Safetensors
qwen2
AALF commited on
Commit
566b1be
·
verified ·
1 Parent(s): 93b1702

Delete trainer_state.json

Browse files
Files changed (1) hide show
  1. trainer_state.json +0 -1557
trainer_state.json DELETED
@@ -1,1557 +0,0 @@
1
- {
2
- "best_metric": null,
3
- "best_model_checkpoint": null,
4
- "epoch": 1.0,
5
- "eval_steps": 500,
6
- "global_step": 501,
7
- "is_hyper_param_search": false,
8
- "is_local_process_zero": true,
9
- "is_world_process_zero": true,
10
- "log_history": [
11
- {
12
- "epoch": 0.001996007984031936,
13
- "grad_norm": 3.4413226850733536,
14
- "learning_rate": 5.88235294117647e-09,
15
- "logits/chosen": -0.2126132994890213,
16
- "logits/rejected": -0.06049485132098198,
17
- "logps/chosen": -312.6929931640625,
18
- "logps/rejected": -303.399658203125,
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": 0.00998003992015968,
28
- "grad_norm": 3.5626484310109503,
29
- "learning_rate": 2.941176470588235e-08,
30
- "logits/chosen": -0.15043149888515472,
31
- "logits/rejected": -0.05884423851966858,
32
- "logps/chosen": -271.6809387207031,
33
- "logps/rejected": -308.14019775390625,
34
- "loss": 0.6932,
35
- "rewards/accuracies": 0.34375,
36
- "rewards/chosen": 0.00032658452983014286,
37
- "rewards/margins": -0.00044720349251292646,
38
- "rewards/rejected": 0.0007737880805507302,
39
- "step": 5
40
- },
41
- {
42
- "epoch": 0.01996007984031936,
43
- "grad_norm": 3.444766226961807,
44
- "learning_rate": 5.88235294117647e-08,
45
- "logits/chosen": -0.04885115846991539,
46
- "logits/rejected": 0.02876092866063118,
47
- "logps/chosen": -314.44207763671875,
48
- "logps/rejected": -324.5089111328125,
49
- "loss": 0.6935,
50
- "rewards/accuracies": 0.4124999940395355,
51
- "rewards/chosen": -0.0015638660406693816,
52
- "rewards/margins": -0.0018882350996136665,
53
- "rewards/rejected": 0.0003243688843213022,
54
- "step": 10
55
- },
56
- {
57
- "epoch": 0.029940119760479042,
58
- "grad_norm": 3.351776759283806,
59
- "learning_rate": 8.823529411764706e-08,
60
- "logits/chosen": -0.14645501971244812,
61
- "logits/rejected": -0.10253484547138214,
62
- "logps/chosen": -325.3197021484375,
63
- "logps/rejected": -335.1691589355469,
64
- "loss": 0.6937,
65
- "rewards/accuracies": 0.3375000059604645,
66
- "rewards/chosen": -0.0009325200808234513,
67
- "rewards/margins": -0.0027469159103929996,
68
- "rewards/rejected": 0.0018143958877772093,
69
- "step": 15
70
- },
71
- {
72
- "epoch": 0.03992015968063872,
73
- "grad_norm": 3.2486760323756765,
74
- "learning_rate": 1.176470588235294e-07,
75
- "logits/chosen": -0.17513605952262878,
76
- "logits/rejected": -0.14125120639801025,
77
- "logps/chosen": -236.529052734375,
78
- "logps/rejected": -224.99609375,
79
- "loss": 0.6932,
80
- "rewards/accuracies": 0.44999998807907104,
81
- "rewards/chosen": -0.00048039742978289723,
82
- "rewards/margins": -0.0017443184042349458,
83
- "rewards/rejected": 0.0012639210326597095,
84
- "step": 20
85
- },
86
- {
87
- "epoch": 0.0499001996007984,
88
- "grad_norm": 3.682895506825886,
89
- "learning_rate": 1.4705882352941175e-07,
90
- "logits/chosen": 0.02582230605185032,
91
- "logits/rejected": -0.026322145015001297,
92
- "logps/chosen": -276.4084777832031,
93
- "logps/rejected": -271.5510559082031,
94
- "loss": 0.6928,
95
- "rewards/accuracies": 0.5375000238418579,
96
- "rewards/chosen": -0.00010870690312003717,
97
- "rewards/margins": -2.2810349037172273e-05,
98
- "rewards/rejected": -8.589646313339472e-05,
99
- "step": 25
100
- },
101
- {
102
- "epoch": 0.059880239520958084,
103
- "grad_norm": 3.521568498969969,
104
- "learning_rate": 1.764705882352941e-07,
105
- "logits/chosen": -0.029633093625307083,
106
- "logits/rejected": 0.017462745308876038,
107
- "logps/chosen": -241.04345703125,
108
- "logps/rejected": -254.0708465576172,
109
- "loss": 0.6929,
110
- "rewards/accuracies": 0.5,
111
- "rewards/chosen": 0.001156697398982942,
112
- "rewards/margins": 0.0012088593794032931,
113
- "rewards/rejected": -5.2162260544719175e-05,
114
- "step": 30
115
- },
116
- {
117
- "epoch": 0.06986027944111776,
118
- "grad_norm": 3.4756100770970852,
119
- "learning_rate": 2.0588235294117647e-07,
120
- "logits/chosen": -0.09809896349906921,
121
- "logits/rejected": -0.12783308327198029,
122
- "logps/chosen": -316.04541015625,
123
- "logps/rejected": -303.2275085449219,
124
- "loss": 0.6926,
125
- "rewards/accuracies": 0.6499999761581421,
126
- "rewards/chosen": 0.0019062187056988478,
127
- "rewards/margins": 0.0034712664783000946,
128
- "rewards/rejected": -0.0015650471905246377,
129
- "step": 35
130
- },
131
- {
132
- "epoch": 0.07984031936127745,
133
- "grad_norm": 3.5251485554920383,
134
- "learning_rate": 2.352941176470588e-07,
135
- "logits/chosen": -0.10866469144821167,
136
- "logits/rejected": -0.08842261880636215,
137
- "logps/chosen": -300.94866943359375,
138
- "logps/rejected": -301.57635498046875,
139
- "loss": 0.6927,
140
- "rewards/accuracies": 0.48750001192092896,
141
- "rewards/chosen": 0.0005061564734205604,
142
- "rewards/margins": 0.0017319575417786837,
143
- "rewards/rejected": -0.0012258009519428015,
144
- "step": 40
145
- },
146
- {
147
- "epoch": 0.08982035928143713,
148
- "grad_norm": 3.562421126179782,
149
- "learning_rate": 2.6470588235294114e-07,
150
- "logits/chosen": -0.04412439838051796,
151
- "logits/rejected": 0.046757772564888,
152
- "logps/chosen": -276.1053771972656,
153
- "logps/rejected": -299.3190612792969,
154
- "loss": 0.692,
155
- "rewards/accuracies": 0.574999988079071,
156
- "rewards/chosen": -0.002857887651771307,
157
- "rewards/margins": 0.0004688422195613384,
158
- "rewards/rejected": -0.003326730104163289,
159
- "step": 45
160
- },
161
- {
162
- "epoch": 0.0998003992015968,
163
- "grad_norm": 3.174201376940309,
164
- "learning_rate": 2.941176470588235e-07,
165
- "logits/chosen": -0.08824320137500763,
166
- "logits/rejected": -0.043245576322078705,
167
- "logps/chosen": -271.13604736328125,
168
- "logps/rejected": -272.74102783203125,
169
- "loss": 0.6914,
170
- "rewards/accuracies": 0.6000000238418579,
171
- "rewards/chosen": -0.0016361953457817435,
172
- "rewards/margins": 0.005965009797364473,
173
- "rewards/rejected": -0.007601206190884113,
174
- "step": 50
175
- },
176
- {
177
- "epoch": 0.10978043912175649,
178
- "grad_norm": 3.3426951285606368,
179
- "learning_rate": 2.999415172560531e-07,
180
- "logits/chosen": 0.032454706728458405,
181
- "logits/rejected": 0.15314094722270966,
182
- "logps/chosen": -301.4625244140625,
183
- "logps/rejected": -304.8877258300781,
184
- "loss": 0.691,
185
- "rewards/accuracies": 0.612500011920929,
186
- "rewards/chosen": -0.007659287191927433,
187
- "rewards/margins": 0.0023522169794887304,
188
- "rewards/rejected": -0.010011503472924232,
189
- "step": 55
190
- },
191
- {
192
- "epoch": 0.11976047904191617,
193
- "grad_norm": 3.2266501701625283,
194
- "learning_rate": 2.997040092642407e-07,
195
- "logits/chosen": -0.13825412094593048,
196
- "logits/rejected": -0.06802596896886826,
197
