k-r-l commited on
Commit
adfc06e
1 Parent(s): 7e9ed06

Training in progress, step 94, checkpoint

Browse files
last-checkpoint/adapter_config.json CHANGED
@@ -20,8 +20,8 @@
20
  "rank_pattern": {},
21
  "revision": null,
22
  "target_modules": [
23
- "q_proj",
24
- "v_proj"
25
  ],
26
  "task_type": "CAUSAL_LM",
27
  "use_dora": false,
 
20
  "rank_pattern": {},
21
  "revision": null,
22
  "target_modules": [
23
+ "v_proj",
24
+ "q_proj"
25
  ],
26
  "task_type": "CAUSAL_LM",
27
  "use_dora": false,
last-checkpoint/adapter_model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:47a08a6e864e2ee1968177f93a78f8afab9ebca87fe279fae45b39b1d4790815
3
  size 13648432
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7cee89f39da97dc062cb36b8f188e9b91c658143ca1a266a82d88c66b2d5eae2
3
  size 13648432
last-checkpoint/optimizer.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:7711dccfe7177717c6c0d73a485a5f5d766f81d9dbdd88ae79b64c25ed23c69e
3
  size 7309882
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a7fbec71a209019c10f1851144c04ee11bfa4ca9f447ab4cd0f45c50a94b49ef
3
  size 7309882
last-checkpoint/scheduler.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:46e4c05515a68694cd891bd31cda7371a27ef3d19d366250cc2850e7de57780a
3
  size 1064
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e91d6ce8e7ae0854f9cd8296ea65e09795565f2d67a3314e42e7f7cca576ea75
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": 0.6318471337579618,
5
  "eval_steps": 500,
6
- "global_step": 62,
7
  "is_hyper_param_search": false,
8
  "is_local_process_zero": true,
9
  "is_world_process_zero": true,
@@ -1123,6 +1123,582 @@
1123
  "rewards/margins": 0.011071886867284775,
1124
  "rewards/rejected": -0.07814602553844452,
1125
  "step": 62
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1126
  }
1127
  ],
1128
  "logging_steps": 1,
 
1
  {
2
  "best_metric": null,
3
  "best_model_checkpoint": null,
4
+ "epoch": 0.9579617834394905,
5
  "eval_steps": 500,
6
+ "global_step": 94,
7
  "is_hyper_param_search": false,
8
  "is_local_process_zero": true,
9
  "is_world_process_zero": true,
 
1123
  "rewards/margins": 0.011071886867284775,
1124
  "rewards/rejected": -0.07814602553844452,
1125
  "step": 62
1126
+ },
1127
+ {
1128
+ "epoch": 0.6420382165605095,
1129
+ "grad_norm": 0.637529730796814,
1130
+ "learning_rate": 7.857142857142858e-05,
1131
+ "log_odds_chosen": 0.33163022994995117,
1132
+ "log_odds_ratio": -0.5596598982810974,
1133
+ "logits/chosen": -0.26017192006111145,
1134
+ "logits/rejected": -0.27084898948669434,
1135
+ "logps/chosen": -0.6162149906158447,
1136
+ "logps/rejected": -0.7858166694641113,
1137
+ "loss": 0.7619,
1138
+ "nll_loss": 0.7059016227722168,
1139
+ "rewards/accuracies": 0.71875,
1140
+ "rewards/chosen": -0.06162149831652641,
1141
+ "rewards/margins": 0.016960175707936287,
1142
+ "rewards/rejected": -0.07858168333768845,
1143
+ "step": 63
1144
+ },
1145
+ {
1146
+ "epoch": 0.6522292993630573,
1147
+ "grad_norm": 0.4790436029434204,
1148
+ "learning_rate": 7.823129251700681e-05,
