seablue commited on
Commit
c2612f2
1 Parent(s): 85e8284

Upload trainer_state.json with huggingface_hub

Browse files
Files changed (1) hide show
  1. trainer_state.json +1467 -0
trainer_state.json ADDED
@@ -0,0 +1,1467 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 0.998691442030882,
5
+ "eval_steps": 500,
6
+ "global_step": 477,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.010468463752944255,
13
+ "grad_norm": 3.843380266792351,
14
+ "learning_rate": 5.208333333333333e-08,
15
+ "logits/chosen": -2.8176448345184326,
16
+ "logits/rejected": -2.775940418243408,
17
+ "logps/chosen": -1.0736682415008545,
18
+ "logps/rejected": -1.1332499980926514,
19
+ "loss": -0.0244,
20
+ "rewards/accuracies": 0.512499988079071,
21
+ "rewards/chosen": -1.0736682415008545,
22
+ "rewards/margins": 0.05958189815282822,
23
+ "rewards/rejected": -1.1332499980926514,
24
+ "step": 5
25
+ },
26
+ {
27
+ "epoch": 0.02093692750588851,
28
+ "grad_norm": 4.13922138458207,
29
+ "learning_rate": 1.0416666666666667e-07,
30
+ "logits/chosen": -2.9982171058654785,
31
+ "logits/rejected": -2.952279806137085,
32
+ "logps/chosen": -1.052896499633789,
33
+ "logps/rejected": -1.194782018661499,
34
+ "loss": -0.0274,
35
+ "rewards/accuracies": 0.5625,
36
+ "rewards/chosen": -1.052896499633789,
37
+ "rewards/margins": 0.1418854296207428,
38
+ "rewards/rejected": -1.194782018661499,
39
+ "step": 10
40
+ },
41
+ {
42
+ "epoch": 0.031405391258832765,
43
+ "grad_norm": 4.388036242315528,
44
+ "learning_rate": 1.5624999999999999e-07,
45
+ "logits/chosen": -2.905001401901245,
46
+ "logits/rejected": -2.881582021713257,
47
+ "logps/chosen": -1.0749495029449463,
48
+ "logps/rejected": -1.1714584827423096,
49
+ "loss": -0.0311,
50
+ "rewards/accuracies": 0.5375000238418579,
51
+ "rewards/chosen": -1.0749495029449463,
52
+ "rewards/margins": 0.09650889784097672,
53
+ "rewards/rejected": -1.1714584827423096,
54
+ "step": 15
55
+ },
56
+ {
57
+ "epoch": 0.04187385501177702,
58
+ "grad_norm": 4.170757799961158,
59
+ "learning_rate": 2.0833333333333333e-07,
60
+ "logits/chosen": -2.9079184532165527,
61
+ "logits/rejected": -2.850598096847534,
62
+ "logps/chosen": -0.9996750950813293,
63
+ "logps/rejected": -1.0603716373443604,
64
+ "loss": -0.0326,
65
+ "rewards/accuracies": 0.512499988079071,
66
+ "rewards/chosen": -0.9996750950813293,
67
+ "rewards/margins": 0.060696445405483246,
68
+ "rewards/rejected": -1.0603716373443604,
69
+ "step": 20
70
+ },
71
+ {
72
+ "epoch": 0.05234231876472128,
73
+ "grad_norm": 4.2140301502962565,
74
+ "learning_rate": 2.604166666666667e-07,
75
+ "logits/chosen": -2.9368786811828613,
76
+ "logits/rejected": -2.880016565322876,
77
+ "logps/chosen": -0.8968712687492371,
78
+ "logps/rejected": -1.018384337425232,
79
+ "loss": -0.0413,
80
+ "rewards/accuracies": 0.5625,
81
+ "rewards/chosen": -0.8968712687492371,
82
+ "rewards/margins": 0.12151306867599487,
83
+ "rewards/rejected": -1.018384337425232,
84
+ "step": 25
85
+ },
86
+ {
87
+ "epoch": 0.06281078251766553,
88
+ "grad_norm": 5.648380861586244,
89
+ "learning_rate": 3.1249999999999997e-07,
90
+ "logits/chosen": -2.9270145893096924,
91
+ "logits/rejected": -2.897721767425537,
92
+ "logps/chosen": -0.8545770645141602,
93
+ "logps/rejected": -0.9005380868911743,
94
+ "loss": -0.0369,
95
+ "rewards/accuracies": 0.574999988079071,
96
+ "rewards/chosen": -0.8545770645141602,
97
+ "rewards/margins": 0.04596105217933655,
98
+ "rewards/rejected": -0.9005380868911743,
99
+ "step": 30
100
+ },
101
+ {
102
+ "epoch": 0.07327924627060979,
103
+ "grad_norm": 5.194083890339375,
104
+ "learning_rate": 3.645833333333333e-07,
105
+ "logits/chosen": -3.003455638885498,
106
+ "logits/rejected": -2.9747724533081055,
107
+ "logps/chosen": -0.8473809957504272,
108
+ "logps/rejected": -0.9919956922531128,
109
+ "loss": -0.0475,
110
+ "rewards/accuracies": 0.6000000238418579,
111
+ "rewards/chosen": -0.8473809957504272,
112
+ "rewards/margins": 0.14461484551429749,
113
+ "rewards/rejected": -0.9919956922531128,
114
+ "step": 35
115
+ },
116
+ {
117
+ "epoch": 0.08374771002355404,
118
+ "grad_norm": 4.598851505341545,
119
+ "learning_rate": 4.1666666666666667e-07,
120
+ "logits/chosen": -2.9274966716766357,
121
+ "logits/rejected": -2.9125096797943115,
122
+ "logps/chosen": -0.875938892364502,
123
+ "logps/rejected": -1.0193151235580444,
124
+ "loss": -0.0484,
125
+ "rewards/accuracies": 0.574999988079071,
126
+ "rewards/chosen": -0.875938892364502,
127
+ "rewards/margins": 0.1433762013912201,
128
+ "rewards/rejected": -1.0193151235580444,
129
+ "step": 40
130
+ },
131
+ {
132
+ "epoch": 0.0942161737764983,
133
+ "grad_norm": 4.789508927144662,
134
+ "learning_rate": 4.6874999999999996e-07,
135
+ "logits/chosen": -2.9352455139160156,
136
+ "logits/rejected": -2.911196231842041,
137
+ "logps/chosen": -0.8146079182624817,
138
+ "logps/rejected": -0.9217915534973145,
139
+ "loss": -0.0617,
140
+ "rewards/accuracies": 0.5874999761581421,
141
+ "rewards/chosen": -0.8146079182624817,
142
+ "rewards/margins": 0.10718371719121933,
143
+ "rewards/rejected": -0.9217915534973145,
144
+ "step": 45
145
+ },
146
+ {
147
+ "epoch": 0.10468463752944256,
148
+ "grad_norm": 5.456564382160052,
149
+ "learning_rate": 4.999731868769026e-07,
150
+ "logits/chosen": -2.9189658164978027,
151
+ "logits/rejected": -2.9572436809539795,
152
+ "logps/chosen": -0.8167168498039246,
153
+ "logps/rejected": -1.0531059503555298,
154
+ "loss": -0.0526,
155
+ "rewards/accuracies": 0.675000011920929,
156
+ "rewards/chosen": -0.8167168498039246,
157
+ "rewards/margins": 0.23638899624347687,
158
+ "rewards/rejected": -1.0531059503555298,
159
+ "step": 50
160
+ },
161
+ {
162
+ "epoch": 0.11515310128238682,
163
+ "grad_norm": 7.50323556887671,
164
+ "learning_rate": 4.996716052911017e-07,
165
+ "logits/chosen": -2.9651036262512207,
166
+ "logits/rejected": -2.993089199066162,
167
+ "logps/chosen": -0.9721449017524719,
168
+ "logps/rejected": -1.0521472692489624,
169
+ "loss": -0.0482,
170
+ "rewards/accuracies": 0.6000000238418579,
171
+ "rewards/chosen": -0.9721449017524719,
172
+ "rewards/margins": 0.08000238239765167,
173
+ "rewards/rejected": -1.0521472692489624,
174
+ "step": 55
175
+ },
176
+ {
177
+ "epoch": 0.12562156503533106,
178
+ "grad_norm": 8.485162916006143,
179
+ "learning_rate": 4.990353313429303e-07,
180
+ "logits/chosen": -2.9055917263031006,
181
+ "logits/rejected": -2.9019391536712646,
182
+ "logps/chosen": -0.9157769083976746,
183
+ "logps/rejected": -1.0977026224136353,
184
+ "loss": -0.0682,
185
+ "rewards/accuracies": 0.5625,
