "logps/chosen": -232.88339233398438,
486
+ "logps/rejected": -248.91647338867188,
487
+ "loss": 0.4626,
488
+ "rewards/accuracies": 0.875,
489
+ "rewards/chosen": 0.02513885498046875,
490
+ "rewards/margins": 0.03001151978969574,
491
+ "rewards/rejected": -0.004872665274888277,
492
+ "step": 23
493
+ },
494
+ {
495
+ "debug/policy_chosen_logits": 1.990630030632019,
496
+ "debug/policy_chosen_logps": -231.47027587890625,
497
+ "debug/policy_rejected_logits": 2.0270304679870605,
498
+ "debug/policy_rejected_logps": -280.219482421875,
499
+ "debug/reference_chosen_logps": -235.265380859375,
500
+ "debug/reference_rejected_logps": -279.20245361328125,
501
+ "epoch": 0.631578947368421,
502
+ "grad_norm": 5.476293220693162,
503
+ "learning_rate": 1e-06,
504
+ "logits/chosen": 1.990630030632019,
505
+ "logits/rejected": 2.0270304679870605,
506
+ "logps/chosen": -231.47027587890625,
507
+ "logps/rejected": -280.219482421875,
508
+ "loss": 0.4767,
509
+ "rewards/accuracies": 0.625,
510
+ "rewards/chosen": 0.037951141595840454,
511
+ "rewards/margins": 0.04812118411064148,
512
+ "rewards/rejected": -0.010170038789510727,
513
+ "step": 24
514
+ },
515
+ {
516
+ "debug/policy_chosen_logits": 1.9207803010940552,
517
+ "debug/policy_chosen_logps": -227.99293518066406,
518
+ "debug/policy_rejected_logits": 1.9825800657272339,
519
+ "debug/policy_rejected_logps": -258.94189453125,
520
+ "debug/reference_chosen_logps": -232.88571166992188,
521
+ "debug/reference_rejected_logps": -259.89483642578125,
522
+ "epoch": 0.6578947368421053,
523
+ "grad_norm": 4.410674654649863,
524
+ "learning_rate": 1e-06,
525
+ "logits/chosen": 1.9207803010940552,
526
+ "logits/rejected": 1.9825800657272339,
527
+ "logps/chosen": -227.99293518066406,
528
+ "logps/rejected": -258.94189453125,
529
+ "loss": 0.4653,
530
+ "rewards/accuracies": 0.875,
531
+ "rewards/chosen": 0.04892778396606445,
532
+ "rewards/margins": 0.039398498833179474,
533
+ "rewards/rejected": 0.00952928513288498,
534
+ "step": 25
535
+ },
536
+ {
537
+ "debug/policy_chosen_logits": 1.6559851169586182,
538
+ "debug/policy_chosen_logps": -242.3477020263672,
539
+ "debug/policy_rejected_logits": 1.9231914281845093,
540
+ "debug/policy_rejected_logps": -339.9230041503906,
541
+ "debug/reference_chosen_logps": -245.39537048339844,
542
+ "debug/reference_rejected_logps": -330.4933166503906,
543
+ "epoch": 0.6842105263157895,
544
+ "grad_norm": 5.841309481968776,
545
+ "learning_rate": 1e-06,
546
+ "logits/chosen": 1.6559851169586182,
547
+ "logits/rejected": 1.9231914281845093,
548
+ "logps/chosen": -242.3477020263672,
549
+ "logps/rejected": -339.9230041503906,
550
+ "loss": 0.4356,
551
+ "rewards/accuracies": 0.875,
552
+ "rewards/chosen": 0.030476529151201248,
553
+ "rewards/margins": 0.12477347999811172,
554
+ "rewards/rejected": -0.09429694712162018,
555
+ "step": 26
556
+ },
557
+ {
558
+ "debug/policy_chosen_logits": 1.379979133605957,
559
+ "debug/policy_chosen_logps": -238.91958618164062,
560
+ "debug/policy_rejected_logits": 1.2741044759750366,
561
+ "debug/policy_rejected_logps": -281.7842712402344,
562
+ "debug/reference_chosen_logps": -237.88525390625,
563
+ "debug/reference_rejected_logps": -274.2608947753906,
564
+ "epoch": 0.7105263157894737,
565
+ "grad_norm": 4.822057229893986,
566
+ "learning_rate": 1e-06,
567
+ "logits/chosen": 1.379979133605957,
568
+ "logits/rejected": 1.2741044759750366,
569
+ "logps/chosen": -238.91958618164062,
570
