Hugging Face's logo

taicheng
/
zephyr-7b-align-scan-0.0-0.3-polynomial-3

Text Generation
Transformers
TensorBoard
Safetensors
mistral
alignment-handbook
trl
dpo
Generated from Trainer
conversational
text-generation-inference
Inference Endpoints
Model card Files Metrics Community
taicheng commited on
Commit
e69a6d6
•
1 Parent(s): e97779d

Model save

Browse files
Files changed (9) hide show
  1. README.md +75 -0
  2. all_results.json +9 -0
  3. generation_config.json +6 -0
  4. model-00001-of-00003.safetensors +1 -1
  5. model-00002-of-00003.safetensors +1 -1
  6. model-00003-of-00003.safetensors +1 -1
  7. runs/Sep24_08-15-53_qa-a100-004.crc.nd.edu/events.out.tfevents.1727180221.qa-a100-004.crc.nd.edu.1794966.0 +2 -2
  8. train_results.json +9 -0
  9. trainer_state.json +509 -0
README.md ADDED
@@ -0,0 +1,75 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: transformers
3
+ license: apache-2.0
4
+ base_model: alignment-handbook/zephyr-7b-sft-full
5
+ tags:
6
+ - trl
7
+ - dpo
8
+ - generated_from_trainer
9
+ model-index:
10
+ - name: zephyr-7b-align-scan-0.0-0.3-polynomial-3
11
+ results: []
12
+ ---
13
+
14
+ <!-- This model card has been generated automatically according to the information the Trainer had access to. You
15
+ should probably proofread and complete it, then remove this comment. -->
16
+
17
+ # zephyr-7b-align-scan-0.0-0.3-polynomial-3
18
+
19
+ This model is a fine-tuned version of [alignment-handbook/zephyr-7b-sft-full](https://huggingface.co/alignment-handbook/zephyr-7b-sft-full) on an unknown dataset.
20
+ It achieves the following results on the evaluation set:
21
+ - Loss: 0.6854
22
+ - Rewards/chosen: -0.2626
23
+ - Rewards/rejected: -0.8523
24
+ - Rewards/accuracies: 0.3254
25
+ - Rewards/margins: 0.5897
26
+ - Logps/rejected: -84.2922
27
+ - Logps/chosen: -75.4660
28
+ - Logits/rejected: -2.5243
29
+ - Logits/chosen: -2.5414
30
+
31
+ ## Model description
32
+
33
+ More information needed
34
+
35
+ ## Intended uses & limitations
36
+
37
+ More information needed
38
+
39
+ ## Training and evaluation data
40
+
41
+ More information needed
42
+
43
+ ## Training procedure
44
+
45
+ ### Training hyperparameters
46
+
47
+ The following hyperparameters were used during training:
48
+ - learning_rate: 2.9336856595926955e-07
49
+ - train_batch_size: 8
50
+ - eval_batch_size: 8
51
+ - seed: 42
52
+ - distributed_type: multi-GPU
53
+ - num_devices: 4
54
+ - gradient_accumulation_steps: 2
55
+ - total_train_batch_size: 64
56
+ - total_eval_batch_size: 32
57
+ - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
58
+ - lr_scheduler_type: polynomial
59
+ - lr_scheduler_warmup_ratio: 0.1
60
+ - num_epochs: 3
61
+
62
+ ### Training results
63
+
64
+ | Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
65
+ |:-------------:|:------:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
66
+ | 0.529 | 1.0417 | 100 | 0.6391 | 0.3319 | 0.0135 | 0.3353 | 0.3184 | -81.0782 | -73.2591 | -2.5045 | -2.5211 |
67
+ | 0.3646 | 2.0833 | 200 | 0.6854 | -0.2626 | -0.8523 | 0.3254 | 0.5897 | -84.2922 | -75.4660 | -2.5243 | -2.5414 |
68
+
69
+
70
+ ### Framework versions
71
+
72
+ - Transformers 4.44.2
73
+ - Pytorch 2.4.0
74
+ - Datasets 2.21.0
75
+ - Tokenizers 0.19.1