1149
+ "log_odds_chosen": 0.35003671050071716,
1150
+ "log_odds_ratio": -0.5487669110298157,
1151
+ "logits/chosen": -0.1195572018623352,
1152
+ "logits/rejected": -0.18973661959171295,
1153
+ "logps/chosen": -0.5801494121551514,
1154
+ "logps/rejected": -0.7505877017974854,
1155
+ "loss": 0.7456,
1156
+ "nll_loss": 0.6906896829605103,
1157
+ "rewards/accuracies": 0.84375,
1158
+ "rewards/chosen": -0.05801495164632797,
1159
+ "rewards/margins": 0.0170438215136528,
1160
+ "rewards/rejected": -0.07505877315998077,
1161
+ "step": 64
1162
+ },
1163
+ {
1164
+ "epoch": 0.6624203821656051,
1165
+ "grad_norm": 0.47054964303970337,
1166
+ "learning_rate": 7.789115646258504e-05,
1167
+ "log_odds_chosen": 0.2697327733039856,
1168
+ "log_odds_ratio": -0.588991641998291,
1169
+ "logits/chosen": -0.12229520082473755,
1170
+ "logits/rejected": -0.12661772966384888,
1171
+ "logps/chosen": -0.6129696369171143,
1172
+ "logps/rejected": -0.7457621693611145,
1173
+ "loss": 0.7732,
1174
+ "nll_loss": 0.7143112421035767,
1175
+ "rewards/accuracies": 0.65625,
1176
+ "rewards/chosen": -0.061296962201595306,
1177
+ "rewards/margins": 0.01327925268560648,
1178
+ "rewards/rejected": -0.07457621395587921,
1179
+ "step": 65
1180
+ },
1181
+ {
1182
+ "epoch": 0.6726114649681528,
1183
+ "grad_norm": 0.5436939597129822,
1184
+ "learning_rate": 7.755102040816327e-05,
1185
+ "log_odds_chosen": 0.3452521860599518,
1186
+ "log_odds_ratio": -0.5557309985160828,
1187
+ "logits/chosen": -0.08714023232460022,
1188
+ "logits/rejected": -0.1457855999469757,
1189
+ "logps/chosen": -0.5788965225219727,
1190
+ "logps/rejected": -0.7455580234527588,
1191
+ "loss": 0.7397,
1192
+ "nll_loss": 0.6841052174568176,
1193
+ "rewards/accuracies": 0.78125,
1194
+ "rewards/chosen": -0.057889651507139206,
1195
+ "rewards/margins": 0.01666616089642048,
1196
+ "rewards/rejected": -0.07455581426620483,
1197
+ "step": 66
1198
+ },
1199
+ {
1200
+ "epoch": 0.6828025477707006,
1201
+ "grad_norm": 0.5666665434837341,
1202
+ "learning_rate": 7.72108843537415e-05,
1203
+ "log_odds_chosen": 0.3019857406616211,
1204
+ "log_odds_ratio": -0.5755807757377625,
1205
+ "logits/chosen": -0.07147038727998734,
1206
+ "logits/rejected": -0.09695108234882355,
1207
+ "logps/chosen": -0.5981124639511108,
1208
+ "logps/rejected": -0.7489673495292664,
1209
+ "loss": 0.754,
1210
+ "nll_loss": 0.6964120864868164,
1211
+ "rewards/accuracies": 0.78125,
1212
+ "rewards/chosen": -0.059811241924762726,
1213
+ "rewards/margins": 0.015085484832525253,
1214
+ "rewards/rejected": -0.07489673048257828,
1215
+ "step": 67
1216
+ },
1217
+ {
1218
+ "epoch": 0.6929936305732484,
1219
+ "grad_norm": 0.6614330410957336,
1220
+ "learning_rate": 7.687074829931973e-05,
1221
+ "log_odds_chosen": 0.24785318970680237,
1222
+ "log_odds_ratio": -0.5993584990501404,
1223
+ "logits/chosen": -0.1450255811214447,
1224
+ "logits/rejected": -0.1640571653842926,
1225
+ "logps/chosen": -0.6308612823486328,
1226
+ "logps/rejected": -0.7485529184341431,
1227
+ "loss": 0.79,
1228
