diff --git "a/trainer_state.json" "b/trainer_state.json" new file mode 100644--- /dev/null +++ "b/trainer_state.json" @@ -0,0 +1,4244 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 1.0, + "eval_steps": 500, + "global_step": 3000, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.0, + "learning_rate": 1.666666666666667e-08, + "logits/chosen": -1.8473987579345703, + "logits/rejected": -1.8474841117858887, + "logps/chosen": -628.1939086914062, + "logps/rejected": -1017.0994873046875, + "loss": 0.1606, + "rewards/accuracies": 0.0, + "rewards/chosen": 0.0, + "rewards/margins": 0.0, + "rewards/rejected": 0.0, + "step": 1 + }, + { + "epoch": 0.0, + "learning_rate": 1.6666666666666668e-07, + "logits/chosen": -1.4316620826721191, + "logits/rejected": -0.9766409397125244, + "logps/chosen": -518.1865234375, + "logps/rejected": -875.9132690429688, + "loss": 0.2429, + "rewards/accuracies": 0.5, + "rewards/chosen": -6.813897016400006e-06, + "rewards/margins": 6.345117071759887e-06, + "rewards/rejected": -1.3159031368559226e-05, + "step": 10 + }, + { + "epoch": 0.01, + "learning_rate": 3.3333333333333335e-07, + "logits/chosen": -1.6879074573516846, + "logits/rejected": -1.2949936389923096, + "logps/chosen": -449.43621826171875, + "logps/rejected": -834.7189331054688, + "loss": 0.2148, + "rewards/accuracies": 0.574999988079071, + "rewards/chosen": 0.00029412214644253254, + "rewards/margins": 0.00020992716599721462, + "rewards/rejected": 8.419497316936031e-05, + "step": 20 + }, + { + "epoch": 0.01, + "learning_rate": 5.000000000000001e-07, + "logits/chosen": -1.7546484470367432, + "logits/rejected": -1.0084898471832275, + "logps/chosen": -496.73309326171875, + "logps/rejected": -889.1583862304688, + "loss": 0.2307, + "rewards/accuracies": 0.875, + "rewards/chosen": 0.0016126197297126055, + "rewards/margins": 0.004062547814100981, + "rewards/rejected": -0.002449928317219019, + "step": 30 + }, + { + "epoch": 0.01, + "learning_rate": 6.666666666666667e-07, + "logits/chosen": -1.8569256067276, + "logits/rejected": -1.1899112462997437, + "logps/chosen": -448.177490234375, + "logps/rejected": -852.029296875, + "loss": 0.2095, + "rewards/accuracies": 0.824999988079071, + "rewards/chosen": 0.003779575927183032, + "rewards/margins": 0.012063580565154552, + "rewards/rejected": -0.008284004405140877, + "step": 40 + }, + { + "epoch": 0.02, + "learning_rate": 8.333333333333333e-07, + "logits/chosen": -1.5573432445526123, + "logits/rejected": -1.0242127180099487, + "logps/chosen": -441.22540283203125, + "logps/rejected": -806.1510009765625, + "loss": 0.1942, + "rewards/accuracies": 0.824999988079071, + "rewards/chosen": 0.011921902187168598, + "rewards/margins": 0.029057607054710388, + "rewards/rejected": -0.017135705798864365, + "step": 50 + }, + { + "epoch": 0.02, + "learning_rate": 1.0000000000000002e-06, + "logits/chosen": -1.7608438730239868, + "logits/rejected": -1.3249375820159912, + "logps/chosen": -435.998779296875, + "logps/rejected": -766.541015625, + "loss": 0.1739, + "rewards/accuracies": 0.800000011920929, + "rewards/chosen": 0.01454467885196209, + "rewards/margins": 0.036728017032146454, + "rewards/rejected": -0.022183334454894066, + "step": 60 + }, + { + "epoch": 0.02, + "learning_rate": 1.1666666666666668e-06, + "logits/chosen": -1.7882951498031616, + "logits/rejected": -0.9916883707046509, + "logps/chosen": -463.21746826171875, + "logps/rejected": -798.3153076171875, + "loss": 0.1613, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": 0.01092723198235035, + "rewards/margins": 0.05515831708908081, + "rewards/rejected": -0.04423108696937561, + "step": 70 + }, + { + "epoch": 0.03, + "learning_rate": 1.3333333333333334e-06, + "logits/chosen": -1.6877580881118774, + "logits/rejected": -1.1066850423812866, + "logps/chosen": -412.07342529296875, + "logps/rejected": -912.7047119140625, + "loss": 0.1534, + "rewards/accuracies": 0.8500000238418579, + "rewards/chosen": -0.0010939339408650994, + "rewards/margins": 0.0824560895562172, + "rewards/rejected": -0.08355002105236053, + "step": 80 + }, + { + "epoch": 0.03, + "learning_rate": 1.5e-06, + "logits/chosen": -1.5878902673721313, + "logits/rejected": -1.1904621124267578, + "logps/chosen": -521.6202392578125, + "logps/rejected": -905.14794921875, + "loss": 0.1709, + "rewards/accuracies": 0.800000011920929, + "rewards/chosen": -0.05081214755773544, + "rewards/margins": 0.08134042471647263, + "rewards/rejected": -0.13215258717536926, + "step": 90 + }, + { + "epoch": 0.03, + "learning_rate": 1.6666666666666667e-06, + "logits/chosen": -1.6791446208953857, + "logits/rejected": -1.182875394821167, + "logps/chosen": -579.6087036132812, + "logps/rejected": -993.3094482421875, + "loss": 0.149, + "rewards/accuracies": 0.800000011920929, + "rewards/chosen": -0.09510762244462967, + "rewards/margins": 0.09448549151420593, + "rewards/rejected": -0.1895931363105774, + "step": 100 + }, + { + "epoch": 0.04, + "learning_rate": 1.8333333333333333e-06, + "logits/chosen": -1.6629585027694702, + "logits/rejected": -1.0604465007781982, + "logps/chosen": -638.2183837890625, + "logps/rejected": -1162.047119140625, + "loss": 0.1087, + "rewards/accuracies": 0.875, + "rewards/chosen": -0.12550528347492218, + "rewards/margins": 0.14394623041152954, + "rewards/rejected": -0.2694514989852905, + "step": 110 + }, + { + "epoch": 0.04, + "learning_rate": 2.0000000000000003e-06, + "logits/chosen": -1.751172423362732, + "logits/rejected": -1.2760882377624512, + "logps/chosen": -573.3961181640625, + "logps/rejected": -981.23486328125, + "loss": 0.1234, + "rewards/accuracies": 0.7749999761581421, + "rewards/chosen": -0.09364907443523407, + "rewards/margins": 0.13833266496658325, + "rewards/rejected": -0.23198175430297852, + "step": 120 + }, + { + "epoch": 0.04, + "learning_rate": 2.166666666666667e-06, + "logits/chosen": -1.813155174255371, + "logits/rejected": -1.2474279403686523, + "logps/chosen": -560.3488159179688, + "logps/rejected": -1005.7208251953125, + "loss": 0.1478, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.08083701133728027, + "rewards/margins": 0.14707627892494202, + "rewards/rejected": -0.2279132902622223, + "step": 130 + }, + { + "epoch": 0.05, + "learning_rate": 2.3333333333333336e-06, + "logits/chosen": -1.9064534902572632, + "logits/rejected": -1.087799072265625, + "logps/chosen": -576.6889038085938, + "logps/rejected": -1124.29345703125, + "loss": 0.1001, + "rewards/accuracies": 0.8500000238418579, + "rewards/chosen": -0.11049157381057739, + "rewards/margins": 0.1889657974243164, + "rewards/rejected": -0.2994573712348938, + "step": 140 + }, + { + "epoch": 0.05, + "learning_rate": 2.5e-06, + "logits/chosen": -1.498268723487854, + "logits/rejected": -0.9696500897407532, + "logps/chosen": -623.1941528320312, + "logps/rejected": -1241.6993408203125, + "loss": 0.0989, + "rewards/accuracies": 0.824999988079071, + "rewards/chosen": -0.18307581543922424, + "rewards/margins": 0.2211984097957611, + "rewards/rejected": -0.40427422523498535, + "step": 150 + }, + { + "epoch": 0.05, + "learning_rate": 2.666666666666667e-06, + "logits/chosen": -1.6134288311004639, + "logits/rejected": -0.7816992998123169, + "logps/chosen": -624.9563598632812, + "logps/rejected": -1204.363037109375, + "loss": 0.0923, + "rewards/accuracies": 0.8999999761581421, + "rewards/chosen": -0.17922857403755188, + "rewards/margins": 0.23642221093177795, + "rewards/rejected": -0.41565078496932983, + "step": 160 + }, + { + "epoch": 0.06, + "learning_rate": 2.8333333333333335e-06, + "logits/chosen": -1.5277217626571655, + "logits/rejected": -1.1026668548583984, + "logps/chosen": -653.5545654296875, + "logps/rejected": -1159.62841796875, + "loss": 0.0954, + "rewards/accuracies": 