JoshMe1 commited on
Commit
ca65f30
·
verified ·
1 Parent(s): 5d03e8a

Training in progress, step 300, checkpoint

Browse files
last-checkpoint/adapter_model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:90745c2fd8deb80481d23669448df93518c8c2b5bb9a500861242deb9c56cc5e
3
  size 1163996488
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:50fe11177652aa9d6b7c63ce8a3223583a8366c6aa627eb6a76b5e99fce17457
3
  size 1163996488
last-checkpoint/optimizer.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:f51d208ea34cd9a59611fb747f66bb8ef62da95fb4738c9fda1b1da724411534
3
- size 325339796
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c01db8d734fa37798dd69e2081a0b4bb8d776e4341c275f0fd35dca95428f749
3
+ size 325340244
last-checkpoint/scheduler.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:ca9c387002ad973acde429f214002dbdd1b969d41a7575de301ee649510de4ee
3
  size 1064
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f0659469b0a60d00d2d4f29c95ae15b33970c49516464d67ceae7cbeb0e2e805
3
  size 1064
last-checkpoint/trainer_state.json CHANGED
@@ -1,9 +1,9 @@
1
  {
2
  "best_metric": null,
3
  "best_model_checkpoint": null,
4
- "epoch": 0.010749509553626616,
5
  "eval_steps": 500,
6
- "global_step": 200,
7
  "is_hyper_param_search": false,
8
  "is_local_process_zero": true,
9
  "is_world_process_zero": true,
@@ -3007,6 +3007,1506 @@
3007
  "rewards/margins": 0.255648672580719,
3008
  "rewards/rejected": -0.12920860946178436,
3009
  "step": 200
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
3010
  }
3011
  ],
3012
  "logging_steps": 1,
 
1
  {
2
  "best_metric": null,
3
  "best_model_checkpoint": null,
4
+ "epoch": 0.016124264330439925,
5
  "eval_steps": 500,
6
+ "global_step": 300,
7
  "is_hyper_param_search": false,
8
  "is_local_process_zero": true,
9
  "is_world_process_zero": true,
 
3007
  "rewards/margins": 0.255648672580719,
3008
  "rewards/rejected": -0.12920860946178436,
3009
  "step": 200
3010
+ },
3011
+ {
3012
+ "epoch": 0.010803257101394749,
3013
+ "grad_norm": 9.018311500549316,
3014
+ "learning_rate": 9.999961446907352e-07,
3015
+ "logits/chosen": -0.38394927978515625,
3016
+ "logits/rejected": -0.8213692903518677,
3017
+ "logps/chosen": -96.14962768554688,
3018
+ "logps/rejected": -125.2713623046875,
3019
+ "loss": 1.215,
3020
+ "rewards/accuracies": 1.0,
3021
+ "rewards/chosen": 0.13162073493003845,
3022
+ "rewards/margins": 0.18227419257164001,
3023
+ "rewards/rejected": -0.05065345764160156,
3024
+ "step": 201
3025
+ },
3026
+ {
3027
+ "epoch": 0.010857004649162881,
3028
+ "grad_norm": 9.991246223449707,
3029
+ "learning_rate": 9.999845788223948e-07,
3030
+ "logits/chosen": -0.6805392503738403,
3031
+ "logits/rejected": -0.36724549531936646,
3032
+ "logps/chosen": -96.92276763916016,
3033
+ "logps/rejected": -110.201416015625,
3034
+ "loss": 1.1747,
3035
+ "rewards/accuracies": 0.875,
3036
+ "rewards/chosen": 0.133368581533432,
3037
+ "rewards/margins": 0.23651599884033203,
3038
+ "rewards/rejected": -0.10314741730690002,
3039
+ "step": 202
3040
+ },
3041
+ {
3042
+ "epoch": 0.010910752196931016,
3043
+ "grad_norm": 8.22262191772461,
3044
+ "learning_rate": 9.999653025733385e-07,
3045
+ "logits/chosen": -0.6609019637107849,
3046
+ "logits/rejected": -0.6106843948364258,
3047
+ "logps/chosen": -97.962646484375,
3048
+ "logps/rejected": -123.79737854003906,
3049
+ "loss": 1.1606,
3050
+ "rewards/accuracies": 0.875,
3051
+ "rewards/chosen": 0.12442468106746674,
3052
+ "rewards/margins": 0.2542736530303955,
3053
+ "rewards/rejected": -0.12984895706176758,
3054
+ "step": 203
3055
+ },
3056
+ {
3057
+ "epoch": 0.010964499744699149,
3058
+ "grad_norm": 8.588205337524414,
3059
+ "learning_rate": 9.999383162408303e-07,
3060
+ "logits/chosen": -0.40745532512664795,
3061
+ "logits/rejected": -0.5700491070747375,
3062
+ "logps/chosen": -86.30723571777344,
3063
+ "logps/rejected": -121.02368927001953,
3064
+ "loss": 1.2478,
3065
+ "rewards/accuracies": 0.875,
3066
+ "rewards/chosen": 0.10299758613109589,
3067
+ "rewards/margins": 0.14931106567382812,
3068
+ "rewards/rejected": -0.04631347581744194,
3069
+ "step": 204
3070
+ },
3071
+ {
3072
+ "epoch": 0.011018247292467281,
3073
+ "grad_norm": 9.101134300231934,
3074
+ "learning_rate": 9.999036202410323e-07,
3075
+ "logits/chosen": -0.6683564186096191,
3076
+ "logits/rejected": -0.826968789100647,
3077
+ "logps/chosen": -80.37091064453125,
3078
+ "logps/rejected": -96.48709869384766,
3079
+ "loss": 1.2045,
3080
+ "rewards/accuracies": 1.0,
3081
+ "rewards/chosen": 0.12043257057666779,
3082
+ "rewards/margins": 0.19962263107299805,
3083
+ "rewards/rejected": -0.07919006049633026,
3084
+ "step": 205
3085
+ },
3086
+ {
3087
+ "epoch": 0.011071994840235414,
3088
+ "grad_norm": 8.764567375183105,
3089
+ "learning_rate": 9.998612151090002e-07,
3090
+ "logits/chosen": -0.4916820824146271,
3091
+ "logits/rejected": -0.6521626114845276,
3092
+ "logps/chosen": -90.51093292236328,
3093
+ "logps/rejected": -133.3006134033203,
3094
+ "loss": 1.1356,
3095
+ "rewards/accuracies": 0.75,
3096
+ "rewards/chosen": 0.15498992800712585,
3097
+ "rewards/margins": 0.2916005253791809,
3098
+ "rewards/rejected": -0.13661059737205505,
3099
+ "step": 206
3100
+ },
3101
+ {
3102
+ "epoch": 0.011125742388003547,
3103
+ "grad_norm": 8.882989883422852,
3104
+ "learning_rate": 9.998111014986734e-07,
3105
+ "logits/chosen": -0.34713417291641235,
3106
+ "logits/rejected": -0.6887156367301941,
3107
+ "logps/chosen": -116.66211700439453,
3108
+ "logps/rejected": -141.1327667236328,
3109
+ "loss": 1.2164,
3110
+ "rewards/accuracies": 1.0,
3111
+ "rewards/chosen": 0.12109574675559998,
3112
+ "rewards/margins": 0.18433637917041779,
3113
+ "rewards/rejected": -0.06324061751365662,
3114
+ "step": 207
3115
+ },
3116
+ {
3117
+ "epoch": 0.01117948993577168,
3118
+ "grad_norm": 8.962393760681152,
3119
+ "learning_rate": 9.997532801828658e-07,
3120
+ "logits/chosen": -0.45506107807159424,
3121
+ "logits/rejected": -0.7329839468002319,
3122
+ "logps/chosen": -102.06887817382812,
3123
+ "logps/rejected": -111.65015411376953,
3124
+ "loss": 1.2346,
3125
+ "rewards/accuracies": 0.875,
3126
+ "rewards/chosen": 0.09608335793018341,
3127
+ "rewards/margins": 0.162027508020401,
3128
+ "rewards/rejected": -0.06594415009021759,
3129
+ "step": 208
3130
+ },
3131
+ {
3132
+ "epoch": 0.011233237483539814,
3133
+ "grad_norm": 7.876529693603516,
3134
+ "learning_rate": 9.996877520532534e-07,
3135
+ "logits/chosen": -0.4077056050300598,
3136
+ "logits/rejected": -0.6110527515411377,
3137
+ "logps/chosen": -86.82136535644531,
3138
+ "logps/rejected": -108.43330383300781,
3139
+ "loss": 1.1838,
3140
+ "rewards/accuracies": 1.0,
3141
+ "rewards/chosen": 0.17611068487167358,
3142
+ "rewards/margins": 0.21694111824035645,
3143
+ "rewards/rejected": -0.040830425918102264,
3144
+ "step": 209
3145
+ },
3146
+ {
3147
+ "epoch": 0.011286985031307947,
3148
+ "grad_norm": 6.917963981628418,
3149
+ "learning_rate": 9.996145181203615e-07,
3150
+ "logits/chosen": -0.6416649222373962,