all_results.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 3.0,
3
+ "total_flos": 0.0,
4
+ "train_loss": 0.4762688593731986,
5
+ "train_runtime": 3179.8752,
6
+ "train_samples": 6113,
7
+ "train_samples_per_second": 5.767,
8
+ "train_steps_per_second": 0.091
9
+ }
generation_config.json ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 1,
4
+ "eos_token_id": 2,
5
+ "transformers_version": "4.44.2"
6
+ }
model-00001-of-00003.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:2a5b7df26918a6b9440d1c80ff4c9525c805890a6b16658b6e68631d69242c7c
3
  size 4943162336
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b70bf9cbd611098eadb5f3a595cf7c11629cb8cc53ad796e0957d17b7a84bd29
3
  size 4943162336
model-00002-of-00003.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:5835055283a7d8ece38fa9f52108b6ef58386cbeeb1e6a6d707927b3f726605d
3
  size 4999819336
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c52a69a032209cb1889fa6ca29590babfc53ad04c3aa47386db9f16950cfe1b3
3
  size 4999819336
model-00003-of-00003.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:77a96e35b4d72091c21721f64cb6a8f4065b2271b1f00296be8e56d02956dae8
3
  size 4540516344
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9109f04f39754d0828524015feabe3fd69f924ce3771d3750b9e43202a1f232f
3
  size 4540516344
runs/Sep24_08-15-53_qa-a100-004.crc.nd.edu/events.out.tfevents.1727180221.qa-a100-004.crc.nd.edu.1794966.0 CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:ab014ba501ac68c2144897ab4256a7ea233b05f35cb0b305bb211a6cddd5619b
3
- size 14637
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:01543967337b490674358b6139645cee37084efa2a399303ecd66ccc17717cae
3
+ size 28091
train_results.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 3.0,
3
+ "total_flos": 0.0,
4
+ "train_loss": 0.4762688593731986,
5
+ "train_runtime": 3179.8752,
6
+ "train_samples": 6113,
7
+ "train_samples_per_second": 5.767,
8
+ "train_steps_per_second": 0.091
9
+ }
trainer_state.json ADDED
@@ -0,0 +1,509 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 3.0,
5
+ "eval_steps": 100,
6
+ "global_step": 288,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.010416666666666666,
13
+ "grad_norm": 177.44975633333866,
14
+ "learning_rate": 1.0116157446871364e-08,
15
+ "logits/chosen": -2.590585231781006,
16
+ "logits/rejected": -2.5664222240448,
17
+ "logps/chosen": -80.29847717285156,
18
+ "logps/rejected": -53.10200881958008,
19
+ "loss": 0.6931,
20
+ "rewards/accuracies": 0.0,
21
+ "rewards/chosen": 0.0,
22
+ "rewards/margins": 0.0,
23
+ "rewards/rejected": 0.0,
24
+ "step": 1
25
+ },
26
+ {
27
+ "epoch": 0.10416666666666667,
28
+ "grad_norm": 157.38886271050058,
29
+ "learning_rate": 1.0116157446871364e-07,
30
+ "logits/chosen": -2.556668758392334,
31
+ "logits/rejected": -2.5388104915618896,
32
+ "logps/chosen": -87.92106628417969,
33
+ "logps/rejected": -81.02932739257812,
34
+ "loss": 0.6933,
35
+ "rewards/accuracies": 0.25,
36
+ "rewards/chosen": -0.0015910979127511382,
37
+ "rewards/margins": -0.004063541069626808,
38
+ "rewards/rejected": 0.002472442574799061,
39
+ "step": 10
40
+ },
41
+ {
42
+ "epoch": 0.20833333333333334,
43
+ "grad_norm": 138.46519913851827,
44