186
+ "rewards/chosen": -0.9157769083976746,
187
+ "rewards/margins": 0.18192583322525024,
188
+ "rewards/rejected": -1.0977026224136353,
189
+ "step": 60
190
+ },
191
+ {
192
+ "epoch": 0.1360900287882753,
193
+ "grad_norm": 6.60792180172325,
194
+ "learning_rate": 4.980652179769217e-07,
195
+ "logits/chosen": -2.966867446899414,
196
+ "logits/rejected": -2.991152048110962,
197
+ "logps/chosen": -0.953170895576477,
198
+ "logps/rejected": -1.0807805061340332,
199
+ "loss": -0.0605,
200
+ "rewards/accuracies": 0.550000011920929,
201
+ "rewards/chosen": -0.953170895576477,
202
+ "rewards/margins": 0.12760967016220093,
203
+ "rewards/rejected": -1.0807805061340332,
204
+ "step": 65
205
+ },
206
+ {
207
+ "epoch": 0.14655849254121958,
208
+ "grad_norm": 7.134153653486866,
209
+ "learning_rate": 4.967625656594781e-07,
210
+ "logits/chosen": -2.9338037967681885,
211
+ "logits/rejected": -2.921638011932373,
212
+ "logps/chosen": -0.9601171612739563,
213
+ "logps/rejected": -1.170008659362793,
214
+ "loss": -0.0809,
215
+ "rewards/accuracies": 0.637499988079071,
216
+ "rewards/chosen": -0.9601171612739563,
217
+ "rewards/margins": 0.20989160239696503,
218
+ "rewards/rejected": -1.170008659362793,
219
+ "step": 70
220
+ },
221
+ {
222
+ "epoch": 0.15702695629416383,
223
+ "grad_norm": 8.86078567599782,
224
+ "learning_rate": 4.951291206355559e-07,
225
+ "logits/chosen": -2.9449570178985596,
226
+ "logits/rejected": -2.863473415374756,
227
+ "logps/chosen": -1.019167423248291,
228
+ "logps/rejected": -1.3563578128814697,
229
+ "loss": -0.0632,
230
+ "rewards/accuracies": 0.6000000238418579,
231
+ "rewards/chosen": -1.019167423248291,
232
+ "rewards/margins": 0.3371904492378235,
233
+ "rewards/rejected": -1.3563578128814697,
234
+ "step": 75
235
+ },
236
+ {
237
+ "epoch": 0.16749542004710807,
238
+ "grad_norm": 6.323638159923026,
239
+ "learning_rate": 4.93167072587771e-07,
240
+ "logits/chosen": -2.9299821853637695,
241
+ "logits/rejected": -2.91919207572937,
242
+ "logps/chosen": -0.993580162525177,
243
+ "logps/rejected": -1.4255390167236328,
244
+ "loss": -0.0782,
245
+ "rewards/accuracies": 0.75,
246
+ "rewards/chosen": -0.993580162525177,
247
+ "rewards/margins": 0.43195900321006775,
248
+ "rewards/rejected": -1.4255390167236328,
249
+ "step": 80
250
+ },
251
+ {
252
+ "epoch": 0.17796388380005235,
253
+ "grad_norm": 10.148627109182378,
254
+ "learning_rate": 4.908790517010636e-07,
255
+ "logits/chosen": -2.9560999870300293,
256
+ "logits/rejected": -2.9236154556274414,
257
+ "logps/chosen": -0.9319014549255371,
258
+ "logps/rejected": -1.2665565013885498,
259
+ "loss": -0.0829,
260
+ "rewards/accuracies": 0.6875,
261
+ "rewards/chosen": -0.9319014549255371,
262
+ "rewards/margins": 0.3346549868583679,
263
+ "rewards/rejected": -1.2665565013885498,
264
+ "step": 85
265
+ },
266
+ {
267
+ "epoch": 0.1884323475529966,
268
+ "grad_norm": 6.282748689638336,
269
+ "learning_rate": 4.882681251368548e-07,
270
+ "logits/chosen": -3.0003504753112793,
271
+ "logits/rejected": -2.9371325969696045,
272
+ "logps/chosen": -0.9956916570663452,
273
+ "logps/rejected": -1.5206801891326904,
274
+ "loss": -0.085,
275
+ "rewards/accuracies": 0.6499999761581421,
276
+ "rewards/chosen": -0.9956916570663452,
277
+ "rewards/margins": 0.5249885320663452,
278
+ "rewards/rejected": -1.5206801891326904,
279
+ "step": 90
280
+ },
281
+ {
282
+ "epoch": 0.19890081130594087,
283
+ "grad_norm": 6.0015528777597265,
284
+ "learning_rate": 4.853377929214243e-07,
285
+ "logits/chosen": -2.938925266265869,
286
+ "logits/rejected": -2.9378859996795654,
287
+ "logps/chosen": -0.9370375871658325,
288
+ "logps/rejected": -1.2533988952636719,
289
+ "loss": -0.073,
290
+ "rewards/accuracies": 0.675000011920929,
291
+ "rewards/chosen": -0.9370375871658325,
292
+ "rewards/margins": 0.3163612484931946,
293
+ "rewards/rejected": -1.2533988952636719,
294
+ "step": 95
295
+ },
296
+ {
297
+ "epoch": 0.2093692750588851,
298
+ "grad_norm": 9.968256329447078,
299
+ "learning_rate": 4.820919832540181e-07,
300
+ "logits/chosen": -3.1243631839752197,
301
+ "logits/rejected": -3.087338924407959,
302
+ "logps/chosen": -0.9904219508171082,
303
+ "logps/rejected": -1.245534062385559,
304
+ "loss": -0.0777,
305
+ "rewards/accuracies": 0.5625,
306
+ "rewards/chosen": -0.9904219508171082,
307
+ "rewards/margins": 0.2551119923591614,
308
+ "rewards/rejected": -1.245534062385559,
309
+ "step": 100
310
+ },
311
+ {
312
+ "epoch": 0.21983773881182936,
313
+ "grad_norm": 13.342769400160067,
314
+ "learning_rate": 4.785350472409791e-07,
315
+ "logits/chosen": -2.987121105194092,
316
+ "logits/rejected": -3.0028722286224365,
317
+ "logps/chosen": -1.1532871723175049,
318
+ "logps/rejected": -1.6190894842147827,
319
+ "loss": -0.1063,
320
+ "rewards/accuracies": 0.625,
321
+ "rewards/chosen": -1.1532871723175049,
322
+ "rewards/margins": 0.4658023715019226,
323
+ "rewards/rejected": -1.6190894842147827,
324
+ "step": 105
325
+ },
326
+ {
327
+ "epoch": 0.23030620256477363,
328
+ "grad_norm": 11.26110556096259,
329
+ "learning_rate": 4.7467175306295647e-07,
330
+ "logits/chosen": -2.966667652130127,
331
+ "logits/rejected": -3.0194077491760254,
332
+ "logps/chosen": -1.0292152166366577,
333
+ "logps/rejected": -1.59182608127594,
334
+ "loss": -0.1066,
335
+ "rewards/accuracies": 0.699999988079071,
336
+ "rewards/chosen": -1.0292152166366577,
337
+ "rewards/margins": 0.562610924243927,
338
+ "rewards/rejected": -1.59182608127594,
339
+ "step": 110
340
+ },
341
+ {
342
+ "epoch": 0.24077466631771788,
343
+ "grad_norm": 8.176863863279703,
344
+ "learning_rate": 4.70507279583015e-07,
345
+ "logits/chosen": -3.0338950157165527,
346
+ "logits/rejected": -3.0541911125183105,
347
+ "logps/chosen": -1.1478749513626099,
348
+ "logps/rejected": -1.4170184135437012,
349
+ "loss": -0.0877,
350
+ "rewards/accuracies": 0.6000000238418579,
351
+ "rewards/chosen": -1.1478749513626099,
352
+ "rewards/margins": 0.2691434919834137,
353
+ "rewards/rejected": -1.4170184135437012,
354
+ "step": 115
355
+ },
356
+ {
357
+ "epoch": 0.2512431300706621,
358
+ "grad_norm": 8.289689216777566,
359
+ "learning_rate": 4.6604720940421207e-07,
360
+ "logits/chosen": -3.0846564769744873,
361
+ "logits/rejected": -3.0871291160583496,
362
+ "logps/chosen": -1.1712749004364014,
363
+ "logps/rejected": -1.6515331268310547,
364
+ "loss": -0.0702,
365
+ "rewards/accuracies": 0.637499988079071,
366
+ "rewards/chosen": -1.1712749004364014,
367
+ "rewards/margins": 0.48025816679000854,
368
+ "rewards/rejected": -1.6515331268310547,
369
+ "step": 120