+ "logps/rejected": -281.7842712402344,
571
+ "loss": 0.4389,
572
+ "rewards/accuracies": 0.5,
573
+ "rewards/chosen": -0.010343209840357304,
574
+ "rewards/margins": 0.06489060819149017,
575
+ "rewards/rejected": -0.0752338171005249,
576
+ "step": 27
577
+ },
578
+ {
579
+ "debug/policy_chosen_logits": 1.4645960330963135,
580
+ "debug/policy_chosen_logps": -242.5520477294922,
581
+ "debug/policy_rejected_logits": 1.5396175384521484,
582
+ "debug/policy_rejected_logps": -264.1595458984375,
583
+ "debug/reference_chosen_logps": -245.91339111328125,
584
+ "debug/reference_rejected_logps": -262.7462158203125,
585
+ "epoch": 0.7368421052631579,
586
+ "grad_norm": 6.865807807832916,
587
+ "learning_rate": 1e-06,
588
+ "logits/chosen": 1.4645960330963135,
589
+ "logits/rejected": 1.5396175384521484,
590
+ "logps/chosen": -242.5520477294922,
591
+ "logps/rejected": -264.1595458984375,
592
+ "loss": 0.4486,
593
+ "rewards/accuracies": 0.875,
594
+ "rewards/chosen": 0.033613525331020355,
595
+ "rewards/margins": 0.047746483236551285,
596
+ "rewards/rejected": -0.01413295604288578,
597
+ "step": 28
598
+ },
599
+ {
600
+ "debug/policy_chosen_logits": 1.6584521532058716,
601
+ "debug/policy_chosen_logps": -222.9864501953125,
602
+ "debug/policy_rejected_logits": 1.551990032196045,
603
+ "debug/policy_rejected_logps": -224.31863403320312,
604
+ "debug/reference_chosen_logps": -225.62203979492188,
605
+ "debug/reference_rejected_logps": -227.37957763671875,
606
+ "epoch": 0.7631578947368421,
607
+ "grad_norm": 15.12560724905019,
608
+ "learning_rate": 1e-06,
609
+ "logits/chosen": 1.6584521532058716,
610
+ "logits/rejected": 1.551990032196045,
611
+ "logps/chosen": -222.9864501953125,
612
+ "logps/rejected": -224.31863403320312,
613
+ "loss": 0.5096,
614
+ "rewards/accuracies": 0.5,
615
+ "rewards/chosen": 0.026355896145105362,
616
+ "rewards/margins": -0.0042535788379609585,
617
+ "rewards/rejected": 0.03060947358608246,
618
+ "step": 29
619
+ },
620
+ {
621
+ "debug/policy_chosen_logits": 1.8109740018844604,
622
+ "debug/policy_chosen_logps": -227.47349548339844,
623
+ "debug/policy_rejected_logits": 2.0046234130859375,
624
+ "debug/policy_rejected_logps": -261.7398986816406,
625
+ "debug/reference_chosen_logps": -233.89767456054688,
626
+ "debug/reference_rejected_logps": -262.44354248046875,
627
+ "epoch": 0.7894736842105263,
628
+ "grad_norm": 4.507027381203014,
629
+ "learning_rate": 1e-06,
630
+ "logits/chosen": 1.8109740018844604,
631
+ "logits/rejected": 2.0046234130859375,
632
+ "logps/chosen": -227.47349548339844,
633
+ "logps/rejected": -261.7398986816406,
634
+ "loss": 0.4523,
635
+ "rewards/accuracies": 0.75,
636
+ "rewards/chosen": 0.06424173712730408,
637
+ "rewards/margins": 0.05720517784357071,
638
+ "rewards/rejected": 0.007036550901830196,
639
+ "step": 30
640
+ },
641
+ {
642
+ "debug/policy_chosen_logits": 1.5493196249008179,
643
+ "debug/policy_chosen_logps": -274.53570556640625,
644
+ "debug/policy_rejected_logits": 1.4743030071258545,
645
+ "debug/policy_rejected_logps": -228.75210571289062,
646
+ "debug/reference_chosen_logps": -273.8138122558594,
647
+ "debug/reference_rejected_logps": -232.58795166015625,
648
+ "epoch": 0.8157894736842105,
649
+ "grad_norm": 4.855213218926067,
650
+ "learning_rate": 1e-06,
651
+ "logits/chosen": 1.5493196249008179,
652
+ "logits/rejected": 1.4743030071258545,
653
+ "logps/chosen": -274.53570556640625,
654
+ "logps/rejected": -228.75210571289062,
655