3151
+ "logits/rejected": -0.870232105255127,
3152
+ "logps/chosen": -74.03877258300781,
3153
+ "logps/rejected": -100.16549682617188,
3154
+ "loss": 1.2437,
3155
+ "rewards/accuracies": 0.75,
3156
+ "rewards/chosen": 0.18118181824684143,
3157
+ "rewards/margins": 0.15956702828407288,
3158
+ "rewards/rejected": 0.021614791825413704,
3159
+ "step": 210
3160
+ },
3161
+ {
3162
+ "epoch": 0.01134073257907608,
3163
+ "grad_norm": 7.945036888122559,
3164
+ "learning_rate": 9.995335795135475e-07,
3165
+ "logits/chosen": -0.35453060269355774,
3166
+ "logits/rejected": -0.6333085894584656,
3167
+ "logps/chosen": -89.64887237548828,
3168
+ "logps/rejected": -103.50331115722656,
3169
+ "loss": 1.198,
3170
+ "rewards/accuracies": 0.75,
3171
+ "rewards/chosen": 0.1395391970872879,
3172
+ "rewards/margins": 0.2095623016357422,
3173
+ "rewards/rejected": -0.07002310454845428,
3174
+ "step": 211
3175
+ },
3176
+ {
3177
+ "epoch": 0.011394480126844212,
3178
+ "grad_norm": 7.781386375427246,
3179
+ "learning_rate": 9.99444937480985e-07,
3180
+ "logits/chosen": -0.8427431583404541,
3181
+ "logits/rejected": -1.0646154880523682,
3182
+ "logps/chosen": -71.52605438232422,
3183
+ "logps/rejected": -101.33668518066406,
3184
+ "loss": 1.1841,
3185
+ "rewards/accuracies": 0.875,
3186
+ "rewards/chosen": 0.1986706256866455,
3187
+ "rewards/margins": 0.2231091856956482,
3188
+ "rewards/rejected": -0.02443857118487358,
3189
+ "step": 212
3190
+ },
3191
+ {
3192
+ "epoch": 0.011448227674612345,
3193
+ "grad_norm": 8.750081062316895,
3194
+ "learning_rate": 9.993485933896437e-07,
3195
+ "logits/chosen": -0.6073715090751648,
3196
+ "logits/rejected": -0.86269211769104,
3197
+ "logps/chosen": -97.16561126708984,
3198
+ "logps/rejected": -121.15423583984375,
3199
+ "loss": 1.2527,
3200
+ "rewards/accuracies": 0.75,
3201
+ "rewards/chosen": 0.06649146229028702,
3202
+ "rewards/margins": 0.145115464925766,
3203
+ "rewards/rejected": -0.07862401008605957,
3204
+ "step": 213
3205
+ },
3206
+ {
3207
+ "epoch": 0.01150197522238048,
3208
+ "grad_norm": 8.668082237243652,
3209
+ "learning_rate": 9.99244548725269e-07,
3210
+ "logits/chosen": -0.6057989597320557,
3211
+ "logits/rejected": -0.7912963628768921,
3212
+ "logps/chosen": -86.06175231933594,
3213
+ "logps/rejected": -133.17581176757812,
3214
+ "loss": 1.1756,
3215
+ "rewards/accuracies": 0.75,
3216
+ "rewards/chosen": 0.0874997079372406,
3217
+ "rewards/margins": 0.24749141931533813,
3218
+ "rewards/rejected": -0.15999169647693634,
3219
+ "step": 214
3220
+ },
3221
+ {
3222
+ "epoch": 0.011555722770148612,
3223
+ "grad_norm": 9.294448852539062,
3224
+ "learning_rate": 9.99132805092358e-07,
3225
+ "logits/chosen": -0.487201452255249,
3226
+ "logits/rejected": -0.7378743290901184,
3227
+ "logps/chosen": -109.46067810058594,
3228
+ "logps/rejected": -152.04519653320312,
3229
+ "loss": 1.0844,
3230
+ "rewards/accuracies": 0.875,
3231
+ "rewards/chosen": 0.1331624537706375,
3232
+ "rewards/margins": 0.3561188280582428,
3233
+ "rewards/rejected": -0.22295637428760529,
3234
+ "step": 215
3235
+ },
3236
+ {
3237
+ "epoch": 0.011609470317916745,
3238
+ "grad_norm": 14.424534797668457,
3239
+ "learning_rate": 9.990133642141357e-07,
3240
+ "logits/chosen": -0.5480656623840332,
3241
+ "logits/rejected": -0.9329161643981934,
3242
+ "logps/chosen": -81.80020904541016,
3243
+ "logps/rejected": -126.82513427734375,
3244
+ "loss": 1.1742,
3245
+ "rewards/accuracies": 0.75,
3246
+ "rewards/chosen": 0.14331316947937012,
3247
+ "rewards/margins": 0.2515562176704407,
3248
+ "rewards/rejected": -0.10824304819107056,
3249
+ "step": 216
3250
+ },
3251
+ {
3252
+ "epoch": 0.011663217865684878,
3253
+ "grad_norm": 7.97386360168457,
3254
+ "learning_rate": 9.988862279325284e-07,
3255
+ "logits/chosen": -0.46164822578430176,
3256
+ "logits/rejected": -1.1121896505355835,
3257
+ "logps/chosen": -82.22102355957031,
3258
+ "logps/rejected": -95.99891662597656,
3259
+ "loss": 1.1936,
3260
+ "rewards/accuracies": 0.875,
3261
+ "rewards/chosen": 0.0752662718296051,
3262
+ "rewards/margins": 0.21177086234092712,
3263
+ "rewards/rejected": -0.13650460541248322,
3264
+ "step": 217
3265
+ },
3266
+ {
3267
+ "epoch": 0.01171696541345301,
3268
+ "grad_norm": 7.369724273681641,
3269
+ "learning_rate": 9.98751398208135e-07,
3270
+ "logits/chosen": -0.5404901504516602,
3271
+ "logits/rejected": -0.790448784828186,
3272
+ "logps/chosen": -72.71309661865234,
3273
+ "logps/rejected": -108.60038757324219,
3274
+ "loss": 1.182,
3275
+ "rewards/accuracies": 0.875,
3276
+ "rewards/chosen": 0.2458677738904953,
3277
+ "rewards/margins": 0.23527422547340393,
3278
+ "rewards/rejected": 0.010593554005026817,
3279
+ "step": 218
3280
+ },
3281
+ {
3282
+ "epoch": 0.011770712961221143,
3283
+ "grad_norm": 7.535937786102295,
3284
+ "learning_rate": 9.986088771201963e-07,
3285
+ "logits/chosen": -0.5325116515159607,
3286
+ "logits/rejected": -0.7334097623825073,
3287
+ "logps/chosen": -66.14689636230469,
3288
+ "logps/rejected": -84.96763610839844,
3289
+ "loss": 1.3108,
3290
+ "rewards/accuracies": 0.875,
3291
+ "rewards/chosen": 0.10276611149311066,
3292
+ "rewards/margins": 0.08222854137420654,
3293
+ "rewards/rejected": 0.020537566393613815,
3294
+ "step": 219
3295
+ },
3296
+ {
3297
+ "epoch": 0.011824460508989278,
3298
+ "grad_norm": 8.013921737670898,
3299
+ "learning_rate": 9.98458666866564e-07,
3300
+ "logits/chosen": -0.42551806569099426,
3301
+ "logits/rejected": -1.0276601314544678,
3302
+ "logps/chosen": -80.58146667480469,
3303
+ "logps/rejected": -114.80330657958984,
3304
+ "loss": 1.071,
3305
+ "rewards/accuracies": 1.0,
3306
+ "rewards/chosen": 0.18189138174057007,
3307
+ "rewards/margins": 0.36933794617652893,
3308
+ "rewards/rejected": -0.18744659423828125,
3309
+ "step": 220
3310
+ },
3311
+ {
3312
+ "epoch": 0.01187820805675741,
3313
+ "grad_norm": 7.9433088302612305,
3314
+ "learning_rate": 9.983007697636658e-07,
3315
+ "logits/chosen": -0.5696620941162109,
3316
+ "logits/rejected": -0.822013258934021,
3317
+ "logps/chosen": -74.19483184814453,
3318
+ "logps/rejected": -113.86673736572266,
3319
+ "loss": 1.2464,
3320
+ "rewards/accuracies": 0.625,
3321
+ "rewards/chosen": 0.08465752750635147,
3322
+ "rewards/margins": 0.15495048463344574,
3323
+ "rewards/rejected": -0.07029294967651367,
3324
+ "step": 221
3325
+ },
3326
+ {
3327
+ "epoch": 0.011931955604525544,
3328
+ "grad_norm": 7.609518527984619,
3329
+ "learning_rate": 9.981351882464707e-07,
3330
+ "logits/chosen": -0.6711792945861816,
3331
+ "logits/rejected": -0.7068201303482056,
3332
+ "logps/chosen": -72.17758178710938,
3333
+ "logps/rejected": -85.7686767578125,
3334
+ "loss": 1.271,
3335
+ "rewards/accuracies": 1.0,
3336
+ "rewards/chosen": 0.11152859032154083,
3337
+ "rewards/margins": 0.12282519787549973,
3338
+ "rewards/rejected": -0.011296607553958893,
3339
+ "step": 222
3340
+ },
3341
+ {
3342
+ "epoch": 0.011985703152293676,
3343
+ "grad_norm": 8.004049301147461,
3344
+ "learning_rate": 9.979619248684501e-07,
3345