+ "learning_rate": 2.0232314893742728e-07,
45
+ "logits/chosen": -2.614779233932495,
46
+ "logits/rejected": -2.566148042678833,
47
+ "logps/chosen": -103.30540466308594,
48
+ "logps/rejected": -89.8634262084961,
49
+ "loss": 0.6884,
50
+ "rewards/accuracies": 0.3125,
51
+ "rewards/chosen": 0.012269504368305206,
52
+ "rewards/margins": 0.006630963645875454,
53
+ "rewards/rejected": 0.005638539791107178,
54
+ "step": 20
55
+ },
56
+ {
57
+ "epoch": 0.3125,
58
+ "grad_norm": 182.67677013741292,
59
+ "learning_rate": 2.92621969179504e-07,
60
+ "logits/chosen": -2.5195202827453613,
61
+ "logits/rejected": -2.5337605476379395,
62
+ "logps/chosen": -67.20966339111328,
63
+ "logps/rejected": -74.8441162109375,
64
+ "loss": 0.68,
65
+ "rewards/accuracies": 0.3125,
66
+ "rewards/chosen": 0.06131690740585327,
67
+ "rewards/margins": 0.03928915411233902,
68
+ "rewards/rejected": 0.0220277551561594,
69
+ "step": 30
70
+ },
71
+ {
72
+ "epoch": 0.4166666666666667,
73
+ "grad_norm": 122.9129643173066,
74
+ "learning_rate": 2.851560013818488e-07,
75
+ "logits/chosen": -2.575892448425293,
76
+ "logits/rejected": -2.563741683959961,
77
+ "logps/chosen": -71.55787658691406,
78
+ "logps/rejected": -70.44486236572266,
79
+ "loss": 0.6694,
80
+ "rewards/accuracies": 0.26875001192092896,
81
+ "rewards/chosen": 0.15858440101146698,
82
+ "rewards/margins": 0.08626840263605118,
83
+ "rewards/rejected": 0.0723160058259964,
84
+ "step": 40
85
+ },
86
+ {
87
+ "epoch": 0.5208333333333334,
88
+ "grad_norm": 125.21896044373509,
89
+ "learning_rate": 2.776900335841936e-07,
90
+ "logits/chosen": -2.4858391284942627,
91
+ "logits/rejected": -2.4980547428131104,
92
+ "logps/chosen": -50.068031311035156,
93
+ "logps/rejected": -58.41581344604492,
94
+ "loss": 0.6626,
95
+ "rewards/accuracies": 0.23125000298023224,
96
+ "rewards/chosen": 0.22443357110023499,
97
+ "rewards/margins": 0.09153502434492111,
98
+ "rewards/rejected": 0.13289853930473328,
99
+ "step": 50
100
+ },
101
+ {
102
+ "epoch": 0.625,
103
+ "grad_norm": 129.16794274257964,
104
+ "learning_rate": 2.702240657865384e-07,
105
+ "logits/chosen": -2.5691611766815186,
106
+ "logits/rejected": -2.550184726715088,
107
+ "logps/chosen": -77.5196762084961,
108
+ "logps/rejected": -78.63575744628906,
109
+ "loss": 0.651,
110
+ "rewards/accuracies": 0.33125001192092896,
111
+ "rewards/chosen": 0.4174376130104065,
112
+ "rewards/margins": 0.22838692367076874,
113
+ "rewards/rejected": 0.18905068933963776,
114
+ "step": 60
115
+ },
116
+ {
117
+ "epoch": 0.7291666666666666,
118
+ "grad_norm": 127.48114805532535,
119
+ "learning_rate": 2.627580979888832e-07,
120
+ "logits/chosen": -2.5489492416381836,
121
+ "logits/rejected": -2.539600133895874,
122
+ "logps/chosen": -94.18212890625,
123
+ "logps/rejected": -84.33738708496094,
124
+ "loss": 0.6407,
125
+ "rewards/accuracies": 0.3375000059604645,
126
+ "rewards/chosen": 0.5826455354690552,
127
+ "rewards/margins": 0.28286492824554443,
128
+ "rewards/rejected": 0.2997806966304779,
129
+ "step": 70
130
+ },
131
+ {
132
+ "epoch": 0.8333333333333334,
133
+ "grad_norm": 204.05352307435749,
134
+ "learning_rate": 2.55292130191228e-07,
135
+ "logits/chosen": -2.549272298812866,
136
+ "logits/rejected": -2.494861602783203,