+ "loss": 0.4778,
656
+ "rewards/accuracies": 0.5,
657
+ "rewards/chosen": -0.007218838203698397,
658
+ "rewards/margins": -0.045577142387628555,
659
+ "rewards/rejected": 0.038358308374881744,
660
+ "step": 31
661
+ },
662
+ {
663
+ "debug/policy_chosen_logits": 1.7340623140335083,
664
+ "debug/policy_chosen_logps": -245.35580444335938,
665
+ "debug/policy_rejected_logits": 1.691189169883728,
666
+ "debug/policy_rejected_logps": -267.03338623046875,
667
+ "debug/reference_chosen_logps": -249.91921997070312,
668
+ "debug/reference_rejected_logps": -264.17169189453125,
669
+ "epoch": 0.8421052631578947,
670
+ "grad_norm": 5.812254879101492,
671
+ "learning_rate": 1e-06,
672
+ "logits/chosen": 1.7340623140335083,
673
+ "logits/rejected": 1.691189169883728,
674
+ "logps/chosen": -245.35580444335938,
675
+ "logps/rejected": -267.03338623046875,
676
+ "loss": 0.4633,
677
+ "rewards/accuracies": 0.75,
678
+ "rewards/chosen": 0.04563392326235771,
679
+ "rewards/margins": 0.07425111532211304,
680
+ "rewards/rejected": -0.028617190197110176,
681
+ "step": 32
682
+ },
683
+ {
684
+ "debug/policy_chosen_logits": 1.8911277055740356,
685
+ "debug/policy_chosen_logps": -278.0496826171875,
686
+ "debug/policy_rejected_logits": 1.451691746711731,
687
+ "debug/policy_rejected_logps": -334.9330749511719,
688
+ "debug/reference_chosen_logps": -279.70513916015625,
689
+ "debug/reference_rejected_logps": -325.73175048828125,
690
+ "epoch": 0.868421052631579,
691
+ "grad_norm": 10.06399047030507,
692
+ "learning_rate": 1e-06,
693
+ "logits/chosen": 1.8911277055740356,
694
+ "logits/rejected": 1.451691746711731,
695
+ "logps/chosen": -278.0496826171875,
696
+ "logps/rejected": -334.9330749511719,
697
+ "loss": 0.4856,
698
+ "rewards/accuracies": 0.75,
699
+ "rewards/chosen": 0.0165544506162405,
700
+ "rewards/margins": 0.10856754332780838,
701
+ "rewards/rejected": -0.09201309084892273,
702
+ "step": 33
703
+ },
704
+ {
705
+ "debug/policy_chosen_logits": 1.5646555423736572,
706
+ "debug/policy_chosen_logps": -196.27777099609375,
707
+ "debug/policy_rejected_logits": 1.442307710647583,
708
+ "debug/policy_rejected_logps": -303.38470458984375,
709
+ "debug/reference_chosen_logps": -207.76614379882812,
710
+ "debug/reference_rejected_logps": -296.6465759277344,
711
+ "epoch": 0.8947368421052632,
712
+ "grad_norm": 4.5718455604745625,
713
+ "learning_rate": 1e-06,
714
+ "logits/chosen": 1.5646555423736572,
715
+ "logits/rejected": 1.442307710647583,
716
+ "logps/chosen": -196.27777099609375,
717
+ "logps/rejected": -303.38470458984375,
718
+ "loss": 0.4644,
719
+ "rewards/accuracies": 0.875,
720
+ "rewards/chosen": 0.11488360911607742,
721
+ "rewards/margins": 0.18226474523544312,
722
+ "rewards/rejected": -0.0673811286687851,
723
+ "step": 34
724
+ },
725
+ {
726
+ "debug/policy_chosen_logits": 1.9010899066925049,
727
+ "debug/policy_chosen_logps": -238.37425231933594,
728
+ "debug/policy_rejected_logits": 1.8657723665237427,
729
+ "debug/policy_rejected_logps": -261.541015625,
730
+ "debug/reference_chosen_logps": -242.17909240722656,
731
+ "debug/reference_rejected_logps": -260.426513671875,
732
+ "epoch": 0.9210526315789473,
733
+ "grad_norm": 10.019901070632033,
734
+ "learning_rate": 1e-06,
735
+ "logits/chosen": 1.9010899066925049,
736
+ "logits/rejected": 1.8657723665237427,
737
+ "logps/chosen": -238.37425231933594,
738
+ "logps/rejected": -261.541015625,
739
+ "loss": 0.4921,
740
+ "rewards/accuracies": 0.75,
741