+ "logits/chosen": -0.724856436252594,
3346
+ "logits/rejected": -0.7913885712623596,
3347
+ "logps/chosen": -94.92695617675781,
3348
+ "logps/rejected": -129.05795288085938,
3349
+ "loss": 1.0848,
3350
+ "rewards/accuracies": 0.75,
3351
+ "rewards/chosen": 0.19596806168556213,
3352
+ "rewards/margins": 0.36456066370010376,
3353
+ "rewards/rejected": -0.16859260201454163,
3354
+ "step": 223
3355
+ },
3356
+ {
3357
+ "epoch": 0.012039450700061809,
3358
+ "grad_norm": 9.519268035888672,
3359
+ "learning_rate": 9.9778098230154e-07,
3360
+ "logits/chosen": -0.4081488847732544,
3361
+ "logits/rejected": -0.9913891553878784,
3362
+ "logps/chosen": -66.12991333007812,
3363
+ "logps/rejected": -126.65169525146484,
3364
+ "loss": 1.232,
3365
+ "rewards/accuracies": 0.75,
3366
+ "rewards/chosen": 0.19431248307228088,
3367
+ "rewards/margins": 0.19488951563835144,
3368
+ "rewards/rejected": -0.0005770251154899597,
3369
+ "step": 224
3370
+ },
3371
+ {
3372
+ "epoch": 0.012093198247829944,
3373
+ "grad_norm": 8.631113052368164,
3374
+ "learning_rate": 9.975923633360984e-07,
3375
+ "logits/chosen": -0.37461525201797485,
3376
+ "logits/rejected": -0.8124234080314636,
3377
+ "logps/chosen": -91.42263793945312,
3378
+ "logps/rejected": -120.60844421386719,
3379
+ "loss": 1.0942,
3380
+ "rewards/accuracies": 0.875,
3381
+ "rewards/chosen": 0.16796961426734924,
3382
+ "rewards/margins": 0.33035793900489807,
3383
+ "rewards/rejected": -0.16238832473754883,
3384
+ "step": 225
3385
+ },
3386
+ {
3387
+ "epoch": 0.012146945795598076,
3388
+ "grad_norm": 8.081117630004883,
3389
+ "learning_rate": 9.973960708808631e-07,
3390
+ "logits/chosen": -0.4398130774497986,
3391
+ "logits/rejected": -0.6970310211181641,
3392
+ "logps/chosen": -82.72227478027344,
3393
+ "logps/rejected": -120.73685455322266,
3394
+ "loss": 1.07,
3395
+ "rewards/accuracies": 0.875,
3396
+ "rewards/chosen": 0.13794194161891937,
3397
+ "rewards/margins": 0.38850194215774536,
3398
+ "rewards/rejected": -0.2505600154399872,
3399
+ "step": 226
3400
+ },
3401
+ {
3402
+ "epoch": 0.012200693343366209,
3403
+ "grad_norm": 7.0437750816345215,
3404
+ "learning_rate": 9.971921079629069e-07,
3405
+ "logits/chosen": -0.5556354522705078,
3406
+ "logits/rejected": -0.864974856376648,
3407
+ "logps/chosen": -73.24920654296875,
3408
+ "logps/rejected": -90.26593017578125,
3409
+ "loss": 1.1827,
3410
+ "rewards/accuracies": 0.875,
3411
+ "rewards/chosen": 0.1013791561126709,
3412
+ "rewards/margins": 0.22670727968215942,
3413
+ "rewards/rejected": -0.12532812356948853,
3414
+ "step": 227
3415
+ },
3416
+ {
3417
+ "epoch": 0.012254440891134342,
3418
+ "grad_norm": 8.556796073913574,
3419
+ "learning_rate": 9.969804777275898e-07,
3420
+ "logits/chosen": -0.35300546884536743,
3421
+ "logits/rejected": -0.6973484754562378,
3422
+ "logps/chosen": -95.61795806884766,
3423
+ "logps/rejected": -130.8153533935547,
3424
+ "loss": 1.0892,
3425
+ "rewards/accuracies": 0.875,
3426
+ "rewards/chosen": 0.12840043008327484,
3427
+ "rewards/margins": 0.3624500334262848,
3428
+ "rewards/rejected": -0.23404961824417114,
3429
+ "step": 228
3430
+ },
3431
+ {
3432
+ "epoch": 0.012308188438902475,
3433
+ "grad_norm": 7.119718551635742,
3434
+ "learning_rate": 9.967611834385122e-07,
3435
+ "logits/chosen": -0.3875174820423126,
3436
+ "logits/rejected": -0.5986741781234741,
3437
+ "logps/chosen": -67.04378509521484,
3438
+ "logps/rejected": -96.16484832763672,
3439
+ "loss": 1.1726,
3440
+ "rewards/accuracies": 0.875,
3441
+ "rewards/chosen": 0.1416163444519043,
3442
+ "rewards/margins": 0.23682399094104767,
3443
+ "rewards/rejected": -0.09520764648914337,
3444
+ "step": 229
3445
+ },
3446
+ {
3447
+ "epoch": 0.012361935986670607,
3448
+ "grad_norm": 7.879711151123047,
3449
+ "learning_rate": 9.965342284774631e-07,
3450
+ "logits/chosen": -0.48159751296043396,
3451
+ "logits/rejected": -0.6574340462684631,
3452
+ "logps/chosen": -80.20025634765625,
3453
+ "logps/rejected": -125.88029479980469,
3454
+ "loss": 1.1334,
3455
+ "rewards/accuracies": 1.0,
3456
+ "rewards/chosen": 0.11776607483625412,
3457
+ "rewards/margins": 0.2744428515434265,
3458
+ "rewards/rejected": -0.1566767692565918,
3459
+ "step": 230
3460
+ },
3461
+ {
3462
+ "epoch": 0.012415683534438742,
3463
+ "grad_norm": 9.587478637695312,
3464
+ "learning_rate": 9.962996163443688e-07,
3465
+ "logits/chosen": -0.37069642543792725,
3466
+ "logits/rejected": -0.8089166879653931,
3467
+ "logps/chosen": -98.1978988647461,
3468
+ "logps/rejected": -132.13845825195312,
3469
+ "loss": 1.1718,
3470
+ "rewards/accuracies": 0.875,
3471
+ "rewards/chosen": 0.18037399649620056,
3472
+ "rewards/margins": 0.26343971490859985,
3473
+ "rewards/rejected": -0.0830657035112381,
3474
+ "step": 231
3475
+ },
3476
+ {
3477
+ "epoch": 0.012469431082206875,
3478
+ "grad_norm": 9.129337310791016,
3479
+ "learning_rate": 9.960573506572389e-07,
3480
+ "logits/chosen": -0.5636274814605713,
3481
+ "logits/rejected": -1.1714890003204346,
3482
+ "logps/chosen": -93.2618408203125,
3483
+ "logps/rejected": -97.5640869140625,
3484
+ "loss": 1.1944,
3485
+ "rewards/accuracies": 0.75,
3486
+ "rewards/chosen": 0.12527954578399658,
3487
+ "rewards/margins": 0.21808025240898132,
3488
+ "rewards/rejected": -0.09280071407556534,
3489
+ "step": 232
3490
+ },
3491
+ {
3492
+ "epoch": 0.012523178629975007,
3493
+ "grad_norm": 8.361261367797852,
3494
+ "learning_rate": 9.958074351521096e-07,
3495
+ "logits/chosen": -0.47556164860725403,
3496
+ "logits/rejected": -0.7843296527862549,
3497
+ "logps/chosen": -75.67303466796875,
3498
+ "logps/rejected": -99.92214965820312,
3499
+ "loss": 1.1867,
3500
+ "rewards/accuracies": 0.875,
3501
+ "rewards/chosen": 0.20813600718975067,
3502
+ "rewards/margins": 0.21530234813690186,
3503
+ "rewards/rejected": -0.007166337221860886,
3504
+ "step": 233
3505
+ },
3506
+ {
3507
+ "epoch": 0.01257692617774314,
3508
+ "grad_norm": 8.374642372131348,
3509
+ "learning_rate": 9.955498736829874e-07,
3510
+ "logits/chosen": -0.6883563995361328,
3511
+ "logits/rejected": -1.486838459968567,
3512
+ "logps/chosen": -76.17497253417969,
3513
+ "logps/rejected": -96.84264373779297,
3514
+ "loss": 1.1193,
3515
+ "rewards/accuracies": 0.875,
3516
+ "rewards/chosen": 0.22805273532867432,
3517
+ "rewards/margins": 0.32633835077285767,
3518
+ "rewards/rejected": -0.09828563034534454,
3519
+ "step": 234
3520
+ },
3521
+ {
3522
+ "epoch": 0.012630673725511273,
3523
+ "grad_norm": 9.872260093688965,
3524
+ "learning_rate": 9.952846702217885e-07,
3525
+ "logits/chosen": -0.6219691038131714,
3526
+ "logits/rejected": -0.8194465637207031,
3527
+ "logps/chosen": -74.91395568847656,
3528
+ "logps/rejected": -106.57698822021484,
3529
+ "loss": 1.1828,
3530
+ "rewards/accuracies": 0.75,
3531
+ "rewards/chosen": 0.2802756428718567,
3532
+ "rewards/margins": 0.22285690903663635,
3533
+ "rewards/rejected": 0.05741872638463974,
3534
+ "step": 235
3535
+ },
3536
+ {
3537
+ "epoch": 0.012684421273279407,
3538
+ "grad_norm": 9.46263599395752,
3539
+ "learning_rate": 9.950118288582787e-07,