0.800000011920929, + "rewards/chosen": -0.13700352609157562, + "rewards/margins": 0.22152018547058105, + "rewards/rejected": -0.35852372646331787, + "step": 170 + }, + { + "epoch": 0.06, + "learning_rate": 3e-06, + "logits/chosen": -1.6350257396697998, + "logits/rejected": -1.2443767786026, + "logps/chosen": -666.1326904296875, + "logps/rejected": -1370.72412109375, + "loss": 0.0655, + "rewards/accuracies": 0.925000011920929, + "rewards/chosen": -0.19158519804477692, + "rewards/margins": 0.27321988344192505, + "rewards/rejected": -0.46480506658554077, + "step": 180 + }, + { + "epoch": 0.06, + "learning_rate": 3.1666666666666667e-06, + "logits/chosen": -1.6068347692489624, + "logits/rejected": -1.0125865936279297, + "logps/chosen": -535.7831420898438, + "logps/rejected": -1120.1278076171875, + "loss": 0.1016, + "rewards/accuracies": 0.8500000238418579, + "rewards/chosen": -0.1586967408657074, + "rewards/margins": 0.20938825607299805, + "rewards/rejected": -0.36808499693870544, + "step": 190 + }, + { + "epoch": 0.07, + "learning_rate": 3.3333333333333333e-06, + "logits/chosen": -1.7533979415893555, + "logits/rejected": -1.1328980922698975, + "logps/chosen": -582.9685668945312, + "logps/rejected": -1131.854736328125, + "loss": 0.0781, + "rewards/accuracies": 0.800000011920929, + "rewards/chosen": -0.11236470937728882, + "rewards/margins": 0.22253580391407013, + "rewards/rejected": -0.33490046858787537, + "step": 200 + }, + { + "epoch": 0.07, + "learning_rate": 3.5e-06, + "logits/chosen": -1.719315528869629, + "logits/rejected": -0.9589029550552368, + "logps/chosen": -651.3718872070312, + "logps/rejected": -1270.794677734375, + "loss": 0.1192, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.1939801275730133, + "rewards/margins": 0.23451688885688782, + "rewards/rejected": -0.4284970164299011, + "step": 210 + }, + { + "epoch": 0.07, + "learning_rate": 3.6666666666666666e-06, + "logits/chosen": -1.8104807138442993, + "logits/rejected": -1.117309808731079, + "logps/chosen": -525.1863403320312, + "logps/rejected": -1155.667724609375, + "loss": 0.0791, + "rewards/accuracies": 0.824999988079071, + "rewards/chosen": -0.10046984255313873, + "rewards/margins": 0.22820088267326355, + "rewards/rejected": -0.3286706805229187, + "step": 220 + }, + { + "epoch": 0.08, + "learning_rate": 3.833333333333334e-06, + "logits/chosen": -1.6729333400726318, + "logits/rejected": -1.3031710386276245, + "logps/chosen": -526.7484741210938, + "logps/rejected": -1042.1409912109375, + "loss": 0.1099, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.13879971206188202, + "rewards/margins": 0.17394515872001648, + "rewards/rejected": -0.3127448856830597, + "step": 230 + }, + { + "epoch": 0.08, + "learning_rate": 4.000000000000001e-06, + "logits/chosen": -1.7155532836914062, + "logits/rejected": -1.3224143981933594, + "logps/chosen": -568.672607421875, + "logps/rejected": -1282.6646728515625, + "loss": 0.0618, + "rewards/accuracies": 0.8999999761581421, + "rewards/chosen": -0.1310817301273346, + "rewards/margins": 0.2731386721134186, + "rewards/rejected": -0.4042204022407532, + "step": 240 + }, + { + "epoch": 0.08, + "learning_rate": 4.166666666666667e-06, + "logits/chosen": -1.694848656654358, + "logits/rejected": -1.1057406663894653, + "logps/chosen": -608.3328247070312, + "logps/rejected": -1172.0714111328125, + "loss": 0.1015, + "rewards/accuracies": 0.824999988079071, + "rewards/chosen": -0.1783088594675064, + "rewards/margins": 0.22445425391197205, + "rewards/rejected": -0.40276312828063965, + "step": 250 + }, + { + "epoch": 0.09, + "learning_rate": 4.333333333333334e-06, + "logits/chosen": -1.4572826623916626, + "logits/rejected": -0.9631636738777161, + "logps/chosen": -678.6790161132812, + "logps/rejected": -1274.690185546875, + "loss": 0.0955, + "rewards/accuracies": 0.875, + "rewards/chosen": -0.18504425883293152, + "rewards/margins": 0.24944782257080078, + "rewards/rejected": -0.4344921112060547, + "step": 260 + }, + { + "epoch": 0.09, + "learning_rate": 4.5e-06, + "logits/chosen": -1.5380445718765259, + "logits/rejected": -0.9687334299087524, + "logps/chosen": -598.2360229492188, + "logps/rejected": -1067.4052734375, + "loss": 0.1052, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.15487895905971527, + "rewards/margins": 0.20970329642295837, + "rewards/rejected": -0.36458227038383484, + "step": 270 + }, + { + "epoch": 0.09, + "learning_rate": 4.666666666666667e-06, + "logits/chosen": -1.5993750095367432, + "logits/rejected": -1.2316395044326782, + "logps/chosen": -559.2089233398438, + "logps/rejected": -1135.50048828125, + "loss": 0.0949, + "rewards/accuracies": 0.8500000238418579, + "rewards/chosen": -0.16124552488327026, + "rewards/margins": 0.21940717101097107, + "rewards/rejected": -0.38065269589424133, + "step": 280 + }, + { + "epoch": 0.1, + "learning_rate": 4.833333333333333e-06, + "logits/chosen": -1.785839319229126, + "logits/rejected": -1.2803369760513306, + "logps/chosen": -712.4893188476562, + "logps/rejected": -1194.8753662109375, + "loss": 0.1221, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.2521515488624573, + "rewards/margins": 0.1936561018228531, + "rewards/rejected": -0.44580763578414917, + "step": 290 + }, + { + "epoch": 0.1, + "learning_rate": 5e-06, + "logits/chosen": -1.8430877923965454, + "logits/rejected": -1.1291439533233643, + "logps/chosen": -703.2359619140625, + "logps/rejected": -1348.4500732421875, + "loss": 0.0795, + "rewards/accuracies": 0.875, + "rewards/chosen": -0.2468656599521637, + "rewards/margins": 0.2653115391731262, + "rewards/rejected": -0.5121771693229675, + "step": 300 + }, + { + "epoch": 0.1, + "learning_rate": 4.999830770009406e-06, + "logits/chosen": -1.5511696338653564, + "logits/rejected": -0.8779551386833191, + "logps/chosen": -645.1519775390625, + "logps/rejected": -1245.902587890625, + "loss": 0.1052, + "rewards/accuracies": 0.800000011920929, + "rewards/chosen": -0.16054175794124603, + "rewards/margins": 0.2655673027038574, + "rewards/rejected": -0.4261090159416199, + "step": 310 + }, + { + "epoch": 0.11, + "learning_rate": 4.999323102948655e-06, + "logits/chosen": -1.9738438129425049, + "logits/rejected": -1.4079262018203735, + "logps/chosen": -632.3164672851562, + "logps/rejected": -1090.2205810546875, + "loss": 0.1445, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.1474119871854782, + "rewards/margins": 0.1809072643518448, + "rewards/rejected": -0.328319251537323, + "step": 320 + }, + { + "epoch": 0.11, + "learning_rate": 4.99847706754774e-06, + "logits/chosen": -1.6942188739776611, + "logits/rejected": -1.3499834537506104, + "logps/chosen": -545.2244262695312, + "logps/rejected": -1028.6619873046875, + "loss": 0.0807, + "rewards/accuracies": 0.8500000238418579, + "rewards/chosen": -0.09895402193069458, + "rewards/margins": 0.21848241984844208, + "rewards/rejected": -0.31743642687797546, + "step": 330 + }, + { + "epoch": 0.11, + "learning_rate": 4.997292778346312e-06, + "logits/chosen": -1.8349497318267822, + "logits/rejected": -1.173274278640747, + "logps/chosen": -638.5476684570312, + "logps/rejected": -1342.106201171875, + "loss": 0.0821, + "rewards/accuracies": 0.875, + "rewards/chosen": -0.15458893775939941, + "rewards/margins": 0.2725256085395813, + "rewards/rejected": -0.4271145761013031, + "step": 340 + }, + { + "epoch": 0.12, + "learning_rate": 4.995770395678171e-06, + "logits/chosen": -1.6314140558242798, + "logits/rejected": -1.0475269556045532, + "logps/chosen": -508.9056701660156, + "logps/rejected": -1216.2740478515625, + "loss": 0.0992, + "rewards/accuracies": 0.8500000238418579, + "rewards/chosen": -0.08030419796705246, + "rewards/margins": 0.28404951095581055, + "rewards/rejected": -0.3643537163734436, + "step": 