370
+ },
371
+ {
372
+ "epoch": 0.26171159382360637,
373
+ "grad_norm": 8.632430270339013,
374
+ "learning_rate": 4.612975213859487e-07,
375
+ "logits/chosen": -3.0736565589904785,
376
+ "logits/rejected": -3.0235087871551514,
377
+ "logps/chosen": -1.246620535850525,
378
+ "logps/rejected": -1.603502869606018,
379
+ "loss": -0.0801,
380
+ "rewards/accuracies": 0.6875,
381
+ "rewards/chosen": -1.246620535850525,
382
+ "rewards/margins": 0.3568824827671051,
383
+ "rewards/rejected": -1.603502869606018,
384
+ "step": 125
385
+ },
386
+ {
387
+ "epoch": 0.2721800575765506,
388
+ "grad_norm": 6.951210195379089,
389
+ "learning_rate": 4.5626458262912735e-07,
390
+ "logits/chosen": -3.0484471321105957,
391
+ "logits/rejected": -3.0786657333374023,
392
+ "logps/chosen": -1.2631396055221558,
393
+ "logps/rejected": -1.8509140014648438,
394
+ "loss": -0.1013,
395
+ "rewards/accuracies": 0.6625000238418579,
396
+ "rewards/chosen": -1.2631396055221558,
397
+ "rewards/margins": 0.5877745747566223,
398
+ "rewards/rejected": -1.8509140014648438,
399
+ "step": 130
400
+ },
401
+ {
402
+ "epoch": 0.2826485213294949,
403
+ "grad_norm": 6.6831446876498966,
404
+ "learning_rate": 4.5095513994085974e-07,
405
+ "logits/chosen": -2.9543702602386475,
406
+ "logits/rejected": -2.9393744468688965,
407
+ "logps/chosen": -1.0893765687942505,
408
+ "logps/rejected": -1.7171491384506226,
409
+ "loss": -0.0945,
410
+ "rewards/accuracies": 0.6875,
411
+ "rewards/chosen": -1.0893765687942505,
412
+ "rewards/margins": 0.6277726292610168,
413
+ "rewards/rejected": -1.7171491384506226,
414
+ "step": 135
415
+ },
416
+ {
417
+ "epoch": 0.29311698508243916,
418
+ "grad_norm": 11.237918474787836,
419
+ "learning_rate": 4.453763107901675e-07,
420
+ "logits/chosen": -3.0231282711029053,
421
+ "logits/rejected": -3.0941128730773926,
422
+ "logps/chosen": -1.3833149671554565,
423
+ "logps/rejected": -1.7585477828979492,
424
+ "loss": -0.0791,
425
+ "rewards/accuracies": 0.5874999761581421,
426
+ "rewards/chosen": -1.3833149671554565,
427
+ "rewards/margins": 0.37523263692855835,
428
+ "rewards/rejected": -1.7585477828979492,
429
+ "step": 140
430
+ },
431
+ {
432
+ "epoch": 0.3035854488353834,
433
+ "grad_norm": 11.485846701864144,
434
+ "learning_rate": 4.395355737667985e-07,
435
+ "logits/chosen": -3.0041990280151367,
436
+ "logits/rejected": -3.0510060787200928,
437
+ "logps/chosen": -1.1283174753189087,
438
+ "logps/rejected": -1.8526252508163452,
439
+ "loss": -0.1012,
440
+ "rewards/accuracies": 0.675000011920929,
441
+ "rewards/chosen": -1.1283174753189087,
442
+ "rewards/margins": 0.7243077158927917,
443
+ "rewards/rejected": -1.8526252508163452,
444
+ "step": 145
445
+ },
446
+ {
447
+ "epoch": 0.31405391258832765,
448
+ "grad_norm": 11.357824867739286,
449
+ "learning_rate": 4.3344075855595097e-07,
450
+ "logits/chosen": -3.0792462825775146,
451
+ "logits/rejected": -3.0518386363983154,
452
+ "logps/chosen": -1.247040033340454,
453
+ "logps/rejected": -1.7509548664093018,
454
+ "loss": -0.0889,
455
+ "rewards/accuracies": 0.625,
456
+ "rewards/chosen": -1.247040033340454,
457
+ "rewards/margins": 0.5039147734642029,
458
+ "rewards/rejected": -1.7509548664093018,
459
+ "step": 150
460
+ },
461
+ {
462
+ "epoch": 0.3245223763412719,
463
+ "grad_norm": 11.335355380541145,
464
+ "learning_rate": 4.271000354423425e-07,
465
+ "logits/chosen": -2.9956650733947754,
466
+ "logits/rejected": -3.0046334266662598,
467
+ "logps/chosen": -1.3198177814483643,
468
+ "logps/rejected": -2.0559017658233643,
469
+ "loss": -0.111,
470
+ "rewards/accuracies": 0.675000011920929,
471
+ "rewards/chosen": -1.3198177814483643,
472
+ "rewards/margins": 0.7360838651657104,
473
+ "rewards/rejected": -2.0559017658233643,
474
+ "step": 155
475
+ },
476
+ {
477
+ "epoch": 0.33499084009421615,
478
+ "grad_norm": 16.693424155067344,
479
+ "learning_rate": 4.2052190435769554e-07,
480
+ "logits/chosen": -2.9203574657440186,
481
+ "logits/rejected": -2.9545798301696777,
482
+ "logps/chosen": -1.1988856792449951,
483
+ "logps/rejected": -1.823240876197815,
484
+ "loss": -0.1107,
485
+ "rewards/accuracies": 0.675000011920929,
486
+ "rewards/chosen": -1.1988856792449951,
487
+ "rewards/margins": 0.6243550181388855,
488
+ "rewards/rejected": -1.823240876197815,
489
+ "step": 160
490
+ },
491
+ {
492
+ "epoch": 0.34545930384716045,
493
+ "grad_norm": 26.190973492572102,
494
+ "learning_rate": 4.137151834863213e-07,
495
+ "logits/chosen": -2.992180347442627,
496
+ "logits/rejected": -2.9608774185180664,
497
+ "logps/chosen": -1.3537108898162842,
498
+ "logps/rejected": -1.587428331375122,
499
+ "loss": -0.114,
500
+ "rewards/accuracies": 0.5874999761581421,
501
+ "rewards/chosen": -1.3537108898162842,
502
+ "rewards/margins": 0.2337172031402588,
503
+ "rewards/rejected": -1.587428331375122,
504
+ "step": 165
505
+ },
506
+ {
507
+ "epoch": 0.3559277676001047,
508
+ "grad_norm": 11.938074204901223,
509
+ "learning_rate": 4.0668899744407567e-07,
510
+ "logits/chosen": -2.943603992462158,
511
+ "logits/rejected": -2.9042210578918457,
512
+ "logps/chosen": -1.1778740882873535,
513
+ "logps/rejected": -1.7121427059173584,
514
+ "loss": -0.1062,
515
+ "rewards/accuracies": 0.625,
516
+ "rewards/chosen": -1.1778740882873535,
517
+ "rewards/margins": 0.5342684984207153,
518
+ "rewards/rejected": -1.7121427059173584,
519
+ "step": 170
520
+ },
521
+ {
522
+ "epoch": 0.36639623135304894,
523
+ "grad_norm": 9.013768613990127,
524
+ "learning_rate": 3.994527650465352e-07,
525
+ "logits/chosen": -2.767575263977051,
526
+ "logits/rejected": -2.7348756790161133,
527
+ "logps/chosen": -1.2128514051437378,
528
+ "logps/rejected": -2.0194027423858643,
529
+ "loss": -0.1222,
530
+ "rewards/accuracies": 0.6625000238418579,
531
+ "rewards/chosen": -1.2128514051437378,
532
+ "rewards/margins": 0.8065516352653503,
533
+ "rewards/rejected": -2.0194027423858643,
534
+ "step": 175
535
+ },
536
+ {
537
+ "epoch": 0.3768646951059932,
538
+ "grad_norm": 20.304059925871332,
539
+ "learning_rate": 3.920161866827889e-07,
540
+ "logits/chosen": -2.6296277046203613,
541
+ "logits/rejected": -2.5583887100219727,
542
+ "logps/chosen": -1.4558690786361694,
543
+ "logps/rejected": -2.660147190093994,
544
+ "loss": -0.104,
545
+ "rewards/accuracies": 0.6499999761581421,
546
+ "rewards/chosen": -1.4558690786361694,
547
+ "rewards/margins": 1.2042779922485352,
548
+ "rewards/rejected": -2.660147190093994,
549
+ "step": 180
550
+ },
551
+ {
552
+ "epoch": 0.38733315885893743,
553
+ "grad_norm": 12.56852650611151,
554
+ "learning_rate": 3.8438923131177237e-07,