3540
+ "logits/chosen": -0.5394343137741089,
3541
+ "logits/rejected": -0.8429723381996155,
3542
+ "logps/chosen": -78.40309143066406,
3543
+ "logps/rejected": -119.71903991699219,
3544
+ "loss": 1.2401,
3545
+ "rewards/accuracies": 0.875,
3546
+ "rewards/chosen": 0.16660980880260468,
3547
+ "rewards/margins": 0.15729470551013947,
3548
+ "rewards/rejected": 0.009315107017755508,
3549
+ "step": 236
3550
+ },
3551
+ {
3552
+ "epoch": 0.01273816882104754,
3553
+ "grad_norm": 7.394505500793457,
3554
+ "learning_rate": 9.947313538000092e-07,
3555
+ "logits/chosen": -0.5725371837615967,
3556
+ "logits/rejected": -0.6441730260848999,
3557
+ "logps/chosen": -69.10836029052734,
3558
+ "logps/rejected": -93.74408721923828,
3559
+ "loss": 1.1444,
3560
+ "rewards/accuracies": 1.0,
3561
+ "rewards/chosen": 0.11338989436626434,
3562
+ "rewards/margins": 0.2632651925086975,
3563
+ "rewards/rejected": -0.14987531304359436,
3564
+ "step": 237
3565
+ },
3566
+ {
3567
+ "epoch": 0.012791916368815673,
3568
+ "grad_norm": 7.916663646697998,
3569
+ "learning_rate": 9.944432493722524e-07,
3570
+ "logits/chosen": -0.38268229365348816,
3571
+ "logits/rejected": -0.5627045035362244,
3572
+ "logps/chosen": -68.636474609375,
3573
+ "logps/rejected": -103.48701477050781,
3574
+ "loss": 1.1002,
3575
+ "rewards/accuracies": 0.875,
3576
+ "rewards/chosen": 0.22320647537708282,
3577
+ "rewards/margins": 0.32132166624069214,
3578
+ "rewards/rejected": -0.09811516106128693,
3579
+ "step": 238
3580
+ },
3581
+ {
3582
+ "epoch": 0.012845663916583806,
3583
+ "grad_norm": 8.479836463928223,
3584
+ "learning_rate": 9.941475200179346e-07,
3585
+ "logits/chosen": -0.5663607716560364,
3586
+ "logits/rejected": -0.7409512996673584,
3587
+ "logps/chosen": -89.14552307128906,
3588
+ "logps/rejected": -122.3517837524414,
3589
+ "loss": 1.1899,
3590
+ "rewards/accuracies": 0.75,
3591
+ "rewards/chosen": 0.13951130211353302,
3592
+ "rewards/margins": 0.23840561509132385,
3593
+ "rewards/rejected": -0.09889431297779083,
3594
+ "step": 239
3595
+ },
3596
+ {
3597
+ "epoch": 0.012899411464351938,
3598
+ "grad_norm": 8.626564979553223,
3599
+ "learning_rate": 9.938441702975689e-07,
3600
+ "logits/chosen": -0.47271132469177246,
3601
+ "logits/rejected": -0.7084850668907166,
3602
+ "logps/chosen": -98.41950225830078,
3603
+ "logps/rejected": -148.84288024902344,
3604
+ "loss": 1.1146,
3605
+ "rewards/accuracies": 0.75,
3606
+ "rewards/chosen": 0.10083074867725372,
3607
+ "rewards/margins": 0.3182566463947296,
3608
+ "rewards/rejected": -0.21742592751979828,
3609
+ "step": 240
3610
+ },
3611
+ {
3612
+ "epoch": 0.012953159012120073,
3613
+ "grad_norm": 8.52356243133545,
3614
+ "learning_rate": 9.935332048891826e-07,
3615
+ "logits/chosen": -0.5194116234779358,
3616
+ "logits/rejected": -0.6223669052124023,
3617
+ "logps/chosen": -90.4648666381836,
3618
+ "logps/rejected": -140.77615356445312,
3619
+ "loss": 1.1332,
3620
+ "rewards/accuracies": 1.0,
3621
+ "rewards/chosen": 0.15672340989112854,
3622
+ "rewards/margins": 0.27920961380004883,
3623
+ "rewards/rejected": -0.12248621881008148,
3624
+ "step": 241
3625
+ },
3626
+ {
3627
+ "epoch": 0.013006906559888206,
3628
+ "grad_norm": 9.13849925994873,
3629
+ "learning_rate": 9.932146285882476e-07,
3630
+ "logits/chosen": -0.43376827239990234,
3631
+ "logits/rejected": -0.9578706622123718,
3632
+ "logps/chosen": -96.71780395507812,
3633
+ "logps/rejected": -124.5494613647461,
3634
+ "loss": 1.1259,
3635
+ "rewards/accuracies": 0.75,
3636
+ "rewards/chosen": 0.20085996389389038,
3637
+ "rewards/margins": 0.29483500123023987,
3638
+ "rewards/rejected": -0.0939750224351883,
3639
+ "step": 242
3640
+ },
3641
+ {
3642
+ "epoch": 0.013060654107656338,
3643
+ "grad_norm": 8.863753318786621,
3644
+ "learning_rate": 9.928884463076043e-07,
3645
+ "logits/chosen": -0.37400951981544495,
3646
+ "logits/rejected": -0.6572523713111877,
3647
+ "logps/chosen": -86.31878662109375,
3648
+ "logps/rejected": -139.85658264160156,
3649
+ "loss": 0.9777,
3650
+ "rewards/accuracies": 1.0,
3651
+ "rewards/chosen": 0.28231894969940186,
3652
+ "rewards/margins": 0.5099464058876038,
3653
+ "rewards/rejected": -0.2276274710893631,
3654
+ "step": 243
3655
+ },
3656
+ {
3657
+ "epoch": 0.013114401655424471,
3658
+ "grad_norm": 8.157885551452637,
3659
+ "learning_rate": 9.925546630773868e-07,
3660
+ "logits/chosen": -0.5754011273384094,
3661
+ "logits/rejected": -0.7451348900794983,
3662
+ "logps/chosen": -68.57931518554688,
3663
+ "logps/rejected": -114.21456909179688,
3664
+ "loss": 1.0157,
3665
+ "rewards/accuracies": 1.0,
3666
+ "rewards/chosen": 0.31994327902793884,
3667
+ "rewards/margins": 0.436386376619339,
3668
+ "rewards/rejected": -0.11644310504198074,
3669
+ "step": 244
3670
+ },
3671
+ {
3672
+ "epoch": 0.013168149203192604,
3673
+ "grad_norm": 8.603972434997559,
3674
+ "learning_rate": 9.922132840449458e-07,
3675
+ "logits/chosen": -0.4437783360481262,
3676
+ "logits/rejected": -0.7457876801490784,
3677
+ "logps/chosen": -104.11923217773438,
3678
+ "logps/rejected": -125.06718444824219,
3679
+ "loss": 1.0947,
3680
+ "rewards/accuracies": 0.875,
3681
+ "rewards/chosen": 0.25450336933135986,
3682
+ "rewards/margins": 0.3329158425331116,
3683
+ "rewards/rejected": -0.0784124881029129,
3684
+ "step": 245
3685
+ },
3686
+ {
3687
+ "epoch": 0.013221896750960737,
3688
+ "grad_norm": 7.682473182678223,
3689
+ "learning_rate": 9.91864314474768e-07,
3690
+ "logits/chosen": -0.7631401419639587,
3691
+ "logits/rejected": -0.7120974659919739,
3692
+ "logps/chosen": -57.73262405395508,
3693
+ "logps/rejected": -72.64689636230469,
3694
+ "loss": 1.2067,
3695
+ "rewards/accuracies": 0.75,
3696
+ "rewards/chosen": 0.10729515552520752,
3697
+ "rewards/margins": 0.19412139058113098,
3698
+ "rewards/rejected": -0.08682622015476227,
3699
+ "step": 246
3700
+ },
3701
+ {
3702
+ "epoch": 0.013275644298728871,
3703
+ "grad_norm": 9.09015941619873,
3704
+ "learning_rate": 9.915077597483958e-07,
3705
+ "logits/chosen": -0.691004753112793,
3706
+ "logits/rejected": -0.8626288175582886,
3707
+ "logps/chosen": -90.16621398925781,
3708
+ "logps/rejected": -135.4027862548828,
3709
+ "loss": 1.2029,
3710
+ "rewards/accuracies": 0.75,
3711
+ "rewards/chosen": 0.18838340044021606,
3712
+ "rewards/margins": 0.21971102058887482,
3713
+ "rewards/rejected": -0.03132762759923935,
3714
+ "step": 247
3715
+ },
3716
+ {
3717
+ "epoch": 0.013329391846497004,
3718
+ "grad_norm": 6.7129621505737305,
3719
+ "learning_rate": 9.911436253643443e-07,
3720
+ "logits/chosen": -0.6010236740112305,
3721
+ "logits/rejected": -0.9694432020187378,
3722
+ "logps/chosen": -51.10399627685547,
3723
+ "logps/rejected": -74.51303100585938,
3724
+ "loss": 1.3418,
3725
+ "rewards/accuracies": 0.75,
3726
+ "rewards/chosen": 0.2066667526960373,
3727
+ "rewards/margins": 0.05391811951994896,
3728
+ "rewards/rejected": 0.15274862945079803,
3729
+ "step": 248
3730
+ },
3731
+ {
3732
+ "epoch": 0.013383139394265137,
3733
+ "grad_norm": 7.616147994995117,
