Hugging Face's logo

YYYYYYibo
/
vanilla_doff_iter_3

PEFT
Safetensors
mistral
alignment-handbook
Generated from Trainer
trl
dpo
Model card Files Community
YYYYYYibo commited on
Commit
a983845
·
verified ·
1 Parent(s): 26cf893

Model save

Browse files
Files changed (4) hide show
  1. README.md +75 -0
  2. all_results.json +8 -0
  3. train_results.json +8 -0
  4. trainer_state.json +284 -0
README.md ADDED
@@ -0,0 +1,75 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: apache-2.0
3
+ library_name: peft
4
+ tags:
5
+ - trl
6
+ - dpo
7
+ - generated_from_trainer
8
+ base_model: alignment-handbook/zephyr-7b-sft-full
9
+ model-index:
10
+ - name: vanilla_doff_iter_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
+ # vanilla_doff_iter_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 the None dataset.
20
+ It achieves the following results on the evaluation set:
21
+ - Loss: 0.6558
22
+ - Rewards/chosen: 0.0487
23
+ - Rewards/rejected: -0.0535
24
+ - Rewards/accuracies: 0.5980
25
+ - Rewards/margins: 0.1022
26
+ - Logps/rejected: -292.4649
27
+ - Logps/chosen: -305.4743
28
+ - Logits/rejected: -2.5270
29
+ - Logits/chosen: -2.6306
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: 5e-06
49
+ - train_batch_size: 2
50
+ - eval_batch_size: 2
51
+ - seed: 42
52
+ - distributed_type: multi-GPU
53
+ - num_devices: 4
54
+ - gradient_accumulation_steps: 16
55
+ - total_train_batch_size: 128
56
+ - total_eval_batch_size: 8
57
+ - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
58
+ - lr_scheduler_type: cosine
59
+ - lr_scheduler_warmup_ratio: 0.1
60
+ - num_epochs: 1
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
+ | 0.6561 | 0.62 | 100 | 0.6558 | 0.0487 | -0.0535 | 0.5980 | 0.1022 | -292.4649 | -305.4743 | -2.5270 | -2.6306 |
67
+
68
+
69
+ ### Framework versions
70
+
71
+ - PEFT 0.7.1
72
+ - Transformers 4.36.2
73
+ - Pytorch 2.3.0+cu121
74
+ - Datasets 2.14.6
75
+ - Tokenizers 0.15.2
all_results.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 1.0,
3
+ "train_loss": 0.6587186426292231,
4
+ "train_runtime": 6701.6263,
5
+ "train_samples": 20735,
6
+ "train_samples_per_second": 3.094,
7
+ "train_steps_per_second": 0.024
8
+ }
train_results.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 1.0,
3
+ "train_loss": 0.6587186426292231,
4
+ "train_runtime": 6701.6263,
5
+ "train_samples": 20735,
6
+ "train_samples_per_second": 3.094,
7
+ "train_steps_per_second": 0.024
8
+ }
trainer_state.json ADDED
@@ -0,0 +1,284 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 1.0,
5
+ "eval_steps": 100,
6
+ "global_step": 162,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.01,
13
+ "learning_rate": 2.9411764705882356e-07,
14
+ "logits/chosen": -2.7916789054870605,
15
+ "logits/rejected": -2.7116382122039795,
16
+ "logps/chosen": -323.9847717285156,
17
+ "logps/rejected": -313.6904296875,
18
+ "loss": 0.6931,
19
+ "rewards/accuracies": 0.0,
20
+ "rewards/chosen": 0.0,
21