- "logps/chosen": -273.85198974609375,
198
- "logps/rejected": -256.1399841308594,
199
- "loss": 0.6905,
200
- "rewards/accuracies": 0.699999988079071,
201
- "rewards/chosen": -0.0074735768139362335,
202
- "rewards/margins": 0.005476811900734901,
203
- "rewards/rejected": -0.012950388714671135,
204
- "step": 60
205
- },
206
- {
207
- "epoch": 0.12974051896207583,
208
- "grad_norm": 3.129826931598691,
209
- "learning_rate": 2.9928410999727467e-07,
210
- "logits/chosen": -0.03806266561150551,
211
- "logits/rejected": -0.06114925071597099,
212
- "logps/chosen": -273.815673828125,
213
- "logps/rejected": -257.2196350097656,
214
- "loss": 0.6897,
215
- "rewards/accuracies": 0.6625000238418579,
216
- "rewards/chosen": -0.010186719708144665,
217
- "rewards/margins": 0.00584446731954813,
218
- "rewards/rejected": -0.016031187027692795,
219
- "step": 65
220
- },
221
- {
222
- "epoch": 0.13972055888223553,
223
- "grad_norm": 3.2696148017727764,
224
- "learning_rate": 2.9868233103773125e-07,
225
- "logits/chosen": -0.19126296043395996,
226
- "logits/rejected": -0.1384088099002838,
227
- "logps/chosen": -289.3074645996094,
228
- "logps/rejected": -304.941650390625,
229
- "loss": 0.6901,
230
- "rewards/accuracies": 0.6499999761581421,
231
- "rewards/chosen": -0.012994937598705292,
232
- "rewards/margins": 0.006138815078884363,
233
- "rewards/rejected": -0.019133752211928368,
234
- "step": 70
235
- },
236
- {
237
- "epoch": 0.1497005988023952,
238
- "grad_norm": 3.158441752144902,
239
- "learning_rate": 2.978994055605757e-07,
240
- "logits/chosen": -0.3353852927684784,
241
- "logits/rejected": -0.20957405865192413,
242
- "logps/chosen": -282.4247131347656,
243
- "logps/rejected": -297.1376647949219,
244
- "loss": 0.6887,
245
- "rewards/accuracies": 0.7749999761581421,
246
- "rewards/chosen": -0.013628537766635418,
247
- "rewards/margins": 0.012240052223205566,
248
- "rewards/rejected": -0.02586858905851841,
249
- "step": 75
250
- },
251
- {
252
- "epoch": 0.1596806387225549,
253
- "grad_norm": 3.2922551208958932,
254
- "learning_rate": 2.9693628743990157e-07,
255
- "logits/chosen": -0.2210439145565033,
256
- "logits/rejected": -0.13906380534172058,
257
- "logps/chosen": -286.2383728027344,
258
- "logps/rejected": -284.0565490722656,
259
- "loss": 0.6885,
260
- "rewards/accuracies": 0.824999988079071,
261
- "rewards/chosen": -0.012788718566298485,
262
- "rewards/margins": 0.01518750749528408,
263
- "rewards/rejected": -0.027976226061582565,
264
- "step": 80
265
- },
266
- {
267
- "epoch": 0.16966067864271456,
268
- "grad_norm": 3.143746580193799,
269
- "learning_rate": 2.9579415008678196e-07,
270
- "logits/chosen": 0.061027251183986664,
271
- "logits/rejected": 0.020937766879796982,
272
- "logps/chosen": -261.85797119140625,
273
- "logps/rejected": -245.4117889404297,
274
- "loss": 0.6879,
275
- "rewards/accuracies": 0.7124999761581421,
276
- "rewards/chosen": -0.020817553624510765,
277
- "rewards/margins": 0.01146668754518032,
278
- "rewards/rejected": -0.032284241169691086,
279
- "step": 85
280
- },
281
- {
282
- "epoch": 0.17964071856287425,
283
- "grad_norm": 3.1363791165769634,
284
- "learning_rate": 2.944743850196487e-07,
285
- "logits/chosen": -0.18295350670814514,
286
- "logits/rejected": -0.138901025056839,
287
- "logps/chosen": -302.30377197265625,
288
- "logps/rejected": -297.4572448730469,
289
- "loss": 0.6867,
290
- "rewards/accuracies": 0.7250000238418579,
291
- "rewards/chosen": -0.022789286449551582,
292
- "rewards/margins": 0.014456982724368572,
293
- "rewards/rejected": -0.03724626824259758,
294
- "step": 90
295
- },
296
- {
297
- "epoch": 0.18962075848303392,
298
- "grad_norm": 3.375624228841244,
299
- "learning_rate": 2.92978600168942e-07,
300
- "logits/chosen": -0.21953599154949188,
301
- "logits/rejected": -0.19249823689460754,
302
- "logps/chosen": -282.0645446777344,
303
- "logps/rejected": -293.2350158691406,
304
- "loss": 0.6865,
305
- "rewards/accuracies": 0.8374999761581421,
306
- "rewards/chosen": -0.028345201164484024,
307
- "rewards/margins": 0.015181821770966053,
308
- "rewards/rejected": -0.0435270220041275,
309
- "step": 95
310
- },
311
- {
312
- "epoch": 0.1996007984031936,
313
- "grad_norm": 3.4004680540887673,
314
- "learning_rate": 2.913086179180945e-07,
315
- "logits/chosen": -0.26934295892715454,
316
- "logits/rejected": -0.24918439984321594,
317
- "logps/chosen": -267.2822570800781,
318
- "logps/rejected": -273.8163757324219,
319
- "loss": 0.6847,
320
- "rewards/accuracies": 0.762499988079071,
321
- "rewards/chosen": -0.02366602048277855,
322
- "rewards/margins": 0.018390221521258354,
323
- "rewards/rejected": -0.042056240141391754,
324
- "step": 100
325
- },
326
- {
327
- "epoch": 0.20958083832335328,
328
- "grad_norm": 3.348944145808141,
329
- "learning_rate": 2.894664728832377e-07,
330
- "logits/chosen": -0.12691783905029297,
331
- "logits/rejected": -0.10880570113658905,
332
- "logps/chosen": -306.7061462402344,
333
- "logps/rejected": -296.7961120605469,
334
- "loss": 0.684,
335
- "rewards/accuracies": 0.8125,
336
- "rewards/chosen": -0.02831300161778927,
337
- "rewards/margins": 0.024948222562670708,
338
- "rewards/rejected": -0.053261227905750275,
339
- "step": 105
340
- },
341
- {
342
- "epoch": 0.21956087824351297,
343
- "grad_norm": 3.3068624575557677,
344
- "learning_rate": 2.8745440943433595e-07,
345
- "logits/chosen": -0.39724451303482056,
346
- "logits/rejected": -0.3233721852302551,
347
- "logps/chosen": -299.71356201171875,
348
- "logps/rejected": -307.81011962890625,
349
- "loss": 0.6826,
350
- "rewards/accuracies": 0.675000011920929,
351
- "rewards/chosen": -0.037553686648607254,
352
- "rewards/margins": 0.01620485447347164,
353
- "rewards/rejected": -0.053758542984724045,
354
- "step": 110
355
- },
356
- {
357
- "epoch": 0.22954091816367264,
358
- "grad_norm": 3.5399975333931057,
359
- "learning_rate": 2.8527487896076705e-07,
360
- "logits/chosen": -0.2971861958503723,
361
- "logits/rejected": -0.26923665404319763,
362
- "logps/chosen": -323.73760986328125,
363
- "logps/rejected": -317.88238525390625,
364
- "loss": 0.6812,
365
- "rewards/accuracies": 0.8125,
366
- "rewards/chosen": -0.03361167758703232,
367
- "rewards/margins": 0.025083836168050766,
368
- "rewards/rejected": -0.058695513755083084,
369
- "step": 115
370
- },
371
- {
372
- "epoch": 0.23952095808383234,
373
- "grad_norm": 3.387251098081717,
374
- "learning_rate": 2.829305368846822e-07,
375
- "logits/chosen": -0.08930722624063492,
376
- "logits/rejected": -0.093043752014637,
377
- "logps/chosen": -298.3551330566406,
378
- "logps/rejected": -301.565185546875,
379
- "loss": 0.6807,
380
- "rewards/accuracies": 0.762499988079071,
381
- "rewards/chosen": -0.04334830120205879,
382
- "rewards/margins": 0.02416048012673855,
383
- "rewards/rejected": -0.06750877946615219,
384
- "step": 120
385
- },
386
- {
387
- "epoch": 0.249500998003992,
388
- "grad_norm": 3.458206539528906,
389
- "learning_rate": 2.8042423942578286e-07,
390
- "logits/chosen": -0.06799415498971939,