3734
+ "learning_rate": 9.907719169380162e-07,
3735
+ "logits/chosen": -0.5555158853530884,
3736
+ "logits/rejected": -1.0062698125839233,
3737
+ "logps/chosen": -85.89315795898438,
3738
+ "logps/rejected": -127.45658874511719,
3739
+ "loss": 1.1302,
3740
+ "rewards/accuracies": 0.75,
3741
+ "rewards/chosen": 0.13865943253040314,
3742
+ "rewards/margins": 0.3135596811771393,
3743
+ "rewards/rejected": -0.17490024864673615,
3744
+ "step": 249
3745
+ },
3746
+ {
3747
+ "epoch": 0.01343688694203327,
3748
+ "grad_norm": 7.474664211273193,
3749
+ "learning_rate": 9.90392640201615e-07,
3750
+ "logits/chosen": -0.3878192603588104,
3751
+ "logits/rejected": -0.804552435874939,
3752
+ "logps/chosen": -80.22721862792969,
3753
+ "logps/rejected": -111.23281860351562,
3754
+ "loss": 1.08,
3755
+ "rewards/accuracies": 0.875,
3756
+ "rewards/chosen": 0.10335011780261993,
3757
+ "rewards/margins": 0.39268797636032104,
3758
+ "rewards/rejected": -0.2893378436565399,
3759
+ "step": 250
3760
+ },
3761
+ {
3762
+ "epoch": 0.013490634489801402,
3763
+ "grad_norm": 8.29160213470459,
3764
+ "learning_rate": 9.900058010040577e-07,
3765
+ "logits/chosen": -0.43254974484443665,
3766
+ "logits/rejected": -0.7992380261421204,
3767
+ "logps/chosen": -85.9269027709961,
3768
+ "logps/rejected": -113.69029235839844,
3769
+ "loss": 1.257,
3770
+ "rewards/accuracies": 0.75,
3771
+ "rewards/chosen": 0.14866304397583008,
3772
+ "rewards/margins": 0.16407746076583862,
3773
+ "rewards/rejected": -0.015414435416460037,
3774
+ "step": 251
3775
+ },
3776
+ {
3777
+ "epoch": 0.013544382037569537,
3778
+ "grad_norm": 7.953009605407715,
3779
+ "learning_rate": 9.89611405310883e-07,
3780
+ "logits/chosen": -0.43862664699554443,
3781
+ "logits/rejected": -0.7818647623062134,
3782
+ "logps/chosen": -88.93862915039062,
3783
+ "logps/rejected": -120.42898559570312,
3784
+ "loss": 0.9718,
3785
+ "rewards/accuracies": 1.0,
3786
+ "rewards/chosen": 0.20921139419078827,
3787
+ "rewards/margins": 0.5064070820808411,
3788
+ "rewards/rejected": -0.2971956729888916,
3789
+ "step": 252
3790
+ },
3791
+ {
3792
+ "epoch": 0.01359812958533767,
3793
+ "grad_norm": 10.189489364624023,
3794
+ "learning_rate": 9.8920945920416e-07,
3795
+ "logits/chosen": -0.5023062825202942,
3796
+ "logits/rejected": -0.982239842414856,
3797
+ "logps/chosen": -74.873779296875,
3798
+ "logps/rejected": -108.82559204101562,
3799
+ "loss": 1.0636,
3800
+ "rewards/accuracies": 0.875,
3801
+ "rewards/chosen": 0.2425529956817627,
3802
+ "rewards/margins": 0.3734268844127655,
3803
+ "rewards/rejected": -0.1308738738298416,
3804
+ "step": 253
3805
+ },
3806
+ {
3807
+ "epoch": 0.013651877133105802,
3808
+ "grad_norm": 11.489689826965332,
3809
+ "learning_rate": 9.887999688823954e-07,
3810
+ "logits/chosen": -0.30702269077301025,
3811
+ "logits/rejected": -0.6260428428649902,
3812
+ "logps/chosen": -92.19865417480469,
3813
+ "logps/rejected": -124.37353515625,
3814
+ "loss": 1.1542,
3815
+ "rewards/accuracies": 0.875,
3816
+ "rewards/chosen": 0.3126932382583618,
3817
+ "rewards/margins": 0.26121002435684204,
3818
+ "rewards/rejected": 0.05148324370384216,
3819
+ "step": 254
3820
+ },
3821
+ {
3822
+ "epoch": 0.013705624680873935,
3823
+ "grad_norm": 7.918270587921143,
3824
+ "learning_rate": 9.883829406604361e-07,
3825
+ "logits/chosen": -0.5766696929931641,
3826
+ "logits/rejected": -1.1000299453735352,
3827
+ "logps/chosen": -87.10360717773438,
3828
+ "logps/rejected": -154.98800659179688,
3829
+ "loss": 0.9432,
3830
+ "rewards/accuracies": 1.0,
3831
+ "rewards/chosen": 0.18990127742290497,
3832
+ "rewards/margins": 0.5358986854553223,
3833
+ "rewards/rejected": -0.3459973931312561,
3834
+ "step": 255
3835
+ },
3836
+ {
3837
+ "epoch": 0.013759372228642068,
3838
+ "grad_norm": 8.557303428649902,
3839
+ "learning_rate": 9.879583809693736e-07,
3840
+ "logits/chosen": -0.22856351733207703,
3841
+ "logits/rejected": -0.864416241645813,
3842
+ "logps/chosen": -103.28632354736328,
3843
+ "logps/rejected": -154.03309631347656,
3844
+ "loss": 0.8591,
3845
+ "rewards/accuracies": 0.875,
3846
+ "rewards/chosen": 0.2975628972053528,
3847
+ "rewards/margins": 0.6666656732559204,
3848
+ "rewards/rejected": -0.3691027760505676,
3849
+ "step": 256
3850
+ },
3851
+ {
3852
+ "epoch": 0.0138131197764102,
3853
+ "grad_norm": 8.320441246032715,
3854
+ "learning_rate": 9.875262963564435e-07,
3855
+ "logits/chosen": -0.5290526151657104,
3856
+ "logits/rejected": -0.7232424020767212,
3857
+ "logps/chosen": -82.62203979492188,
3858
+ "logps/rejected": -149.920654296875,
3859
+ "loss": 1.046,
3860
+ "rewards/accuracies": 1.0,
3861
+ "rewards/chosen": 0.16208958625793457,
3862
+ "rewards/margins": 0.39549511671066284,
3863
+ "rewards/rejected": -0.23340550065040588,
3864
+ "step": 257
3865
+ },
3866
+ {
3867
+ "epoch": 0.013866867324178335,
3868
+ "grad_norm": 7.353522777557373,
3869
+ "learning_rate": 9.870866934849246e-07,
3870
+ "logits/chosen": -0.6042935252189636,
3871
+ "logits/rejected": -0.9732577800750732,
3872
+ "logps/chosen": -67.07771301269531,
3873
+ "logps/rejected": -84.36785125732422,
3874
+ "loss": 1.1253,
3875
+ "rewards/accuracies": 0.875,
3876
+ "rewards/chosen": 0.16771726310253143,
3877
+ "rewards/margins": 0.3118589520454407,
3878
+ "rewards/rejected": -0.14414167404174805,
3879
+ "step": 258
3880
+ },
3881
+ {
3882
+ "epoch": 0.013920614871946468,
3883
+ "grad_norm": 9.230646133422852,
3884
+ "learning_rate": 9.866395791340374e-07,
3885
+ "logits/chosen": -0.5574221014976501,
3886
+ "logits/rejected": -0.6890866756439209,
3887
+ "logps/chosen": -87.2138442993164,
3888
+ "logps/rejected": -133.9090576171875,
3889
+ "loss": 1.1809,
3890
+ "rewards/accuracies": 0.75,
3891
+ "rewards/chosen": 0.09173579514026642,
3892
+ "rewards/margins": 0.2554128170013428,
3893
+ "rewards/rejected": -0.16367703676223755,
3894
+ "step": 259
3895
+ },
3896
+ {
3897
+ "epoch": 0.0139743624197146,
3898
+ "grad_norm": 7.673941612243652,
3899
+ "learning_rate": 9.861849601988383e-07,
3900
+ "logits/chosen": -0.4745372533798218,
3901
+ "logits/rejected": -0.5913327932357788,
3902
+ "logps/chosen": -82.71144104003906,
3903
+ "logps/rejected": -130.95370483398438,
3904
+ "loss": 0.9595,
3905
+ "rewards/accuracies": 0.875,
3906
+ "rewards/chosen": 0.32120776176452637,
3907
+ "rewards/margins": 0.5713176131248474,
3908
+ "rewards/rejected": -0.2501099109649658,
3909
+ "step": 260
3910
+ },
3911
+ {
3912
+ "epoch": 0.014028109967482733,
3913
+ "grad_norm": 5.969098091125488,
3914
+ "learning_rate": 9.857228436901134e-07,
3915
+ "logits/chosen": -0.5343005657196045,
3916
+ "logits/rejected": -0.44417738914489746,
3917
+ "logps/chosen": -58.944358825683594,
3918
+ "logps/rejected": -89.23226165771484,
3919
+ "loss": 1.1487,
3920
+ "rewards/accuracies": 0.875,
3921
+ "rewards/chosen": 0.13331182301044464,
3922
+ "rewards/margins": 0.26338741183280945,
3923
+ "rewards/rejected": -0.1300755888223648,
3924
+ "step": 261
3925
+ },
3926
+ {
3927
+ "epoch": 0.014081857515250866,
3928
+ "grad_norm": 8.642870903015137,
