Hugging Face's logo

jz666
/
simpo-train-large-correct

Text Generation
Transformers
Safetensors
gemma2
alignment-handbook
trl
simpo
Generated from Trainer
conversational
text-generation-inference
Model card Files
xet
Community
jz666 commited on
Commit
1c0e83b
·
verified ·
1 Parent(s): f96fe7c

Model save

Browse files
Files changed (5) hide show
  1. README.md +61 -0
  2. all_results.json +9 -0
  3. generation_config.json +8 -0
  4. train_results.json +9 -0
  5. trainer_state.json +447 -0
README.md ADDED
@@ -0,0 +1,61 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: transformers
3
+ license: gemma
4
+ base_model: google/gemma-2-9b-it
5
+ tags:
6
+ - trl
7
+ - simpo
8
+ - generated_from_trainer
9
+ model-index:
10
+ - name: simpo-train-large-correct
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
+ # simpo-train-large-correct
18
+
19
+ This model is a fine-tuned version of [google/gemma-2-9b-it](https://huggingface.co/google/gemma-2-9b-it) on an unknown dataset.
20
+
21
+ ## Model description
22
+
23
+ More information needed
24
+
25
+ ## Intended uses & limitations
26
+
27
+ More information needed
28
+
29
+ ## Training and evaluation data
30
+
31
+ More information needed
32
+
33
+ ## Training procedure
34
+
35
+ ### Training hyperparameters
36
+
37
+ The following hyperparameters were used during training:
38
+ - learning_rate: 8e-07
39
+ - train_batch_size: 2
40
+ - eval_batch_size: 4
41
+ - seed: 42
42
+ - distributed_type: multi-GPU
43
+ - num_devices: 4
44
+ - gradient_accumulation_steps: 16
45
+ - total_train_batch_size: 128
46
+ - total_eval_batch_size: 16
47
+ - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
48
+ - lr_scheduler_type: cosine
49
+ - lr_scheduler_warmup_ratio: 0.1
50
+ - num_epochs: 1
51
+
52
+ ### Training results
53
+
54
+
55
+
56
+ ### Framework versions
57
+
58
+ - Transformers 4.44.2
59
+ - Pytorch 2.7.0+cu128
60
+ - Datasets 2.18.0
61
+ - Tokenizers 0.19.1
all_results.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 0.995910949568378,
3
+ "total_flos": 0.0,
4
+ "train_loss": 2.6376401560149922,
5
+ "train_runtime": 2565.593,
6
+ "train_samples": 17605,
7
+ "train_samples_per_second": 6.862,
8
+ "train_steps_per_second": 0.053
9
+ }
generation_config.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 2,
4
+ "cache_implementation": "hybrid",
5
+ "eos_token_id": 1,
6
+ "pad_token_id": 0,
7
+ "transformers_version": "4.44.2"
8
+ }
train_results.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 0.995910949568378,
3
+ "total_flos": 0.0,
4
+ "train_loss": 2.6376401560149922,
5
+ "train_runtime": 2565.593,
6
+ "train_samples": 17605,
7
+ "train_samples_per_second": 6.862,
8
+ "train_steps_per_second": 0.053
9
+ }
trainer_state.json ADDED
@@ -0,0 +1,447 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 0.995910949568378,
5
+ "eval_steps": 400,
6
+ "global_step": 137,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.03634711494775102,
13
+ "grad_norm": 73.50294131392823,
14