137
+ "logps/chosen": -83.9472427368164,
138
+ "logps/rejected": -79.15623474121094,
139
+ "loss": 0.6197,
140
+ "rewards/accuracies": 0.41874998807907104,
141
+ "rewards/chosen": 0.45742616057395935,
142
+ "rewards/margins": 0.39647966623306274,
143
+ "rewards/rejected": 0.06094648689031601,
144
+ "step": 80
145
+ },
146
+ {
147
+ "epoch": 0.9375,
148
+ "grad_norm": 177.818739531135,
149
+ "learning_rate": 2.4782616239357287e-07,
150
+ "logits/chosen": -2.481269359588623,
151
+ "logits/rejected": -2.476210355758667,
152
+ "logps/chosen": -53.21185302734375,
153
+ "logps/rejected": -63.6622200012207,
154
+ "loss": 0.6402,
155
+ "rewards/accuracies": 0.28125,
156
+ "rewards/chosen": 0.20689637959003448,
157
+ "rewards/margins": 0.25917547941207886,
158
+ "rewards/rejected": -0.05227910727262497,
159
+ "step": 90
160
+ },
161
+ {
162
+ "epoch": 1.0416666666666667,
163
+ "grad_norm": 42.3875191112844,
164
+ "learning_rate": 2.403601945959176e-07,
165
+ "logits/chosen": -2.4865407943725586,
166
+ "logits/rejected": -2.4648594856262207,
167
+ "logps/chosen": -71.56224060058594,
168
+ "logps/rejected": -73.2033920288086,
169
+ "loss": 0.529,
170
+ "rewards/accuracies": 0.40625,
171
+ "rewards/chosen": 0.5279342532157898,
172
+ "rewards/margins": 0.8453754186630249,
173
+ "rewards/rejected": -0.3174411952495575,
174
+ "step": 100
175
+ },
176
+ {
177
+ "epoch": 1.0416666666666667,
178
+ "eval_logits/chosen": -2.5210909843444824,
179
+ "eval_logits/rejected": -2.504479169845581,
180
+ "eval_logps/chosen": -73.25910949707031,
181
+ "eval_logps/rejected": -81.07815551757812,
182
+ "eval_loss": 0.6391177773475647,
183
+ "eval_rewards/accuracies": 0.335317462682724,
184
+ "eval_rewards/chosen": 0.33193930983543396,
185
+ "eval_rewards/margins": 0.3184153139591217,
186
+ "eval_rewards/rejected": 0.01352396234869957,
187
+ "eval_runtime": 113.5699,
188
+ "eval_samples_per_second": 17.61,
189
+ "eval_steps_per_second": 0.555,
190
+ "step": 100
191
+ },
192
+ {
193
+ "epoch": 1.1458333333333333,
194
+ "grad_norm": 32.2449798694013,
195
+ "learning_rate": 2.3289422679826245e-07,
196
+ "logits/chosen": -2.494868516921997,
197
+ "logits/rejected": -2.528102159500122,
198
+ "logps/chosen": -61.715614318847656,
199
+ "logps/rejected": -84.56453704833984,
200
+ "loss": 0.4071,
201
+ "rewards/accuracies": 0.46875,
202
+ "rewards/chosen": 0.9841395616531372,
203
+ "rewards/margins": 1.814012885093689,
204
+ "rewards/rejected": -0.8298734426498413,
205
+ "step": 110
206
+ },
207
+ {
208
+ "epoch": 1.25,
209
+ "grad_norm": 34.815563904415676,
210
+ "learning_rate": 2.2542825900060728e-07,
211
+ "logits/chosen": -2.561702251434326,
212
+ "logits/rejected": -2.5297372341156006,
213
+ "logps/chosen": -103.7978744506836,
214
+ "logps/rejected": -98.90431213378906,
215
+ "loss": 0.4202,
216
+ "rewards/accuracies": 0.581250011920929,
217
+ "rewards/chosen": 1.1336109638214111,
218
+ "rewards/margins": 2.633657932281494,
219
+ "rewards/rejected": -1.5000473260879517,
220
+ "step": 120
221
+ },
222
+ {
223
+ "epoch": 1.3541666666666667,
224
+ "grad_norm": 36.65231392815614,
225
+ "learning_rate": 2.1796229120295209e-07,
226
+ "logits/chosen": -2.4902865886688232,
227
+ "logits/rejected": -2.4940361976623535,