+ "rewards/margins": 0.0,
22
+ "rewards/rejected": 0.0,
23
+ "step": 1
24
+ },
25
+ {
26
+ "epoch": 0.06,
27
+ "learning_rate": 2.9411764705882355e-06,
28
+ "logits/chosen": -2.673537492752075,
29
+ "logits/rejected": -2.636620044708252,
30
+ "logps/chosen": -280.41143798828125,
31
+ "logps/rejected": -293.9028625488281,
32
+ "loss": 0.6926,
33
+ "rewards/accuracies": 0.4305555522441864,
34
+ "rewards/chosen": 0.0071474602445960045,
35
+ "rewards/margins": 0.0006154652219265699,
36
+ "rewards/rejected": 0.006531994789838791,
37
+ "step": 10
38
+ },
39
+ {
40
+ "epoch": 0.12,
41
+ "learning_rate": 4.994720857837211e-06,
42
+ "logits/chosen": -2.6557788848876953,
43
+ "logits/rejected": -2.6616427898406982,
44
+ "logps/chosen": -242.90036010742188,
45
+ "logps/rejected": -262.7171630859375,
46
+ "loss": 0.6876,
47
+ "rewards/accuracies": 0.5687500238418579,
48
+ "rewards/chosen": 0.17924346029758453,
49
+ "rewards/margins": 0.011152910068631172,
50
+ "rewards/rejected": 0.1680905520915985,
51
+ "step": 20
52
+ },
53
+ {
54
+ "epoch": 0.19,
55
+ "learning_rate": 4.901488388458247e-06,
56
+ "logits/chosen": -2.6518654823303223,
57
+ "logits/rejected": -2.661100387573242,
58
+ "logps/chosen": -248.0233612060547,
59
+ "logps/rejected": -281.92254638671875,
60
+ "loss": 0.6776,
61
+ "rewards/accuracies": 0.640625,
62
+ "rewards/chosen": 0.21578983962535858,
63
+ "rewards/margins": 0.03539184108376503,
64
+ "rewards/rejected": 0.18039801716804504,
65
+ "step": 30
66
+ },
67
+ {
68
+ "epoch": 0.25,
69
+ "learning_rate": 4.6959649910976165e-06,
70
+ "logits/chosen": -2.6537535190582275,
71
+ "logits/rejected": -2.5938658714294434,
72
+ "logps/chosen": -256.3780212402344,
73
+ "logps/rejected": -289.9882507324219,
74
+ "loss": 0.6675,
75
+ "rewards/accuracies": 0.625,
76
+ "rewards/chosen": 0.15942983329296112,
77
+ "rewards/margins": 0.06988275796175003,
78
+ "rewards/rejected": 0.08954708278179169,
79
+ "step": 40
80
+ },
81
+ {
82
+ "epoch": 0.31,
83
+ "learning_rate": 4.387760711393052e-06,
84
+ "logits/chosen": -2.5633559226989746,
85
+ "logits/rejected": -2.5739550590515137,
86
+ "logps/chosen": -240.4188232421875,
87
+ "logps/rejected": -295.7885437011719,
88
+ "loss": 0.6634,
89
+ "rewards/accuracies": 0.6156250238418579,
90
+ "rewards/chosen": 0.020841460675001144,
91
+ "rewards/margins": 0.06278014928102493,
92
+ "rewards/rejected": -0.04193869233131409,
93
+ "step": 50
94
+ },
95
+ {
96
+ "epoch": 0.37,
97
+ "learning_rate": 3.991286838919086e-06,
98
+ "logits/chosen": -2.5727829933166504,
99
+ "logits/rejected": -2.5552239418029785,
100
+ "logps/chosen": -271.25531005859375,
101
+ "logps/rejected": -284.837646484375,
102
+ "loss": 0.6597,
103
+ "rewards/accuracies": 0.596875011920929,
104
+ "rewards/chosen": 0.10798764228820801,
105
+ "rewards/margins": 0.07805077731609344,
106
+ "rewards/rejected": 0.029936855658888817,
107
+ "step": 60
108
+ },
109
+ {
110
+ "epoch": 0.43,
111