350 + }, + { + "epoch": 0.12, + "learning_rate": 4.993910125649561e-06, + "logits/chosen": -1.792074203491211, + "logits/rejected": -1.2048676013946533, + "logps/chosen": -606.4808959960938, + "logps/rejected": -1278.70703125, + "loss": 0.0583, + "rewards/accuracies": 0.925000011920929, + "rewards/chosen": -0.14861299097537994, + "rewards/margins": 0.27981966733932495, + "rewards/rejected": -0.4284326136112213, + "step": 360 + }, + { + "epoch": 0.12, + "learning_rate": 4.991712220111265e-06, + "logits/chosen": -1.8713674545288086, + "logits/rejected": -1.2337167263031006, + "logps/chosen": -554.6063232421875, + "logps/rejected": -996.7093505859375, + "loss": 0.1127, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.12071527540683746, + "rewards/margins": 0.2035401612520218, + "rewards/rejected": -0.32425546646118164, + "step": 370 + }, + { + "epoch": 0.13, + "learning_rate": 4.989176976624511e-06, + "logits/chosen": -1.7384974956512451, + "logits/rejected": -1.1386005878448486, + "logps/chosen": -556.4752807617188, + "logps/rejected": -1155.403564453125, + "loss": 0.0799, + "rewards/accuracies": 0.824999988079071, + "rewards/chosen": -0.13758501410484314, + "rewards/margins": 0.2377336472272873, + "rewards/rejected": -0.3753186762332916, + "step": 380 + }, + { + "epoch": 0.13, + "learning_rate": 4.986304738420684e-06, + "logits/chosen": -1.8614715337753296, + "logits/rejected": -1.1824796199798584, + "logps/chosen": -628.7012329101562, + "logps/rejected": -1232.29296875, + "loss": 0.0766, + "rewards/accuracies": 0.875, + "rewards/chosen": -0.13570387661457062, + "rewards/margins": 0.2571316361427307, + "rewards/rejected": -0.3928355276584625, + "step": 390 + }, + { + "epoch": 0.13, + "learning_rate": 4.983095894354858e-06, + "logits/chosen": -1.7244367599487305, + "logits/rejected": -1.2317596673965454, + "logps/chosen": -643.6784057617188, + "logps/rejected": -1263.03173828125, + "loss": 0.0899, + "rewards/accuracies": 0.800000011920929, + "rewards/chosen": -0.20891329646110535, + "rewards/margins": 0.2505062222480774, + "rewards/rejected": -0.4594195485115051, + "step": 400 + }, + { + "epoch": 0.14, + "learning_rate": 4.979550878853154e-06, + "logits/chosen": -1.7407877445220947, + "logits/rejected": -0.9090331196784973, + "logps/chosen": -650.4755859375, + "logps/rejected": -1323.707763671875, + "loss": 0.0489, + "rewards/accuracies": 0.875, + "rewards/chosen": -0.15291151404380798, + "rewards/margins": 0.3299728333950043, + "rewards/rejected": -0.48288434743881226, + "step": 410 + }, + { + "epoch": 0.14, + "learning_rate": 4.975670171853926e-06, + "logits/chosen": -1.802283525466919, + "logits/rejected": -1.0241405963897705, + "logps/chosen": -637.089599609375, + "logps/rejected": -1355.4974365234375, + "loss": 0.0541, + "rewards/accuracies": 0.949999988079071, + "rewards/chosen": -0.1302250325679779, + "rewards/margins": 0.318683922290802, + "rewards/rejected": -0.4489089548587799, + "step": 420 + }, + { + "epoch": 0.14, + "learning_rate": 4.971454298742779e-06, + "logits/chosen": -1.6149762868881226, + "logits/rejected": -1.1530131101608276, + "logps/chosen": -572.8499755859375, + "logps/rejected": -1171.6473388671875, + "loss": 0.1168, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.14183470606803894, + "rewards/margins": 0.26052767038345337, + "rewards/rejected": -0.4023623466491699, + "step": 430 + }, + { + "epoch": 0.15, + "learning_rate": 4.966903830281449e-06, + "logits/chosen": -1.600284218788147, + "logits/rejected": -1.0997772216796875, + "logps/chosen": -637.7472534179688, + "logps/rejected": -1324.5953369140625, + "loss": 0.073, + "rewards/accuracies": 0.8500000238418579, + "rewards/chosen": -0.16721096634864807, + "rewards/margins": 0.30269843339920044, + "rewards/rejected": -0.4699093699455261, + "step": 440 + }, + { + "epoch": 0.15, + "learning_rate": 4.962019382530521e-06, + "logits/chosen": -1.4888650178909302, + "logits/rejected": -0.9744981527328491, + "logps/chosen": -489.33538818359375, + "logps/rejected": -1064.3255615234375, + "loss": 0.1045, + "rewards/accuracies": 0.800000011920929, + "rewards/chosen": -0.1382354497909546, + "rewards/margins": 0.23734359443187714, + "rewards/rejected": -0.3755790591239929, + "step": 450 + }, + { + "epoch": 0.15, + "learning_rate": 4.956801616766033e-06, + "logits/chosen": -1.4963598251342773, + "logits/rejected": -1.1484572887420654, + "logps/chosen": -499.119384765625, + "logps/rejected": -1084.587158203125, + "loss": 0.1161, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.14311179518699646, + "rewards/margins": 0.24761180579662323, + "rewards/rejected": -0.3907236158847809, + "step": 460 + }, + { + "epoch": 0.16, + "learning_rate": 4.951251239389949e-06, + "logits/chosen": -1.7252105474472046, + "logits/rejected": -1.16670823097229, + "logps/chosen": -446.770263671875, + "logps/rejected": -1193.3294677734375, + "loss": 0.0715, + "rewards/accuracies": 0.8500000238418579, + "rewards/chosen": -0.07284434884786606, + "rewards/margins": 0.26454949378967285, + "rewards/rejected": -0.3373938202857971, + "step": 470 + }, + { + "epoch": 0.16, + "learning_rate": 4.9453690018345144e-06, + "logits/chosen": -1.9960441589355469, + "logits/rejected": -1.3124287128448486, + "logps/chosen": -519.2340087890625, + "logps/rejected": -1184.5723876953125, + "loss": 0.064, + "rewards/accuracies": 0.8999999761581421, + "rewards/chosen": -0.0516168549656868, + "rewards/margins": 0.296741783618927, + "rewards/rejected": -0.3483586013317108, + "step": 480 + }, + { + "epoch": 0.16, + "learning_rate": 4.939155700460536e-06, + "logits/chosen": -1.7496163845062256, + "logits/rejected": -1.1100032329559326, + "logps/chosen": -532.7080078125, + "logps/rejected": -1199.1121826171875, + "loss": 0.1005, + "rewards/accuracies": 0.800000011920929, + "rewards/chosen": -0.13813085854053497, + "rewards/margins": 0.2582959532737732, + "rewards/rejected": -0.396426796913147, + "step": 490 + }, + { + "epoch": 0.17, + "learning_rate": 4.93261217644956e-06, + "logits/chosen": -1.8717424869537354, + "logits/rejected": -1.1161689758300781, + "logps/chosen": -709.0521240234375, + "logps/rejected": -1247.1685791015625, + "loss": 0.0938, + "rewards/accuracies": 0.800000011920929, + "rewards/chosen": -0.21577858924865723, + "rewards/margins": 0.2319638729095459, + "rewards/rejected": -0.44774240255355835, + "step": 500 + }, + { + "epoch": 0.17, + "learning_rate": 4.925739315689991e-06, + "logits/chosen": -1.7890201807022095, + "logits/rejected": -1.2021349668502808, + "logps/chosen": -548.541748046875, + "logps/rejected": -1329.703857421875, + "loss": 0.0815, + "rewards/accuracies": 0.8999999761581421, + "rewards/chosen": -0.11823161691427231, + "rewards/margins": 0.31565752625465393, + "rewards/rejected": -0.43388909101486206, + "step": 510 + }, + { + "epoch": 0.17, + "learning_rate": 4.91853804865716e-06, + "logits/chosen": -2.0370450019836426, + "logits/rejected": -1.1039373874664307, + "logps/chosen": -516.0086669921875, + "logps/rejected": -1140.8289794921875, + "loss": 0.1045, + "rewards/accuracies": 0.8999999761581421, + "rewards/chosen": -0.031420618295669556, + "rewards/margins": 0.28368836641311646, + "rewards/rejected": -0.3151089549064636, + "step": 520 + }, + { + "epoch": 0.18, + "learning_rate": 4.911009350287348e-06, + "logits/chosen": -2.096604824066162, + "logits/rejected": -1.4794657230377197, + "logps/chosen": -484.7252502441406, + "logps/rejected": -1161.0413818359375, + "loss": 0.0601, + "rewards/accuracies": 0.824999988079071, + "rewards/chosen": -0.031848691403865814, + "rewards/margins": 0.29393482208251953, + "rewards/rejected": -0.32578352093696594, + "step": 530 + }, + { + "epoch": 0.18, + "learning_rate": 