391
- "logits/rejected": 0.05015769600868225,
392
- "logps/chosen": -252.72793579101562,
393
- "logps/rejected": -279.40631103515625,
394
- "loss": 0.6788,
395
- "rewards/accuracies": 0.887499988079071,
396
- "rewards/chosen": -0.0386417880654335,
397
- "rewards/margins": 0.03730592131614685,
398
- "rewards/rejected": -0.07594770938158035,
399
- "step": 125
400
- },
401
- {
402
- "epoch": 0.25948103792415167,
403
- "grad_norm": 3.4555138461844757,
404
- "learning_rate": 2.7775904012145714e-07,
405
- "logits/chosen": -0.24486279487609863,
406
- "logits/rejected": -0.1943987011909485,
407
- "logps/chosen": -326.3341369628906,
408
- "logps/rejected": -326.3017272949219,
409
- "loss": 0.6792,
410
- "rewards/accuracies": 0.75,
411
- "rewards/chosen": -0.04846924915909767,
412
- "rewards/margins": 0.03331124037504196,
413
- "rewards/rejected": -0.08178049325942993,
414
- "step": 130
415
- },
416
- {
417
- "epoch": 0.2694610778443114,
418
- "grad_norm": 3.4103149118409104,
419
- "learning_rate": 2.749381861065149e-07,
420
- "logits/chosen": -0.11615544557571411,
421
- "logits/rejected": -0.02477203868329525,
422
- "logps/chosen": -281.5689697265625,
423
- "logps/rejected": -289.6168518066406,
424
- "loss": 0.6756,
425
- "rewards/accuracies": 0.762499988079071,
426
- "rewards/chosen": -0.04677049443125725,
427
- "rewards/margins": 0.03843165561556816,
428
- "rewards/rejected": -0.085202157497406,
429
- "step": 135
430
- },
431
- {
432
- "epoch": 0.27944111776447106,
433
- "grad_norm": 3.3294732309430946,
434
- "learning_rate": 2.719651141570541e-07,
435
- "logits/chosen": -0.07356293499469757,
436
- "logits/rejected": 0.01894739270210266,
437
- "logps/chosen": -295.6517333984375,
438
- "logps/rejected": -294.22064208984375,
439
- "loss": 0.6733,
440
- "rewards/accuracies": 0.800000011920929,
441
- "rewards/chosen": -0.04726024717092514,
442
- "rewards/margins": 0.05050128698348999,
443
- "rewards/rejected": -0.09776153415441513,
444
- "step": 140
445
- },
446
- {
447
- "epoch": 0.2894211576846307,
448
- "grad_norm": 3.335787112423438,
449
- "learning_rate": 2.6884344650327857e-07,
450
- "logits/chosen": -0.3144836127758026,
451
- "logits/rejected": -0.2471332997083664,
452
- "logps/chosen": -275.6163635253906,
453
- "logps/rejected": -283.7267150878906,
454
- "loss": 0.6741,
455
- "rewards/accuracies": 0.7250000238418579,
456
- "rewards/chosen": -0.06026310846209526,
457
- "rewards/margins": 0.03206663206219673,
458
- "rewards/rejected": -0.0923297330737114,
459
- "step": 145
460
- },
461
- {
462
- "epoch": 0.2994011976047904,
463
- "grad_norm": 3.351780879830391,
464
- "learning_rate": 2.6557698641636836e-07,
465
- "logits/chosen": -0.09479612112045288,
466
- "logits/rejected": -0.05155174061655998,
467
- "logps/chosen": -273.46527099609375,
468
- "logps/rejected": -285.8160095214844,
469
- "loss": 0.6719,
470
- "rewards/accuracies": 0.737500011920929,
471
- "rewards/chosen": -0.057442475110292435,
472
- "rewards/margins": 0.04023780673742294,
473
- "rewards/rejected": -0.09768027812242508,
474
- "step": 150
475
- },
476
- {
477
- "epoch": 0.3093812375249501,
478
- "grad_norm": 3.3953812787691744,
479
- "learning_rate": 2.6216971357477977e-07,
480
- "logits/chosen": -0.20542898774147034,
481
- "logits/rejected": -0.16760793328285217,
482
- "logps/chosen": -319.43438720703125,
483
- "logps/rejected": -337.89678955078125,
484
- "loss": 0.6677,
485
- "rewards/accuracies": 0.7124999761581421,
486
- "rewards/chosen": -0.0877736285328865,
487
- "rewards/margins": 0.05791449546813965,
488
- "rewards/rejected": -0.14568811655044556,
489
- "step": 155
490
- },
491
- {
492
- "epoch": 0.3193612774451098,
493
- "grad_norm": 3.7295101426966957,
494
- "learning_rate": 2.5862577921562015e-07,
495
- "logits/chosen": -0.26064497232437134,
496
- "logits/rejected": -0.1645033359527588,
497
- "logps/chosen": -292.02484130859375,
498
- "logps/rejected": -300.4161376953125,
499
- "loss": 0.665,
500
- "rewards/accuracies": 0.75,
501
- "rewards/chosen": -0.08468987792730331,
502
- "rewards/margins": 0.062365107238292694,
503
- "rewards/rejected": -0.1470550000667572,
504
- "step": 160
505
- },
506
- {
507
- "epoch": 0.32934131736526945,
508
- "grad_norm": 3.5278102635310775,
509
- "learning_rate": 2.549495010770048e-07,
510
- "logits/chosen": -0.2758394777774811,
511
- "logits/rejected": -0.20927894115447998,
512
- "logps/chosen": -288.8243713378906,
513
- "logps/rejected": -291.8514404296875,
514
- "loss": 0.6655,
515
- "rewards/accuracies": 0.7749999761581421,
516
- "rewards/chosen": -0.10788210481405258,
517
- "rewards/margins": 0.06151789426803589,
518
- "rewards/rejected": -0.16940000653266907,
519
- "step": 165
520
- },
521
- {
522
- "epoch": 0.3393213572854291,
523
- "grad_norm": 3.512512373193113,
524
- "learning_rate": 2.511453581375585e-07,
525
- "logits/chosen": -0.39803841710090637,
526
- "logits/rejected": -0.3213597238063812,
527
- "logps/chosen": -258.41326904296875,
528
- "logps/rejected": -263.5000915527344,
529
- "loss": 0.6625,
530
- "rewards/accuracies": 0.7250000238418579,
531
- "rewards/chosen": -0.10974812507629395,
532
- "rewards/margins": 0.04202640801668167,
533
- "rewards/rejected": -0.15177452564239502,
534
- "step": 170
535
- },
536
- {
537
- "epoch": 0.34930139720558884,
538
- "grad_norm": 3.7824536020137383,
539
- "learning_rate": 2.4721798515946963e-07,
540
- "logits/chosen": -0.2360338717699051,
541
- "logits/rejected": -0.17899933457374573,
542
- "logps/chosen": -274.8690185546875,
543
- "logps/rejected": -291.98419189453125,
544
- "loss": 0.6587,
545
- "rewards/accuracies": 0.8125,
546
- "rewards/chosen": -0.1216902956366539,
547
- "rewards/margins": 0.0732160434126854,
548
- "rewards/rejected": -0.1949063539505005,
549
- "step": 175
550
- },
551
- {
552
- "epoch": 0.3592814371257485,
553
- "grad_norm": 3.6570477679381033,
554
- "learning_rate": 2.4317216704174655e-07,
555
- "logits/chosen": -0.2678098976612091,
556
- "logits/rejected": -0.22654588520526886,
557
- "logps/chosen": -285.76885986328125,
558
- "logps/rejected": -300.3975524902344,
559
- "loss": 0.6575,
560
- "rewards/accuracies": 0.7749999761581421,
561
- "rewards/chosen": -0.13574646413326263,
562
- "rewards/margins": 0.06750482320785522,
563
- "rewards/rejected": -0.20325128734111786,
564
- "step": 180
565
- },
566
- {
567
- "epoch": 0.36926147704590817,
568
- "grad_norm": 3.705774667801494,
569
- "learning_rate": 2.3901283299055524e-07,
570
- "logits/chosen": -0.4732998013496399,
571
- "logits/rejected": -0.39881661534309387,
572
- "logps/chosen": -321.14984130859375,
573
- "logps/rejected": -326.15032958984375,
574
- "loss": 0.6514,
575
- "rewards/accuracies": 0.800000011920929,
576
- "rewards/chosen": -0.17002463340759277,
577
- "rewards/margins": 0.08681143075227737,
578
- "rewards/rejected": -0.25683605670928955,
579
- "step": 185
580
- },
581
- {
582
- "epoch": 0.37924151696606784,
583
- "grad_norm": 3.4125215861302176,
584
- "learning_rate": 2.3474505051374067e-07,
