File size: 10,659 Bytes
f44f70b
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9984301412872841,
  "eval_steps": 500,
  "global_step": 159,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.006279434850863423,
      "grad_norm": 5.557445139643298,
      "learning_rate": 3.125e-08,
      "logits/chosen": 0.18015038967132568,
      "logits/rejected": 0.2519298493862152,
      "logps/chosen": -297.10906982421875,
      "logps/pi_response": -130.58929443359375,
      "logps/ref_response": -130.58929443359375,
      "logps/rejected": -316.44769287109375,
      "loss": 0.6931,
      "rewards/accuracies": 0.0,
      "rewards/chosen": 0.0,
      "rewards/margins": 0.0,
      "rewards/rejected": 0.0,
      "step": 1
    },
    {
      "epoch": 0.06279434850863422,
      "grad_norm": 5.880023460230945,
      "learning_rate": 3.1249999999999997e-07,
      "logits/chosen": 0.16638809442520142,
      "logits/rejected": 0.3159521222114563,
      "logps/chosen": -243.77159118652344,
      "logps/pi_response": -120.18633270263672,
      "logps/ref_response": -120.15902709960938,
      "logps/rejected": -281.09716796875,
      "loss": 0.6928,
      "rewards/accuracies": 0.4444444477558136,
      "rewards/chosen": -0.0007160517852753401,
      "rewards/margins": 0.00040519109461456537,
      "rewards/rejected": -0.0011212429963052273,
      "step": 10
    },
    {
      "epoch": 0.12558869701726844,
      "grad_norm": 6.011032264913819,
      "learning_rate": 4.990353313429303e-07,
      "logits/chosen": 0.1316775530576706,
      "logits/rejected": 0.32217010855674744,
      "logps/chosen": -244.0759735107422,
      "logps/pi_response": -121.6043701171875,
      "logps/ref_response": -121.85536193847656,
      "logps/rejected": -266.6847229003906,
      "loss": 0.6884,
      "rewards/accuracies": 0.6937500238418579,
      "rewards/chosen": -0.011644470505416393,
      "rewards/margins": 0.011915634386241436,
      "rewards/rejected": -0.02356010302901268,
      "step": 20
    },
    {
      "epoch": 0.18838304552590268,
      "grad_norm": 6.296068063682766,
      "learning_rate": 4.882681251368548e-07,
      "logits/chosen": 0.17213600873947144,
      "logits/rejected": 0.3042981028556824,
      "logps/chosen": -244.4438018798828,
      "logps/pi_response": -109.73341369628906,
      "logps/ref_response": -110.8894271850586,
      "logps/rejected": -290.1441650390625,
      "loss": 0.6685,
      "rewards/accuracies": 0.6812499761581421,
      "rewards/chosen": -0.043184880167245865,
      "rewards/margins": 0.054663728922605515,
      "rewards/rejected": -0.09784860908985138,
      "step": 30
    },
    {
      "epoch": 0.25117739403453687,
      "grad_norm": 5.762840753548806,
      "learning_rate": 4.6604720940421207e-07,
      "logits/chosen": 0.2107941210269928,
      "logits/rejected": 0.39838385581970215,
      "logps/chosen": -287.46002197265625,
      "logps/pi_response": -125.36665344238281,
      "logps/ref_response": -129.86325073242188,
      "logps/rejected": -316.40423583984375,
      "loss": 0.6349,
      "rewards/accuracies": 0.7124999761581421,
      "rewards/chosen": -0.08985555917024612,
      "rewards/margins": 0.1335289627313614,
      "rewards/rejected": -0.22338449954986572,
      "step": 40
    },
    {
      "epoch": 0.3139717425431711,
      "grad_norm": 5.876442100853928,
      "learning_rate": 4.3344075855595097e-07,
      "logits/chosen": 0.3732234835624695,
      "logits/rejected": 0.5105798840522766,
      "logps/chosen": -247.15914916992188,
      "logps/pi_response": -109.07597351074219,
      "logps/ref_response": -116.5090560913086,
      "logps/rejected": -310.7102966308594,
      "loss": 0.606,
      "rewards/accuracies": 0.737500011920929,
      "rewards/chosen": -0.11287806183099747,
      "rewards/margins": 0.24187734723091125,
      "rewards/rejected": -0.3547554314136505,
      "step": 50
    },
    {
      "epoch": 0.37676609105180536,
      "grad_norm": 9.315356312481537,
      "learning_rate": 3.920161866827889e-07,
      "logits/chosen": 0.5162631273269653,
      "logits/rejected": 0.6798213124275208,
      "logps/chosen": -268.0201721191406,
      "logps/pi_response": -116.46971130371094,
      "logps/ref_response": -119.4989242553711,
      "logps/rejected": -347.75079345703125,
      "loss": 0.5814,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -0.20100387930870056,
      "rewards/margins": 0.41021671891212463,
      "rewards/rejected": -0.6112205386161804,
      "step": 60
    },
    {
      "epoch": 0.43956043956043955,
      "grad_norm": 7.144146562974543,
      "learning_rate": 3.4376480090239047e-07,
      "logits/chosen": 0.6009117960929871,
      "logits/rejected": 0.767359733581543,
      "logps/chosen": -236.7908172607422,
      "logps/pi_response": -114.82955169677734,
      "logps/ref_response": -116.70068359375,
      "logps/rejected": -368.27850341796875,
      "loss": 0.5568,
      "rewards/accuracies": 0.699999988079071,
      "rewards/chosen": -0.2643741965293884,
      "rewards/margins": 0.5534776449203491,
      "rewards/rejected": -0.8178518414497375,
      "step": 70
    },
    {
      "epoch": 0.5023547880690737,
      "grad_norm": 7.105190568030611,
      "learning_rate": 2.910060778827554e-07,