555
+ "logits/chosen": -2.646308660507202,
556
+ "logits/rejected": -2.5967037677764893,
557
+ "logps/chosen": -1.2808468341827393,
558
+ "logps/rejected": -2.6707491874694824,
559
+ "loss": -0.1264,
560
+ "rewards/accuracies": 0.7250000238418579,
561
+ "rewards/chosen": -1.2808468341827393,
562
+ "rewards/margins": 1.3899024724960327,
563
+ "rewards/rejected": -2.6707491874694824,
564
+ "step": 185
565
+ },
566
+ {
567
+ "epoch": 0.39780162261188173,
568
+ "grad_norm": 10.671381825602081,
569
+ "learning_rate": 3.765821230985757e-07,
570
+ "logits/chosen": -2.6582224369049072,
571
+ "logits/rejected": -2.643031358718872,
572
+ "logps/chosen": -1.2568540573120117,
573
+ "logps/rejected": -2.3855767250061035,
574
+ "loss": -0.1248,
575
+ "rewards/accuracies": 0.675000011920929,
576
+ "rewards/chosen": -1.2568540573120117,
577
+ "rewards/margins": 1.1287227869033813,
578
+ "rewards/rejected": -2.3855767250061035,
579
+ "step": 190
580
+ },
581
+ {
582
+ "epoch": 0.408270086364826,
583
+ "grad_norm": 12.305445384315465,
584
+ "learning_rate": 3.6860532770864005e-07,
585
+ "logits/chosen": -2.6588711738586426,
586
+ "logits/rejected": -2.650085926055908,
587
+ "logps/chosen": -1.2373688220977783,
588
+ "logps/rejected": -2.440690755844116,
589
+ "loss": -0.1086,
590
+ "rewards/accuracies": 0.737500011920929,
591
+ "rewards/chosen": -1.2373688220977783,
592
+ "rewards/margins": 1.2033220529556274,
593
+ "rewards/rejected": -2.440690755844116,
594
+ "step": 195
595
+ },
596
+ {
597
+ "epoch": 0.4187385501177702,
598
+ "grad_norm": 114.17793963840649,
599
+ "learning_rate": 3.604695382782159e-07,
600
+ "logits/chosen": -2.5837745666503906,
601
+ "logits/rejected": -2.5190796852111816,
602
+ "logps/chosen": -1.535466194152832,
603
+ "logps/rejected": -2.662421703338623,
604
+ "loss": -0.1231,
605
+ "rewards/accuracies": 0.637499988079071,
606
+ "rewards/chosen": -1.535466194152832,
607
+ "rewards/margins": 1.126955509185791,
608
+ "rewards/rejected": -2.662421703338623,
609
+ "step": 200
610
+ },
611
+ {
612
+ "epoch": 0.42920701387071447,
613
+ "grad_norm": 29.258244612720908,
614
+ "learning_rate": 3.5218566107988867e-07,
615
+ "logits/chosen": -2.4933407306671143,
616
+ "logits/rejected": -2.3973355293273926,
617
+ "logps/chosen": -1.6263624429702759,
618
+ "logps/rejected": -3.2516510486602783,
619
+ "loss": -0.1144,
620
+ "rewards/accuracies": 0.6499999761581421,
621
+ "rewards/chosen": -1.6263624429702759,
622
+ "rewards/margins": 1.6252882480621338,
623
+ "rewards/rejected": -3.2516510486602783,
624
+ "step": 205
625
+ },
626
+ {
627
+ "epoch": 0.4396754776236587,
628
+ "grad_norm": 39.526967700856005,
629
+ "learning_rate": 3.4376480090239047e-07,
630
+ "logits/chosen": -2.342941999435425,
631
+ "logits/rejected": -2.3288662433624268,
632
+ "logps/chosen": -2.3925395011901855,
633
+ "logps/rejected": -3.234570264816284,
634
+ "loss": -0.0999,
635
+ "rewards/accuracies": 0.637499988079071,
636
+ "rewards/chosen": -2.3925395011901855,
637
+ "rewards/margins": 0.8420308232307434,
638
+ "rewards/rejected": -3.234570264816284,
639
+ "step": 210
640
+ },
641
+ {
642
+ "epoch": 0.45014394137660296,
643
+ "grad_norm": 15.657987448676648,
644
+ "learning_rate": 3.3521824616429284e-07,
645
+ "logits/chosen": -2.3377509117126465,
646
+ "logits/rejected": -2.3475136756896973,
647
+ "logps/chosen": -2.209704875946045,
648
+ "logps/rejected": -3.1227197647094727,
649
+ "loss": -0.1112,
650
+ "rewards/accuracies": 0.699999988079071,
651
+ "rewards/chosen": -2.209704875946045,
652
+ "rewards/margins": 0.9130150079727173,
653
+ "rewards/rejected": -3.1227197647094727,
654
+ "step": 215
655
+ },
656
+ {
657
+ "epoch": 0.46061240512954726,
658
+ "grad_norm": 24.63970678663639,
659
+ "learning_rate": 3.265574537815398e-07,
660
+ "logits/chosen": -2.3190903663635254,
661
+ "logits/rejected": -2.222909927368164,
662
+ "logps/chosen": -1.5355461835861206,
663
+ "logps/rejected": -2.3923590183258057,
664
+ "loss": -0.125,
665
+ "rewards/accuracies": 0.6875,
666
+ "rewards/chosen": -1.5355461835861206,
667
+ "rewards/margins": 0.8568128347396851,
668
+ "rewards/rejected": -2.3923590183258057,
669
+ "step": 220
670
+ },
671
+ {
672
+ "epoch": 0.4710808688824915,
673
+ "grad_norm": 10.677074157142071,
674
+ "learning_rate": 3.1779403380910425e-07,
675
+ "logits/chosen": -2.237248182296753,
676
+ "logits/rejected": -2.051560163497925,
677
+ "logps/chosen": -1.662973165512085,
678
+ "logps/rejected": -2.656738519668579,
679
+ "loss": -0.1184,
680
+ "rewards/accuracies": 0.75,
681
+ "rewards/chosen": -1.662973165512085,
682
+ "rewards/margins": 0.9937652349472046,
683
+ "rewards/rejected": -2.656738519668579,
684
+ "step": 225
685
+ },
686
+ {
687
+ "epoch": 0.48154933263543576,
688
+ "grad_norm": 63.22829689035937,
689
+ "learning_rate": 3.0893973387735683e-07,
690
+ "logits/chosen": -2.1220011711120605,
691
+ "logits/rejected": -1.9981296062469482,
692
+ "logps/chosen": -1.8179805278778076,
693
+ "logps/rejected": -3.1246554851531982,
694
+ "loss": -0.1233,
695
+ "rewards/accuracies": 0.6875,
696
+ "rewards/chosen": -1.8179805278778076,
697
+ "rewards/margins": 1.3066750764846802,
698
+ "rewards/rejected": -3.1246554851531982,
699
+ "step": 230
700
+ },
701
+ {
702
+ "epoch": 0.49201779638838,
703
+ "grad_norm": 16.030732819551737,
704
+ "learning_rate": 3.000064234440111e-07,
705
+ "logits/chosen": -2.058353900909424,
706
+ "logits/rejected": -1.8866300582885742,
707
+ "logps/chosen": -1.891436219215393,
708
+ "logps/rejected": -2.721259832382202,
709
+ "loss": -0.1073,
710
+ "rewards/accuracies": 0.637499988079071,
711
+ "rewards/chosen": -1.891436219215393,
712
+ "rewards/margins": 0.8298239707946777,
713
+ "rewards/rejected": -2.721259832382202,
714
+ "step": 235
715
+ },
716
+ {
717
+ "epoch": 0.5024862601413242,
718
+ "grad_norm": 13.634356149094991,
719
+ "learning_rate": 2.910060778827554e-07,
720
+ "logits/chosen": -1.8825008869171143,
721
+ "logits/rejected": -1.8133189678192139,
722
+ "logps/chosen": -2.548449993133545,
723
+ "logps/rejected": -3.235548496246338,
724
+ "loss": -0.1088,
725
+ "rewards/accuracies": 0.6499999761581421,
726
+ "rewards/chosen": -2.548449993133545,
727
+ "rewards/margins": 0.6870983839035034,
728
+ "rewards/rejected": -3.235548496246338,
729
+ "step": 240
730
+ },
731
+ {
732
+ "epoch": 0.5129547238942685,
733
+ "grad_norm": 29.272649537767588,
734
+ "learning_rate": 2.8195076242990116e-07,
735
+ "logits/chosen": -2.0280842781066895,
736
+ "logits/rejected": -1.8670088052749634,
737
+ "logps/chosen": -1.8938684463500977,