+ "learning_rate": 2.857142857142857e-07,
15
+ "logits/chosen": -10.441385269165039,
16
+ "logits/rejected": -9.520084381103516,
17
+ "logps/chosen": -0.9301475286483765,
18
+ "logps/rejected": -1.348732590675354,
19
+ "loss": 3.7321,
20
+ "rewards/accuracies": 0.893750011920929,
21
+ "rewards/chosen": -9.301475524902344,
22
+ "rewards/margins": 4.185849189758301,
23
+ "rewards/rejected": -13.487324714660645,
24
+ "step": 5
25
+ },
26
+ {
27
+ "epoch": 0.07269422989550205,
28
+ "grad_norm": 49.41339223972053,
29
+ "learning_rate": 5.714285714285714e-07,
30
+ "logits/chosen": -10.112415313720703,
31
+ "logits/rejected": -9.321776390075684,
32
+ "logps/chosen": -1.0208427906036377,
33
+ "logps/rejected": -1.3525886535644531,
34
+ "loss": 3.5537,
35
+ "rewards/accuracies": 0.893750011920929,
36
+ "rewards/chosen": -10.208429336547852,
37
+ "rewards/margins": 3.317457914352417,
38
+ "rewards/rejected": -13.525888442993164,
39
+ "step": 10
40
+ },
41
+ {
42
+ "epoch": 0.10904134484325306,
43
+ "grad_norm": 56.21490730930501,
44
+ "learning_rate": 7.998695344323425e-07,
45
+ "logits/chosen": -10.125561714172363,
46
+ "logits/rejected": -9.403921127319336,
47
+ "logps/chosen": -1.0114651918411255,
48
+ "logps/rejected": -1.4378917217254639,
49
+ "loss": 3.688,
50
+ "rewards/accuracies": 0.8187500238418579,
51
+ "rewards/chosen": -10.114651679992676,
52
+ "rewards/margins": 4.264266014099121,
53
+ "rewards/rejected": -14.37891674041748,
54
+ "step": 15
55
+ },
56
+ {
57
+ "epoch": 0.1453884597910041,
58
+ "grad_norm": 49.2098026266453,
59
+ "learning_rate": 7.953121695121394e-07,
60
+ "logits/chosen": -9.852151870727539,
61
+ "logits/rejected": -8.917947769165039,
62
+ "logps/chosen": -0.917424201965332,
63
+ "logps/rejected": -1.2778595685958862,
64
+ "loss": 3.3675,
65
+ "rewards/accuracies": 0.84375,
66
+ "rewards/chosen": -9.17424201965332,
67
+ "rewards/margins": 3.6043541431427,
68
+ "rewards/rejected": -12.778596878051758,
69
+ "step": 20
70
+ },
71
+ {
72
+ "epoch": 0.18173557473875512,
73
+ "grad_norm": 65.87909071485771,
74
+ "learning_rate": 7.843163833184991e-07,
75
+ "logits/chosen": -10.04487419128418,
76
+ "logits/rejected": -9.234810829162598,
77
+ "logps/chosen": -1.029579520225525,
78
+ "logps/rejected": -1.2623463869094849,
79
+ "loss": 3.3763,
80
+ "rewards/accuracies": 0.800000011920929,
81
+ "rewards/chosen": -10.295795440673828,
82
+ "rewards/margins": 2.3276684284210205,
83
+ "rewards/rejected": -12.623464584350586,
84
+ "step": 25
85
+ },
86
+ {
87
+ "epoch": 0.21808268968650613,
88
+ "grad_norm": 53.32133564579517,
89
+ "learning_rate": 7.670612634414511e-07,
90
+ "logits/chosen": -9.879186630249023,
91
+ "logits/rejected": -8.854208946228027,
92
+ "logps/chosen": -0.9565294981002808,
93
+ "logps/rejected": -1.3326746225357056,
94
+ "loss": 3.2331,
95
+ "rewards/accuracies": 0.893750011920929,
96
+ "rewards/chosen": -9.565296173095703,
97
+ "rewards/margins": 3.7614502906799316,
98
+ "rewards/rejected": -13.326745986938477,
99
+ "step": 30