+ "learning_rate": 3.5250820513035403e-06,
112
+ "logits/chosen": -2.5731635093688965,
113
+ "logits/rejected": -2.561147451400757,
114
+ "logps/chosen": -277.0789794921875,
115
+ "logps/rejected": -301.56134033203125,
116
+ "loss": 0.6577,
117
+ "rewards/accuracies": 0.625,
118
+ "rewards/chosen": -0.024448225274682045,
119
+ "rewards/margins": 0.09842626750469208,
120
+ "rewards/rejected": -0.12287448346614838,
121
+ "step": 70
122
+ },
123
+ {
124
+ "epoch": 0.49,
125
+ "learning_rate": 3.0109455662659126e-06,
126
+ "logits/chosen": -2.563314914703369,
127
+ "logits/rejected": -2.529156446456909,
128
+ "logps/chosen": -258.3672790527344,
129
+ "logps/rejected": -269.1060791015625,
130
+ "loss": 0.6578,
131
+ "rewards/accuracies": 0.625,
132
+ "rewards/chosen": 0.09700657427310944,
133
+ "rewards/margins": 0.09656216204166412,
134
+ "rewards/rejected": 0.0004444071964826435,
135
+ "step": 80
136
+ },
137
+ {
138
+ "epoch": 0.56,
139
+ "learning_rate": 2.4729178344249007e-06,
140
+ "logits/chosen": -2.61881422996521,
141
+ "logits/rejected": -2.595266580581665,
142
+ "logps/chosen": -295.1117858886719,
143
+ "logps/rejected": -314.3699035644531,
144
+ "loss": 0.6503,
145
+ "rewards/accuracies": 0.6187499761581421,
146
+ "rewards/chosen": -0.054836757481098175,
147
+ "rewards/margins": 0.10771781206130981,
148
+ "rewards/rejected": -0.1625545769929886,
149
+ "step": 90
150
+ },
151
+ {
152
+ "epoch": 0.62,
153
+ "learning_rate": 1.936156434546515e-06,
154
+ "logits/chosen": -2.587148904800415,
155
+ "logits/rejected": -2.5676586627960205,
156
+ "logps/chosen": -283.98406982421875,
157
+ "logps/rejected": -304.22027587890625,
158
+ "loss": 0.6561,
159
+ "rewards/accuracies": 0.59375,
160
+ "rewards/chosen": -0.020496565848588943,
161
+ "rewards/margins": 0.08849820494651794,
162
+ "rewards/rejected": -0.10899476706981659,
163
+ "step": 100
164
+ },
165
+ {
166
+ "epoch": 0.62,
167
+ "eval_logits/chosen": -2.6306071281433105,
168
+ "eval_logits/rejected": -2.5270211696624756,
169
+ "eval_logps/chosen": -305.47430419921875,
170
+ "eval_logps/rejected": -292.46490478515625,
171
+ "eval_loss": 0.6558266878128052,
172
+ "eval_rewards/accuracies": 0.5979999899864197,
173
+ "eval_rewards/chosen": 0.048699021339416504,
174
+ "eval_rewards/margins": 0.10221949964761734,
175
+ "eval_rewards/rejected": -0.053520478308200836,
176
+ "eval_runtime": 384.3017,
177
+ "eval_samples_per_second": 5.204,
178
+ "eval_steps_per_second": 0.651,
179
+ "step": 100
180
+ },
181
+ {
182
+ "epoch": 0.68,
183
+ "learning_rate": 1.4257597331216211e-06,
184
+ "logits/chosen": -2.5318920612335205,
185
+ "logits/rejected": -2.544257879257202,
186
+ "logps/chosen": -247.9283447265625,
187
+ "logps/rejected": -287.6404724121094,
188
+ "loss": 0.6463,
189
+ "rewards/accuracies": 0.6187499761581421,
190
+ "rewards/chosen": 0.021015625447034836,
191
+ "rewards/margins": 0.11108793318271637,
192
+ "rewards/rejected": -0.09007229655981064,