4.903154239845798e-06, + "logits/chosen": -1.6835973262786865, + "logits/rejected": -1.0456864833831787, + "logps/chosen": -528.0047607421875, + "logps/rejected": -1127.9058837890625, + "loss": 0.1008, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.11576949059963226, + "rewards/margins": 0.23724015057086945, + "rewards/rejected": -0.3530096709728241, + "step": 540 + }, + { + "epoch": 0.18, + "learning_rate": 4.894973780788722e-06, + "logits/chosen": -1.7007629871368408, + "logits/rejected": -1.2928742170333862, + "logps/chosen": -464.3915100097656, + "logps/rejected": -941.9600830078125, + "loss": 0.0885, + "rewards/accuracies": 0.6499999761581421, + "rewards/chosen": -0.10153596103191376, + "rewards/margins": 0.20636403560638428, + "rewards/rejected": -0.30790001153945923, + "step": 550 + }, + { + "epoch": 0.19, + "learning_rate": 4.88646908061933e-06, + "logits/chosen": -1.7848955392837524, + "logits/rejected": -1.275628924369812, + "logps/chosen": -571.953369140625, + "logps/rejected": -1022.10595703125, + "loss": 0.1112, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.12155290693044662, + "rewards/margins": 0.1930900365114212, + "rewards/rejected": -0.3146429657936096, + "step": 560 + }, + { + "epoch": 0.19, + "learning_rate": 4.8776412907378845e-06, + "logits/chosen": -1.714331030845642, + "logits/rejected": -1.2135049104690552, + "logps/chosen": -649.1993408203125, + "logps/rejected": -1231.960205078125, + "loss": 0.0826, + "rewards/accuracies": 0.8999999761581421, + "rewards/chosen": -0.11209288984537125, + "rewards/margins": 0.2815489172935486, + "rewards/rejected": -0.39364179968833923, + "step": 570 + }, + { + "epoch": 0.19, + "learning_rate": 4.868491606285823e-06, + "logits/chosen": -1.8973315954208374, + "logits/rejected": -1.380531907081604, + "logps/chosen": -590.170166015625, + "logps/rejected": -1119.1627197265625, + "loss": 0.0901, + "rewards/accuracies": 0.7749999761581421, + "rewards/chosen": -0.12205713987350464, + "rewards/margins": 0.2438439577817917, + "rewards/rejected": -0.36590105295181274, + "step": 580 + }, + { + "epoch": 0.2, + "learning_rate": 4.859021265983959e-06, + "logits/chosen": -1.8590736389160156, + "logits/rejected": -1.3631782531738281, + "logps/chosen": -622.0940551757812, + "logps/rejected": -1332.389404296875, + "loss": 0.0855, + "rewards/accuracies": 0.8999999761581421, + "rewards/chosen": -0.13945552706718445, + "rewards/margins": 0.28879013657569885, + "rewards/rejected": -0.4282456338405609, + "step": 590 + }, + { + "epoch": 0.2, + "learning_rate": 4.849231551964771e-06, + "logits/chosen": -1.587874174118042, + "logits/rejected": -1.3274825811386108, + "logps/chosen": -697.1937255859375, + "logps/rejected": -1232.4580078125, + "loss": 0.0898, + "rewards/accuracies": 0.824999988079071, + "rewards/chosen": -0.2024591863155365, + "rewards/margins": 0.22971025109291077, + "rewards/rejected": -0.43216943740844727, + "step": 600 + }, + { + "epoch": 0.2, + "learning_rate": 4.839123789598829e-06, + "logits/chosen": -1.5191724300384521, + "logits/rejected": -1.2849657535552979, + "logps/chosen": -543.1875610351562, + "logps/rejected": -1340.810791015625, + "loss": 0.065, + "rewards/accuracies": 0.824999988079071, + "rewards/chosen": -0.11399342864751816, + "rewards/margins": 0.3168831467628479, + "rewards/rejected": -0.43087658286094666, + "step": 610 + }, + { + "epoch": 0.21, + "learning_rate": 4.828699347315357e-06, + "logits/chosen": -1.8808469772338867, + "logits/rejected": -1.1103613376617432, + "logps/chosen": -547.0265502929688, + "logps/rejected": -1164.1422119140625, + "loss": 0.0665, + "rewards/accuracies": 0.875, + "rewards/chosen": -0.07433022558689117, + "rewards/margins": 0.2945881187915802, + "rewards/rejected": -0.36891835927963257, + "step": 620 + }, + { + "epoch": 0.21, + "learning_rate": 4.817959636416969e-06, + "logits/chosen": -1.5736801624298096, + "logits/rejected": -0.8019571304321289, + "logps/chosen": -586.7124633789062, + "logps/rejected": -1286.5223388671875, + "loss": 0.076, + "rewards/accuracies": 0.8500000238418579, + "rewards/chosen": -0.1401849091053009, + "rewards/margins": 0.3048226535320282, + "rewards/rejected": -0.4450075030326843, + "step": 630 + }, + { + "epoch": 0.21, + "learning_rate": 4.806906110888606e-06, + "logits/chosen": -1.6860675811767578, + "logits/rejected": -1.2921956777572632, + "logps/chosen": -588.60986328125, + "logps/rejected": -1157.620361328125, + "loss": 0.0857, + "rewards/accuracies": 0.824999988079071, + "rewards/chosen": -0.1366177797317505, + "rewards/margins": 0.2525423467159271, + "rewards/rejected": -0.3891601264476776, + "step": 640 + }, + { + "epoch": 0.22, + "learning_rate": 4.7955402672006855e-06, + "logits/chosen": -1.8331724405288696, + "logits/rejected": -1.2644875049591064, + "logps/chosen": -634.9255981445312, + "logps/rejected": -1305.0550537109375, + "loss": 0.084, + "rewards/accuracies": 0.8500000238418579, + "rewards/chosen": -0.14361020922660828, + "rewards/margins": 0.3017338216304779, + "rewards/rejected": -0.4453439712524414, + "step": 650 + }, + { + "epoch": 0.22, + "learning_rate": 4.783863644106502e-06, + "logits/chosen": -1.7160533666610718, + "logits/rejected": -1.132188320159912, + "logps/chosen": -611.103759765625, + "logps/rejected": -1045.3682861328125, + "loss": 0.1126, + "rewards/accuracies": 0.6000000238418579, + "rewards/chosen": -0.1587737798690796, + "rewards/margins": 0.1975192278623581, + "rewards/rejected": -0.3562930226325989, + "step": 660 + }, + { + "epoch": 0.22, + "learning_rate": 4.7718778224339115e-06, + "logits/chosen": -1.7684398889541626, + "logits/rejected": -1.1700496673583984, + "logps/chosen": -532.6051025390625, + "logps/rejected": -1173.128662109375, + "loss": 0.0925, + "rewards/accuracies": 0.8500000238418579, + "rewards/chosen": -0.08183015137910843, + "rewards/margins": 0.2822849452495575, + "rewards/rejected": -0.3641151189804077, + "step": 670 + }, + { + "epoch": 0.23, + "learning_rate": 4.759584424871302e-06, + "logits/chosen": -1.9278764724731445, + "logits/rejected": -0.9649882316589355, + "logps/chosen": -626.4827270507812, + "logps/rejected": -1222.0137939453125, + "loss": 0.0645, + "rewards/accuracies": 0.9750000238418579, + "rewards/chosen": -0.106475330889225, + "rewards/margins": 0.29441386461257935, + "rewards/rejected": -0.40088921785354614, + "step": 680 + }, + { + "epoch": 0.23, + "learning_rate": 4.746985115747918e-06, + "logits/chosen": -1.6022624969482422, + "logits/rejected": -1.3960731029510498, + "logps/chosen": -610.5640869140625, + "logps/rejected": -1229.7557373046875, + "loss": 0.0951, + "rewards/accuracies": 0.7749999761581421, + "rewards/chosen": -0.1899951994419098, + "rewards/margins": 0.2415318489074707, + "rewards/rejected": -0.4315270483493805, + "step": 690 + }, + { + "epoch": 0.23, + "learning_rate": 4.734081600808531e-06, + "logits/chosen": -1.6148655414581299, + "logits/rejected": -1.0519297122955322, + "logps/chosen": -587.4749755859375, + "logps/rejected": -1153.4019775390625, + "loss": 0.0781, + "rewards/accuracies": 0.824999988079071, + "rewards/chosen": -0.1272963136434555, + "rewards/margins": 0.2538054585456848, + "rewards/rejected": -0.3811017572879791, + "step": 700 + }, + { + "epoch": 0.24, + "learning_rate": 4.720875626982511e-06, + "logits/chosen": -1.5970103740692139, + "logits/rejected": -1.1024632453918457, + "logps/chosen": -622.9388427734375, + "logps/rejected": -1346.39453125, + "loss": 0.0675, + "rewards/accuracies": 0.875, + "rewards/chosen": -0.17155757546424866, + "rewards/margins": 0.3096774220466614, + "rewards/rejected": -0.4812350273132324, + "step": 710 + }, + { + "epoch": 0.24, + "learning_rate": 