585
- "logits/chosen": -0.34785977005958557,
586
- "logits/rejected": -0.24581901729106903,
587
- "logps/chosen": -323.81732177734375,
588
- "logps/rejected": -328.8360900878906,
589
- "loss": 0.6496,
590
- "rewards/accuracies": 0.699999988079071,
591
- "rewards/chosen": -0.20683076977729797,
592
- "rewards/margins": 0.07259462773799896,
593
- "rewards/rejected": -0.2794254422187805,
594
- "step": 190
595
- },
596
- {
597
- "epoch": 0.38922155688622756,
598
- "grad_norm": 3.6035947439545715,
599
- "learning_rate": 2.3037401924684947e-07,
600
- "logits/chosen": -0.2815292775630951,
601
- "logits/rejected": -0.2620851397514343,
602
- "logps/chosen": -319.5675048828125,
603
- "logps/rejected": -332.9228210449219,
604
- "loss": 0.6501,
605
- "rewards/accuracies": 0.762499988079071,
606
- "rewards/chosen": -0.2412305772304535,
607
- "rewards/margins": 0.09057402610778809,
608
- "rewards/rejected": -0.3318046033382416,
609
- "step": 195
610
- },
611
- {
612
- "epoch": 0.3992015968063872,
613
- "grad_norm": 3.6822468215438624,
614
- "learning_rate": 2.2590506461817452e-07,
615
- "logits/chosen": -0.259944349527359,
616
- "logits/rejected": -0.20725861191749573,
617
- "logps/chosen": -298.30303955078125,
618
- "logps/rejected": -305.5474548339844,
619
- "loss": 0.6419,
620
- "rewards/accuracies": 0.800000011920929,
621
- "rewards/chosen": -0.24381069839000702,
622
- "rewards/margins": 0.11996382474899292,
623
- "rewards/rejected": -0.36377447843551636,
624
- "step": 200
625
- },
626
- {
627
- "epoch": 0.4091816367265469,
628
- "grad_norm": 3.7132773714988927,
629
- "learning_rate": 2.213436313605413e-07,
630
- "logits/chosen": -0.5560285449028015,
631
- "logits/rejected": -0.49163347482681274,
632
- "logps/chosen": -387.44171142578125,
633
- "logps/rejected": -402.449462890625,
634
- "loss": 0.6434,
635
- "rewards/accuracies": 0.7250000238418579,
636
- "rewards/chosen": -0.3169836699962616,
637
- "rewards/margins": 0.07618891447782516,
638
- "rewards/rejected": -0.39317259192466736,
639
- "step": 205
640
- },
641
- {
642
- "epoch": 0.41916167664670656,
643
- "grad_norm": 3.775451248470663,
644
- "learning_rate": 2.166952768777391e-07,
645
- "logits/chosen": -0.2941240966320038,
646
- "logits/rejected": -0.3131045997142792,
647
- "logps/chosen": -303.090087890625,
648
- "logps/rejected": -298.85699462890625,
649
- "loss": 0.6404,
650
- "rewards/accuracies": 0.75,
651
- "rewards/chosen": -0.2724608778953552,
652
- "rewards/margins": 0.1073354110121727,
653
- "rewards/rejected": -0.3797963261604309,
654
- "step": 210
655
- },
656
- {
657
- "epoch": 0.4291417165668663,
658
- "grad_norm": 3.7925188623151422,
659
- "learning_rate": 2.1196566447368126e-07,
660
- "logits/chosen": -0.26538053154945374,
661
- "logits/rejected": -0.25145062804222107,
662
- "logps/chosen": -334.455078125,
663
- "logps/rejected": -356.2786865234375,
664
- "loss": 0.6355,
665
- "rewards/accuracies": 0.75,
666
- "rewards/chosen": -0.34816819429397583,
667
- "rewards/margins": 0.12635044753551483,
668
- "rewards/rejected": -0.4745185971260071,
669
- "step": 215
670
- },
671
- {
672
- "epoch": 0.43912175648702595,
673
- "grad_norm": 3.8981243362393565,
674
- "learning_rate": 2.0716055645254113e-07,
675
- "logits/chosen": -0.46777939796447754,
676
- "logits/rejected": -0.4300519824028015,
677
- "logps/chosen": -325.1937255859375,
678
- "logps/rejected": -333.54022216796875,
679
- "loss": 0.641,
680
- "rewards/accuracies": 0.737500011920929,
681
- "rewards/chosen": -0.4105430245399475,
682
- "rewards/margins": 0.10119415819644928,
683
- "rewards/rejected": -0.5117372274398804,
684
- "step": 220
685
- },
686
- {
687
- "epoch": 0.4491017964071856,
688
- "grad_norm": 3.7392987472200354,
689
- "learning_rate": 2.0228580709827226e-07,
690
- "logits/chosen": -0.5368104577064514,
691
- "logits/rejected": -0.45491117238998413,
692
- "logps/chosen": -349.7928771972656,
693
- "logps/rejected": -371.95050048828125,
694
- "loss": 0.6326,
695
- "rewards/accuracies": 0.800000011920929,
696
- "rewards/chosen": -0.3743346631526947,
697
- "rewards/margins": 0.1426324099302292,
698
- "rewards/rejected": -0.5169671177864075,
699
- "step": 225
700
- },
701
- {
702
- "epoch": 0.4590818363273453,
703
- "grad_norm": 3.6974655152807174,
704
- "learning_rate": 1.9734735554206537e-07,
705
- "logits/chosen": -0.31408676505088806,
706
- "logits/rejected": -0.2642524242401123,
707
- "logps/chosen": -349.4893798828125,
708
- "logps/rejected": -369.08990478515625,
709
- "loss": 0.6238,
710
- "rewards/accuracies": 0.75,
711
- "rewards/chosen": -0.47439342737197876,
712
- "rewards/margins": 0.15798419713974,
713
- "rewards/rejected": -0.632377564907074,
714
- "step": 230
715
- },
716
- {
717
- "epoch": 0.469061876247505,
718
- "grad_norm": 3.9074441662298662,
719
- "learning_rate": 1.9235121852643148e-07,
720
- "logits/chosen": -0.4889460504055023,
721
- "logits/rejected": -0.48752936720848083,
722
- "logps/chosen": -331.08624267578125,
723
- "logps/rejected": -348.8202819824219,
724
- "loss": 0.6259,
725
- "rewards/accuracies": 0.7875000238418579,
726
- "rewards/chosen": -0.41649022698402405,
727
- "rewards/margins": 0.1930549144744873,
728
- "rewards/rejected": -0.609545111656189,
729
- "step": 235
730
- },
731
- {
732
- "epoch": 0.47904191616766467,
733
- "grad_norm": 3.7710978575315757,
734
- "learning_rate": 1.8730348307472826e-07,
735
- "logits/chosen": -0.4119279384613037,
736
- "logits/rejected": -0.444758802652359,
737
- "logps/chosen": -348.2624206542969,
738
- "logps/rejected": -369.9598083496094,
739
- "loss": 0.6249,
740
- "rewards/accuracies": 0.7250000238418579,
741
- "rewards/chosen": -0.5225319862365723,
742
- "rewards/margins": 0.13163016736507416,
743
- "rewards/rejected": -0.6541622281074524,
744
- "step": 240
745
- },
746
- {
747
- "epoch": 0.48902195608782434,
748
- "grad_norm": 4.08348789093204,
749
- "learning_rate": 1.822102990750595e-07,
750
- "logits/chosen": -0.5410420298576355,
751
- "logits/rejected": -0.4581461548805237,
752
- "logps/chosen": -300.3963317871094,
753
- "logps/rejected": -326.97357177734375,
754
- "loss": 0.6205,
755
- "rewards/accuracies": 0.824999988079071,
756
- "rewards/chosen": -0.4268406927585602,
757
- "rewards/margins": 0.17870576679706573,
758
- "rewards/rejected": -0.6055464744567871,
759
- "step": 245
760
- },
761
- {
762
- "epoch": 0.499001996007984,
763
- "grad_norm": 3.8249633059365977,
764
- "learning_rate": 1.7707787178758402e-07,
765
- "logits/chosen": -0.5376532673835754,
766
- "logits/rejected": -0.5166374444961548,
767
- "logps/chosen": -334.3974609375,
768
- "logps/rejected": -363.59832763671875,
769
- "loss": 0.6097,
770
- "rewards/accuracies": 0.824999988079071,
771
- "rewards/chosen": -0.5467029809951782,
772
- "rewards/margins": 0.219395712018013,
773
- "rewards/rejected": -0.7660986185073853,
774
- "step": 250
775
- },
776
- {
777
- "epoch": 0.5089820359281437,
778
- "grad_norm": 4.084608415013939,
779
- "learning_rate": 1.7191245428436176e-07,