193
+ "step": 110
194
+ },
195
+ {
196
+ "epoch": 0.74,
197
+ "learning_rate": 9.655933126436565e-07,
198
+ "logits/chosen": -2.5357818603515625,
199
+ "logits/rejected": -2.534702777862549,
200
+ "logps/chosen": -273.9105224609375,
201
+ "logps/rejected": -312.25885009765625,
202
+ "loss": 0.6432,
203
+ "rewards/accuracies": 0.59375,
204
+ "rewards/chosen": -0.0325644388794899,
205
+ "rewards/margins": 0.10732172429561615,
206
+ "rewards/rejected": -0.13988614082336426,
207
+ "step": 120
208
+ },
209
+ {
210
+ "epoch": 0.8,
211
+ "learning_rate": 5.771740434959278e-07,
212
+ "logits/chosen": -2.556398630142212,
213
+ "logits/rejected": -2.560457229614258,
214
+ "logps/chosen": -263.99798583984375,
215
+ "logps/rejected": -307.7806701660156,
216
+ "loss": 0.6483,
217
+ "rewards/accuracies": 0.6468750238418579,
218
+ "rewards/chosen": -0.007414107210934162,
219
+ "rewards/margins": 0.11074688285589218,
220
+ "rewards/rejected": -0.11816100031137466,
221
+ "step": 130
222
+ },
223
+ {
224
+ "epoch": 0.86,
225
+ "learning_rate": 2.786639790067719e-07,
226
+ "logits/chosen": -2.563178539276123,
227
+ "logits/rejected": -2.5608487129211426,
228
+ "logps/chosen": -264.00750732421875,
229
+ "logps/rejected": -310.0502014160156,
230
+ "loss": 0.6492,
231
+ "rewards/accuracies": 0.6499999761581421,
232
+ "rewards/chosen": -0.03133418411016464,
233
+ "rewards/margins": 0.12591516971588135,
234
+ "rewards/rejected": -0.1572493612766266,
235
+ "step": 140
236
+ },
237
+ {
238
+ "epoch": 0.93,
239
+ "learning_rate": 8.402111802159413e-08,
240
+ "logits/chosen": -2.5827817916870117,
241
+ "logits/rejected": -2.5521092414855957,
242
+ "logps/chosen": -257.99468994140625,
243
+ "logps/rejected": -305.74700927734375,
244
+ "loss": 0.6457,
245
+ "rewards/accuracies": 0.6499999761581421,
246
+ "rewards/chosen": 0.012706765905022621,
247
+ "rewards/margins": 0.15555641055107117,
248
+ "rewards/rejected": -0.1428496539592743,
249
+ "step": 150
250
+ },
251
+ {
252
+ "epoch": 0.99,
253
+ "learning_rate": 2.34674439005822e-09,
254
+ "logits/chosen": -2.63196063041687,
255
+ "logits/rejected": -2.579667568206787,
256
+ "logps/chosen": -291.2770080566406,
257
+ "logps/rejected": -327.9423828125,
258
+ "loss": 0.6383,
259
+ "rewards/accuracies": 0.6625000238418579,
260
+ "rewards/chosen": -0.003916625864803791,
261
+ "rewards/margins": 0.17247094213962555,
262
+ "rewards/rejected": -0.17638754844665527,
263
+ "step": 160
264
+ },
265
+ {
266
+ "epoch": 1.0,
267
+ "step": 162,
268
+ "total_flos": 0.0,
269
+ "train_loss": 0.6587186426292231,
270
+ "train_runtime": 6701.6263,
271
+ "train_samples_per_second": 3.094,
272
+ "train_steps_per_second": 0.024
273
+ }
274
+ ],
275
+ "logging_steps": 10,
276
+ "max_steps": 162,
277
+ "num_input_tokens_seen": 0,
278
+ "num_train_epochs": 1,
279
+ "save_steps": 100,
280
+ "total_flos": 0.0,
281
+ "train_batch_size": 2,
282
+ "trial_name": null,
283
+ "trial_params": null
284
+ }