+ "nll_loss": 0.7300571799278259,
1229
+ "rewards/accuracies": 0.75,
1230
+ "rewards/chosen": -0.06308613717556,
1231
+ "rewards/margins": 0.011769158765673637,
1232
+ "rewards/rejected": -0.07485529780387878,
1233
+ "step": 68
1234
+ },
1235
+ {
1236
+ "epoch": 0.7031847133757961,
1237
+ "grad_norm": 0.5287584662437439,
1238
+ "learning_rate": 7.653061224489796e-05,
1239
+ "log_odds_chosen": 0.2588488459587097,
1240
+ "log_odds_ratio": -0.5906943082809448,
1241
+ "logits/chosen": -0.09049258381128311,
1242
+ "logits/rejected": -0.09309229999780655,
1243
+ "logps/chosen": -0.6527013778686523,
1244
+ "logps/rejected": -0.7823808789253235,
1245
+ "loss": 0.7924,
1246
+ "nll_loss": 0.7333185076713562,
1247
+ "rewards/accuracies": 0.78125,
1248
+ "rewards/chosen": -0.06527013331651688,
1249
+ "rewards/margins": 0.012967949733138084,
1250
+ "rewards/rejected": -0.07823808491230011,
1251
+ "step": 69
1252
+ },
1253
+ {
1254
+ "epoch": 0.7133757961783439,
1255
+ "grad_norm": 0.6998805999755859,
1256
+ "learning_rate": 7.619047619047618e-05,
1257
+ "log_odds_chosen": 0.21674904227256775,
1258
+ "log_odds_ratio": -0.6123349666595459,
1259
+ "logits/chosen": -0.11004475504159927,
1260
+ "logits/rejected": -0.11400632560253143,
1261
+ "logps/chosen": -0.6173443794250488,
1262
+ "logps/rejected": -0.7168747782707214,
1263
+ "loss": 0.768,
1264
+ "nll_loss": 0.7068023085594177,
1265
+ "rewards/accuracies": 0.65625,
1266
+ "rewards/chosen": -0.06173444539308548,
1267
+ "rewards/margins": 0.009953041560947895,
1268
+ "rewards/rejected": -0.0716874822974205,
1269
+ "step": 70
1270
+ },
1271
+ {
1272
+ "epoch": 0.7235668789808917,
1273
+ "grad_norm": 0.742544412612915,
1274
+ "learning_rate": 7.585034013605443e-05,
1275
+ "log_odds_chosen": 0.19548919796943665,
1276
+ "log_odds_ratio": -0.6275305151939392,
1277
+ "logits/chosen": -0.12460804730653763,
1278
+ "logits/rejected": -0.12356013804674149,
1279
+ "logps/chosen": -0.7157447934150696,
1280
+ "logps/rejected": -0.799144983291626,
1281
+ "loss": 0.8176,
1282
+ "nll_loss": 0.7548049688339233,
1283
+ "rewards/accuracies": 0.6875,
1284
+ "rewards/chosen": -0.07157447934150696,
1285
+ "rewards/margins": 0.008340023458003998,
1286
+ "rewards/rejected": -0.07991450279951096,
1287
+ "step": 71
1288
+ },
1289
+ {
1290
+ "epoch": 0.7337579617834394,
1291
+ "grad_norm": 0.5506495237350464,
1292
+ "learning_rate": 7.551020408163266e-05,
1293
+ "log_odds_chosen": 0.2938334345817566,
1294
+ "log_odds_ratio": -0.5719183087348938,
1295
+ "logits/chosen": -0.09276354312896729,
1296
+ "logits/rejected": -0.09323181211948395,
1297
+ "logps/chosen": -0.6002932786941528,
1298
+ "logps/rejected": -0.7459129691123962,
1299
+ "loss": 0.7346,
1300
+ "nll_loss": 0.6773681640625,
1301
+ "rewards/accuracies": 0.75,
1302
+ "rewards/chosen": -0.060029324144124985,
1303
+ "rewards/margins": 0.014561973512172699,
1304
+ "rewards/rejected": -0.07459130138158798,
1305
+ "step": 72
1306
+ },
1307
+ {
1308
+ "epoch": 0.7439490445859872,
1309
+ "grad_norm": 0.5797027349472046,