780
- "logits/chosen": -0.6120297908782959,
781
- "logits/rejected": -0.5665668249130249,
782
- "logps/chosen": -388.49237060546875,
783
- "logps/rejected": -404.17913818359375,
784
- "loss": 0.6066,
785
- "rewards/accuracies": 0.8125,
786
- "rewards/chosen": -0.5541552305221558,
787
- "rewards/margins": 0.18654166162014008,
788
- "rewards/rejected": -0.7406969666481018,
789
- "step": 255
790
- },
791
- {
792
- "epoch": 0.5189620758483033,
793
- "grad_norm": 4.127013078261958,
794
- "learning_rate": 1.667203398309488e-07,
795
- "logits/chosen": -0.602649450302124,
796
- "logits/rejected": -0.47091540694236755,
797
- "logps/chosen": -342.5652770996094,
798
- "logps/rejected": -363.1390380859375,
799
- "loss": 0.6099,
800
- "rewards/accuracies": 0.762499988079071,
801
- "rewards/chosen": -0.5999665260314941,
802
- "rewards/margins": 0.22458717226982117,
803
- "rewards/rejected": -0.8245537877082825,
804
- "step": 260
805
- },
806
- {
807
- "epoch": 0.5289421157684631,
808
- "grad_norm": 4.181662944251744,
809
- "learning_rate": 1.6150785421902278e-07,
810
- "logits/chosen": -0.49136123061180115,
811
- "logits/rejected": -0.4355231821537018,
812
- "logps/chosen": -355.0814514160156,
813
- "logps/rejected": -400.48486328125,
814
- "loss": 0.6027,
815
- "rewards/accuracies": 0.762499988079071,
816
- "rewards/chosen": -0.6992485523223877,
817
- "rewards/margins": 0.2647705674171448,
818
- "rewards/rejected": -0.9640190005302429,
819
- "step": 265
820
- },
821
- {
822
- "epoch": 0.5389221556886228,
823
- "grad_norm": 4.187328944312369,
824
- "learning_rate": 1.5628134805937993e-07,
825
- "logits/chosen": -0.6901322603225708,
826
- "logits/rejected": -0.7155795693397522,
827
- "logps/chosen": -338.4058532714844,
828
- "logps/rejected": -357.05145263671875,
829
- "loss": 0.6044,
830
- "rewards/accuracies": 0.800000011920929,
831
- "rewards/chosen": -0.6499531865119934,
832
- "rewards/margins": 0.2078942060470581,
833
- "rewards/rejected": -0.8578473329544067,
834
- "step": 270
835
- },
836
- {
837
- "epoch": 0.5489021956087824,
838
- "grad_norm": 4.480820141822546,
839
- "learning_rate": 1.5104718904469424e-07,
840
- "logits/chosen": -0.49210062623023987,
841
- "logits/rejected": -0.4995419979095459,
842
- "logps/chosen": -362.78936767578125,
843
- "logps/rejected": -376.6024475097656,
844
- "loss": 0.6022,
845
- "rewards/accuracies": 0.7749999761581421,
846
- "rewards/chosen": -0.684400200843811,
847
- "rewards/margins": 0.22897310554981232,
848
- "rewards/rejected": -0.9133733510971069,
849
- "step": 275
850
- },
851
- {
852
- "epoch": 0.5588822355289421,
853
- "grad_norm": 4.335560690928003,
854
- "learning_rate": 1.458117541914647e-07,
855
- "logits/chosen": -0.6205674409866333,
856
- "logits/rejected": -0.5415416955947876,
857
- "logps/chosen": -363.15484619140625,
858
- "logps/rejected": -380.423828125,
859
- "loss": 0.601,
860
- "rewards/accuracies": 0.762499988079071,
861
- "rewards/chosen": -0.7319628000259399,
862
- "rewards/margins": 0.22972945868968964,
863
- "rewards/rejected": -0.9616923332214355,
864
- "step": 280
865
- },
866
- {
867
- "epoch": 0.5688622754491018,
868
- "grad_norm": 4.236266469806635,
869
- "learning_rate": 1.4058142207060298e-07,
870
- "logits/chosen": -0.5227392315864563,
871
- "logits/rejected": -0.44109734892845154,
872
- "logps/chosen": -350.81268310546875,
873
- "logps/rejected": -387.339111328125,
874
- "loss": 0.599,
875
- "rewards/accuracies": 0.8125,
876
- "rewards/chosen": -0.7962437272071838,
877
- "rewards/margins": 0.2677839994430542,
878
- "rewards/rejected": -1.0640277862548828,
879
- "step": 285
880
- },
881
- {
882
- "epoch": 0.5788423153692615,
883
- "grad_norm": 4.1859368563558546,
884
- "learning_rate": 1.353625650361276e-07,
885
- "logits/chosen": -0.4567599296569824,
886
- "logits/rejected": -0.37081751227378845,
887
- "logps/chosen": -292.90582275390625,
888
- "logps/rejected": -351.08428955078125,
889
- "loss": 0.5976,
890
- "rewards/accuracies": 0.800000011920929,
891
- "rewards/chosen": -0.7117597460746765,
892
- "rewards/margins": 0.23347418010234833,
893
- "rewards/rejected": -0.9452340006828308,
894
- "step": 290
895
- },
896
- {
897
- "epoch": 0.5888223552894212,
898
- "grad_norm": 4.963202872238042,
899
- "learning_rate": 1.301615414614316e-07,
900
- "logits/chosen": -0.5814245939254761,
901
- "logits/rejected": -0.477074533700943,
902
- "logps/chosen": -359.9764404296875,
903
- "logps/rejected": -408.649169921875,
904
- "loss": 0.5854,
905
- "rewards/accuracies": 0.8125,
906
- "rewards/chosen": -0.8252065777778625,
907
- "rewards/margins": 0.3438906669616699,
908
- "rewards/rejected": -1.1690973043441772,
909
- "step": 295
910
- },
911
- {
912
- "epoch": 0.5988023952095808,
913
- "grad_norm": 4.522965998928778,
914
- "learning_rate": 1.2498468799258466e-07,
915
- "logits/chosen": -0.6931585669517517,
916
- "logits/rejected": -0.6726911067962646,
917
- "logps/chosen": -447.56134033203125,
918
- "logps/rejected": -467.0315856933594,
919
- "loss": 0.5951,
920
- "rewards/accuracies": 0.737500011920929,
921
- "rewards/chosen": -1.0262882709503174,
922
- "rewards/margins": 0.3147868514060974,
923
- "rewards/rejected": -1.3410751819610596,
924
- "step": 300
925
- },
926
- {
927
- "epoch": 0.6087824351297405,
928
- "grad_norm": 4.334372580127937,
929
- "learning_rate": 1.1983831182810532e-07,
930
- "logits/chosen": -0.7098149657249451,
931
- "logits/rejected": -0.6338291764259338,
932
- "logps/chosen": -370.42584228515625,
933
- "logps/rejected": -416.33673095703125,
934
- "loss": 0.5741,
935
- "rewards/accuracies": 0.7250000238418579,
936
- "rewards/chosen": -0.9695831537246704,
937
- "rewards/margins": 0.28274181485176086,
938
- "rewards/rejected": -1.2523248195648193,
939
- "step": 305
940
- },
941
- {
942
- "epoch": 0.6187624750499002,
943
- "grad_norm": 4.645155562753339,
944
- "learning_rate": 1.147286830346115e-07,
945
- "logits/chosen": -0.7523954510688782,
946
- "logits/rejected": -0.6753584146499634,
947
- "logps/chosen": -415.63629150390625,
948
- "logps/rejected": -454.6825256347656,
949
- "loss": 0.5819,
950
- "rewards/accuracies": 0.7749999761581421,
951
- "rewards/chosen": -1.1591014862060547,
952
- "rewards/margins": 0.3371506929397583,
953
- "rewards/rejected": -1.496252179145813,
954
- "step": 310
955
- },
956
- {
957
- "epoch": 0.6287425149700598,
958
- "grad_norm": 4.669200509826971,
959
- "learning_rate": 1.0966202690771015e-07,
960
- "logits/chosen": -0.668272852897644,
961
- "logits/rejected": -0.5711073875427246,
962
- "logps/chosen": -412.66424560546875,
963
- "logps/rejected": -463.01971435546875,
964
- "loss": 0.5802,
965
- "rewards/accuracies": 0.800000011920929,
966
- "rewards/chosen": -1.1411970853805542,
967
- "rewards/margins": 0.34684187173843384,
968
- "rewards/rejected": -1.4880390167236328,
969
- "step": 315
970
- },
971
- {
972
- "epoch": 0.6387225548902196,
973
- "grad_norm": 4.60028355094582,
974
- "learning_rate": 1.0464451638743333e-07,
975