      "logits/chosen": 0.47289925813674927,
      "logits/rejected": 0.8142975568771362,
      "logps/chosen": -325.69622802734375,
      "logps/pi_response": -129.21812438964844,
      "logps/ref_response": -127.53900146484375,
      "logps/rejected": -348.5531311035156,
      "loss": 0.5849,
      "rewards/accuracies": 0.6499999761581421,
      "rewards/chosen": -0.3693317174911499,
      "rewards/margins": 0.37722334265708923,
      "rewards/rejected": -0.7465550303459167,
      "step": 80
    },
    {
      "epoch": 0.565149136577708,
      "grad_norm": 6.295924145315453,
      "learning_rate": 2.3627616503391812e-07,
      "logits/chosen": 0.43639689683914185,
      "logits/rejected": 0.8138043284416199,
      "logps/chosen": -301.4584045410156,
      "logps/pi_response": -135.66107177734375,
      "logps/ref_response": -129.38760375976562,
      "logps/rejected": -421.0269470214844,
      "loss": 0.5432,
      "rewards/accuracies": 0.7562500238418579,
      "rewards/chosen": -0.29649025201797485,
      "rewards/margins": 0.6172370910644531,
      "rewards/rejected": -0.9137271642684937,
      "step": 90
    },
    {
      "epoch": 0.6279434850863422,
      "grad_norm": 7.528550263483296,
      "learning_rate": 1.8220596619089573e-07,
      "logits/chosen": 0.6732058525085449,
      "logits/rejected": 0.8459190130233765,
      "logps/chosen": -283.48883056640625,
      "logps/pi_response": -121.66845703125,
      "logps/ref_response": -114.0061264038086,
      "logps/rejected": -304.6356201171875,
      "loss": 0.5426,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": -0.30830463767051697,
      "rewards/margins": 0.3349376320838928,
      "rewards/rejected": -0.6432422995567322,
      "step": 100
    },
    {
      "epoch": 0.6907378335949764,
      "grad_norm": 7.109656663434607,
      "learning_rate": 1.3139467229135998e-07,
      "logits/chosen": 0.6394161581993103,
      "logits/rejected": 0.9435567855834961,
      "logps/chosen": -298.015869140625,
      "logps/pi_response": -136.70449829101562,
      "logps/ref_response": -125.8144760131836,
      "logps/rejected": -383.9209899902344,
      "loss": 0.5314,
      "rewards/accuracies": 0.7562500238418579,
      "rewards/chosen": -0.33838868141174316,
      "rewards/margins": 0.5101521015167236,
      "rewards/rejected": -0.8485407829284668,
      "step": 110
    },
    {
      "epoch": 0.7535321821036107,
      "grad_norm": 7.158165163184529,
      "learning_rate": 8.628481651367875e-08,
      "logits/chosen": 0.5716279745101929,
      "logits/rejected": 0.8841035962104797,
      "logps/chosen": -320.6312255859375,
      "logps/pi_response": -131.89306640625,
      "logps/ref_response": -120.58707427978516,
      "logps/rejected": -369.5261535644531,
      "loss": 0.5541,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -0.4013099670410156,
      "rewards/margins": 0.5214625597000122,
      "rewards/rejected": -0.9227724075317383,
      "step": 120
    },
    {
      "epoch": 0.8163265306122449,
      "grad_norm": 6.942114883986432,
      "learning_rate": 4.904486005914027e-08,
      "logits/chosen": 0.5996646881103516,
      "logits/rejected": 0.8270283937454224,
      "logps/chosen": -277.0096130371094,
      "logps/pi_response": -136.15554809570312,
      "logps/ref_response": -123.1449966430664,
      "logps/rejected": -372.5554504394531,
      "loss": 0.5271,
      "rewards/accuracies": 0.737500011920929,
      "rewards/chosen": -0.3302404284477234,
      "rewards/margins": 0.5928428173065186,
      "rewards/rejected": -0.9230831861495972,
      "step": 130
    },
    {
      "epoch": 0.8791208791208791,
      "grad_norm": 7.301927243964412,
      "learning_rate": 2.1464952759020856e-08,
      "logits/chosen": 0.5675501823425293,
      "logits/rejected": 0.8693594932556152,
      "logps/chosen": -284.0587158203125,
      "logps/pi_response": -131.0335235595703,
      "logps/ref_response": -121.63087463378906,
      "logps/rejected": -404.29278564453125,
      "loss": 0.53,
      "rewards/accuracies": 0.78125,
      "rewards/chosen": -0.3401293456554413,
      "rewards/margins": 0.748315155506134,
      "rewards/rejected": -1.088444471359253,
      "step": 140
    },
    {
      "epoch": 0.9419152276295133,
      "grad_norm": 8.46328336009499,
      "learning_rate": 4.8708793644441086e-09,
      "logits/chosen": 0.5060345530509949,
      "logits/rejected": 0.8951950073242188,
      "logps/chosen": -317.99163818359375,
      "logps/pi_response": -145.62229919433594,
      "logps/ref_response": -132.86119079589844,
      "logps/rejected": -398.7874755859375,
      "loss": 0.5193,
      "rewards/accuracies": 0.762499988079071,
      "rewards/chosen": -0.39088305830955505,
      "rewards/margins": 0.5848211050033569,
      "rewards/rejected": -0.9757040739059448,
      "step": 150
    },
    {
      "epoch": 0.9984301412872841,
      "step": 159,
      "total_flos": 0.0,
      "train_loss": 0.5789602717513558,
      "train_runtime": 4365.2801,
      "train_samples_per_second": 4.668,
      "train_steps_per_second": 0.036
    }
  ],
  "logging_steps": 10,
  "max_steps": 159,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 100,
  "total_flos": 0.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}