228
+ "logps/chosen": -85.04226684570312,
229
+ "logps/rejected": -96.18626403808594,
230
+ "loss": 0.4074,
231
+ "rewards/accuracies": 0.4625000059604645,
232
+ "rewards/chosen": 0.41049623489379883,
233
+ "rewards/margins": 2.2558376789093018,
234
+ "rewards/rejected": -1.8453413248062134,
235
+ "step": 130
236
+ },
237
+ {
238
+ "epoch": 1.4583333333333333,
239
+ "grad_norm": 77.16009856421289,
240
+ "learning_rate": 2.104963234052969e-07,
241
+ "logits/chosen": -2.4936976432800293,
242
+ "logits/rejected": -2.509275197982788,
243
+ "logps/chosen": -47.730525970458984,
244
+ "logps/rejected": -70.30806732177734,
245
+ "loss": 0.414,
246
+ "rewards/accuracies": 0.4124999940395355,
247
+ "rewards/chosen": 0.41952085494995117,
248
+ "rewards/margins": 1.8167709112167358,
249
+ "rewards/rejected": -1.3972498178482056,
250
+ "step": 140
251
+ },
252
+ {
253
+ "epoch": 1.5625,
254
+ "grad_norm": 26.196738217281826,
255
+ "learning_rate": 2.030303556076417e-07,
256
+ "logits/chosen": -2.5342118740081787,
257
+ "logits/rejected": -2.5107994079589844,
258
+ "logps/chosen": -68.2138671875,
259
+ "logps/rejected": -77.74583435058594,
260
+ "loss": 0.4107,
261
+ "rewards/accuracies": 0.4625000059604645,
262
+ "rewards/chosen": 1.0598348379135132,
263
+ "rewards/margins": 2.18320894241333,
264
+ "rewards/rejected": -1.1233742237091064,
265
+ "step": 150
266
+ },
267
+ {
268
+ "epoch": 1.6666666666666665,
269
+ "grad_norm": 17.275725311131126,
270
+ "learning_rate": 1.955643878099865e-07,
271
+ "logits/chosen": -2.5043556690216064,
272
+ "logits/rejected": -2.513214588165283,
273
+ "logps/chosen": -74.29962921142578,
274
+ "logps/rejected": -91.91645050048828,
275
+ "loss": 0.399,
276
+ "rewards/accuracies": 0.5375000238418579,
277
+ "rewards/chosen": 1.7849496603012085,
278
+ "rewards/margins": 2.8569164276123047,
279
+ "rewards/rejected": -1.0719670057296753,
280
+ "step": 160
281
+ },
282
+ {
283
+ "epoch": 1.7708333333333335,
284
+ "grad_norm": 41.27028375416643,
285
+ "learning_rate": 1.880984200123313e-07,
286
+ "logits/chosen": -2.47822904586792,
287
+ "logits/rejected": -2.469449281692505,
288
+ "logps/chosen": -68.61357879638672,
289
+ "logps/rejected": -85.9351806640625,
290
+ "loss": 0.4081,
291
+ "rewards/accuracies": 0.5249999761581421,
292
+ "rewards/chosen": 1.7798223495483398,
293
+ "rewards/margins": 2.5742733478546143,
294
+ "rewards/rejected": -0.794451117515564,
295
+ "step": 170
296
+ },
297
+ {
298
+ "epoch": 1.875,
299
+ "grad_norm": 59.43887874577346,
300
+ "learning_rate": 1.806324522146761e-07,
301
+ "logits/chosen": -2.500208616256714,
302
+ "logits/rejected": -2.481628656387329,
303
+ "logps/chosen": -68.31395721435547,
304
+ "logps/rejected": -78.98301696777344,
305
+ "loss": 0.3901,
306
+ "rewards/accuracies": 0.48124998807907104,
307
+ "rewards/chosen": 0.9541478157043457,
308
+ "rewards/margins": 2.4010345935821533,
309
+ "rewards/rejected": -1.4468867778778076,
310
+ "step": 180
311
+ },
312
+ {
313
+ "epoch": 1.9791666666666665,
314
+ "grad_norm": 36.95333305414057,
315
+ "learning_rate": 1.731664844170209e-07,
316
+ "logits/chosen": -2.5415730476379395,
317
+ "logits/rejected": -2.540619373321533,
318