- "logits/chosen": -0.720443606376648,
976
- "logits/rejected": -0.633614182472229,
977
- "logps/chosen": -381.11358642578125,
978
- "logps/rejected": -422.4686584472656,
979
- "loss": 0.5803,
980
- "rewards/accuracies": 0.824999988079071,
981
- "rewards/chosen": -1.1198880672454834,
982
- "rewards/margins": 0.3332558274269104,
983
- "rewards/rejected": -1.4531439542770386,
984
- "step": 320
985
- },
986
- {
987
- "epoch": 0.6487025948103793,
988
- "grad_norm": 4.7117497209054315,
989
- "learning_rate": 9.968226453746175e-08,
990
- "logits/chosen": -0.6728391647338867,
991
- "logits/rejected": -0.6361491084098816,
992
- "logps/chosen": -388.78375244140625,
993
- "logps/rejected": -422.4482421875,
994
- "loss": 0.5744,
995
- "rewards/accuracies": 0.6875,
996
- "rewards/chosen": -1.175586462020874,
997
- "rewards/margins": 0.2871216833591461,
998
- "rewards/rejected": -1.4627082347869873,
999
- "step": 325
1000
- },
1001
- {
1002
- "epoch": 0.6586826347305389,
1003
- "grad_norm": 4.348227355176717,
1004
- "learning_rate": 9.478131709729831e-08,
1005
- "logits/chosen": -0.7032121419906616,
1006
- "logits/rejected": -0.6237956285476685,
1007
- "logps/chosen": -377.9942321777344,
1008
- "logps/rejected": -404.4933166503906,
1009
- "loss": 0.5795,
1010
- "rewards/accuracies": 0.75,
1011
- "rewards/chosen": -1.1823105812072754,
1012
- "rewards/margins": 0.23070795834064484,
1013
- "rewards/rejected": -1.4130184650421143,
1014
- "step": 330
1015
- },
1016
- {
1017
- "epoch": 0.6686626746506986,
1018
- "grad_norm": 4.567093235085812,
1019
- "learning_rate": 8.994764511646577e-08,
1020
- "logits/chosen": -0.696010410785675,
1021
- "logits/rejected": -0.6754131317138672,
1022
- "logps/chosen": -400.3501892089844,
1023
- "logps/rejected": -406.4004821777344,
1024
- "loss": 0.5887,
1025
- "rewards/accuracies": 0.75,
1026
- "rewards/chosen": -1.2299916744232178,
1027
- "rewards/margins": 0.2420133799314499,
1028
- "rewards/rejected": -1.4720051288604736,
1029
- "step": 335
1030
- },
1031
- {
1032
- "epoch": 0.6786427145708582,
1033
- "grad_norm": 5.005664260886447,
1034
- "learning_rate": 8.518713767970262e-08,
1035
- "logits/chosen": -0.7765355706214905,
1036
- "logits/rejected": -0.7291234135627747,
1037
- "logps/chosen": -440.3502502441406,
1038
- "logps/rejected": -463.7428283691406,
1039
- "loss": 0.5654,
1040
- "rewards/accuracies": 0.699999988079071,
1041
- "rewards/chosen": -1.2532676458358765,
1042
- "rewards/margins": 0.2593846023082733,
1043
- "rewards/rejected": -1.5126522779464722,
1044
- "step": 340
1045
- },
1046
- {
1047
- "epoch": 0.688622754491018,
1048
- "grad_norm": 5.103097065108184,
1049
- "learning_rate": 8.050559473202076e-08,
1050
- "logits/chosen": -0.5867692232131958,
1051
- "logits/rejected": -0.5882567167282104,
1052
- "logps/chosen": -486.0328063964844,
1053
- "logps/rejected": -492.4873046875,
1054
- "loss": 0.5909,
1055
- "rewards/accuracies": 0.7124999761581421,
1056
- "rewards/chosen": -1.4522783756256104,
1057
- "rewards/margins": 0.20934677124023438,
1058
- "rewards/rejected": -1.6616249084472656,
1059
- "step": 345
1060
- },
1061
- {
1062
- "epoch": 0.6986027944111777,
1063
- "grad_norm": 4.912813992114836,
1064
- "learning_rate": 7.590872001236538e-08,
1065
- "logits/chosen": -0.7335643768310547,
1066
- "logits/rejected": -0.6834427714347839,
1067
- "logps/chosen": -402.9820861816406,
1068
- "logps/rejected": -425.92156982421875,
1069
- "loss": 0.5716,
1070
- "rewards/accuracies": 0.7875000238418579,
1071
- "rewards/chosen": -1.3429086208343506,
1072
- "rewards/margins": 0.27442997694015503,
1073
- "rewards/rejected": -1.6173385381698608,
1074
- "step": 350
1075
- },
1076
- {
1077
- "epoch": 0.7085828343313373,
1078
- "grad_norm": 5.102923641063365,
1079
- "learning_rate": 7.140211410448815e-08,
1080
- "logits/chosen": -0.721831202507019,
1081
- "logits/rejected": -0.61592036485672,
1082
- "logps/chosen": -370.572265625,
1083
- "logps/rejected": -414.7581481933594,
1084
- "loss": 0.5721,
1085
- "rewards/accuracies": 0.7124999761581421,
1086
- "rewards/chosen": -1.2241812944412231,
1087
- "rewards/margins": 0.3305676281452179,
1088
- "rewards/rejected": -1.554748773574829,
1089
- "step": 355
1090
- },
1091
- {
1092
- "epoch": 0.718562874251497,
1093
- "grad_norm": 5.015860169175414,
1094
- "learning_rate": 6.69912676134984e-08,
1095
- "logits/chosen": -0.7092261910438538,
1096
- "logits/rejected": -0.6547318696975708,
1097
- "logps/chosen": -440.85711669921875,
1098
- "logps/rejected": -457.0215759277344,
1099
- "loss": 0.5707,
1100
- "rewards/accuracies": 0.7250000238418579,
1101
- "rewards/chosen": -1.397318720817566,
1102
- "rewards/margins": 0.24624857306480408,
1103
- "rewards/rejected": -1.6435673236846924,
1104
- "step": 360
1105
- },
1106
- {
1107
- "epoch": 0.7285429141716567,
1108
- "grad_norm": 4.774816172341646,
1109
- "learning_rate": 6.26815544764066e-08,
1110
- "logits/chosen": -0.8451409339904785,
1111
- "logits/rejected": -0.7582667469978333,
1112
- "logps/chosen": -390.70233154296875,
1113
- "logps/rejected": -433.895263671875,
1114
- "loss": 0.5661,
1115
- "rewards/accuracies": 0.675000011920929,
1116
- "rewards/chosen": -1.2642087936401367,
1117
- "rewards/margins": 0.3393135070800781,
1118
- "rewards/rejected": -1.6035223007202148,
1119
- "step": 365
1120
- },
1121
- {
1122
- "epoch": 0.7385229540918163,
1123
- "grad_norm": 4.965964761489328,
1124
- "learning_rate": 5.847822541480985e-08,
1125
- "logits/chosen": -0.6884382963180542,
1126
- "logits/rejected": -0.6832146644592285,
1127
- "logps/chosen": -505.1039123535156,
1128
- "logps/rejected": -543.939697265625,
1129
- "loss": 0.5737,
1130
- "rewards/accuracies": 0.7250000238418579,
1131
- "rewards/chosen": -1.5997720956802368,
1132
- "rewards/margins": 0.34007787704467773,
1133
- "rewards/rejected": -1.939850091934204,
1134
- "step": 370
1135
- },
1136
- {
1137
- "epoch": 0.7485029940119761,
1138
- "grad_norm": 5.257593216933955,
1139
- "learning_rate": 5.4386401537696535e-08,
1140
- "logits/chosen": -0.654364287853241,
1141
- "logits/rejected": -0.618787407875061,
1142
- "logps/chosen": -424.26031494140625,
1143
- "logps/rejected": -464.0692443847656,
1144
- "loss": 0.5656,
1145
- "rewards/accuracies": 0.75,
1146
- "rewards/chosen": -1.4769487380981445,
1147
- "rewards/margins": 0.3530046045780182,
1148
- "rewards/rejected": -1.8299534320831299,
1149
- "step": 375
1150
- },
1151
- {
1152
- "epoch": 0.7584830339321357,
1153
- "grad_norm": 5.018164548165559,
1154
- "learning_rate": 5.0411068102163756e-08,
1155
- "logits/chosen": -0.7129963636398315,
1156
- "logits/rejected": -0.6934917569160461,
1157
- "logps/chosen": -338.116943359375,
1158
- "logps/rejected": -385.53192138671875,
1159
- "loss": 0.5812,
1160
- "rewards/accuracies": 0.7875000238418579,
1161
- "rewards/chosen": -1.1324220895767212,
1162
- "rewards/margins": 0.3832218050956726,
1163
- "rewards/rejected": -1.515643835067749,
1164
- "step": 380
1165
- },
1166
- {
1167
- "epoch": 0.7684630738522954,