738
+ "logps/rejected": -3.3442654609680176,
739
+ "loss": -0.1234,
740
+ "rewards/accuracies": 0.675000011920929,
741
+ "rewards/chosen": -1.8938684463500977,
742
+ "rewards/margins": 1.4503968954086304,
743
+ "rewards/rejected": -3.3442654609680176,
744
+ "step": 245
745
+ },
746
+ {
747
+ "epoch": 0.5234231876472127,
748
+ "grad_norm": 20.57218560544621,
749
+ "learning_rate": 2.7285261601056697e-07,
750
+ "logits/chosen": -2.137603759765625,
751
+ "logits/rejected": -1.8585504293441772,
752
+ "logps/chosen": -1.5400477647781372,
753
+ "logps/rejected": -3.192748546600342,
754
+ "loss": -0.1016,
755
+ "rewards/accuracies": 0.762499988079071,
756
+ "rewards/chosen": -1.5400477647781372,
757
+ "rewards/margins": 1.6527007818222046,
758
+ "rewards/rejected": -3.192748546600342,
759
+ "step": 250
760
+ },
761
+ {
762
+ "epoch": 0.533891651400157,
763
+ "grad_norm": 32.87314594164391,
764
+ "learning_rate": 2.6372383496608186e-07,
765
+ "logits/chosen": -2.2034177780151367,
766
+ "logits/rejected": -2.07490873336792,
767
+ "logps/chosen": -1.890310525894165,
768
+ "logps/rejected": -2.4908692836761475,
769
+ "loss": -0.1104,
770
+ "rewards/accuracies": 0.5874999761581421,
771
+ "rewards/chosen": -1.890310525894165,
772
+ "rewards/margins": 0.6005589365959167,
773
+ "rewards/rejected": -2.4908692836761475,
774
+ "step": 255
775
+ },
776
+ {
777
+ "epoch": 0.5443601151531012,
778
+ "grad_norm": 13.702145049518174,
779
+ "learning_rate": 2.5457665670441937e-07,
780
+ "logits/chosen": -1.9207513332366943,
781
+ "logits/rejected": -1.8922897577285767,
782
+ "logps/chosen": -1.7071243524551392,
783
+ "logps/rejected": -2.4078123569488525,
784
+ "loss": -0.1183,
785
+ "rewards/accuracies": 0.6499999761581421,
786
+ "rewards/chosen": -1.7071243524551392,
787
+ "rewards/margins": 0.7006877064704895,
788
+ "rewards/rejected": -2.4078123569488525,
789
+ "step": 260
790
+ },
791
+ {
792
+ "epoch": 0.5548285789060455,
793
+ "grad_norm": 38.3020358782431,
794
+ "learning_rate": 2.454233432955807e-07,
795
+ "logits/chosen": -2.05800199508667,
796
+ "logits/rejected": -2.0371463298797607,
797
+ "logps/chosen": -1.9121761322021484,
798
+ "logps/rejected": -2.3142733573913574,
799
+ "loss": -0.1309,
800
+ "rewards/accuracies": 0.675000011920929,
801
+ "rewards/chosen": -1.9121761322021484,
802
+ "rewards/margins": 0.4020973742008209,
803
+ "rewards/rejected": -2.3142733573913574,
804
+ "step": 265
805
+ },
806
+ {
807
+ "epoch": 0.5652970426589898,
808
+ "grad_norm": 24.19175613282827,
809
+ "learning_rate": 2.3627616503391812e-07,
810
+ "logits/chosen": -1.966522455215454,
811
+ "logits/rejected": -2.0543503761291504,
812
+ "logps/chosen": -1.5488941669464111,
813
+ "logps/rejected": -2.404010534286499,
814
+ "loss": -0.1062,
815
+ "rewards/accuracies": 0.6625000238418579,
816
+ "rewards/chosen": -1.5488941669464111,
817
+ "rewards/margins": 0.8551164865493774,
818
+ "rewards/rejected": -2.404010534286499,
819
+ "step": 270
820
+ },
821
+ {
822
+ "epoch": 0.575765506411934,
823
+ "grad_norm": 32.544264458959944,
824
+ "learning_rate": 2.2714738398943308e-07,
825
+ "logits/chosen": -2.1266486644744873,
826
+ "logits/rejected": -1.9693682193756104,
827
+ "logps/chosen": -1.345821499824524,
828
+ "logps/rejected": -2.7840895652770996,
829
+ "loss": -0.1459,
830
+ "rewards/accuracies": 0.75,
831
+ "rewards/chosen": -1.345821499824524,
832
+ "rewards/margins": 1.4382681846618652,
833
+ "rewards/rejected": -2.7840895652770996,
834
+ "step": 275
835
+ },
836
+ {
837
+ "epoch": 0.5862339701648783,
838
+ "grad_norm": 21.822579332484686,
839
+ "learning_rate": 2.1804923757009882e-07,
840
+ "logits/chosen": -2.1691737174987793,
841
+ "logits/rejected": -1.9860219955444336,
842
+ "logps/chosen": -1.2002053260803223,
843
+ "logps/rejected": -2.559629201889038,
844
+ "loss": -0.1202,
845
+ "rewards/accuracies": 0.6625000238418579,
846
+ "rewards/chosen": -1.2002053260803223,
847
+ "rewards/margins": 1.359424114227295,
848
+ "rewards/rejected": -2.559629201889038,
849
+ "step": 280
850
+ },
851
+ {
852
+ "epoch": 0.5967024339178225,
853
+ "grad_norm": 35.5262923825776,
854
+ "learning_rate": 2.089939221172446e-07,
855
+ "logits/chosen": -2.1386303901672363,
856
+ "logits/rejected": -2.101440906524658,
857
+ "logps/chosen": -1.337780237197876,
858
+ "logps/rejected": -2.1833016872406006,
859
+ "loss": -0.1175,
860
+ "rewards/accuracies": 0.6499999761581421,
861
+ "rewards/chosen": -1.337780237197876,
862
+ "rewards/margins": 0.8455215692520142,
863
+ "rewards/rejected": -2.1833016872406006,
864
+ "step": 285
865
+ },
866
+ {
867
+ "epoch": 0.6071708976707668,
868
+ "grad_norm": 27.60629251004708,
869
+ "learning_rate": 1.9999357655598891e-07,
870
+ "logits/chosen": -2.098567008972168,
871
+ "logits/rejected": -1.9679734706878662,
872
+ "logps/chosen": -1.6509898900985718,
873
+ "logps/rejected": -2.216376781463623,
874
+ "loss": -0.1164,
875
+ "rewards/accuracies": 0.6000000238418579,
876
+ "rewards/chosen": -1.6509898900985718,
877
+ "rewards/margins": 0.5653868913650513,
878
+ "rewards/rejected": -2.216376781463623,
879
+ "step": 290
880
+ },
881
+ {
882
+ "epoch": 0.6176393614237111,
883
+ "grad_norm": 29.13174757981831,
884
+ "learning_rate": 1.9106026612264315e-07,
885
+ "logits/chosen": -1.9900572299957275,
886
+ "logits/rejected": -1.8516021966934204,
887
+ "logps/chosen": -1.4829915761947632,
888
+ "logps/rejected": -2.072486400604248,
889
+ "loss": -0.1277,
890
+ "rewards/accuracies": 0.6625000238418579,
891
+ "rewards/chosen": -1.4829915761947632,
892
+ "rewards/margins": 0.5894945859909058,
893
+ "rewards/rejected": -2.072486400604248,
894
+ "step": 295
895
+ },
896
+ {
897
+ "epoch": 0.6281078251766553,
898
+ "grad_norm": 18.174081755957985,
899
+ "learning_rate": 1.8220596619089573e-07,
900
+ "logits/chosen": -1.8426250219345093,
901
+ "logits/rejected": -1.7023963928222656,
902
+ "logps/chosen": -1.5617187023162842,
903
+ "logps/rejected": -2.6344046592712402,
904
+ "loss": -0.1179,
905
+ "rewards/accuracies": 0.7250000238418579,
906
+ "rewards/chosen": -1.5617187023162842,
907
+ "rewards/margins": 1.072685956954956,
908
+ "rewards/rejected": -2.6344046592712402,
909
+ "step": 300
910
+ },
911
+ {
912
+ "epoch": 0.6385762889295996,
913
+ "grad_norm": 23.872227813445928,
914
+ "learning_rate": 1.7344254621846017e-07,
915
+ "logits/chosen": -1.8805307149887085,
916
+ "logits/rejected": -1.748567819595337,
917
+ "logps/chosen": -1.3314863443374634,
918
+ "logps/rejected": -1.9546124935150146,
919
+ "loss": -0.1335,
920
+ "rewards/accuracies": 0.637499988079071,