+ "logps/chosen": -85.87374114990234,
319
+ "logps/rejected": -99.43651580810547,
320
+ "loss": 0.3978,
321
+ "rewards/accuracies": 0.53125,
322
+ "rewards/chosen": 0.6972099542617798,
323
+ "rewards/margins": 3.0720162391662598,
324
+ "rewards/rejected": -2.3748061656951904,
325
+ "step": 190
326
+ },
327
+ {
328
+ "epoch": 2.0833333333333335,
329
+ "grad_norm": 1.2362916215000161,
330
+ "learning_rate": 1.657005166193657e-07,
331
+ "logits/chosen": -2.4930806159973145,
332
+ "logits/rejected": -2.5152084827423096,
333
+ "logps/chosen": -85.92092895507812,
334
+ "logps/rejected": -112.84880065917969,
335
+ "loss": 0.3646,
336
+ "rewards/accuracies": 0.5375000238418579,
337
+ "rewards/chosen": 0.8306751251220703,
338
+ "rewards/margins": 3.8782944679260254,
339
+ "rewards/rejected": -3.047619342803955,
340
+ "step": 200
341
+ },
342
+ {
343
+ "epoch": 2.0833333333333335,
344
+ "eval_logits/chosen": -2.541405439376831,
345
+ "eval_logits/rejected": -2.5242767333984375,
346
+ "eval_logps/chosen": -75.4660415649414,
347
+ "eval_logps/rejected": -84.29215240478516,
348
+ "eval_loss": 0.6853565573692322,
349
+ "eval_rewards/accuracies": 0.3253968358039856,
350
+ "eval_rewards/chosen": -0.2626197934150696,
351
+ "eval_rewards/margins": 0.589722752571106,
352
+ "eval_rewards/rejected": -0.8523425459861755,
353
+ "eval_runtime": 113.3948,
354
+ "eval_samples_per_second": 17.637,
355
+ "eval_steps_per_second": 0.556,
356
+ "step": 200
357
+ },
358
+ {
359
+ "epoch": 2.1875,
360
+ "grad_norm": 2.0446238829126155,
361
+ "learning_rate": 1.5823454882171052e-07,
362
+ "logits/chosen": -2.539818286895752,
363
+ "logits/rejected": -2.533048391342163,
364
+ "logps/chosen": -59.99456787109375,
365
+ "logps/rejected": -76.70134735107422,
366
+ "loss": 0.3844,
367
+ "rewards/accuracies": 0.42500001192092896,
368
+ "rewards/chosen": 0.9697097539901733,
369
+ "rewards/margins": 3.1277875900268555,
370
+ "rewards/rejected": -2.1580779552459717,
371
+ "step": 210
372
+ },
373
+ {
374
+ "epoch": 2.2916666666666665,
375
+ "grad_norm": 1.7408027431292374,
376
+ "learning_rate": 1.5076858102405535e-07,
377
+ "logits/chosen": -2.5433623790740967,
378
+ "logits/rejected": -2.4930663108825684,
379
+ "logps/chosen": -97.69389343261719,
380
+ "logps/rejected": -106.0047607421875,
381
+ "loss": 0.332,
382
+ "rewards/accuracies": 0.574999988079071,
383
+ "rewards/chosen": 1.1734516620635986,
384
+ "rewards/margins": 4.781111240386963,
385
+ "rewards/rejected": -3.607659101486206,
386
+ "step": 220
387
+ },
388
+ {
389
+ "epoch": 2.3958333333333335,
390
+ "grad_norm": 1.4060304822058385,
391
+ "learning_rate": 1.4330261322640012e-07,
392
+ "logits/chosen": -2.526247978210449,
393
+ "logits/rejected": -2.49235200881958,
394
+ "logps/chosen": -56.86894607543945,
395
+ "logps/rejected": -65.83305358886719,
396
+ "loss": 0.3811,
397
+ "rewards/accuracies": 0.4312500059604645,
398
+ "rewards/chosen": 0.7982641458511353,
399
+ "rewards/margins": 3.461902618408203,
400
+ "rewards/rejected": -2.663638114929199,
401
+ "step": 230
402
+ },
403
+ {
404
+ "epoch": 2.5,
405
+ "grad_norm": 2.2210458558484496,
406
+ "learning_rate": 1.3583664542874493e-07,
407
+ "logits/chosen": -2.505704164505005,
408
+ "logits/rejected": -2.455660104751587,