1310
+ "learning_rate": 7.517006802721089e-05,
1311
+ "log_odds_chosen": 0.1955791562795639,
1312
+ "log_odds_ratio": -0.6199466586112976,
1313
+ "logits/chosen": -0.07904835790395737,
1314
+ "logits/rejected": -0.12583205103874207,
1315
+ "logps/chosen": -0.6262507438659668,
1316
+ "logps/rejected": -0.7190106511116028,
1317
+ "loss": 0.7983,
1318
+ "nll_loss": 0.7363026142120361,
1319
+ "rewards/accuracies": 0.65625,
1320
+ "rewards/chosen": -0.06262507289648056,
1321
+ "rewards/margins": 0.009275996126234531,
1322
+ "rewards/rejected": -0.07190106809139252,
1323
+ "step": 73
1324
+ },
1325
+ {
1326
+ "epoch": 0.7541401273885351,
1327
+ "grad_norm": 0.6255090832710266,
1328
+ "learning_rate": 7.482993197278913e-05,
1329
+ "log_odds_chosen": 0.24435332417488098,
1330
+ "log_odds_ratio": -0.5920006632804871,
1331
+ "logits/chosen": -0.08072115480899811,
1332
+ "logits/rejected": -0.0795532613992691,
1333
+ "logps/chosen": -0.6259615421295166,
1334
+ "logps/rejected": -0.7442739009857178,
1335
+ "loss": 0.7805,
1336
+ "nll_loss": 0.7213104963302612,
1337
+ "rewards/accuracies": 0.8125,
1338
+ "rewards/chosen": -0.0625961497426033,
1339
+ "rewards/margins": 0.011831245385110378,
1340
+ "rewards/rejected": -0.07442739605903625,
1341
+ "step": 74
1342
+ },
1343
+ {
1344
+ "epoch": 0.7643312101910829,
1345
+ "grad_norm": 0.500413715839386,
1346
+ "learning_rate": 7.448979591836736e-05,
1347
+ "log_odds_chosen": 0.1923215687274933,
1348
+ "log_odds_ratio": -0.6174032688140869,
1349
+ "logits/chosen": -0.09215761721134186,
1350
+ "logits/rejected": -0.08426235616207123,
1351
+ "logps/chosen": -0.6419164538383484,
1352
+ "logps/rejected": -0.7315772771835327,
1353
+ "loss": 0.7981,
1354
+ "nll_loss": 0.7363995313644409,
1355
+ "rewards/accuracies": 0.71875,
1356
+ "rewards/chosen": -0.06419164687395096,
1357
+ "rewards/margins": 0.008966086432337761,
1358
+ "rewards/rejected": -0.07315773516893387,
1359
+ "step": 75
1360
+ },
1361
+ {
1362
+ "epoch": 0.7745222929936306,
1363
+ "grad_norm": 0.47041788697242737,
1364
+ "learning_rate": 7.414965986394559e-05,
1365
+ "log_odds_chosen": 0.32406049966812134,
1366
+ "log_odds_ratio": -0.558575451374054,
1367
+ "logits/chosen": -0.0999608039855957,
1368
+ "logits/rejected": -0.10914846509695053,
1369
+ "logps/chosen": -0.6198108196258545,
1370
+ "logps/rejected": -0.7787230014801025,
1371
+ "loss": 0.7178,
1372
+ "nll_loss": 0.6619524359703064,
1373
+ "rewards/accuracies": 0.84375,
1374
+ "rewards/chosen": -0.06198108196258545,
1375
+ "rewards/margins": 0.015891216695308685,
1376
+ "rewards/rejected": -0.07787229120731354,
1377
+ "step": 76
1378
+ },
1379
+ {
1380
+ "epoch": 0.7847133757961784,
1381
+ "grad_norm": 0.5418070554733276,
1382
+ "learning_rate": 7.380952380952382e-05,
1383
+ "log_odds_chosen": 0.1873037964105606,
1384
+ "log_odds_ratio": -0.6218591928482056,
1385
+ "logits/chosen": -0.09966853260993958,
1386
+ "logits/rejected": -0.0808413028717041,
1387
+ "logps/chosen": -0.6605861186981201,
1388
+ "logps/rejected": -0.7546226978302002,