1168
- "grad_norm": 4.842940867729204,
1169
- "learning_rate": 4.6557068439649526e-08,
1170
- "logits/chosen": -0.6886343955993652,
1171
- "logits/rejected": -0.6565831303596497,
1172
- "logps/chosen": -475.0184631347656,
1173
- "logps/rejected": -506.1280212402344,
1174
- "loss": 0.5773,
1175
- "rewards/accuracies": 0.762499988079071,
1176
- "rewards/chosen": -1.5225403308868408,
1177
- "rewards/margins": 0.3815816044807434,
1178
- "rewards/rejected": -1.90412175655365,
1179
- "step": 385
1180
- },
1181
- {
1182
- "epoch": 0.7784431137724551,
1183
- "grad_norm": 4.981718868828084,
1184
- "learning_rate": 4.2829098055079526e-08,
1185
- "logits/chosen": -0.8457885980606079,
1186
- "logits/rejected": -0.8267861604690552,
1187
- "logps/chosen": -402.27032470703125,
1188
- "logps/rejected": -441.7845153808594,
1189
- "loss": 0.5706,
1190
- "rewards/accuracies": 0.7250000238418579,
1191
- "rewards/chosen": -1.30136239528656,
1192
- "rewards/margins": 0.37314456701278687,
1193
- "rewards/rejected": -1.6745071411132812,
1194
- "step": 390
1195
- },
1196
- {
1197
- "epoch": 0.7884231536926147,
1198
- "grad_norm": 4.76229443747579,
1199
- "learning_rate": 3.923169890611741e-08,
1200
- "logits/chosen": -0.7744909524917603,
1201
- "logits/rejected": -0.7276748418807983,
1202
- "logps/chosen": -407.0968322753906,
1203
- "logps/rejected": -451.7494201660156,
1204
- "loss": 0.5605,
1205
- "rewards/accuracies": 0.762499988079071,
1206
- "rewards/chosen": -1.4497464895248413,
1207
- "rewards/margins": 0.40129631757736206,
1208
- "rewards/rejected": -1.8510427474975586,
1209
- "step": 395
1210
- },
1211
- {
1212
- "epoch": 0.7984031936127745,
1213
- "grad_norm": 4.572364850180581,
1214
- "learning_rate": 3.576925386948951e-08,
1215
- "logits/chosen": -0.9308286905288696,
1216
- "logits/rejected": -0.8524009585380554,
1217
- "logps/chosen": -406.74578857421875,
1218
- "logps/rejected": -458.7312927246094,
1219
- "loss": 0.5563,
1220
- "rewards/accuracies": 0.762499988079071,
1221
- "rewards/chosen": -1.3046013116836548,
1222
- "rewards/margins": 0.4329892098903656,
1223
- "rewards/rejected": -1.7375905513763428,
1224
- "step": 400
1225
- },
1226
- {
1227
- "epoch": 0.8083832335329342,
1228
- "grad_norm": 4.905353303890107,
1229
- "learning_rate": 3.244598140112404e-08,
1230
- "logits/chosen": -0.8015610575675964,
1231
- "logits/rejected": -0.7380157709121704,
1232
- "logps/chosen": -399.67218017578125,
1233
- "logps/rejected": -457.02825927734375,
1234
- "loss": 0.5685,
1235
- "rewards/accuracies": 0.7124999761581421,
1236
- "rewards/chosen": -1.3612743616104126,
1237
- "rewards/margins": 0.4470387399196625,
1238
- "rewards/rejected": -1.808313012123108,
1239
- "step": 405
1240
- },
1241
- {
1242
- "epoch": 0.8183632734530938,
1243
- "grad_norm": 5.311955246309319,
1244
- "learning_rate": 2.9265930396612894e-08,
1245
- "logits/chosen": -0.7535597085952759,
1246
- "logits/rejected": -0.6952261328697205,
1247
- "logps/chosen": -399.0675964355469,
1248
- "logps/rejected": -452.47344970703125,
1249
- "loss": 0.582,
1250
- "rewards/accuracies": 0.800000011920929,
1251
- "rewards/chosen": -1.2877695560455322,
1252
- "rewards/margins": 0.3449626564979553,
1253
- "rewards/rejected": -1.6327321529388428,
1254
- "step": 410
1255
- },
1256
- {
1257
- "epoch": 0.8283433133732535,
1258
- "grad_norm": 5.053420598356344,
1259
- "learning_rate": 2.623297525825574e-08,
1260
- "logits/chosen": -0.8561833500862122,
1261
- "logits/rejected": -0.8156620264053345,
1262
- "logps/chosen": -430.49725341796875,
1263
- "logps/rejected": -483.80364990234375,
1264
- "loss": 0.5686,
1265
- "rewards/accuracies": 0.800000011920929,
1266
- "rewards/chosen": -1.4236512184143066,
1267
- "rewards/margins": 0.3616659343242645,
1268
- "rewards/rejected": -1.7853171825408936,
1269
- "step": 415
1270
- },
1271
- {
1272
- "epoch": 0.8383233532934131,
1273
- "grad_norm": 5.004297553699464,
1274
- "learning_rate": 2.335081117469777e-08,
1275
- "logits/chosen": -0.8136135339736938,
1276
- "logits/rejected": -0.7686566114425659,
1277
- "logps/chosen": -444.23419189453125,
1278
- "logps/rejected": -508.3873596191406,
1279
- "loss": 0.5616,
1280
- "rewards/accuracies": 0.824999988079071,
1281
- "rewards/chosen": -1.4251264333724976,
1282
- "rewards/margins": 0.4452531933784485,
1283
- "rewards/rejected": -1.8703796863555908,
1284
- "step": 420
1285
- },
1286
- {
1287
- "epoch": 0.8483033932135728,
1288
- "grad_norm": 5.171033512433695,
1289
- "learning_rate": 2.062294961891138e-08,
1290
- "logits/chosen": -0.9262305498123169,
1291
- "logits/rejected": -0.8602676391601562,
1292
- "logps/chosen": -419.66705322265625,
1293
- "logps/rejected": -458.7422790527344,
1294
- "loss": 0.5601,
1295
- "rewards/accuracies": 0.824999988079071,
1296
- "rewards/chosen": -1.3270304203033447,
1297
- "rewards/margins": 0.39937418699264526,
1298
- "rewards/rejected": -1.7264049053192139,
1299
- "step": 425
1300
- },
1301
- {
1302
- "epoch": 0.8582834331337326,
1303
- "grad_norm": 5.325797888088927,
1304
- "learning_rate": 1.8052714070007545e-08,
1305
- "logits/chosen": -0.7420898675918579,
1306
- "logits/rejected": -0.7444819211959839,
1307
- "logps/chosen": -404.44879150390625,
1308
- "logps/rejected": -432.81304931640625,
1309
- "loss": 0.5725,
1310
- "rewards/accuracies": 0.7124999761581421,
1311
- "rewards/chosen": -1.3653631210327148,
1312
- "rewards/margins": 0.33520203828811646,
1313
- "rewards/rejected": -1.7005650997161865,
1314
- "step": 430
1315
- },
1316
- {
1317
- "epoch": 0.8682634730538922,
1318
- "grad_norm": 5.010002107069334,
1319
- "learning_rate": 1.5643235964088065e-08,
1320
- "logits/chosen": -0.8383352160453796,
1321
- "logits/rejected": -0.8191484212875366,
1322
- "logps/chosen": -455.4537048339844,
1323
- "logps/rejected": -499.7396545410156,
1324
- "loss": 0.5761,
1325
- "rewards/accuracies": 0.824999988079071,
1326
- "rewards/chosen": -1.4779622554779053,
1327
- "rewards/margins": 0.4129580557346344,
1328
- "rewards/rejected": -1.8909204006195068,
1329
- "step": 435
1330
- },
1331
- {
1332
- "epoch": 0.8782435129740519,
1333
- "grad_norm": 4.772084864042891,
1334
- "learning_rate": 1.3397450879073446e-08,
1335
- "logits/chosen": -0.8404937982559204,
1336
- "logits/rejected": -0.8073346018791199,
1337
- "logps/chosen": -468.2997131347656,
1338
- "logps/rejected": -504.8995666503906,
1339
- "loss": 0.5648,
1340
- "rewards/accuracies": 0.762499988079071,
1341
- "rewards/chosen": -1.6386680603027344,
1342
- "rewards/margins": 0.4446341395378113,
1343
- "rewards/rejected": -2.0833022594451904,
1344
- "step": 440
1345
- },
1346
- {
1347
- "epoch": 0.8882235528942116,
1348
- "grad_norm": 4.956615120474398,
1349
- "learning_rate": 1.1318094958153047e-08,
1350
- "logits/chosen": -0.9180776476860046,
1351
- "logits/rejected": -0.8899502754211426,
1352
- "logps/chosen": -371.7776794433594,
1353
- "logps/rejected": -428.5679626464844,
1354
- "loss": 0.5901,
1355
- "rewards/accuracies": 0.7250000238418579,
1356