4.707368982147318e-06, + "logits/chosen": -1.564493179321289, + "logits/rejected": -1.182403564453125, + "logps/chosen": -557.9251098632812, + "logps/rejected": -1197.14697265625, + "loss": 0.085, + "rewards/accuracies": 0.875, + "rewards/chosen": -0.13098809123039246, + "rewards/margins": 0.27282702922821045, + "rewards/rejected": -0.4038150906562805, + "step": 720 + }, + { + "epoch": 0.24, + "learning_rate": 4.693563494886455e-06, + "logits/chosen": -1.5731178522109985, + "logits/rejected": -1.2045758962631226, + "logps/chosen": -589.6763916015625, + "logps/rejected": -1163.5670166015625, + "loss": 0.0924, + "rewards/accuracies": 0.7749999761581421, + "rewards/chosen": -0.1421278566122055, + "rewards/margins": 0.26834186911582947, + "rewards/rejected": -0.4104697108268738, + "step": 730 + }, + { + "epoch": 0.25, + "learning_rate": 4.679461034241906e-06, + "logits/chosen": -1.5476421117782593, + "logits/rejected": -1.2368953227996826, + "logps/chosen": -624.2233276367188, + "logps/rejected": -1224.5426025390625, + "loss": 0.0832, + "rewards/accuracies": 0.7749999761581421, + "rewards/chosen": -0.1516999900341034, + "rewards/margins": 0.26089781522750854, + "rewards/rejected": -0.41259780526161194, + "step": 740 + }, + { + "epoch": 0.25, + "learning_rate": 4.665063509461098e-06, + "logits/chosen": -1.521240234375, + "logits/rejected": -0.895469069480896, + "logps/chosen": -604.10205078125, + "logps/rejected": -1188.8580322265625, + "loss": 0.0574, + "rewards/accuracies": 0.800000011920929, + "rewards/chosen": -0.1560906320810318, + "rewards/margins": 0.2723260521888733, + "rewards/rejected": -0.4284166693687439, + "step": 750 + }, + { + "epoch": 0.25, + "learning_rate": 4.650372869738415e-06, + "logits/chosen": -1.7068440914154053, + "logits/rejected": -1.068950891494751, + "logps/chosen": -696.2512817382812, + "logps/rejected": -1256.27001953125, + "loss": 0.0588, + "rewards/accuracies": 0.875, + "rewards/chosen": -0.15982361137866974, + "rewards/margins": 0.2860753834247589, + "rewards/rejected": -0.44589900970458984, + "step": 760 + }, + { + "epoch": 0.26, + "learning_rate": 4.635391103951315e-06, + "logits/chosen": -1.7902898788452148, + "logits/rejected": -1.3396075963974, + "logps/chosen": -573.7864990234375, + "logps/rejected": -1184.572998046875, + "loss": 0.086, + "rewards/accuracies": 0.7749999761581421, + "rewards/chosen": -0.14987358450889587, + "rewards/margins": 0.27520817518234253, + "rewards/rejected": -0.4250817894935608, + "step": 770 + }, + { + "epoch": 0.26, + "learning_rate": 4.620120240391065e-06, + "logits/chosen": -1.7508132457733154, + "logits/rejected": -1.1751254796981812, + "logps/chosen": -639.8759765625, + "logps/rejected": -1238.5482177734375, + "loss": 0.0675, + "rewards/accuracies": 0.7749999761581421, + "rewards/chosen": -0.18339064717292786, + "rewards/margins": 0.2874487638473511, + "rewards/rejected": -0.47083941102027893, + "step": 780 + }, + { + "epoch": 0.26, + "learning_rate": 4.604562346488144e-06, + "logits/chosen": -1.7333307266235352, + "logits/rejected": -1.1350371837615967, + "logps/chosen": -610.3933715820312, + "logps/rejected": -1281.648193359375, + "loss": 0.066, + "rewards/accuracies": 0.7749999761581421, + "rewards/chosen": -0.15978248417377472, + "rewards/margins": 0.2836895287036896, + "rewards/rejected": -0.4434719681739807, + "step": 790 + }, + { + "epoch": 0.27, + "learning_rate": 4.588719528532342e-06, + "logits/chosen": -1.5033605098724365, + "logits/rejected": -1.08058500289917, + "logps/chosen": -483.33892822265625, + "logps/rejected": -1186.935546875, + "loss": 0.0829, + "rewards/accuracies": 0.800000011920929, + "rewards/chosen": -0.12199674546718597, + "rewards/margins": 0.32258424162864685, + "rewards/rejected": -0.44458097219467163, + "step": 800 + }, + { + "epoch": 0.27, + "learning_rate": 4.572593931387604e-06, + "logits/chosen": -1.4822036027908325, + "logits/rejected": -1.1556123495101929, + "logps/chosen": -739.7813110351562, + "logps/rejected": -1216.9873046875, + "loss": 0.1328, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.24871714413166046, + "rewards/margins": 0.21633216738700867, + "rewards/rejected": -0.4650493264198303, + "step": 810 + }, + { + "epoch": 0.27, + "learning_rate": 4.556187738201656e-06, + "logits/chosen": -1.767778754234314, + "logits/rejected": -1.1542942523956299, + "logps/chosen": -614.69677734375, + "logps/rejected": -1348.5400390625, + "loss": 0.0518, + "rewards/accuracies": 0.9750000238418579, + "rewards/chosen": -0.14709432423114777, + "rewards/margins": 0.3316415846347809, + "rewards/rejected": -0.47873586416244507, + "step": 820 + }, + { + "epoch": 0.28, + "learning_rate": 4.539503170110431e-06, + "logits/chosen": -1.6926486492156982, + "logits/rejected": -1.1664977073669434, + "logps/chosen": -735.2862548828125, + "logps/rejected": -1265.770263671875, + "loss": 0.0932, + "rewards/accuracies": 0.800000011920929, + "rewards/chosen": -0.18908649682998657, + "rewards/margins": 0.24448461830615997, + "rewards/rejected": -0.4335711598396301, + "step": 830 + }, + { + "epoch": 0.28, + "learning_rate": 4.522542485937369e-06, + "logits/chosen": -1.3090674877166748, + "logits/rejected": -1.0646418333053589, + "logps/chosen": -558.6735229492188, + "logps/rejected": -1291.3309326171875, + "loss": 0.073, + "rewards/accuracies": 0.8500000238418579, + "rewards/chosen": -0.1368141919374466, + "rewards/margins": 0.3053056597709656, + "rewards/rejected": -0.44211989641189575, + "step": 840 + }, + { + "epoch": 0.28, + "learning_rate": 4.50530798188761e-06, + "logits/chosen": -1.625422477722168, + "logits/rejected": -0.9810975790023804, + "logps/chosen": -609.5338134765625, + "logps/rejected": -1308.818115234375, + "loss": 0.0728, + "rewards/accuracies": 0.949999988079071, + "rewards/chosen": -0.09720011800527573, + "rewards/margins": 0.30527880787849426, + "rewards/rejected": -0.4024789333343506, + "step": 850 + }, + { + "epoch": 0.29, + "learning_rate": 4.48780199123712e-06, + "logits/chosen": -1.5866591930389404, + "logits/rejected": -1.1791462898254395, + "logps/chosen": -654.497314453125, + "logps/rejected": -1295.8955078125, + "loss": 0.1033, + "rewards/accuracies": 0.824999988079071, + "rewards/chosen": -0.1393161565065384, + "rewards/margins": 0.2792035937309265, + "rewards/rejected": -0.4185197353363037, + "step": 860 + }, + { + "epoch": 0.29, + "learning_rate": 4.470026884016805e-06, + "logits/chosen": -1.6597150564193726, + "logits/rejected": -1.071593999862671, + "logps/chosen": -675.8863525390625, + "logps/rejected": -1333.055908203125, + "loss": 0.0584, + "rewards/accuracies": 0.8999999761581421, + "rewards/chosen": -0.16408918797969818, + "rewards/margins": 0.300179660320282, + "rewards/rejected": -0.46426883339881897, + "step": 870 + }, + { + "epoch": 0.29, + "learning_rate": 4.451985066691649e-06, + "logits/chosen": -1.533385992050171, + "logits/rejected": -0.9318403005599976, + "logps/chosen": -672.769287109375, + "logps/rejected": -1254.0054931640625, + "loss": 0.0782, + "rewards/accuracies": 0.8500000238418579, + "rewards/chosen": -0.18980053067207336, + "rewards/margins": 0.24227580428123474, + "rewards/rejected": -0.4320763647556305, + "step": 880 + }, + { + "epoch": 0.3, + "learning_rate": 4.4336789818349105e-06, + "logits/chosen": -1.9593079090118408, + "logits/rejected": -1.2628549337387085, + "logps/chosen": -551.6669311523438, + "logps/rejected": -1144.910888671875, + "loss": 0.0791, + "rewards/accuracies": 0.824999988079071, + "rewards/chosen": -0.11993531137704849, + "rewards/margins": 0.2696130871772766, + "rewards/rejected": -0.3895483613014221, + "step": 890 + }, + { + "epoch": 0.3, + "learning_rate": 