100
+ },
101
+ {
102
+ "epoch": 0.25442980463425713,
103
+ "grad_norm": 47.904474253960814,
104
+ "learning_rate": 7.438278427948805e-07,
105
+ "logits/chosen": -10.390070915222168,
106
+ "logits/rejected": -9.588186264038086,
107
+ "logps/chosen": -0.997881293296814,
108
+ "logps/rejected": -1.4048808813095093,
109
+ "loss": 3.1431,
110
+ "rewards/accuracies": 0.84375,
111
+ "rewards/chosen": -9.978813171386719,
112
+ "rewards/margins": 4.069996356964111,
113
+ "rewards/rejected": -14.048810005187988,
114
+ "step": 35
115
+ },
116
+ {
117
+ "epoch": 0.2907769195820082,
118
+ "grad_norm": 52.83351276305546,
119
+ "learning_rate": 7.149945224533862e-07,
120
+ "logits/chosen": -10.330121040344238,
121
+ "logits/rejected": -9.436741828918457,
122
+ "logps/chosen": -1.2074191570281982,
123
+ "logps/rejected": -1.6147959232330322,
124
+ "loss": 3.079,
125
+ "rewards/accuracies": 0.8062499761581421,
126
+ "rewards/chosen": -12.074190139770508,
127
+ "rewards/margins": 4.07376766204834,
128
+ "rewards/rejected": -16.147958755493164,
129
+ "step": 40
130
+ },
131
+ {
132
+ "epoch": 0.3271240345297592,
133
+ "grad_norm": 50.87463012185036,
134
+ "learning_rate": 6.810309086608129e-07,
135
+ "logits/chosen": -10.859004974365234,
136
+ "logits/rejected": -9.854225158691406,
137
+ "logps/chosen": -1.2316290140151978,
138
+ "logps/rejected": -1.8107597827911377,
139
+ "loss": 2.9332,
140
+ "rewards/accuracies": 0.887499988079071,
141
+ "rewards/chosen": -12.316289901733398,
142
+ "rewards/margins": 5.79130744934082,
143
+ "rewards/rejected": -18.10759735107422,
144
+ "step": 45
145
+ },
146
+ {
147
+ "epoch": 0.36347114947751025,
148
+ "grad_norm": 96.65968234360786,
149
+ "learning_rate": 6.424901643866552e-07,
150
+ "logits/chosen": -11.214125633239746,
151
+ "logits/rejected": -10.290956497192383,
152
+ "logps/chosen": -1.4425190687179565,
153
+ "logps/rejected": -1.994137167930603,
154
+ "loss": 2.9568,
155
+ "rewards/accuracies": 0.831250011920929,
156
+ "rewards/chosen": -14.425191879272461,
157
+ "rewards/margins": 5.516180515289307,
158
+ "rewards/rejected": -19.941370010375977,
159
+ "step": 50
160
+ },
161
+ {
162
+ "epoch": 0.39981826442526125,
163
+ "grad_norm": 76.49645663435781,
164
+ "learning_rate": 6e-07,
165
+ "logits/chosen": -11.809680938720703,
166
+ "logits/rejected": -10.86604118347168,
167
+ "logps/chosen": -1.511315107345581,
168
+ "logps/rejected": -2.095630168914795,
169
+ "loss": 2.7726,
170
+ "rewards/accuracies": 0.831250011920929,
171
+ "rewards/chosen": -15.113149642944336,
172
+ "rewards/margins": 5.843151092529297,
173
+ "rewards/rejected": -20.956302642822266,
174
+ "step": 55
175
+ },
176
+ {
177
+ "epoch": 0.43616537937301225,
178
+ "grad_norm": 52.07543926650269,
179
+ "learning_rate": 5.542524497952543e-07,
180
+ "logits/chosen": -12.52156925201416,
181
+ "logits/rejected": -11.367462158203125,
182
+ "logps/chosen": -1.417328119277954,
183
+ "logps/rejected": -1.9500234127044678,
184
+ "loss": 2.7633,
185