1389
+ "loss": 0.7967,
1390
+ "nll_loss": 0.7345606088638306,
1391
+ "rewards/accuracies": 0.6875,
1392
+ "rewards/chosen": -0.06605862081050873,
1393
+ "rewards/margins": 0.009403654374182224,
1394
+ "rewards/rejected": -0.07546227425336838,
1395
+ "step": 77
1396
+ },
1397
+ {
1398
+ "epoch": 0.7949044585987262,
1399
+ "grad_norm": 0.6250921487808228,
1400
+ "learning_rate": 7.346938775510205e-05,
1401
+ "log_odds_chosen": 0.31786370277404785,
1402
+ "log_odds_ratio": -0.5675632357597351,
1403
+ "logits/chosen": -0.15341174602508545,
1404
+ "logits/rejected": -0.13008305430412292,
1405
+ "logps/chosen": -0.5935506224632263,
1406
+ "logps/rejected": -0.7325422763824463,
1407
+ "loss": 0.7511,
1408
+ "nll_loss": 0.6942938566207886,
1409
+ "rewards/accuracies": 0.75,
1410
+ "rewards/chosen": -0.05935506522655487,
1411
+ "rewards/margins": 0.013899162411689758,
1412
+ "rewards/rejected": -0.07325422763824463,
1413
+ "step": 78
1414
+ },
1415
+ {
1416
+ "epoch": 0.8050955414012739,
1417
+ "grad_norm": 0.40636491775512695,
1418
+ "learning_rate": 7.312925170068027e-05,
1419
+ "log_odds_chosen": 0.30228132009506226,
1420
+ "log_odds_ratio": -0.5691430568695068,
1421
+ "logits/chosen": -0.08937245607376099,
1422
+ "logits/rejected": -0.1251370757818222,
1423
+ "logps/chosen": -0.6316071152687073,
1424
+ "logps/rejected": -0.7746922373771667,
1425
+ "loss": 0.7947,
1426
+ "nll_loss": 0.7377853989601135,
1427
+ "rewards/accuracies": 0.8125,
1428
+ "rewards/chosen": -0.06316071003675461,
1429
+ "rewards/margins": 0.014308515936136246,
1430
+ "rewards/rejected": -0.07746922224760056,
1431
+ "step": 79
1432
+ },
1433
+ {
1434
+ "epoch": 0.8152866242038217,
1435
+ "grad_norm": 0.5820095539093018,
1436
+ "learning_rate": 7.27891156462585e-05,
1437
+ "log_odds_chosen": 0.3547573387622833,
1438
+ "log_odds_ratio": -0.5511735081672668,
1439
+ "logits/chosen": -0.11485984176397324,
1440
+ "logits/rejected": -0.11412106454372406,
1441
+ "logps/chosen": -0.586860716342926,
1442
+ "logps/rejected": -0.7574341297149658,
1443
+ "loss": 0.7343,
1444
+ "nll_loss": 0.679178774356842,
1445
+ "rewards/accuracies": 0.71875,
1446
+ "rewards/chosen": -0.05868607014417648,
1447
+ "rewards/margins": 0.017057344317436218,
1448
+ "rewards/rejected": -0.0757434219121933,
1449
+ "step": 80
1450
+ },
1451
+ {
1452
+ "epoch": 0.8254777070063695,
1453
+ "grad_norm": 0.6103206276893616,
1454
+ "learning_rate": 7.244897959183675e-05,
1455
+ "log_odds_chosen": 0.3428616225719452,
1456
+ "log_odds_ratio": -0.547142505645752,
1457
+ "logits/chosen": -0.17640018463134766,
1458
+ "logits/rejected": -0.1795547753572464,
1459
+ "logps/chosen": -0.5735194683074951,
1460
+ "logps/rejected": -0.7396149039268494,
1461
+ "loss": 0.7013,
1462
+ "nll_loss": 0.6465373635292053,
1463
+ "rewards/accuracies": 0.875,
1464
+ "rewards/chosen": -0.05735195428133011,
1465
+ "rewards/margins": 0.016609538346529007,
1466
+ "rewards/rejected": -0.07396149635314941,
1467
+ "step": 81
1468
+ },
1469
+ {
1470
+ "epoch": 0.8356687898089172,