3929
+ "learning_rate": 9.852532367342712e-07,
3930
+ "logits/chosen": -0.5430492758750916,
3931
+ "logits/rejected": -0.919087290763855,
3932
+ "logps/chosen": -93.74098205566406,
3933
+ "logps/rejected": -122.88327026367188,
3934
+ "loss": 1.0657,
3935
+ "rewards/accuracies": 0.875,
3936
+ "rewards/chosen": 0.1623714417219162,
3937
+ "rewards/margins": 0.36680108308792114,
3938
+ "rewards/rejected": -0.20442962646484375,
3939
+ "step": 262
3940
+ },
3941
+ {
3942
+ "epoch": 0.014135605063019,
3943
+ "grad_norm": 8.33414077758789,
3944
+ "learning_rate": 9.847761465732316e-07,
3945
+ "logits/chosen": -0.4385722875595093,
3946
+ "logits/rejected": -0.6532965898513794,
3947
+ "logps/chosen": -79.09358215332031,
3948
+ "logps/rejected": -116.81510162353516,
3949
+ "loss": 1.0642,
3950
+ "rewards/accuracies": 0.875,
3951
+ "rewards/chosen": 0.22798743844032288,
3952
+ "rewards/margins": 0.4548647999763489,
3953
+ "rewards/rejected": -0.2268773764371872,
3954
+ "step": 263
3955
+ },
3956
+ {
3957
+ "epoch": 0.014189352610787133,
3958
+ "grad_norm": 7.47625732421875,
3959
+ "learning_rate": 9.842915805643156e-07,
3960
+ "logits/chosen": -0.48590952157974243,
3961
+ "logits/rejected": -0.7134944200515747,
3962
+ "logps/chosen": -83.9005126953125,
3963
+ "logps/rejected": -124.9341812133789,
3964
+ "loss": 0.9709,
3965
+ "rewards/accuracies": 1.0,
3966
+ "rewards/chosen": 0.2338734120130539,
3967
+ "rewards/margins": 0.4882698655128479,
3968
+ "rewards/rejected": -0.2543964385986328,
3969
+ "step": 264
3970
+ },
3971
+ {
3972
+ "epoch": 0.014243100158555266,
3973
+ "grad_norm": 8.187015533447266,
3974
+ "learning_rate": 9.837995461801299e-07,
3975
+ "logits/chosen": -0.23714786767959595,
3976
+ "logits/rejected": -0.674926221370697,
3977
+ "logps/chosen": -60.35591125488281,
3978
+ "logps/rejected": -77.57908630371094,
3979
+ "loss": 1.1285,
3980
+ "rewards/accuracies": 1.0,
3981
+ "rewards/chosen": 0.24997663497924805,
3982
+ "rewards/margins": 0.2878130078315735,
3983
+ "rewards/rejected": -0.037836357951164246,
3984
+ "step": 265
3985
+ },
3986
+ {
3987
+ "epoch": 0.014296847706323399,
3988
+ "grad_norm": 8.187094688415527,
3989
+ "learning_rate": 9.833000510084537e-07,
3990
+ "logits/chosen": -0.46092841029167175,
3991
+ "logits/rejected": -0.7724924087524414,
3992
+ "logps/chosen": -83.680419921875,
3993
+ "logps/rejected": -112.01083374023438,
3994
+ "loss": 1.1699,
3995
+ "rewards/accuracies": 0.875,
3996
+ "rewards/chosen": 0.14704155921936035,
3997
+ "rewards/margins": 0.2566852867603302,
3998
+ "rewards/rejected": -0.10964371263980865,
3999
+ "step": 266
4000
+ },
4001
+ {
4002
+ "epoch": 0.014350595254091532,
4003
+ "grad_norm": 8.889602661132812,
4004
+ "learning_rate": 9.827931027521203e-07,
4005
+ "logits/chosen": -0.5110130310058594,
4006
+ "logits/rejected": -0.5687258243560791,
4007
+ "logps/chosen": -108.56094360351562,
4008
+ "logps/rejected": -124.67503356933594,
4009
+ "loss": 1.0632,
4010
+ "rewards/accuracies": 0.875,
4011
+ "rewards/chosen": 0.18289077281951904,
4012
+ "rewards/margins": 0.39289993047714233,
4013
+ "rewards/rejected": -0.2100091576576233,
4014
+ "step": 267
4015
+ },
4016
+ {
4017
+ "epoch": 0.014404342801859666,
4018
+ "grad_norm": 9.36739730834961,
4019
+ "learning_rate": 9.82278709228899e-07,
4020
+ "logits/chosen": -0.4345582127571106,
4021
+ "logits/rejected": -0.5341508388519287,
4022
+ "logps/chosen": -86.82331085205078,
4023
+ "logps/rejected": -104.12142181396484,
4024
+ "loss": 1.1031,
4025
+ "rewards/accuracies": 1.0,
4026
+ "rewards/chosen": 0.2522311210632324,
4027
+ "rewards/margins": 0.3170914053916931,
4028
+ "rewards/rejected": -0.0648602545261383,
4029
+ "step": 268
4030
+ },
4031
+ {
4032
+ "epoch": 0.014458090349627799,
4033
+ "grad_norm": 7.332289695739746,
4034
+ "learning_rate": 9.817568783713743e-07,
4035
+ "logits/chosen": -0.768648087978363,
4036
+ "logits/rejected": -0.9154085516929626,
4037
+ "logps/chosen": -85.2851333618164,
4038
+ "logps/rejected": -93.83207702636719,
4039
+ "loss": 1.1129,
4040
+ "rewards/accuracies": 1.0,
4041
+ "rewards/chosen": 0.20519228279590607,
4042
+ "rewards/margins": 0.31278467178344727,
4043
+ "rewards/rejected": -0.1075923889875412,
4044
+ "step": 269
4045
+ },
4046
+ {
4047
+ "epoch": 0.014511837897395932,
4048
+ "grad_norm": 9.34799861907959,
4049
+ "learning_rate": 9.812276182268236e-07,
4050
+ "logits/chosen": -0.5562636256217957,
4051
+ "logits/rejected": -0.6497949361801147,
4052
+ "logps/chosen": -84.54762268066406,
4053
+ "logps/rejected": -142.047607421875,
4054
+ "loss": 0.9431,
4055
+ "rewards/accuracies": 1.0,
4056
+ "rewards/chosen": 0.12430868297815323,
4057
+ "rewards/margins": 0.5213713049888611,
4058
+ "rewards/rejected": -0.39706259965896606,
4059
+ "step": 270
4060
+ },
4061
+ {
4062
+ "epoch": 0.014565585445164064,
4063
+ "grad_norm": 7.788771629333496,
4064
+ "learning_rate": 9.80690936957093e-07,
4065
+ "logits/chosen": -0.5215786099433899,
4066
+ "logits/rejected": -0.6321144104003906,
4067
+ "logps/chosen": -102.1207275390625,
4068
+ "logps/rejected": -114.18605041503906,
4069
+ "loss": 1.147,
4070
+ "rewards/accuracies": 0.875,
4071
+ "rewards/chosen": 0.15436579287052155,
4072
+ "rewards/margins": 0.280081570148468,
4073
+ "rewards/rejected": -0.12571577727794647,
4074
+ "step": 271
4075
+ },
4076
+ {
4077
+ "epoch": 0.014619332992932197,
4078
+ "grad_norm": 6.995871067047119,
4079
+ "learning_rate": 9.801468428384716e-07,
4080
+ "logits/chosen": -0.4826238751411438,
4081
+ "logits/rejected": -1.045019507408142,
4082
+ "logps/chosen": -51.4791259765625,
4083
+ "logps/rejected": -64.26348114013672,
4084
+ "loss": 1.2279,
4085
+ "rewards/accuracies": 0.75,
4086
+ "rewards/chosen": 0.16225357353687286,
4087
+ "rewards/margins": 0.21734732389450073,
4088
+ "rewards/rejected": -0.05509376898407936,
4089
+ "step": 272
4090
+ },
4091
+ {
4092
+ "epoch": 0.01467308054070033,
4093
+ "grad_norm": 8.320442199707031,
4094
+ "learning_rate": 9.795953442615637e-07,
4095
+ "logits/chosen": -0.2795771360397339,
4096
+ "logits/rejected": -0.9269124269485474,
4097
+ "logps/chosen": -83.69010925292969,
4098
+ "logps/rejected": -132.68978881835938,
4099
+ "loss": 0.9347,
4100
+ "rewards/accuracies": 0.875,
4101
+ "rewards/chosen": 0.28460603952407837,
4102
+ "rewards/margins": 0.5781921744346619,
4103
+ "rewards/rejected": -0.2935860753059387,
4104
+ "step": 273
4105
+ },
4106
+ {
4107
+ "epoch": 0.014726828088468464,
4108
+ "grad_norm": 7.224923133850098,
4109
+ "learning_rate": 9.790364497311595e-07,
4110
+ "logits/chosen": -0.516793966293335,
4111
+ "logits/rejected": -0.5459295511245728,
4112
+ "logps/chosen": -71.00935363769531,
4113
+ "logps/rejected": -104.81021118164062,
4114
+ "loss": 0.9845,
4115
+ "rewards/accuracies": 1.0,
4116
+ "rewards/chosen": 0.29926228523254395,
4117
+ "rewards/margins": 0.5231636762619019,
4118
+ "rewards/rejected": -0.2239014208316803,
4119
+ "step": 274
4120
+ },
4121
+ {
4122
+ "epoch": 0.014780575636236597,
4123