+ "rewards/accuracies": 0.8999999761581421,
186
+ "rewards/chosen": -14.1732816696167,
187
+ "rewards/margins": 5.3269548416137695,
188
+ "rewards/rejected": -19.500234603881836,
189
+ "step": 60
190
+ },
191
+ {
192
+ "epoch": 0.4725124943207633,
193
+ "grad_norm": 86.00453174970828,
194
+ "learning_rate": 5.059926008786647e-07,
195
+ "logits/chosen": -13.00245189666748,
196
+ "logits/rejected": -11.82313346862793,
197
+ "logps/chosen": -1.3697974681854248,
198
+ "logps/rejected": -1.979061484336853,
199
+ "loss": 2.6871,
200
+ "rewards/accuracies": 0.893750011920929,
201
+ "rewards/chosen": -13.697975158691406,
202
+ "rewards/margins": 6.092639923095703,
203
+ "rewards/rejected": -19.79061508178711,
204
+ "step": 65
205
+ },
206
+ {
207
+ "epoch": 0.5088596092685143,
208
+ "grad_norm": 74.14149953928305,
209
+ "learning_rate": 4.5600645798745166e-07,
210
+ "logits/chosen": -13.302050590515137,
211
+ "logits/rejected": -12.503473281860352,
212
+ "logps/chosen": -1.3789888620376587,
213
+ "logps/rejected": -1.9646472930908203,
214
+ "loss": 2.6182,
215
+ "rewards/accuracies": 0.8687499761581421,
216
+ "rewards/chosen": -13.789888381958008,
217
+ "rewards/margins": 5.856583595275879,
218
+ "rewards/rejected": -19.646472930908203,
219
+ "step": 70
220
+ },
221
+ {
222
+ "epoch": 0.5452067242162654,
223
+ "grad_norm": 82.09661278070892,
224
+ "learning_rate": 4.051081418863895e-07,
225
+ "logits/chosen": -14.177574157714844,
226
+ "logits/rejected": -12.784692764282227,
227
+ "logps/chosen": -1.523525595664978,
228
+ "logps/rejected": -2.243549346923828,
229
+ "loss": 2.4206,
230
+ "rewards/accuracies": 0.8999999761581421,
231
+ "rewards/chosen": -15.235257148742676,
232
+ "rewards/margins": 7.200235843658447,
233
+ "rewards/rejected": -22.43549156188965,
234
+ "step": 75
235
+ },
236
+ {
237
+ "epoch": 0.5815538391640164,
238
+ "grad_norm": 78.17477380695215,
239
+ "learning_rate": 3.541266298406398e-07,
240
+ "logits/chosen": -13.6174955368042,
241
+ "logits/rejected": -12.198015213012695,
242
+ "logps/chosen": -1.413405418395996,
243
+ "logps/rejected": -2.035590648651123,
244
+ "loss": 2.3449,
245
+ "rewards/accuracies": 0.8687499761581421,
246
+ "rewards/chosen": -14.134053230285645,
247
+ "rewards/margins": 6.221851825714111,
248
+ "rewards/rejected": -20.355907440185547,
249
+ "step": 80
250
+ },
251
+ {
252
+ "epoch": 0.6179009541117674,
253
+ "grad_norm": 161.18580193098623,
254
+ "learning_rate": 3.0389225412181565e-07,
255
+ "logits/chosen": -14.64953899383545,
256
+ "logits/rejected": -13.436065673828125,
257
+ "logps/chosen": -1.577237844467163,
258
+ "logps/rejected": -2.26001238822937,
259
+ "loss": 2.3647,
260
+ "rewards/accuracies": 0.8999999761581421,
261
+ "rewards/chosen": -15.772378921508789,
262
+ "rewards/margins": 6.827744483947754,
263
+ "rewards/rejected": -22.60012435913086,
264
+ "step": 85
265
+ },
266
+ {
267
+ "epoch": 0.6542480690595184,
268
+ "grad_norm": 110.64212883100511,
269
+ "learning_rate": 2.5522317844515273e-07,
270