409
+ "logps/chosen": -83.47601318359375,
410
+ "logps/rejected": -90.42607116699219,
411
+ "loss": 0.3758,
412
+ "rewards/accuracies": 0.5,
413
+ "rewards/chosen": 1.1906569004058838,
414
+ "rewards/margins": 4.154426574707031,
415
+ "rewards/rejected": -2.9637694358825684,
416
+ "step": 240
417
+ },
418
+ {
419
+ "epoch": 2.6041666666666665,
420
+ "grad_norm": 8.8054104852066,
421
+ "learning_rate": 1.2837067763108973e-07,
422
+ "logits/chosen": -2.450004816055298,
423
+ "logits/rejected": -2.437671184539795,
424
+ "logps/chosen": -72.790771484375,
425
+ "logps/rejected": -84.4940185546875,
426
+ "loss": 0.3479,
427
+ "rewards/accuracies": 0.48750001192092896,
428
+ "rewards/chosen": -0.03280387073755264,
429
+ "rewards/margins": 3.7539596557617188,
430
+ "rewards/rejected": -3.7867636680603027,
431
+ "step": 250
432
+ },
433
+ {
434
+ "epoch": 2.7083333333333335,
435
+ "grad_norm": 1.6270602506146907,
436
+ "learning_rate": 1.2090470983343453e-07,
437
+ "logits/chosen": -2.443758726119995,
438
+ "logits/rejected": -2.461164712905884,
439
+ "logps/chosen": -55.83058547973633,
440
+ "logps/rejected": -86.06732940673828,
441
+ "loss": 0.3802,
442
+ "rewards/accuracies": 0.44999998807907104,
443
+ "rewards/chosen": 0.6197060346603394,
444
+ "rewards/margins": 4.14993143081665,
445
+ "rewards/rejected": -3.5302250385284424,
446
+ "step": 260
447
+ },
448
+ {
449
+ "epoch": 2.8125,
450
+ "grad_norm": 6.344724508571099,
451
+ "learning_rate": 1.1343874203577934e-07,
452
+ "logits/chosen": -2.443955659866333,
453
+ "logits/rejected": -2.4808030128479004,
454
+ "logps/chosen": -74.16885375976562,
455
+ "logps/rejected": -91.22604370117188,
456
+ "loss": 0.3551,
457
+ "rewards/accuracies": 0.48750001192092896,
458
+ "rewards/chosen": 0.6749483346939087,
459
+ "rewards/margins": 4.160412788391113,
460
+ "rewards/rejected": -3.4854648113250732,
461
+ "step": 270
462
+ },
463
+ {
464
+ "epoch": 2.9166666666666665,
465
+ "grad_norm": 10.707111276605081,
466
+ "learning_rate": 1.0597277423812416e-07,
467
+ "logits/chosen": -2.5015740394592285,
468
+ "logits/rejected": -2.465491771697998,
469
+ "logps/chosen": -81.94514465332031,
470
+ "logps/rejected": -95.30039978027344,
471
+ "loss": 0.3654,
472
+ "rewards/accuracies": 0.48750001192092896,
473
+ "rewards/chosen": 0.9193007349967957,
474
+ "rewards/margins": 4.401783466339111,
475
+ "rewards/rejected": -3.482482433319092,
476
+ "step": 280
477
+ },
478
+ {
479
+ "epoch": 3.0,
480
+ "step": 288,
481
+ "total_flos": 0.0,
482
+ "train_loss": 0.4762688593731986,
483
+ "train_runtime": 3179.8752,
484
+ "train_samples_per_second": 5.767,
485
+ "train_steps_per_second": 0.091
486
+ }
487
+ ],
488
+ "logging_steps": 10,
489
+ "max_steps": 288,
490
+ "num_input_tokens_seen": 0,
491
+ "num_train_epochs": 3,
492
+ "save_steps": 100,
493
+ "stateful_callbacks": {
494
+ "TrainerControl": {
495
+ "args": {
496
+ "should_epoch_stop": false,
497
+ "should_evaluate": false,
498
+ "should_log": false,
499
+ "should_save": true,
500
+ "should_training_stop": true
501
+ },
502
+ "attributes": {}
503
+ }
504
+ },
505
+ "total_flos": 0.0,
506
+ "train_batch_size": 8,
507
+ "trial_name": null,
508
+ "trial_params": null
509
+ }