+ "rewards/chosen": 0.03804826736450195,
742
+ "rewards/margins": 0.04919330030679703,
743
+ "rewards/rejected": -0.011145040392875671,
744
+ "step": 35
745
+ },
746
+ {
747
+ "debug/policy_chosen_logits": 1.3850834369659424,
748
+ "debug/policy_chosen_logps": -217.4523162841797,
749
+ "debug/policy_rejected_logits": 1.716528296470642,
750
+ "debug/policy_rejected_logps": -294.6310729980469,
751
+ "debug/reference_chosen_logps": -223.49221801757812,
752
+ "debug/reference_rejected_logps": -288.3724365234375,
753
+ "epoch": 0.9473684210526315,
754
+ "grad_norm": 7.371191589299899,
755
+ "learning_rate": 1e-06,
756
+ "logits/chosen": 1.3850834369659424,
757
+ "logits/rejected": 1.716528296470642,
758
+ "logps/chosen": -217.4523162841797,
759
+ "logps/rejected": -294.6310729980469,
760
+ "loss": 0.4547,
761
+ "rewards/accuracies": 0.75,
762
+ "rewards/chosen": 0.06039898097515106,
763
+ "rewards/margins": 0.12298539280891418,
764
+ "rewards/rejected": -0.06258641928434372,
765
+ "step": 36
766
+ },
767
+ {
768
+ "debug/policy_chosen_logits": 1.8597347736358643,
769
+ "debug/policy_chosen_logps": -215.003662109375,
770
+ "debug/policy_rejected_logits": 1.8925625085830688,
771
+ "debug/policy_rejected_logps": -247.18080139160156,
772
+ "debug/reference_chosen_logps": -224.59774780273438,
773
+ "debug/reference_rejected_logps": -249.6659393310547,
774
+ "epoch": 0.9736842105263158,
775
+ "grad_norm": 5.387832295644532,
776
+ "learning_rate": 1e-06,
777
+ "logits/chosen": 1.8597347736358643,
778
+ "logits/rejected": 1.8925625085830688,
779
+ "logps/chosen": -215.003662109375,
780
+ "logps/rejected": -247.18080139160156,
781
+ "loss": 0.4668,
782
+ "rewards/accuracies": 0.75,
783
+ "rewards/chosen": 0.09594079852104187,
784
+ "rewards/margins": 0.0710894912481308,
785
+ "rewards/rejected": 0.024851299822330475,
786
+ "step": 37
787
+ },
788
+ {
789
+ "debug/policy_chosen_logits": 1.3891351222991943,
790
+ "debug/policy_chosen_logps": -240.48060607910156,
791
+ "debug/policy_rejected_logits": 1.173421025276184,
792
+ "debug/policy_rejected_logps": -255.14474487304688,
793
+ "debug/reference_chosen_logps": -245.025390625,
794
+ "debug/reference_rejected_logps": -254.23486328125,
795
+ "epoch": 1.0,
796
+ "grad_norm": 7.832224742495988,
797
+ "learning_rate": 1e-06,
798
+ "logits/chosen": 1.3891351222991943,
799
+ "logits/rejected": 1.173421025276184,
800
+ "logps/chosen": -240.48060607910156,
801
+ "logps/rejected": -255.14474487304688,
802
+ "loss": 0.452,
803
+ "rewards/accuracies": 0.625,
804
+ "rewards/chosen": 0.04544801637530327,
805
+ "rewards/margins": 0.05454694479703903,
806
+ "rewards/rejected": -0.009098930284380913,
807
+ "step": 38
808
+ },
809
+ {
810
+ "epoch": 1.0,
811
+ "step": 38,
812
+ "total_flos": 0.0,
813
+ "train_loss": 0.47894636502391413,
814
+ "train_runtime": 421.3681,
815
+ "train_samples_per_second": 5.644,
816
+ "train_steps_per_second": 0.09
817
+ }
818
+ ],
819
+ "logging_steps": 1,
820
+ "max_steps": 38,
821
+ "num_input_tokens_seen": 0,
822
+ "num_train_epochs": 1,
823
+ "save_steps": 500,
824
+ "stateful_callbacks": {
825
+ "TrainerControl": {
826
+ "args": {
827
+ "should_epoch_stop": false,
828
+ "should_evaluate": false,
829
+ "should_log": false,
830
+ "should_save": true,
831
+ "should_training_stop": true
832
+ },
833
+ "attributes": {}
834
+ }
835
+ },
836
+ "total_flos": 0.0,
837
+ "train_batch_size": 8,
838
+ "trial_name": null,
839
+ "trial_params": null
840
+ }