+ "logits/chosen": -15.540824890136719,
271
+ "logits/rejected": -14.182299613952637,
272
+ "logps/chosen": -1.7375078201293945,
273
+ "logps/rejected": -2.5354347229003906,
274
+ "loss": 2.2847,
275
+ "rewards/accuracies": 0.862500011920929,
276
+ "rewards/chosen": -17.375080108642578,
277
+ "rewards/margins": 7.979270935058594,
278
+ "rewards/rejected": -25.35434913635254,
279
+ "step": 90
280
+ },
281
+ {
282
+ "epoch": 0.6905951840072694,
283
+ "grad_norm": 94.27522984793,
284
+ "learning_rate": 2.0891207259509476e-07,
285
+ "logits/chosen": -16.05794906616211,
286
+ "logits/rejected": -14.873092651367188,
287
+ "logps/chosen": -1.8003565073013306,
288
+ "logps/rejected": -2.508544683456421,
289
+ "loss": 2.1888,
290
+ "rewards/accuracies": 0.8687499761581421,
291
+ "rewards/chosen": -18.003564834594727,
292
+ "rewards/margins": 7.081883907318115,
293
+ "rewards/rejected": -25.085451126098633,
294
+ "step": 95
295
+ },
296
+ {
297
+ "epoch": 0.7269422989550205,
298
+ "grad_norm": 108.23339503673007,
299
+ "learning_rate": 1.6571320226872206e-07,
300
+ "logits/chosen": -15.491415023803711,
301
+ "logits/rejected": -14.448579788208008,
302
+ "logps/chosen": -1.7869226932525635,
303
+ "logps/rejected": -2.6193735599517822,
304
+ "loss": 2.235,
305
+ "rewards/accuracies": 0.918749988079071,
306
+ "rewards/chosen": -17.869226455688477,
307
+ "rewards/margins": 8.324507713317871,
308
+ "rewards/rejected": -26.193735122680664,
309
+ "step": 100
310
+ },
311
+ {
312
+ "epoch": 0.7632894139027715,
313
+ "grad_norm": 98.94867860284775,
314
+ "learning_rate": 1.2633014440382787e-07,
315
+ "logits/chosen": -16.772117614746094,
316
+ "logits/rejected": -15.896406173706055,
317
+ "logps/chosen": -1.7917287349700928,
318
+ "logps/rejected": -2.5536069869995117,
319
+ "loss": 1.8876,
320
+ "rewards/accuracies": 0.918749988079071,
321
+ "rewards/chosen": -17.917287826538086,
322
+ "rewards/margins": 7.61877965927124,
323
+ "rewards/rejected": -25.536067962646484,
324
+ "step": 105
325
+ },
326
+ {
327
+ "epoch": 0.7996365288505225,
328
+ "grad_norm": 116.37014336849657,
329
+ "learning_rate": 9.14043280712228e-08,
330
+ "logits/chosen": -16.703330993652344,
331
+ "logits/rejected": -15.419268608093262,
332
+ "logps/chosen": -1.8466812372207642,
333
+ "logps/rejected": -2.60697340965271,
334
+ "loss": 2.0518,
335
+ "rewards/accuracies": 0.8812500238418579,
336
+ "rewards/chosen": -18.466812133789062,
337
+ "rewards/margins": 7.602923393249512,
338
+ "rewards/rejected": -26.06973648071289,
339
+ "step": 110
340
+ },
341
+ {
342
+ "epoch": 0.8359836437982735,
343
+ "grad_norm": 88.56805919880858,
344
+ "learning_rate": 6.150458756494239e-08,
345
+ "logits/chosen": -17.041057586669922,
346
+ "logits/rejected": -15.86296558380127,
347
+ "logps/chosen": -1.9040367603302002,
348
+ "logps/rejected": -2.855522632598877,
349
+ "loss": 1.8644,
350
+ "rewards/accuracies": 0.9125000238418579,
351
+ "rewards/chosen": -19.040367126464844,
352
+ "rewards/margins": 9.514856338500977,