4.415111107797445e-06, + "logits/chosen": -1.7251970767974854, + "logits/rejected": -1.1421195268630981, + "logps/chosen": -561.43212890625, + "logps/rejected": -1138.663818359375, + "loss": 0.0933, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.08620895445346832, + "rewards/margins": 0.272499144077301, + "rewards/rejected": -0.35870811343193054, + "step": 900 + }, + { + "epoch": 0.3, + "learning_rate": 4.396283958372173e-06, + "logits/chosen": -1.675585150718689, + "logits/rejected": -1.1132071018218994, + "logps/chosen": -531.9155883789062, + "logps/rejected": -1153.8314208984375, + "loss": 0.0866, + "rewards/accuracies": 0.800000011920929, + "rewards/chosen": -0.06681212037801743, + "rewards/margins": 0.28186994791030884, + "rewards/rejected": -0.34868210554122925, + "step": 910 + }, + { + "epoch": 0.31, + "learning_rate": 4.377200082453748e-06, + "logits/chosen": -1.6496893167495728, + "logits/rejected": -1.0332069396972656, + "logps/chosen": -485.20947265625, + "logps/rejected": -1017.17138671875, + "loss": 0.0822, + "rewards/accuracies": 0.824999988079071, + "rewards/chosen": -0.057542771100997925, + "rewards/margins": 0.2658589780330658, + "rewards/rejected": -0.3234017789363861, + "step": 920 + }, + { + "epoch": 0.31, + "learning_rate": 4.357862063693486e-06, + "logits/chosen": -1.699328064918518, + "logits/rejected": -0.9203550219535828, + "logps/chosen": -631.9972534179688, + "logps/rejected": -1354.697509765625, + "loss": 0.0503, + "rewards/accuracies": 0.925000011920929, + "rewards/chosen": -0.16203594207763672, + "rewards/margins": 0.3373655676841736, + "rewards/rejected": -0.4994015097618103, + "step": 930 + }, + { + "epoch": 0.31, + "learning_rate": 4.338272520149572e-06, + "logits/chosen": -1.18538498878479, + "logits/rejected": -0.7905349731445312, + "logps/chosen": -693.11474609375, + "logps/rejected": -1318.9244384765625, + "loss": 0.0951, + "rewards/accuracies": 0.824999988079071, + "rewards/chosen": -0.20799028873443604, + "rewards/margins": 0.2950304448604584, + "rewards/rejected": -0.5030207633972168, + "step": 940 + }, + { + "epoch": 0.32, + "learning_rate": 4.318434103932622e-06, + "logits/chosen": -1.5692521333694458, + "logits/rejected": -1.009714961051941, + "logps/chosen": -633.1151123046875, + "logps/rejected": -1297.8974609375, + "loss": 0.0691, + "rewards/accuracies": 0.875, + "rewards/chosen": -0.19843876361846924, + "rewards/margins": 0.31439024209976196, + "rewards/rejected": -0.5128290057182312, + "step": 950 + }, + { + "epoch": 0.32, + "learning_rate": 4.2983495008466285e-06, + "logits/chosen": -1.5247453451156616, + "logits/rejected": -0.8545368313789368, + "logps/chosen": -540.1340942382812, + "logps/rejected": -1173.1444091796875, + "loss": 0.0566, + "rewards/accuracies": 0.824999988079071, + "rewards/chosen": -0.09455517679452896, + "rewards/margins": 0.2999194264411926, + "rewards/rejected": -0.394474595785141, + "step": 960 + }, + { + "epoch": 0.32, + "learning_rate": 4.278021430025343e-06, + "logits/chosen": -1.7150371074676514, + "logits/rejected": -0.8671354055404663, + "logps/chosen": -531.0009155273438, + "logps/rejected": -1244.0396728515625, + "loss": 0.0531, + "rewards/accuracies": 0.8500000238418579, + "rewards/chosen": -0.11200281232595444, + "rewards/margins": 0.33397001028060913, + "rewards/rejected": -0.44597283005714417, + "step": 970 + }, + { + "epoch": 0.33, + "learning_rate": 4.257452643564155e-06, + "logits/chosen": -1.7381508350372314, + "logits/rejected": -0.9717987179756165, + "logps/chosen": -660.3062744140625, + "logps/rejected": -1395.575927734375, + "loss": 0.0381, + "rewards/accuracies": 0.949999988079071, + "rewards/chosen": -0.16878202557563782, + "rewards/margins": 0.35611775517463684, + "rewards/rejected": -0.5248997807502747, + "step": 980 + }, + { + "epoch": 0.33, + "learning_rate": 4.236645926147493e-06, + "logits/chosen": -1.6484124660491943, + "logits/rejected": -1.1722246408462524, + "logps/chosen": -589.8314208984375, + "logps/rejected": -1167.5753173828125, + "loss": 0.0772, + "rewards/accuracies": 0.800000011920929, + "rewards/chosen": -0.18166238069534302, + "rewards/margins": 0.2549310624599457, + "rewards/rejected": -0.4365933835506439, + "step": 990 + }, + { + "epoch": 0.33, + "learning_rate": 4.215604094671835e-06, + "logits/chosen": -1.650120496749878, + "logits/rejected": -1.1718031167984009, + "logps/chosen": -673.8374633789062, + "logps/rejected": -1340.0517578125, + "loss": 0.0901, + "rewards/accuracies": 0.800000011920929, + "rewards/chosen": -0.21558384597301483, + "rewards/margins": 0.26856911182403564, + "rewards/rejected": -0.4841529428958893, + "step": 1000 + }, + { + "epoch": 0.34, + "learning_rate": 4.194329997864331e-06, + "logits/chosen": -1.612789511680603, + "logits/rejected": -1.0092923641204834, + "logps/chosen": -659.1913452148438, + "logps/rejected": -1288.3807373046875, + "loss": 0.0598, + "rewards/accuracies": 0.875, + "rewards/chosen": -0.1437804251909256, + "rewards/margins": 0.31068724393844604, + "rewards/rejected": -0.45446768403053284, + "step": 1010 + }, + { + "epoch": 0.34, + "learning_rate": 4.172826515897146e-06, + "logits/chosen": -1.5820199251174927, + "logits/rejected": -1.1678297519683838, + "logps/chosen": -549.3736572265625, + "logps/rejected": -1264.0751953125, + "loss": 0.0723, + "rewards/accuracies": 0.8500000238418579, + "rewards/chosen": -0.12197370827198029, + "rewards/margins": 0.32163676619529724, + "rewards/rejected": -0.4436105191707611, + "step": 1020 + }, + { + "epoch": 0.34, + "learning_rate": 4.151096559997519e-06, + "logits/chosen": -1.5460295677185059, + "logits/rejected": -0.9339001774787903, + "logps/chosen": -657.1842041015625, + "logps/rejected": -1248.517822265625, + "loss": 0.0713, + "rewards/accuracies": 0.875, + "rewards/chosen": -0.17015226185321808, + "rewards/margins": 0.2676629424095154, + "rewards/rejected": -0.43781518936157227, + "step": 1030 + }, + { + "epoch": 0.35, + "learning_rate": 4.129143072053639e-06, + "logits/chosen": -1.3850871324539185, + "logits/rejected": -0.6763152480125427, + "logps/chosen": -712.0982666015625, + "logps/rejected": -1295.492431640625, + "loss": 0.0918, + "rewards/accuracies": 0.7749999761581421, + "rewards/chosen": -0.24367225170135498, + "rewards/margins": 0.27682679891586304, + "rewards/rejected": -0.5204989910125732, + "step": 1040 + }, + { + "epoch": 0.35, + "learning_rate": 4.106969024216348e-06, + "logits/chosen": -1.5604678392410278, + "logits/rejected": -1.1585630178451538, + "logps/chosen": -631.67041015625, + "logps/rejected": -1394.1226806640625, + "loss": 0.0878, + "rewards/accuracies": 0.875, + "rewards/chosen": -0.22048361599445343, + "rewards/margins": 0.31476694345474243, + "rewards/rejected": -0.5352505445480347, + "step": 1050 + }, + { + "epoch": 0.35, + "learning_rate": 4.084577418496775e-06, + "logits/chosen": -1.5729732513427734, + "logits/rejected": -0.983799159526825, + "logps/chosen": -644.2813720703125, + "logps/rejected": -1280.1788330078125, + "loss": 0.067, + "rewards/accuracies": 0.875, + "rewards/chosen": -0.21176350116729736, + "rewards/margins": 0.33453696966171265, + "rewards/rejected": -0.54630047082901, + "step": 1060 + }, + { + "epoch": 0.36, + "learning_rate": 4.0619712863599005e-06, + "logits/chosen": -1.5669440031051636, + "logits/rejected": -1.1483709812164307, + "logps/chosen": -664.9472045898438, + "logps/rejected": -1134.748046875, + "loss": 0.1104, + "rewards/accuracies": 0.7749999761581421, + "rewards/chosen": -0.2157289683818817, + "rewards/margins": 0.22911766171455383, + "rewards/rejected": -0.44484663009643555, + "step": 1070 + }, + { + "epoch": 0.36, + "learning_rate": 4.039153688314146e-06, + "logits/chosen": -1.594029188156128, + "logits/rejected": -1.106323003768921, + "logps/chosen": -597.1380615234375, + "logps/rejected": -1102.068115234375, + "loss": 