921
+ "rewards/chosen": -1.3314863443374634,
922
+ "rewards/margins": 0.623126208782196,
923
+ "rewards/rejected": -1.9546124935150146,
924
+ "step": 305
925
+ },
926
+ {
927
+ "epoch": 0.6490447526825438,
928
+ "grad_norm": 29.431070606486642,
929
+ "learning_rate": 1.647817538357072e-07,
930
+ "logits/chosen": -1.8305540084838867,
931
+ "logits/rejected": -1.5959422588348389,
932
+ "logps/chosen": -1.4600447416305542,
933
+ "logps/rejected": -3.613452196121216,
934
+ "loss": -0.1338,
935
+ "rewards/accuracies": 0.737500011920929,
936
+ "rewards/chosen": -1.4600447416305542,
937
+ "rewards/margins": 2.153407573699951,
938
+ "rewards/rejected": -3.613452196121216,
939
+ "step": 310
940
+ },
941
+ {
942
+ "epoch": 0.6595132164354881,
943
+ "grad_norm": 20.33172315066391,
944
+ "learning_rate": 1.562351990976095e-07,
945
+ "logits/chosen": -1.7097995281219482,
946
+ "logits/rejected": -1.444483995437622,
947
+ "logps/chosen": -1.8937146663665771,
948
+ "logps/rejected": -2.9181923866271973,
949
+ "loss": -0.147,
950
+ "rewards/accuracies": 0.762499988079071,
951
+ "rewards/chosen": -1.8937146663665771,
952
+ "rewards/margins": 1.0244777202606201,
953
+ "rewards/rejected": -2.9181923866271973,
954
+ "step": 315
955
+ },
956
+ {
957
+ "epoch": 0.6699816801884323,
958
+ "grad_norm": 11.69882174941769,
959
+ "learning_rate": 1.478143389201113e-07,
960
+ "logits/chosen": -1.5626455545425415,
961
+ "logits/rejected": -1.408778190612793,
962
+ "logps/chosen": -1.8121535778045654,
963
+ "logps/rejected": -2.9141504764556885,
964
+ "loss": -0.1327,
965
+ "rewards/accuracies": 0.6499999761581421,
966
+ "rewards/chosen": -1.8121535778045654,
967
+ "rewards/margins": 1.1019970178604126,
968
+ "rewards/rejected": -2.9141504764556885,
969
+ "step": 320
970
+ },
971
+ {
972
+ "epoch": 0.6804501439413766,
973
+ "grad_norm": 31.4758832690907,
974
+ "learning_rate": 1.3953046172178413e-07,
975
+ "logits/chosen": -1.538895606994629,
976
+ "logits/rejected": -1.3528716564178467,
977
+ "logps/chosen": -1.577697515487671,
978
+ "logps/rejected": -3.2226383686065674,
979
+ "loss": -0.108,
980
+ "rewards/accuracies": 0.7250000238418579,
981
+ "rewards/chosen": -1.577697515487671,
982
+ "rewards/margins": 1.6449410915374756,
983
+ "rewards/rejected": -3.2226383686065674,
984
+ "step": 325
985
+ },
986
+ {
987
+ "epoch": 0.6909186076943209,
988
+ "grad_norm": 30.00262748267467,
989
+ "learning_rate": 1.3139467229135998e-07,
990
+ "logits/chosen": -1.633194923400879,
991
+ "logits/rejected": -1.5735431909561157,
992
+ "logps/chosen": -1.4919952154159546,
993
+ "logps/rejected": -3.3074562549591064,
994
+ "loss": -0.1222,
995
+ "rewards/accuracies": 0.762499988079071,
996
+ "rewards/chosen": -1.4919952154159546,
997
+ "rewards/margins": 1.8154609203338623,
998
+ "rewards/rejected": -3.3074562549591064,
999
+ "step": 330
1000
+ },
1001
+ {
1002
+ "epoch": 0.7013870714472651,
1003
+ "grad_norm": 148.5006608776355,
1004
+ "learning_rate": 1.2341787690142435e-07,
1005
+ "logits/chosen": -1.653297781944275,
1006
+ "logits/rejected": -1.3402432203292847,
1007
+ "logps/chosen": -1.81472647190094,
1008
+ "logps/rejected": -3.3855223655700684,
1009
+ "loss": -0.118,
1010
+ "rewards/accuracies": 0.737500011920929,
1011
+ "rewards/chosen": -1.81472647190094,
1012
+ "rewards/margins": 1.5707957744598389,
1013
+ "rewards/rejected": -3.3855223655700684,
1014
+ "step": 335
1015
+ },
1016
+ {
1017
+ "epoch": 0.7118555352002094,
1018
+ "grad_norm": 19.087270572643252,
1019
+ "learning_rate": 1.1561076868822755e-07,
1020
+ "logits/chosen": -1.3242827653884888,
1021
+ "logits/rejected": -1.1375129222869873,
1022
+ "logps/chosen": -1.8913873434066772,
1023
+ "logps/rejected": -3.8381896018981934,
1024
+ "loss": -0.1367,
1025
+ "rewards/accuracies": 0.6625000238418579,
1026
+ "rewards/chosen": -1.8913873434066772,
1027
+ "rewards/margins": 1.9468021392822266,
1028
+ "rewards/rejected": -3.8381896018981934,
1029
+ "step": 340
1030
+ },
1031
+ {
1032
+ "epoch": 0.7223239989531536,
1033
+ "grad_norm": 12.352384869534044,
1034
+ "learning_rate": 1.0798381331721107e-07,
1035
+ "logits/chosen": -1.622854471206665,
1036
+ "logits/rejected": -1.4534351825714111,
1037
+ "logps/chosen": -1.520581603050232,
1038
+ "logps/rejected": -2.94417142868042,
1039
+ "loss": -0.1259,
1040
+ "rewards/accuracies": 0.7875000238418579,
1041
+ "rewards/chosen": -1.520581603050232,
1042
+ "rewards/margins": 1.423590064048767,
1043
+ "rewards/rejected": -2.94417142868042,
1044
+ "step": 345
1045
+ },
1046
+ {
1047
+ "epoch": 0.7327924627060979,
1048
+ "grad_norm": 14.805439320214807,
1049
+ "learning_rate": 1.0054723495346482e-07,
1050
+ "logits/chosen": -1.6887247562408447,
1051
+ "logits/rejected": -1.1422927379608154,
1052
+ "logps/chosen": -1.5419939756393433,
1053
+ "logps/rejected": -3.1307151317596436,
1054
+ "loss": -0.1286,
1055
+ "rewards/accuracies": 0.75,
1056
+ "rewards/chosen": -1.5419939756393433,
1057
+ "rewards/margins": 1.5887210369110107,
1058
+ "rewards/rejected": -3.1307151317596436,
1059
+ "step": 350
1060
+ },
1061
+ {
1062
+ "epoch": 0.7432609264590422,
1063
+ "grad_norm": 20.09481183401406,
1064
+ "learning_rate": 9.331100255592436e-08,
1065
+ "logits/chosen": -1.481506109237671,
1066
+ "logits/rejected": -1.2393898963928223,
1067
+ "logps/chosen": -1.6499707698822021,
1068
+ "logps/rejected": -2.86677885055542,
1069
+ "loss": -0.1247,
1070
+ "rewards/accuracies": 0.625,
1071
+ "rewards/chosen": -1.6499707698822021,
1072
+ "rewards/margins": 1.2168083190917969,
1073
+ "rewards/rejected": -2.86677885055542,
1074
+ "step": 355
1075
+ },
1076
+ {
1077
+ "epoch": 0.7537293902119864,
1078
+ "grad_norm": 21.363222549077037,
1079
+ "learning_rate": 8.628481651367875e-08,
1080
+ "logits/chosen": -1.531849980354309,
1081
+ "logits/rejected": -1.4370256662368774,
1082
+ "logps/chosen": -1.5883554220199585,
1083
+ "logps/rejected": -3.119661808013916,
1084
+ "loss": -0.1282,
1085
+ "rewards/accuracies": 0.699999988079071,
1086
+ "rewards/chosen": -1.5883554220199585,
1087
+ "rewards/margins": 1.531306505203247,
1088
+ "rewards/rejected": -3.119661808013916,
1089
+ "step": 360
1090
+ },
1091
+ {
1092
+ "epoch": 0.7641978539649307,
1093
+ "grad_norm": 24.973104208608277,
1094
+ "learning_rate": 7.947809564230445e-08,
1095
+ "logits/chosen": -1.7802162170410156,
1096
+ "logits/rejected": -1.3090742826461792,
1097
+ "logps/chosen": -1.4277862310409546,
1098
+ "logps/rejected": -3.4141273498535156,
1099
+ "loss": -0.1417,
1100
+ "rewards/accuracies": 0.737500011920929,
1101