- "rewards/chosen": -1.327950119972229,
1357
- "rewards/margins": 0.3467249274253845,
1358
- "rewards/rejected": -1.6746749877929688,
1359
- "step": 445
1360
- },
1361
- {
1362
- "epoch": 0.8982035928143712,
1363
- "grad_norm": 4.788171459453746,
1364
- "learning_rate": 9.40770157621628e-09,
1365
- "logits/chosen": -0.7842416167259216,
1366
- "logits/rejected": -0.7082861661911011,
1367
- "logps/chosen": -417.77520751953125,
1368
- "logps/rejected": -457.5712890625,
1369
- "loss": 0.5593,
1370
- "rewards/accuracies": 0.7124999761581421,
1371
- "rewards/chosen": -1.3801639080047607,
1372
- "rewards/margins": 0.37675610184669495,
1373
- "rewards/rejected": -1.7569200992584229,
1374
- "step": 450
1375
- },
1376
- {
1377
- "epoch": 0.908183632734531,
1378
- "grad_norm": 4.581072227617621,
1379
- "learning_rate": 7.66859825332552e-09,
1380
- "logits/chosen": -0.80731201171875,
1381
- "logits/rejected": -0.7170153856277466,
1382
- "logps/chosen": -426.0374450683594,
1383
- "logps/rejected": -491.17724609375,
1384
- "loss": 0.5684,
1385
- "rewards/accuracies": 0.762499988079071,
1386
- "rewards/chosen": -1.4778865575790405,
1387
- "rewards/margins": 0.48648372292518616,
1388
- "rewards/rejected": -1.9643701314926147,
1389
- "step": 455
1390
- },
1391
- {
1392
- "epoch": 0.9181636726546906,
1393
- "grad_norm": 5.211747811331229,
1394
- "learning_rate": 6.102903818991373e-09,
1395
- "logits/chosen": -0.7492889761924744,
1396
- "logits/rejected": -0.692005455493927,
1397
- "logps/chosen": -408.1048583984375,
1398
- "logps/rejected": -465.81072998046875,
1399
- "loss": 0.5508,
1400
- "rewards/accuracies": 0.7124999761581421,
1401
- "rewards/chosen": -1.4337399005889893,
1402
- "rewards/margins": 0.4602373242378235,
1403
- "rewards/rejected": -1.893977165222168,
1404
- "step": 460
1405
- },
1406
- {
1407
- "epoch": 0.9281437125748503,
1408
- "grad_norm": 4.735277022639165,
1409
- "learning_rate": 4.712525830705338e-09,
1410
- "logits/chosen": -0.7876361012458801,
1411
- "logits/rejected": -0.7164231538772583,
1412
- "logps/chosen": -456.64654541015625,
1413
- "logps/rejected": -530.7649536132812,
1414
- "loss": 0.5495,
1415
- "rewards/accuracies": 0.8374999761581421,
1416
- "rewards/chosen": -1.4868801832199097,
1417
- "rewards/margins": 0.5220947265625,
1418
- "rewards/rejected": -2.00897479057312,
1419
- "step": 465
1420
- },
1421
- {
1422
- "epoch": 0.93812375249501,
1423
- "grad_norm": 5.092880826292085,
1424
- "learning_rate": 3.4991582498748116e-09,
1425
- "logits/chosen": -0.7580504417419434,
1426
- "logits/rejected": -0.7435745000839233,
1427
- "logps/chosen": -418.53643798828125,
1428
- "logps/rejected": -443.73016357421875,
1429
- "loss": 0.5562,
1430
- "rewards/accuracies": 0.6875,
1431
- "rewards/chosen": -1.392344355583191,
1432
- "rewards/margins": 0.26586177945137024,
1433
- "rewards/rejected": -1.6582062244415283,
1434
- "step": 470
1435
- },
1436
- {
1437
- "epoch": 0.9481037924151696,
1438
- "grad_norm": 4.985840092185972,
1439
- "learning_rate": 2.464279377992129e-09,
1440
- "logits/chosen": -0.7668736577033997,
1441
- "logits/rejected": -0.7452988624572754,
1442
- "logps/chosen": -466.7787170410156,
1443
- "logps/rejected": -506.02178955078125,
1444
- "loss": 0.5624,
1445
- "rewards/accuracies": 0.737500011920929,
1446
- "rewards/chosen": -1.6175203323364258,
1447
- "rewards/margins": 0.31748542189598083,
1448
- "rewards/rejected": -1.9350059032440186,
1449
- "step": 475
1450
- },
1451
- {
1452
- "epoch": 0.9580838323353293,
1453
- "grad_norm": 4.967479585965563,
1454
- "learning_rate": 1.60915005555175e-09,
1455
- "logits/chosen": -0.8129042387008667,
1456
- "logits/rejected": -0.739594578742981,
1457
- "logps/chosen": -447.076171875,
1458
- "logps/rejected": -484.808349609375,
1459
- "loss": 0.5671,
1460
- "rewards/accuracies": 0.824999988079071,
1461
- "rewards/chosen": -1.4572417736053467,
1462
- "rewards/margins": 0.46610623598098755,
1463
- "rewards/rejected": -1.9233481884002686,
1464
- "step": 480
1465
- },
1466
- {
1467
- "epoch": 0.9680638722554891,
1468
- "grad_norm": 5.96217808704067,
1469
- "learning_rate": 9.348121259105446e-10,
1470
- "logits/chosen": -0.7033708691596985,
1471
- "logits/rejected": -0.6699236035346985,
1472
- "logps/chosen": -417.16009521484375,
1473
- "logps/rejected": -451.69482421875,
1474
- "loss": 0.5803,
1475
- "rewards/accuracies": 0.7749999761581421,
1476
- "rewards/chosen": -1.3706567287445068,
1477
- "rewards/margins": 0.40733280777931213,
1478
- "rewards/rejected": -1.777989387512207,
1479
- "step": 485
1480
- },
1481
- {
1482
- "epoch": 0.9780439121756487,
1483
- "grad_norm": 5.734111733284475,
1484
- "learning_rate": 4.4208716596196293e-10,
1485
- "logits/chosen": -0.6810269951820374,
1486
- "logits/rejected": -0.6299699544906616,
1487
- "logps/chosen": -417.22967529296875,
1488
- "logps/rejected": -446.90447998046875,
1489
- "loss": 0.5733,
1490
- "rewards/accuracies": 0.75,
1491
- "rewards/chosen": -1.4043614864349365,
1492
- "rewards/margins": 0.3416087031364441,
1493
- "rewards/rejected": -1.745970368385315,
1494
- "step": 490
1495
- },
1496
- {
1497
- "epoch": 0.9880239520958084,
1498
- "grad_norm": 4.812399430304323,
1499
- "learning_rate": 1.3157548517124251e-10,
1500
- "logits/chosen": -0.7450631260871887,
1501
- "logits/rejected": -0.7431787252426147,
1502
- "logps/chosen": -416.9190979003906,
1503
- "logps/rejected": -472.78607177734375,
1504
- "loss": 0.5482,
1505
- "rewards/accuracies": 0.8125,
1506
- "rewards/chosen": -1.3006139993667603,
1507
- "rewards/margins": 0.49053430557250977,
1508
- "rewards/rejected": -1.7911481857299805,
1509
- "step": 495
1510
- },
1511
- {
1512
- "epoch": 0.998003992015968,
1513
- "grad_norm": 4.968518024512937,
1514
- "learning_rate": 3.6553941907879745e-12,
1515
- "logits/chosen": -0.765219509601593,
1516
- "logits/rejected": -0.7625989317893982,
1517
- "logps/chosen": -421.33843994140625,
1518
- "logps/rejected": -453.9501037597656,
1519
- "loss": 0.5756,
1520
- "rewards/accuracies": 0.699999988079071,
1521
- "rewards/chosen": -1.4310005903244019,
1522
- "rewards/margins": 0.3257448077201843,
1523
- "rewards/rejected": -1.7567455768585205,
1524
- "step": 500
1525
- },
1526
- {
1527
- "epoch": 1.0,
1528
- "step": 501,
1529
- "total_flos": 0.0,
1530
- "train_loss": 0.6225600397872354,
1531
- "train_runtime": 17487.6193,
1532
- "train_samples_per_second": 3.667,
1533
- "train_steps_per_second": 0.029
1534
- }
1535
- ],
1536
- "logging_steps": 5,
1537
- "max_steps": 501,
1538
- "num_input_tokens_seen": 0,
1539
- "num_train_epochs": 1,
1540
- "save_steps": 101,
1541
- "stateful_callbacks": {
1542
- "TrainerControl": {
1543
- "args": {
1544
- "should_epoch_stop": false,
1545
- "should_evaluate": false,
1546
- "should_log": false,
1547
- "should_save": true,
1548
- "should_training_stop": true
1549
- },
1550
- "attributes": {}
1551
- }
1552
- },
1553
- "total_flos": 0.0,
1554
- "train_batch_size": 2,
1555
- "trial_name": null,
1556
- "trial_params": null
1557
- }