0.1373, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.17422744631767273, + "rewards/margins": 0.21986624598503113, + "rewards/rejected": -0.39409369230270386, + "step": 1080 + }, + { + "epoch": 0.36, + "learning_rate": 4.016127713497034e-06, + "logits/chosen": -1.6402870416641235, + "logits/rejected": -0.9547089338302612, + "logps/chosen": -685.6248168945312, + "logps/rejected": -1294.0582275390625, + "loss": 0.0874, + "rewards/accuracies": 0.824999988079071, + "rewards/chosen": -0.16791805624961853, + "rewards/margins": 0.27242207527160645, + "rewards/rejected": -0.440340131521225, + "step": 1090 + }, + { + "epoch": 0.37, + "learning_rate": 3.992896479256966e-06, + "logits/chosen": -1.5232279300689697, + "logits/rejected": -1.086789608001709, + "logps/chosen": -680.0406494140625, + "logps/rejected": -1135.125, + "loss": 0.1149, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.22198696434497833, + "rewards/margins": 0.21400491893291473, + "rewards/rejected": -0.43599191308021545, + "step": 1100 + }, + { + "epoch": 0.37, + "learning_rate": 3.969463130731183e-06, + "logits/chosen": -1.3749518394470215, + "logits/rejected": -0.9445620775222778, + "logps/chosen": -614.1600952148438, + "logps/rejected": -1332.16650390625, + "loss": 0.0712, + "rewards/accuracies": 0.8500000238418579, + "rewards/chosen": -0.16799691319465637, + "rewards/margins": 0.3212730288505554, + "rewards/rejected": -0.4892699122428894, + "step": 1110 + }, + { + "epoch": 0.37, + "learning_rate": 3.945830840419966e-06, + "logits/chosen": -1.5307501554489136, + "logits/rejected": -0.9613451957702637, + "logps/chosen": -572.4707641601562, + "logps/rejected": -1237.0767822265625, + "loss": 0.0536, + "rewards/accuracies": 0.875, + "rewards/chosen": -0.14543718099594116, + "rewards/margins": 0.29924580454826355, + "rewards/rejected": -0.4446829855442047, + "step": 1120 + }, + { + "epoch": 0.38, + "learning_rate": 3.92200280775713e-06, + "logits/chosen": -1.3332005739212036, + "logits/rejected": -0.8951355218887329, + "logps/chosen": -686.0225830078125, + "logps/rejected": -1341.870361328125, + "loss": 0.0835, + "rewards/accuracies": 0.8500000238418579, + "rewards/chosen": -0.22055836021900177, + "rewards/margins": 0.28799134492874146, + "rewards/rejected": -0.508549690246582, + "step": 1130 + }, + { + "epoch": 0.38, + "learning_rate": 3.897982258676867e-06, + "logits/chosen": -1.2923948764801025, + "logits/rejected": -1.0148731470108032, + "logps/chosen": -460.26336669921875, + "logps/rejected": -1217.972412109375, + "loss": 0.067, + "rewards/accuracies": 0.824999988079071, + "rewards/chosen": -0.09934146702289581, + "rewards/margins": 0.30596452951431274, + "rewards/rejected": -0.40530601143836975, + "step": 1140 + }, + { + "epoch": 0.38, + "learning_rate": 3.8737724451770155e-06, + "logits/chosen": -1.660675287246704, + "logits/rejected": -0.7988176345825195, + "logps/chosen": -664.8593139648438, + "logps/rejected": -1191.18017578125, + "loss": 0.0708, + "rewards/accuracies": 0.8500000238418579, + "rewards/chosen": -0.12432744354009628, + "rewards/margins": 0.2866395115852356, + "rewards/rejected": -0.4109669625759125, + "step": 1150 + }, + { + "epoch": 0.39, + "learning_rate": 3.849376644878783e-06, + "logits/chosen": -1.7713876962661743, + "logits/rejected": -0.9808756113052368, + "logps/chosen": -621.3865356445312, + "logps/rejected": -1367.19482421875, + "loss": 0.0447, + "rewards/accuracies": 0.8999999761581421, + "rewards/chosen": -0.1509043574333191, + "rewards/margins": 0.34973520040512085, + "rewards/rejected": -0.5006396174430847, + "step": 1160 + }, + { + "epoch": 0.39, + "learning_rate": 3.824798160583012e-06, + "logits/chosen": -1.5198707580566406, + "logits/rejected": -0.90589439868927, + "logps/chosen": -679.7938232421875, + "logps/rejected": -1217.0797119140625, + "loss": 0.0979, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.20723366737365723, + "rewards/margins": 0.2676594853401184, + "rewards/rejected": -0.47489315271377563, + "step": 1170 + }, + { + "epoch": 0.39, + "learning_rate": 3.8000403198230385e-06, + "logits/chosen": -1.8150522708892822, + "logits/rejected": -0.9454204440116882, + "logps/chosen": -640.8989868164062, + "logps/rejected": -1308.7498779296875, + "loss": 0.0713, + "rewards/accuracies": 0.8999999761581421, + "rewards/chosen": -0.12217221409082413, + "rewards/margins": 0.3217760920524597, + "rewards/rejected": -0.44394832849502563, + "step": 1180 + }, + { + "epoch": 0.4, + "learning_rate": 3.7751064744141886e-06, + "logits/chosen": -1.1891100406646729, + "logits/rejected": -1.013519525527954, + "logps/chosen": -462.0289001464844, + "logps/rejected": -1144.2987060546875, + "loss": 0.0713, + "rewards/accuracies": 0.824999988079071, + "rewards/chosen": -0.057515986263751984, + "rewards/margins": 0.3108125925064087, + "rewards/rejected": -0.36832860112190247, + "step": 1190 + }, + { + "epoch": 0.4, + "learning_rate": 3.7500000000000005e-06, + "logits/chosen": -1.53715980052948, + "logits/rejected": -1.0485864877700806, + "logps/chosen": -620.6614379882812, + "logps/rejected": -1361.7210693359375, + "loss": 0.0534, + "rewards/accuracies": 0.925000011920929, + "rewards/chosen": -0.14007523655891418, + "rewards/margins": 0.3020576536655426, + "rewards/rejected": -0.4421328604221344, + "step": 1200 + }, + { + "epoch": 0.4, + "learning_rate": 3.724724295595218e-06, + "logits/chosen": -1.4643341302871704, + "logits/rejected": -0.7922446131706238, + "logps/chosen": -477.1775817871094, + "logps/rejected": -1089.298583984375, + "loss": 0.0713, + "rewards/accuracies": 0.8500000238418579, + "rewards/chosen": -0.0700678676366806, + "rewards/margins": 0.27943548560142517, + "rewards/rejected": -0.3495033383369446, + "step": 1210 + }, + { + "epoch": 0.41, + "learning_rate": 3.699282783125616e-06, + "logits/chosen": -1.6988475322723389, + "logits/rejected": -1.08052396774292, + "logps/chosen": -519.7030029296875, + "logps/rejected": -1169.624755859375, + "loss": 0.0815, + "rewards/accuracies": 0.8500000238418579, + "rewards/chosen": -0.10375889390707016, + "rewards/margins": 0.2844247817993164, + "rewards/rejected": -0.3881836533546448, + "step": 1220 + }, + { + "epoch": 0.41, + "learning_rate": 3.6736789069647273e-06, + "logits/chosen": -1.5467779636383057, + "logits/rejected": -0.878455638885498, + "logps/chosen": -689.0154418945312, + "logps/rejected": -1260.757568359375, + "loss": 0.0964, + "rewards/accuracies": 0.824999988079071, + "rewards/chosen": -0.20392462611198425, + "rewards/margins": 0.26304885745048523, + "rewards/rejected": -0.46697354316711426, + "step": 1230 + }, + { + "epoch": 0.41, + "learning_rate": 3.6479161334675294e-06, + "logits/chosen": -1.4715616703033447, + "logits/rejected": -0.9548036456108093, + "logps/chosen": -503.00982666015625, + "logps/rejected": -1227.5513916015625, + "loss": 0.06, + "rewards/accuracies": 0.8500000238418579, + "rewards/chosen": -0.09338453412055969, + "rewards/margins": 0.32899850606918335, + "rewards/rejected": -0.42238301038742065, + "step": 1240 + }, + { + "epoch": 0.42, + "learning_rate": 3.621997950501156e-06, + "logits/chosen": -1.564557433128357, + "logits/rejected": -1.0119483470916748, + "logps/chosen": -533.0374145507812, + "logps/rejected": -1133.299072265625, + "loss": 0.0832, + "rewards/accuracies": 0.7749999761581421, + "rewards/chosen": -0.12650765478610992, + "rewards/margins": 0.2633581757545471, + "rewards/rejected": -0.38986584544181824, + "step": 1250 + }, + { + "epoch": 0.42, + "learning_rate": 3.595927866972694e-06, + "logits/chosen": -1.5929651260375977, + "logits/rejected": -0.946239173412323, + "logps/chosen": -796.1531982421875, + "logps/rejected": -1373.961181640625, + "loss": 