1471
+ "grad_norm": 0.5069766640663147,
1472
+ "learning_rate": 7.210884353741498e-05,
1473
+ "log_odds_chosen": 0.12126234918832779,
1474
+ "log_odds_ratio": -0.6486403346061707,
1475
+ "logits/chosen": -0.14951294660568237,
1476
+ "logits/rejected": -0.15072469413280487,
1477
+ "logps/chosen": -0.5951024293899536,
1478
+ "logps/rejected": -0.6498846411705017,
1479
+ "loss": 0.7563,
1480
+ "nll_loss": 0.6914424300193787,
1481
+ "rewards/accuracies": 0.65625,
1482
+ "rewards/chosen": -0.059510238468647,
1483
+ "rewards/margins": 0.005478225648403168,
1484
+ "rewards/rejected": -0.06498846411705017,
1485
+ "step": 82
1486
+ },
1487
+ {
1488
+ "epoch": 0.845859872611465,
1489
+ "grad_norm": 0.4635069966316223,
1490
+ "learning_rate": 7.17687074829932e-05,
1491
+ "log_odds_chosen": 0.2604025900363922,
1492
+ "log_odds_ratio": -0.5959426760673523,
1493
+ "logits/chosen": -0.15265314280986786,
1494
+ "logits/rejected": -0.12722516059875488,
1495
+ "logps/chosen": -0.5640627145767212,
1496
+ "logps/rejected": -0.6875459551811218,
1497
+ "loss": 0.72,
1498
+ "nll_loss": 0.6604282259941101,
1499
+ "rewards/accuracies": 0.75,
1500
+ "rewards/chosen": -0.056406278163194656,
1501
+ "rewards/margins": 0.012348320335149765,
1502
+ "rewards/rejected": -0.06875459849834442,
1503
+ "step": 83
1504
+ },
1505
+ {
1506
+ "epoch": 0.8560509554140128,
1507
+ "grad_norm": 0.5292064547538757,
1508
+ "learning_rate": 7.142857142857143e-05,
1509
+ "log_odds_chosen": 0.23624499142169952,
1510
+ "log_odds_ratio": -0.6043432354927063,
1511
+ "logits/chosen": -0.11054065823554993,
1512
+ "logits/rejected": -0.1269894540309906,
1513
+ "logps/chosen": -0.624123215675354,
1514
+ "logps/rejected": -0.7454429864883423,
1515
+ "loss": 0.7638,
1516
+ "nll_loss": 0.703406035900116,
1517
+ "rewards/accuracies": 0.6875,
1518
+ "rewards/chosen": -0.0624123141169548,
1519
+ "rewards/margins": 0.012131982482969761,
1520
+ "rewards/rejected": -0.07454430311918259,
1521
+ "step": 84
1522
+ },
1523
+ {
1524
+ "epoch": 0.8662420382165605,
1525
+ "grad_norm": 0.45075032114982605,
1526
+ "learning_rate": 7.108843537414966e-05,
1527
+ "log_odds_chosen": 0.24057269096374512,
1528
+ "log_odds_ratio": -0.6011200547218323,
1529
+ "logits/chosen": -0.11038406193256378,
1530
+ "logits/rejected": -0.12801669538021088,
1531
+ "logps/chosen": -0.5973632335662842,
1532
+ "logps/rejected": -0.7181129455566406,
1533
+ "loss": 0.717,
1534
+ "nll_loss": 0.6568683385848999,
1535
+ "rewards/accuracies": 0.65625,
1536
+ "rewards/chosen": -0.05973632633686066,
1537
+ "rewards/margins": 0.012074977159500122,
1538
+ "rewards/rejected": -0.07181129604578018,
1539
+ "step": 85
1540
+ },
1541
+ {
1542
+ "epoch": 0.8764331210191083,
1543
+ "grad_norm": 0.577867865562439,
1544
+ "learning_rate": 7.074829931972789e-05,
1545
+ "log_odds_chosen": 0.25693386793136597,
1546
+ "log_odds_ratio": -0.590366542339325,
1547
+ "logits/chosen": -0.0883067399263382,
1548
+ "logits/rejected": -0.0588250458240509,
1549
+ "logps/chosen": -0.6325963735580444,