+ "grad_norm": 8.644327163696289,
4124
+ "learning_rate": 9.784701678661044e-07,
4125
+ "logits/chosen": -0.29655879735946655,
4126
+ "logits/rejected": -0.696053147315979,
4127
+ "logps/chosen": -73.27018737792969,
4128
+ "logps/rejected": -96.52377319335938,
4129
+ "loss": 1.1951,
4130
+ "rewards/accuracies": 0.875,
4131
+ "rewards/chosen": 0.20520314574241638,
4132
+ "rewards/margins": 0.20678368210792542,
4133
+ "rewards/rejected": -0.001580517739057541,
4134
+ "step": 275
4135
+ },
4136
+ {
4137
+ "epoch": 0.01483432318400473,
4138
+ "grad_norm": 9.182957649230957,
4139
+ "learning_rate": 9.77896507399165e-07,
4140
+ "logits/chosen": -0.5673010349273682,
4141
+ "logits/rejected": -0.6784734725952148,
4142
+ "logps/chosen": -88.32586669921875,
4143
+ "logps/rejected": -138.0816192626953,
4144
+ "loss": 1.1084,
4145
+ "rewards/accuracies": 0.75,
4146
+ "rewards/chosen": 0.19880113005638123,
4147
+ "rewards/margins": 0.33900701999664307,
4148
+ "rewards/rejected": -0.14020586013793945,
4149
+ "step": 276
4150
+ },
4151
+ {
4152
+ "epoch": 0.014888070731772863,
4153
+ "grad_norm": 7.926046848297119,
4154
+ "learning_rate": 9.773154771768955e-07,
4155
+ "logits/chosen": -0.44852402806282043,
4156
+ "logits/rejected": -0.8911378979682922,
4157
+ "logps/chosen": -79.64956665039062,
4158
+ "logps/rejected": -105.65498352050781,
4159
+ "loss": 1.1516,
4160
+ "rewards/accuracies": 0.875,
4161
+ "rewards/chosen": 0.29726237058639526,
4162
+ "rewards/margins": 0.28044331073760986,
4163
+ "rewards/rejected": 0.016819097101688385,
4164
+ "step": 277
4165
+ },
4166
+ {
4167
+ "epoch": 0.014941818279540995,
4168
+ "grad_norm": 8.613622665405273,
4169
+ "learning_rate": 9.767270861595004e-07,
4170
+ "logits/chosen": -0.6433182954788208,
4171
+ "logits/rejected": -0.8262368440628052,
4172
+ "logps/chosen": -75.862060546875,
4173
+ "logps/rejected": -110.22479248046875,
4174
+ "loss": 1.113,
4175
+ "rewards/accuracies": 0.75,
4176
+ "rewards/chosen": 0.17685633897781372,
4177
+ "rewards/margins": 0.3377441465854645,
4178
+ "rewards/rejected": -0.16088780760765076,
4179
+ "step": 278
4180
+ },
4181
+ {
4182
+ "epoch": 0.01499556582730913,
4183
+ "grad_norm": 7.498295307159424,
4184
+ "learning_rate": 9.761313434206977e-07,
4185
+ "logits/chosen": -0.5290209054946899,
4186
+ "logits/rejected": -0.6595941781997681,
4187
+ "logps/chosen": -93.25931549072266,
4188
+ "logps/rejected": -114.23931884765625,
4189
+ "loss": 0.9943,
4190
+ "rewards/accuracies": 0.875,
4191
+ "rewards/chosen": 0.14882898330688477,
4192
+ "rewards/margins": 0.49634385108947754,
4193
+ "rewards/rejected": -0.3475148677825928,
4194
+ "step": 279
4195
+ },
4196
+ {
4197
+ "epoch": 0.015049313375077263,
4198
+ "grad_norm": 7.240914821624756,
4199
+ "learning_rate": 9.755282581475767e-07,
4200
+ "logits/chosen": -0.4989565908908844,
4201
+ "logits/rejected": -0.6776658296585083,
4202
+ "logps/chosen": -82.03950500488281,
4203
+ "logps/rejected": -100.86872863769531,
4204
+ "loss": 1.0541,
4205
+ "rewards/accuracies": 0.875,
4206
+ "rewards/chosen": 0.28837645053863525,
4207
+ "rewards/margins": 0.3993419408798218,
4208
+ "rewards/rejected": -0.11096549034118652,
4209
+ "step": 280
4210
+ },
4211
+ {
4212
+ "epoch": 0.015103060922845395,
4213
+ "grad_norm": 6.950934410095215,
4214
+ "learning_rate": 9.749178396404588e-07,
4215
+ "logits/chosen": -0.3202768564224243,
4216
+ "logits/rejected": -0.8115454912185669,
4217
+ "logps/chosen": -61.45012283325195,
4218
+ "logps/rejected": -85.72331237792969,
4219
+ "loss": 1.1402,
4220
+ "rewards/accuracies": 0.875,
4221
+ "rewards/chosen": 0.24857988953590393,
4222
+ "rewards/margins": 0.2943744361400604,
4223
+ "rewards/rejected": -0.04579455032944679,
4224
+ "step": 281
4225
+ },
4226
+ {
4227
+ "epoch": 0.015156808470613528,
4228
+ "grad_norm": 6.702651023864746,
4229
+ "learning_rate": 9.743000973127523e-07,
4230
+ "logits/chosen": -0.6097660064697266,
4231
+ "logits/rejected": -0.8681362867355347,
4232
+ "logps/chosen": -59.426475524902344,
4233
+ "logps/rejected": -101.32482147216797,
4234
+ "loss": 1.0391,
4235
+ "rewards/accuracies": 0.875,
4236
+ "rewards/chosen": 0.24413172900676727,
4237
+ "rewards/margins": 0.4027537405490875,
4238
+ "rewards/rejected": -0.15862202644348145,
4239
+ "step": 282
4240
+ },
4241
+ {
4242
+ "epoch": 0.01521055601838166,
4243
+ "grad_norm": 7.214409828186035,
4244
+ "learning_rate": 9.73675040690808e-07,
4245
+ "logits/chosen": -0.4117346405982971,
4246
+ "logits/rejected": -0.7110905647277832,
4247
+ "logps/chosen": -96.80953216552734,
4248
+ "logps/rejected": -143.1290283203125,
4249
+ "loss": 0.8017,
4250
+ "rewards/accuracies": 1.0,
4251
+ "rewards/chosen": 0.3847959339618683,
4252
+ "rewards/margins": 0.7773008942604065,
4253
+ "rewards/rejected": -0.39250487089157104,
4254
+ "step": 283
4255
+ },
4256
+ {
4257
+ "epoch": 0.015264303566149794,
4258
+ "grad_norm": 9.447858810424805,
4259
+ "learning_rate": 9.730426794137726e-07,
4260
+ "logits/chosen": -0.46513450145721436,
4261
+ "logits/rejected": -0.7564966678619385,
4262
+ "logps/chosen": -94.57246398925781,
4263
+ "logps/rejected": -120.65567016601562,
4264
+ "loss": 0.999,
4265
+ "rewards/accuracies": 1.0,
4266
+ "rewards/chosen": 0.25270184874534607,
4267
+ "rewards/margins": 0.4643063247203827,
4268
+ "rewards/rejected": -0.21160446107387543,
4269
+ "step": 284
4270
+ },
4271
+ {
4272
+ "epoch": 0.015318051113917928,
4273
+ "grad_norm": 8.516096115112305,
4274
+ "learning_rate": 9.72403023233439e-07,
4275
+ "logits/chosen": -0.16967588663101196,
4276
+ "logits/rejected": -0.6391971111297607,
4277
+ "logps/chosen": -82.84049987792969,
4278
+ "logps/rejected": -115.44432067871094,
4279
+ "loss": 0.8835,
4280
+ "rewards/accuracies": 1.0,
4281
+ "rewards/chosen": 0.24860167503356934,
4282
+ "rewards/margins": 0.6482660174369812,
4283
+ "rewards/rejected": -0.3996642827987671,
4284
+ "step": 285
4285
+ },
4286
+ {
4287
+ "epoch": 0.01537179866168606,
4288
+ "grad_norm": 10.064424514770508,
4289
+ "learning_rate": 9.717560820140968e-07,
4290
+ "logits/chosen": -0.3854745030403137,
4291
+ "logits/rejected": -0.6414456367492676,
4292
+ "logps/chosen": -75.41014099121094,
4293
+ "logps/rejected": -109.79766845703125,
4294
+ "loss": 1.1009,
4295
+ "rewards/accuracies": 0.75,
4296
+ "rewards/chosen": 0.2192663997411728,
4297
+ "rewards/margins": 0.3585866689682007,
4298
+ "rewards/rejected": -0.1393202841281891,
4299
+ "step": 286
4300
+ },
4301
+ {
4302
+ "epoch": 0.015425546209454194,
4303
+ "grad_norm": 6.992526054382324,
4304
+ "learning_rate": 9.711018657323798e-07,
4305
+ "logits/chosen": -0.6344618201255798,
4306
+ "logits/rejected": -0.9946074485778809,
4307
+ "logps/chosen": -68.52722930908203,
4308
+ "logps/rejected": -115.47061920166016,
4309
+ "loss": 1.0299,
4310
+ "rewards/accuracies": 0.875,
4311
+ "rewards/chosen": 0.3005034327507019,
4312
+ "rewards/margins": 0.4479685425758362,
4313
+ "rewards/rejected": -0.1474650800228119,
4314
+ "step": 287
4315
+ },
4316
+ {
4317
+ "epoch": 0.015479293757222326,