353
+ "rewards/rejected": -28.555225372314453,
354
+ "step": 115
355
+ },
356
+ {
357
+ "epoch": 0.8723307587460245,
358
+ "grad_norm": 119.49778162125348,
359
+ "learning_rate": 3.711789783843522e-08,
360
+ "logits/chosen": -16.915624618530273,
361
+ "logits/rejected": -15.977205276489258,
362
+ "logps/chosen": -1.9863815307617188,
363
+ "logps/rejected": -2.819209575653076,
364
+ "loss": 2.0238,
365
+ "rewards/accuracies": 0.8687499761581421,
366
+ "rewards/chosen": -19.863813400268555,
367
+ "rewards/margins": 8.328282356262207,
368
+ "rewards/rejected": -28.192096710205078,
369
+ "step": 120
370
+ },
371
+ {
372
+ "epoch": 0.9086778736937755,
373
+ "grad_norm": 100.49399716119115,
374
+ "learning_rate": 1.8641443178027784e-08,
375
+ "logits/chosen": -17.482728958129883,
376
+ "logits/rejected": -16.11745262145996,
377
+ "logps/chosen": -1.8876680135726929,
378
+ "logps/rejected": -2.652832269668579,
379
+ "loss": 2.0136,
380
+ "rewards/accuracies": 0.862500011920929,
381
+ "rewards/chosen": -18.876678466796875,
382
+ "rewards/margins": 7.6516432762146,
383
+ "rewards/rejected": -26.528324127197266,
384
+ "step": 125
385
+ },
386
+ {
387
+ "epoch": 0.9450249886415266,
388
+ "grad_norm": 78.59767084378562,
389
+ "learning_rate": 6.376148290617145e-09,
390
+ "logits/chosen": -16.445621490478516,
391
+ "logits/rejected": -15.228750228881836,
392
+ "logps/chosen": -1.947279691696167,
393
+ "logps/rejected": -2.799992084503174,
394
+ "loss": 1.9014,
395
+ "rewards/accuracies": 0.8812500238418579,
396
+ "rewards/chosen": -19.472797393798828,
397
+ "rewards/margins": 8.527124404907227,
398
+ "rewards/rejected": -27.999919891357422,
399
+ "step": 130
400
+ },
401
+ {
402
+ "epoch": 0.9813721035892776,
403
+ "grad_norm": 106.19866039415072,
404
+ "learning_rate": 5.217771643080127e-10,
405
+ "logits/chosen": -17.454696655273438,
406
+ "logits/rejected": -16.37453842163086,
407
+ "logps/chosen": -1.9225728511810303,
408
+ "logps/rejected": -2.7656478881835938,
409
+ "loss": 1.9868,
410
+ "rewards/accuracies": 0.875,
411
+ "rewards/chosen": -19.225727081298828,
412
+ "rewards/margins": 8.430750846862793,
413
+ "rewards/rejected": -27.656478881835938,
414
+ "step": 135
415
+ },
416
+ {
417
+ "epoch": 0.995910949568378,
418
+ "step": 137,
419
+ "total_flos": 0.0,
420
+ "train_loss": 2.6376401560149922,
421
+ "train_runtime": 2565.593,
422
+ "train_samples_per_second": 6.862,
423
+ "train_steps_per_second": 0.053
424
+ }
425
+ ],
426
+ "logging_steps": 5,
427
+ "max_steps": 137,
428
+ "num_input_tokens_seen": 0,
429
+ "num_train_epochs": 1,
430
+ "save_steps": 1000000,
431
+ "stateful_callbacks": {
432
+ "TrainerControl": {
433
+ "args": {
434
+ "should_epoch_stop": false,
435
+ "should_evaluate": false,
436
+ "should_log": false,
437
+ "should_save": true,
438
+ "should_training_stop": true
439
+ },
440
+ "attributes": {}
441
+ }
442
+ },
443
+ "total_flos": 0.0,
444
+ "train_batch_size": 2,
445
+ "trial_name": null,
446
+ "trial_params": null
447
+ }