1550
+ "logps/rejected": -0.7457253932952881,
1551
+ "loss": 0.7601,
1552
+ "nll_loss": 0.7010390758514404,
1553
+ "rewards/accuracies": 0.75,
1554
+ "rewards/chosen": -0.06325963884592056,
1555
+ "rewards/margins": 0.011312908492982388,
1556
+ "rewards/rejected": -0.07457254827022552,
1557
+ "step": 86
1558
+ },
1559
+ {
1560
+ "epoch": 0.8866242038216561,
1561
+ "grad_norm": 0.5341058373451233,
1562
+ "learning_rate": 7.040816326530612e-05,
1563
+ "log_odds_chosen": 0.21396467089653015,
1564
+ "log_odds_ratio": -0.6061536073684692,
1565
+ "logits/chosen": -0.0878664180636406,
1566
+ "logits/rejected": -0.10543867945671082,
1567
+ "logps/chosen": -0.5701617002487183,
1568
+ "logps/rejected": -0.6728689670562744,
1569
+ "loss": 0.751,
1570
+ "nll_loss": 0.6903491616249084,
1571
+ "rewards/accuracies": 0.75,
1572
+ "rewards/chosen": -0.057016175240278244,
1573
+ "rewards/margins": 0.010270730592310429,
1574
+ "rewards/rejected": -0.0672869086265564,
1575
+ "step": 87
1576
+ },
1577
+ {
1578
+ "epoch": 0.8968152866242038,
1579
+ "grad_norm": 0.4483081102371216,
1580
+ "learning_rate": 7.006802721088435e-05,
1581
+ "log_odds_chosen": 0.34266385436058044,
1582
+ "log_odds_ratio": -0.5521624088287354,
1583
+ "logits/chosen": -0.07731188088655472,
1584
+ "logits/rejected": -0.10345152020454407,
1585
+ "logps/chosen": -0.5693691372871399,
1586
+ "logps/rejected": -0.7366178631782532,
1587
+ "loss": 0.7171,
1588
+ "nll_loss": 0.6619335412979126,
1589
+ "rewards/accuracies": 0.8125,
1590
+ "rewards/chosen": -0.05693691596388817,
1591
+ "rewards/margins": 0.01672486774623394,
1592
+ "rewards/rejected": -0.07366178184747696,
1593
+ "step": 88
1594
+ },
1595
+ {
1596
+ "epoch": 0.9070063694267516,
1597
+ "grad_norm": 0.727847158908844,
1598
+ "learning_rate": 6.97278911564626e-05,
1599
+ "log_odds_chosen": 0.2518327534198761,
1600
+ "log_odds_ratio": -0.595729410648346,
1601
+ "logits/chosen": -0.092443086206913,
1602
+ "logits/rejected": -0.12563319504261017,
1603
+ "logps/chosen": -0.608110249042511,
1604
+ "logps/rejected": -0.7250441312789917,
1605
+ "loss": 0.751,
1606
+ "nll_loss": 0.6914761066436768,
1607
+ "rewards/accuracies": 0.75,
1608
+ "rewards/chosen": -0.06081102788448334,
1609
+ "rewards/margins": 0.011693387292325497,
1610
+ "rewards/rejected": -0.07250441610813141,
1611
+ "step": 89
1612
+ },
1613
+ {
1614
+ "epoch": 0.9171974522292994,
1615
+ "grad_norm": 0.5988894701004028,
1616
+ "learning_rate": 6.938775510204082e-05,
1617
+ "log_odds_chosen": 0.2530748248100281,
1618
+ "log_odds_ratio": -0.5891579389572144,
1619
+ "logits/chosen": -0.08331869542598724,
1620
+ "logits/rejected": -0.09794420003890991,
1621
+ "logps/chosen": -0.567914605140686,
1622
+ "logps/rejected": -0.6916981339454651,
1623
+ "loss": 0.7225,
1624
+ "nll_loss": 0.6635410785675049,
1625
+ "rewards/accuracies": 0.78125,
1626
+ "rewards/chosen": -0.05679146572947502,
1627
+ "rewards/margins": 0.012378348037600517,
1628
+ "rewards/rejected": -0.06916981935501099,
1629
+ "step": 90
1630
+ },
1631
+ {
1632