+ "rewards/chosen": -1.4277862310409546,
1102
+ "rewards/margins": 1.986340880393982,
1103
+ "rewards/rejected": -3.4141273498535156,
1104
+ "step": 365
1105
+ },
1106
+ {
1107
+ "epoch": 0.7746663177178749,
1108
+ "grad_norm": 20.21283934356993,
1109
+ "learning_rate": 7.289996455765748e-08,
1110
+ "logits/chosen": -1.6334741115570068,
1111
+ "logits/rejected": -1.4152132272720337,
1112
+ "logps/chosen": -1.3029416799545288,
1113
+ "logps/rejected": -2.3641180992126465,
1114
+ "loss": -0.162,
1115
+ "rewards/accuracies": 0.800000011920929,
1116
+ "rewards/chosen": -1.3029416799545288,
1117
+ "rewards/margins": 1.0611765384674072,
1118
+ "rewards/rejected": -2.3641180992126465,
1119
+ "step": 370
1120
+ },
1121
+ {
1122
+ "epoch": 0.7851347814708192,
1123
+ "grad_norm": 19.15939039084446,
1124
+ "learning_rate": 6.655924144404906e-08,
1125
+ "logits/chosen": -1.5635571479797363,
1126
+ "logits/rejected": -1.3937865495681763,
1127
+ "logps/chosen": -1.0526535511016846,
1128
+ "logps/rejected": -3.306544780731201,
1129
+ "loss": -0.1278,
1130
+ "rewards/accuracies": 0.699999988079071,
1131
+ "rewards/chosen": -1.0526535511016846,
1132
+ "rewards/margins": 2.2538914680480957,
1133
+ "rewards/rejected": -3.306544780731201,
1134
+ "step": 375
1135
+ },
1136
+ {
1137
+ "epoch": 0.7956032452237635,
1138
+ "grad_norm": 14.808944976338482,
1139
+ "learning_rate": 6.046442623320145e-08,
1140
+ "logits/chosen": -1.7755823135375977,
1141
+ "logits/rejected": -1.715908408164978,
1142
+ "logps/chosen": -1.3490417003631592,
1143
+ "logps/rejected": -1.8757483959197998,
1144
+ "loss": -0.1173,
1145
+ "rewards/accuracies": 0.637499988079071,
1146
+ "rewards/chosen": -1.3490417003631592,
1147
+ "rewards/margins": 0.5267067551612854,
1148
+ "rewards/rejected": -1.8757483959197998,
1149
+ "step": 380
1150
+ },
1151
+ {
1152
+ "epoch": 0.8060717089767077,
1153
+ "grad_norm": 37.744351087223556,
1154
+ "learning_rate": 5.4623689209832484e-08,
1155
+ "logits/chosen": -1.621198296546936,
1156
+ "logits/rejected": -1.5765188932418823,
1157
+ "logps/chosen": -1.1883819103240967,
1158
+ "logps/rejected": -2.3331074714660645,
1159
+ "loss": -0.1254,
1160
+ "rewards/accuracies": 0.6875,
1161
+ "rewards/chosen": -1.1883819103240967,
1162
+ "rewards/margins": 1.1447254419326782,
1163
+ "rewards/rejected": -2.3331074714660645,
1164
+ "step": 385
1165
+ },
1166
+ {
1167
+ "epoch": 0.816540172729652,
1168
+ "grad_norm": 40.457315431893065,
1169
+ "learning_rate": 4.904486005914027e-08,
1170
+ "logits/chosen": -1.804915189743042,
1171
+ "logits/rejected": -1.42214834690094,
1172
+ "logps/chosen": -1.3127548694610596,
1173
+ "logps/rejected": -2.009843349456787,
1174
+ "loss": -0.1109,
1175
+ "rewards/accuracies": 0.675000011920929,
1176
+ "rewards/chosen": -1.3127548694610596,
1177
+ "rewards/margins": 0.697088360786438,
1178
+ "rewards/rejected": -2.009843349456787,
1179
+ "step": 390
1180
+ },
1181
+ {
1182
+ "epoch": 0.8270086364825961,
1183
+ "grad_norm": 24.94087598352193,
1184
+ "learning_rate": 4.373541737087263e-08,
1185
+ "logits/chosen": -1.7377560138702393,
1186
+ "logits/rejected": -1.3866163492202759,
1187
+ "logps/chosen": -1.1465688943862915,
1188
+ "logps/rejected": -2.7002861499786377,
1189
+ "loss": -0.12,
1190
+ "rewards/accuracies": 0.612500011920929,
1191
+ "rewards/chosen": -1.1465688943862915,
1192
+ "rewards/margins": 1.5537172555923462,
1193
+ "rewards/rejected": -2.7002861499786377,
1194
+ "step": 395
1195
+ },
1196
+ {
1197
+ "epoch": 0.8374771002355405,
1198
+ "grad_norm": 88.79784088043822,
1199
+ "learning_rate": 3.8702478614051345e-08,
1200
+ "logits/chosen": -1.7607088088989258,
1201
+ "logits/rejected": -1.5547138452529907,
1202
+ "logps/chosen": -1.3398617506027222,
1203
+ "logps/rejected": -2.071460008621216,
1204
+ "loss": -0.1154,
1205
+ "rewards/accuracies": 0.637499988079071,
1206
+ "rewards/chosen": -1.3398617506027222,
1207
+ "rewards/margins": 0.7315981388092041,
1208
+ "rewards/rejected": -2.071460008621216,
1209
+ "step": 400
1210
+ },
1211
+ {
1212
+ "epoch": 0.8479455639884846,
1213
+ "grad_norm": 9.248843953744515,
1214
+ "learning_rate": 3.3952790595787986e-08,
1215
+ "logits/chosen": -1.6915082931518555,
1216
+ "logits/rejected": -1.4412751197814941,
1217
+ "logps/chosen": -1.4074960947036743,
1218
+ "logps/rejected": -2.708440065383911,
1219
+ "loss": -0.1226,
1220
+ "rewards/accuracies": 0.699999988079071,
1221
+ "rewards/chosen": -1.4074960947036743,
1222
+ "rewards/margins": 1.3009440898895264,
1223
+ "rewards/rejected": -2.708440065383911,
1224
+ "step": 405
1225
+ },
1226
+ {
1227
+ "epoch": 0.8584140277414289,
1228
+ "grad_norm": 19.94643713892694,
1229
+ "learning_rate": 2.9492720416985e-08,
1230
+ "logits/chosen": -1.532260537147522,
1231
+ "logits/rejected": -1.374741554260254,
1232
+ "logps/chosen": -1.6915867328643799,
1233
+ "logps/rejected": -3.1600654125213623,
1234
+ "loss": -0.137,
1235
+ "rewards/accuracies": 0.637499988079071,
1236
+ "rewards/chosen": -1.6915867328643799,
1237
+ "rewards/margins": 1.4684789180755615,
1238
+ "rewards/rejected": -3.1600654125213623,
1239
+ "step": 410
1240
+ },
1241
+ {
1242
+ "epoch": 0.8688824914943732,
1243
+ "grad_norm": 130.75278775438778,
1244
+ "learning_rate": 2.5328246937043525e-08,
1245
+ "logits/chosen": -1.5330517292022705,
1246
+ "logits/rejected": -1.3144365549087524,
1247
+ "logps/chosen": -1.8018486499786377,
1248
+ "logps/rejected": -2.8836231231689453,
1249
+ "loss": -0.1262,
1250
+ "rewards/accuracies": 0.6499999761581421,
1251
+ "rewards/chosen": -1.8018486499786377,
1252
+ "rewards/margins": 1.0817744731903076,
1253
+ "rewards/rejected": -2.8836231231689453,
1254
+ "step": 415
1255
+ },
1256
+ {
1257
+ "epoch": 0.8793509552473174,
1258
+ "grad_norm": 25.478377748856754,
1259
+ "learning_rate": 2.1464952759020856e-08,
1260
+ "logits/chosen": -1.623173475265503,
1261
+ "logits/rejected": -1.3844332695007324,
1262
+ "logps/chosen": -1.4715114831924438,
1263
+ "logps/rejected": -2.6988463401794434,
1264
+ "loss": -0.1389,
1265
+ "rewards/accuracies": 0.625,
1266
+ "rewards/chosen": -1.4715114831924438,
1267
+ "rewards/margins": 1.2273352146148682,
1268
+ "rewards/rejected": -2.6988463401794434,
1269
+ "step": 420
1270
+ },
1271
+ {
1272
+ "epoch": 0.8898194190002617,
1273
+ "grad_norm": 44.22845937144706,
1274
+ "learning_rate": 1.7908016745981856e-08,
1275
+ "logits/chosen": -1.6746991872787476,
1276
+ "logits/rejected": -1.4371649026870728,
1277
+ "logps/chosen": -1.6984422206878662,
1278
+ "logps/rejected": -2.9744515419006348,
1279
+ "loss": -0.1342,
1280