4318
+ "grad_norm": 9.140570640563965,
4319
+ "learning_rate": 9.704403844771127e-07,
4320
+ "logits/chosen": -0.411197304725647,
4321
+ "logits/rejected": -0.7791616916656494,
4322
+ "logps/chosen": -99.13587188720703,
4323
+ "logps/rejected": -147.218017578125,
4324
+ "loss": 1.0681,
4325
+ "rewards/accuracies": 0.75,
4326
+ "rewards/chosen": 0.1548137664794922,
4327
+ "rewards/margins": 0.39904481172561646,
4328
+ "rewards/rejected": -0.24423104524612427,
4329
+ "step": 288
4330
+ },
4331
+ {
4332
+ "epoch": 0.01553304130499046,
4333
+ "grad_norm": 7.424576282501221,
4334
+ "learning_rate": 9.697716484491545e-07,
4335
+ "logits/chosen": -0.5678108930587769,
4336
+ "logits/rejected": -0.7492731213569641,
4337
+ "logps/chosen": -90.81136322021484,
4338
+ "logps/rejected": -138.5076446533203,
4339
+ "loss": 0.7894,
4340
+ "rewards/accuracies": 1.0,
4341
+ "rewards/chosen": 0.2965516448020935,
4342
+ "rewards/margins": 0.7807052135467529,
4343
+ "rewards/rejected": -0.4841535985469818,
4344
+ "step": 289
4345
+ },
4346
+ {
4347
+ "epoch": 0.015586788852758594,
4348
+ "grad_norm": 7.465813636779785,
4349
+ "learning_rate": 9.69095667961242e-07,
4350
+ "logits/chosen": -0.40255221724510193,
4351
+ "logits/rejected": -0.658112645149231,
4352
+ "logps/chosen": -86.02987670898438,
4353
+ "logps/rejected": -108.27945709228516,
4354
+ "loss": 0.9936,
4355
+ "rewards/accuracies": 1.0,
4356
+ "rewards/chosen": 0.30489006638526917,
4357
+ "rewards/margins": 0.4739183485507965,
4358
+ "rewards/rejected": -0.16902829706668854,
4359
+ "step": 290
4360
+ },
4361
+ {
4362
+ "epoch": 0.015640536400526726,
4363
+ "grad_norm": 8.759163856506348,
4364
+ "learning_rate": 9.684124534378306e-07,
4365
+ "logits/chosen": -0.4031268060207367,
4366
+ "logits/rejected": -0.5718150734901428,
4367
+ "logps/chosen": -72.60466003417969,
4368
+ "logps/rejected": -90.86416625976562,
4369
+ "loss": 1.0946,
4370
+ "rewards/accuracies": 0.75,
4371
+ "rewards/chosen": 0.12894652783870697,
4372
+ "rewards/margins": 0.3612363934516907,
4373
+ "rewards/rejected": -0.2322898656129837,
4374
+ "step": 291
4375
+ },
4376
+ {
4377
+ "epoch": 0.01569428394829486,
4378
+ "grad_norm": 7.793913841247559,
4379
+ "learning_rate": 9.677220154149337e-07,
4380
+ "logits/chosen": -0.5084730386734009,
4381
+ "logits/rejected": -0.9328963756561279,
4382
+ "logps/chosen": -57.33200454711914,
4383
+ "logps/rejected": -103.25164794921875,
4384
+ "loss": 1.088,
4385
+ "rewards/accuracies": 1.0,
4386
+ "rewards/chosen": 0.28789860010147095,
4387
+ "rewards/margins": 0.3288136124610901,
4388
+ "rewards/rejected": -0.04091501235961914,
4389
+ "step": 292
4390
+ },
4391
+ {
4392
+ "epoch": 0.015748031496062992,
4393
+ "grad_norm": 7.355979919433594,
4394
+ "learning_rate": 9.670243645399592e-07,
4395
+ "logits/chosen": -0.23863175511360168,
4396
+ "logits/rejected": -0.6393299102783203,
4397
+ "logps/chosen": -73.53858184814453,
4398
+ "logps/rejected": -106.6146240234375,
4399
+ "loss": 1.1431,
4400
+ "rewards/accuracies": 0.75,
4401
+ "rewards/chosen": 0.17891855537891388,
4402
+ "rewards/margins": 0.29221975803375244,
4403
+ "rewards/rejected": -0.11330118775367737,
4404
+ "step": 293
4405
+ },
4406
+ {
4407
+ "epoch": 0.015801779043831125,
4408
+ "grad_norm": 9.70301342010498,
4409
+ "learning_rate": 9.66319511571547e-07,
4410
+ "logits/chosen": -0.5887746810913086,
4411
+ "logits/rejected": -0.8464258909225464,
4412
+ "logps/chosen": -72.31002807617188,
4413
+ "logps/rejected": -113.12400817871094,
4414
+ "loss": 1.0684,
4415
+ "rewards/accuracies": 1.0,
4416
+ "rewards/chosen": 0.18143503367900848,
4417
+ "rewards/margins": 0.38432347774505615,
4418
+ "rewards/rejected": -0.20288844406604767,
4419
+ "step": 294
4420
+ },
4421
+ {
4422
+ "epoch": 0.015855526591599257,
4423
+ "grad_norm": 8.012235641479492,
4424
+ "learning_rate": 9.656074673794017e-07,
4425
+ "logits/chosen": -0.48777371644973755,
4426
+ "logits/rejected": -0.7305865287780762,
4427
+ "logps/chosen": -76.85704040527344,
4428
+ "logps/rejected": -106.29739379882812,
4429
+ "loss": 1.1032,
4430
+ "rewards/accuracies": 0.75,
4431
+ "rewards/chosen": 0.23623672127723694,
4432
+ "rewards/margins": 0.348111093044281,
4433
+ "rewards/rejected": -0.11187433451414108,
4434
+ "step": 295
4435
+ },
4436
+ {
4437
+ "epoch": 0.01590927413936739,
4438
+ "grad_norm": 8.312561988830566,
4439
+ "learning_rate": 9.648882429441256e-07,
4440
+ "logits/chosen": -0.6219150424003601,
4441
+ "logits/rejected": -0.7722678184509277,
4442
+ "logps/chosen": -81.02972412109375,
4443
+ "logps/rejected": -108.74473571777344,
4444
+ "loss": 1.0751,
4445
+ "rewards/accuracies": 0.875,
4446
+ "rewards/chosen": 0.2990270256996155,
4447
+ "rewards/margins": 0.3803249001502991,
4448
+ "rewards/rejected": -0.0812978744506836,
4449
+ "step": 296
4450
+ },
4451
+ {
4452
+ "epoch": 0.015963021687135523,
4453
+ "grad_norm": 7.4272685050964355,
4454
+ "learning_rate": 9.641618493570494e-07,
4455
+ "logits/chosen": -0.610638439655304,
4456
+ "logits/rejected": -0.6773062944412231,
4457
+ "logps/chosen": -78.98297119140625,
4458
+ "logps/rejected": -98.52257537841797,
4459
+ "loss": 1.0557,
4460
+ "rewards/accuracies": 1.0,
4461
+ "rewards/chosen": 0.2879084348678589,
4462
+ "rewards/margins": 0.3793227970600128,
4463
+ "rewards/rejected": -0.09141434729099274,
4464
+ "step": 297
4465
+ },
4466
+ {
4467
+ "epoch": 0.01601676923490366,
4468
+ "grad_norm": 9.1967191696167,
4469
+ "learning_rate": 9.634282978200603e-07,
4470
+ "logits/chosen": -0.47958338260650635,
4471
+ "logits/rejected": -0.7445369958877563,
4472
+ "logps/chosen": -88.43070983886719,
4473
+ "logps/rejected": -117.46142578125,
4474
+ "loss": 1.0955,
4475
+ "rewards/accuracies": 0.75,
4476
+ "rewards/chosen": 0.21477727591991425,
4477
+ "rewards/margins": 0.353880912065506,
4478
+ "rewards/rejected": -0.13910359144210815,
4479
+ "step": 298
4480
+ },
4481
+ {
4482
+ "epoch": 0.016070516782671792,
4483
+ "grad_norm": 8.55931568145752,
4484
+ "learning_rate": 9.62687599645431e-07,
4485
+ "logits/chosen": -0.5550699830055237,
4486
+ "logits/rejected": -0.8242961168289185,
4487
+ "logps/chosen": -95.43873596191406,
4488
+ "logps/rejected": -119.18885803222656,
4489
+ "loss": 1.0174,
4490
+ "rewards/accuracies": 0.875,
4491
+ "rewards/chosen": 0.2723308205604553,
4492
+ "rewards/margins": 0.4446273446083069,
4493
+ "rewards/rejected": -0.17229653894901276,
4494
+ "step": 299
4495
+ },
4496
+ {
4497
+ "epoch": 0.016124264330439925,
4498
+ "grad_norm": 9.203601837158203,
4499
+ "learning_rate": 9.619397662556433e-07,
4500
+ "logits/chosen": -0.6323695182800293,
4501
+ "logits/rejected": -0.7377463579177856,
4502
+ "logps/chosen": -89.126220703125,
4503
+ "logps/rejected": -110.41032409667969,
4504
+ "loss": 0.8626,
4505
+ "rewards/accuracies": 1.0,
4506
+ "rewards/chosen": 0.28707414865493774,
4507
+ "rewards/margins": 0.6520431637763977,
4508
+ "rewards/rejected": -0.36496901512145996,
4509
+ "step": 300
4510
  }
4511
  ],
4512
  "logging_steps": 1,