+ "epoch": 0.9273885350318471,
1633
+ "grad_norm": 0.6276490092277527,
1634
+ "learning_rate": 6.904761904761905e-05,
1635
+ "log_odds_chosen": 0.1880638748407364,
1636
+ "log_odds_ratio": -0.6190271973609924,
1637
+ "logits/chosen": 0.024591676890850067,
1638
+ "logits/rejected": 0.03547324985265732,
1639
+ "logps/chosen": -0.6322571635246277,
1640
+ "logps/rejected": -0.7290378212928772,
1641
+ "loss": 0.7772,
1642
+ "nll_loss": 0.7153192758560181,
1643
+ "rewards/accuracies": 0.65625,
1644
+ "rewards/chosen": -0.06322571635246277,
1645
+ "rewards/margins": 0.009678068570792675,
1646
+ "rewards/rejected": -0.07290378957986832,
1647
+ "step": 91
1648
+ },
1649
+ {
1650
+ "epoch": 0.9375796178343949,
1651
+ "grad_norm": 0.43707507848739624,
1652
+ "learning_rate": 6.870748299319728e-05,
1653
+ "log_odds_chosen": 0.2878055274486542,
1654
+ "log_odds_ratio": -0.5840734839439392,
1655
+ "logits/chosen": -0.0694851279258728,
1656
+ "logits/rejected": -0.07452072948217392,
1657
+ "logps/chosen": -0.5883352160453796,
1658
+ "logps/rejected": -0.7242968678474426,
1659
+ "loss": 0.7213,
1660
+ "nll_loss": 0.6628736853599548,
1661
+ "rewards/accuracies": 0.78125,
1662
+ "rewards/chosen": -0.0588335283100605,
1663
+ "rewards/margins": 0.013596156612038612,
1664
+ "rewards/rejected": -0.07242967933416367,
1665
+ "step": 92
1666
+ },
1667
+ {
1668
+ "epoch": 0.9477707006369427,
1669
+ "grad_norm": 0.4694580137729645,
1670
+ "learning_rate": 6.836734693877551e-05,
1671
+ "log_odds_chosen": 0.2617676556110382,
1672
+ "log_odds_ratio": -0.5865938663482666,
1673
+ "logits/chosen": -0.031561583280563354,
1674
+ "logits/rejected": -0.025262409821152687,
1675
+ "logps/chosen": -0.5720298290252686,
1676
+ "logps/rejected": -0.695450484752655,
1677
+ "loss": 0.7592,
1678
+ "nll_loss": 0.7005623579025269,
1679
+ "rewards/accuracies": 0.75,
1680
+ "rewards/chosen": -0.05720299109816551,
1681
+ "rewards/margins": 0.012342060916125774,
1682
+ "rewards/rejected": -0.06954504549503326,
1683
+ "step": 93
1684
+ },
1685
+ {
1686
+ "epoch": 0.9579617834394905,
1687
+ "grad_norm": 0.4407201409339905,
1688
+ "learning_rate": 6.802721088435374e-05,
1689
+ "log_odds_chosen": 0.32194840908050537,
1690
+ "log_odds_ratio": -0.5685411095619202,
1691
+ "logits/chosen": -0.07971872389316559,
1692
+ "logits/rejected": -0.07843183726072311,
1693
+ "logps/chosen": -0.6054414510726929,
1694
+ "logps/rejected": -0.7565875053405762,
1695
+ "loss": 0.759,
1696
+ "nll_loss": 0.7021029591560364,
1697
+ "rewards/accuracies": 0.75,
1698
+ "rewards/chosen": -0.060544148087501526,
1699
+ "rewards/margins": 0.015114611014723778,
1700
+ "rewards/rejected": -0.07565876096487045,
1701
+ "step": 94
1702
  }
1703
  ],
1704
  "logging_steps": 1,
last-checkpoint/training_args.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:b6e9bd93f66e1ed0f17812a85321161533aa48011f619aebcabb9399556f35a5
3
  size 5560
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5ace7eb9613605d1a80effde8f5e673e03dc8c398af3367a0d9d550ae1639482
3
  size 5560