+ "rewards/accuracies": 0.637499988079071,
1281
+ "rewards/chosen": -1.6984422206878662,
1282
+ "rewards/margins": 1.2760093212127686,
1283
+ "rewards/rejected": -2.9744515419006348,
1284
+ "step": 425
1285
+ },
1286
+ {
1287
+ "epoch": 0.9002878827532059,
1288
+ "grad_norm": 10.062945084378958,
1289
+ "learning_rate": 1.4662207078575684e-08,
1290
+ "logits/chosen": -1.5444406270980835,
1291
+ "logits/rejected": -1.2706645727157593,
1292
+ "logps/chosen": -1.5169013738632202,
1293
+ "logps/rejected": -3.1792664527893066,
1294
+ "loss": -0.1458,
1295
+ "rewards/accuracies": 0.6625000238418579,
1296
+ "rewards/chosen": -1.5169013738632202,
1297
+ "rewards/margins": 1.6623647212982178,
1298
+ "rewards/rejected": -3.1792664527893066,
1299
+ "step": 430
1300
+ },
1301
+ {
1302
+ "epoch": 0.9107563465061502,
1303
+ "grad_norm": 26.84204726450129,
1304
+ "learning_rate": 1.1731874863145142e-08,
1305
+ "logits/chosen": -1.4832799434661865,
1306
+ "logits/rejected": -1.6065374612808228,
1307
+ "logps/chosen": -1.2497235536575317,
1308
+ "logps/rejected": -2.3406825065612793,
1309
+ "loss": -0.1329,
1310
+ "rewards/accuracies": 0.612500011920929,
1311
+ "rewards/chosen": -1.2497235536575317,
1312
+ "rewards/margins": 1.0909589529037476,
1313
+ "rewards/rejected": -2.3406825065612793,
1314
+ "step": 435
1315
+ },
1316
+ {
1317
+ "epoch": 0.9212248102590945,
1318
+ "grad_norm": 14.511762413527599,
1319
+ "learning_rate": 9.12094829893642e-09,
1320
+ "logits/chosen": -1.6549100875854492,
1321
+ "logits/rejected": -1.6303857564926147,
1322
+ "logps/chosen": -1.6263110637664795,
1323
+ "logps/rejected": -2.7538211345672607,
1324
+ "loss": -0.1285,
1325
+ "rewards/accuracies": 0.6000000238418579,
1326
+ "rewards/chosen": -1.6263110637664795,
1327
+ "rewards/margins": 1.1275101900100708,
1328
+ "rewards/rejected": -2.7538211345672607,
1329
+ "step": 440
1330
+ },
1331
+ {
1332
+ "epoch": 0.9316932740120387,
1333
+ "grad_norm": 115.57689083092143,
1334
+ "learning_rate": 6.832927412229017e-09,
1335
+ "logits/chosen": -1.6242246627807617,
1336
+ "logits/rejected": -1.3819233179092407,
1337
+ "logps/chosen": -1.624022126197815,
1338
+ "logps/rejected": -2.9669742584228516,
1339
+ "loss": -0.1358,
1340
+ "rewards/accuracies": 0.737500011920929,
1341
+ "rewards/chosen": -1.624022126197815,
1342
+ "rewards/margins": 1.3429521322250366,
1343
+ "rewards/rejected": -2.9669742584228516,
1344
+ "step": 445
1345
+ },
1346
+ {
1347
+ "epoch": 0.942161737764983,
1348
+ "grad_norm": 34.008733673808756,
1349
+ "learning_rate": 4.8708793644441086e-09,
1350
+ "logits/chosen": -1.7152725458145142,
1351
+ "logits/rejected": -1.4042075872421265,
1352
+ "logps/chosen": -1.1938213109970093,
1353
+ "logps/rejected": -2.202075958251953,
1354
+ "loss": -0.1383,
1355
+ "rewards/accuracies": 0.574999988079071,
1356
+ "rewards/chosen": -1.1938213109970093,
1357
+ "rewards/margins": 1.0082545280456543,
1358
+ "rewards/rejected": -2.202075958251953,
1359
+ "step": 450
1360
+ },
1361
+ {
1362
+ "epoch": 0.9526302015179272,
1363
+ "grad_norm": 17.121700600376354,
1364
+ "learning_rate": 3.2374343405217884e-09,
1365
+ "logits/chosen": -1.6548484563827515,
1366
+ "logits/rejected": -1.4442580938339233,
1367
+ "logps/chosen": -1.8814382553100586,
1368
+ "logps/rejected": -3.200958251953125,
1369
+ "loss": -0.1119,
1370
+ "rewards/accuracies": 0.7250000238418579,
1371
+ "rewards/chosen": -1.8814382553100586,
1372
+ "rewards/margins": 1.3195202350616455,
1373
+ "rewards/rejected": -3.200958251953125,
1374
+ "step": 455
1375
+ },
1376
+ {
1377
+ "epoch": 0.9630986652708715,
1378
+ "grad_norm": 71.31324195680072,
1379
+ "learning_rate": 1.9347820230782295e-09,
1380
+ "logits/chosen": -1.741633415222168,
1381
+ "logits/rejected": -1.2778360843658447,
1382
+ "logps/chosen": -1.1031286716461182,
1383
+ "logps/rejected": -2.995805025100708,
1384
+ "loss": -0.1387,
1385
+ "rewards/accuracies": 0.737500011920929,
1386
+ "rewards/chosen": -1.1031286716461182,
1387
+ "rewards/margins": 1.892676591873169,
1388
+ "rewards/rejected": -2.995805025100708,
1389
+ "step": 460
1390
+ },
1391
+ {
1392
+ "epoch": 0.9735671290238157,
1393
+ "grad_norm": 17.102606051768348,
1394
+ "learning_rate": 9.64668657069706e-10,
1395
+ "logits/chosen": -1.6080182790756226,
1396
+ "logits/rejected": -1.2144018411636353,
1397
+ "logps/chosen": -1.5425249338150024,
1398
+ "logps/rejected": -2.9966964721679688,
1399
+ "loss": -0.1402,
1400
+ "rewards/accuracies": 0.7749999761581421,
1401
+ "rewards/chosen": -1.5425249338150024,
1402
+ "rewards/margins": 1.454171895980835,
1403
+ "rewards/rejected": -2.9966964721679688,
1404
+ "step": 465
1405
+ },
1406
+ {
1407
+ "epoch": 0.98403559277676,
1408
+ "grad_norm": 13.352840323224417,
1409
+ "learning_rate": 3.2839470889836627e-10,
1410
+ "logits/chosen": -1.6037025451660156,
1411
+ "logits/rejected": -1.2192766666412354,
1412
+ "logps/chosen": -1.357714295387268,
1413
+ "logps/rejected": -3.102246046066284,
1414
+ "loss": -0.1326,
1415
+ "rewards/accuracies": 0.75,
1416
+ "rewards/chosen": -1.357714295387268,
1417
+ "rewards/margins": 1.7445319890975952,
1418
+ "rewards/rejected": -3.102246046066284,
1419
+ "step": 470
1420
+ },
1421
+ {
1422
+ "epoch": 0.9945040565297043,
1423
+ "grad_norm": 57.54326325826418,
1424
+ "learning_rate": 2.6813123097352287e-11,
1425
+ "logits/chosen": -1.570783257484436,
1426
+ "logits/rejected": -1.3488304615020752,
1427
+ "logps/chosen": -1.446059226989746,
1428
+ "logps/rejected": -3.1210341453552246,
1429
+ "loss": -0.1382,
1430
+ "rewards/accuracies": 0.6499999761581421,
1431
+ "rewards/chosen": -1.446059226989746,
1432
+ "rewards/margins": 1.674975037574768,
1433
+ "rewards/rejected": -3.1210341453552246,
1434
+ "step": 475
1435
+ },
1436
+ {
1437
+ "epoch": 0.998691442030882,
1438
+ "step": 477,
1439
+ "total_flos": 0.0,
1440
+ "train_loss": -0.10643900037436616,
1441
+ "train_runtime": 11655.0761,
1442
+ "train_samples_per_second": 5.245,
1443
+ "train_steps_per_second": 0.041
1444
+ }
1445
+ ],
1446
+ "logging_steps": 5,
1447
+ "max_steps": 477,
1448
+ "num_input_tokens_seen": 0,
1449
+ "num_train_epochs": 1,
1450
+ "save_steps": 239,
1451
+ "stateful_callbacks": {
1452
+ "TrainerControl": {
1453
+ "args": {
1454
+ "should_epoch_stop": false,
1455
+ "should_evaluate": false,
1456
+ "should_log": false,
1457
+ "should_save": true,
1458
+ "should_training_stop": false
1459
+ },
1460
+ "attributes": {}
1461
+ }
1462
+ },
1463
+ "total_flos": 0.0,
1464
+ "train_batch_size": 2,
1465
+ "trial_name": null,
1466
+ "trial_params": null
1467
+ }