0.1067, + "rewards/accuracies": 0.7749999761581421, + "rewards/chosen": -0.258171945810318, + "rewards/margins": 0.25018230080604553, + "rewards/rejected": -0.5083541870117188, + "step": 1260 + }, + { + "epoch": 0.42, + "learning_rate": 3.5697094123541357e-06, + "logits/chosen": -1.3169076442718506, + "logits/rejected": -0.6554785370826721, + "logps/chosen": -671.5099487304688, + "logps/rejected": -1274.65576171875, + "loss": 0.072, + "rewards/accuracies": 0.824999988079071, + "rewards/chosen": -0.23343141376972198, + "rewards/margins": 0.258681058883667, + "rewards/rejected": -0.49211248755455017, + "step": 1270 + }, + { + "epoch": 0.43, + "learning_rate": 3.543346136204545e-06, + "logits/chosen": -1.5627593994140625, + "logits/rejected": -1.0817673206329346, + "logps/chosen": -626.169189453125, + "logps/rejected": -1331.075439453125, + "loss": 0.0776, + "rewards/accuracies": 0.824999988079071, + "rewards/chosen": -0.23857741057872772, + "rewards/margins": 0.29800939559936523, + "rewards/rejected": -0.5365867614746094, + "step": 1280 + }, + { + "epoch": 0.43, + "learning_rate": 3.516841607689501e-06, + "logits/chosen": -1.4650442600250244, + "logits/rejected": -0.9328089952468872, + "logps/chosen": -621.0172729492188, + "logps/rejected": -1351.593017578125, + "loss": 0.0719, + "rewards/accuracies": 0.875, + "rewards/chosen": -0.204871267080307, + "rewards/margins": 0.3157797157764435, + "rewards/rejected": -0.5206509828567505, + "step": 1290 + }, + { + "epoch": 0.43, + "learning_rate": 3.4901994150978926e-06, + "logits/chosen": -1.4772984981536865, + "logits/rejected": -1.0536553859710693, + "logps/chosen": -532.893310546875, + "logps/rejected": -1081.2454833984375, + "loss": 0.0756, + "rewards/accuracies": 0.824999988079071, + "rewards/chosen": -0.09208095073699951, + "rewards/margins": 0.2642812132835388, + "rewards/rejected": -0.35636216402053833, + "step": 1300 + }, + { + "epoch": 0.44, + "learning_rate": 3.4634231653561213e-06, + "logits/chosen": -1.5390266180038452, + "logits/rejected": -0.9685665965080261, + "logps/chosen": -569.3391723632812, + "logps/rejected": -1077.9581298828125, + "loss": 0.0826, + "rewards/accuracies": 0.8500000238418579, + "rewards/chosen": -0.08572982251644135, + "rewards/margins": 0.25851181149482727, + "rewards/rejected": -0.3442416191101074, + "step": 1310 + }, + { + "epoch": 0.44, + "learning_rate": 3.436516483539781e-06, + "logits/chosen": -1.7010656595230103, + "logits/rejected": -1.0056097507476807, + "logps/chosen": -629.4781494140625, + "logps/rejected": -1221.6732177734375, + "loss": 0.0792, + "rewards/accuracies": 0.8500000238418579, + "rewards/chosen": -0.13440915942192078, + "rewards/margins": 0.266291081905365, + "rewards/rejected": -0.40070027112960815, + "step": 1320 + }, + { + "epoch": 0.44, + "learning_rate": 3.409483012382879e-06, + "logits/chosen": -1.8139673471450806, + "logits/rejected": -1.0937250852584839, + "logps/chosen": -542.4208984375, + "logps/rejected": -1275.978271484375, + "loss": 0.0647, + "rewards/accuracies": 0.824999988079071, + "rewards/chosen": -0.12924353778362274, + "rewards/margins": 0.3123559355735779, + "rewards/rejected": -0.44159942865371704, + "step": 1330 + }, + { + "epoch": 0.45, + "learning_rate": 3.3823264117846722e-06, + "logits/chosen": -1.34089195728302, + "logits/rejected": -1.0331387519836426, + "logps/chosen": -669.740966796875, + "logps/rejected": -1183.39404296875, + "loss": 0.0927, + "rewards/accuracies": 0.800000011920929, + "rewards/chosen": -0.16455522179603577, + "rewards/margins": 0.24129804968833923, + "rewards/rejected": -0.4058533310890198, + "step": 1340 + }, + { + "epoch": 0.45, + "learning_rate": 3.3550503583141726e-06, + "logits/chosen": -1.4803988933563232, + "logits/rejected": -1.2195205688476562, + "logps/chosen": -631.5969848632812, + "logps/rejected": -1318.831787109375, + "loss": 0.0847, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.24165017902851105, + "rewards/margins": 0.2722802758216858, + "rewards/rejected": -0.5139304399490356, + "step": 1350 + }, + { + "epoch": 0.45, + "learning_rate": 3.3276585447123957e-06, + "logits/chosen": -1.698537826538086, + "logits/rejected": -1.1106715202331543, + "logps/chosen": -758.1787719726562, + "logps/rejected": -1358.7789306640625, + "loss": 0.0781, + "rewards/accuracies": 0.8999999761581421, + "rewards/chosen": -0.2571624517440796, + "rewards/margins": 0.247967928647995, + "rewards/rejected": -0.5051303505897522, + "step": 1360 + }, + { + "epoch": 0.46, + "learning_rate": 3.300154679392429e-06, + "logits/chosen": -1.565079927444458, + "logits/rejected": -0.8933359980583191, + "logps/chosen": -655.6455078125, + "logps/rejected": -1246.6724853515625, + "loss": 0.0855, + "rewards/accuracies": 0.7749999761581421, + "rewards/chosen": -0.2172234058380127, + "rewards/margins": 0.259296715259552, + "rewards/rejected": -0.4765201508998871, + "step": 1370 + }, + { + "epoch": 0.46, + "learning_rate": 3.272542485937369e-06, + "logits/chosen": -1.7766815423965454, + "logits/rejected": -1.0995599031448364, + "logps/chosen": -622.3171997070312, + "logps/rejected": -1246.709716796875, + "loss": 0.0775, + "rewards/accuracies": 0.875, + "rewards/chosen": -0.16058366000652313, + "rewards/margins": 0.2945863604545593, + "rewards/rejected": -0.45517000555992126, + "step": 1380 + }, + { + "epoch": 0.46, + "learning_rate": 3.244825702596205e-06, + "logits/chosen": -1.6319786310195923, + "logits/rejected": -1.2775378227233887, + "logps/chosen": -543.9849853515625, + "logps/rejected": -1321.574462890625, + "loss": 0.0694, + "rewards/accuracies": 0.8500000238418579, + "rewards/chosen": -0.13428989052772522, + "rewards/margins": 0.3316313922405243, + "rewards/rejected": -0.4659213125705719, + "step": 1390 + }, + { + "epoch": 0.47, + "learning_rate": 3.217008081777726e-06, + "logits/chosen": -1.6392805576324463, + "logits/rejected": -1.2436825037002563, + "logps/chosen": -662.4228515625, + "logps/rejected": -1163.439208984375, + "loss": 0.1084, + "rewards/accuracies": 0.7749999761581421, + "rewards/chosen": -0.1475670039653778, + "rewards/margins": 0.2356584370136261, + "rewards/rejected": -0.38322538137435913, + "step": 1400 + }, + { + "epoch": 0.47, + "learning_rate": 3.189093389542498e-06, + "logits/chosen": -1.8173353672027588, + "logits/rejected": -0.9919673204421997, + "logps/chosen": -665.50537109375, + "logps/rejected": -1430.367431640625, + "loss": 0.0367, + "rewards/accuracies": 0.9750000238418579, + "rewards/chosen": -0.16980603337287903, + "rewards/margins": 0.3760393559932709, + "rewards/rejected": -0.5458453893661499, + "step": 1410 + }, + { + "epoch": 0.47, + "learning_rate": 3.1610854050930063e-06, + "logits/chosen": -1.5377254486083984, + "logits/rejected": -0.8976453542709351, + "logps/chosen": -662.4138793945312, + "logps/rejected": -1309.8470458984375, + "loss": 0.0547, + "rewards/accuracies": 0.8999999761581421, + "rewards/chosen": -0.20192308723926544, + "rewards/margins": 0.30307039618492126, + "rewards/rejected": -0.5049934983253479, + "step": 1420 + }, + { + "epoch": 0.48, + "learning_rate": 3.132987920262005e-06, + "logits/chosen": -1.5729992389678955, + "logits/rejected": -0.892600417137146, + "logps/chosen": -780.0401000976562, + "logps/rejected": -1402.983154296875, + "loss": 0.0667, + "rewards/accuracies": 0.8500000238418579, + "rewards/chosen": -0.18477649986743927, + "rewards/margins": 0.33369770646095276, + "rewards/rejected": -0.5184742212295532, + "step": 1430 + }, + { + "epoch": 0.48, + "learning_rate": 3.1048047389991693e-06, + "logits/chosen": -1.5943243503570557, + "logits/rejected": -0.9319471120834351, + "logps/chosen": -730.2401733398438, + "logps/rejected": -1355.317626953125, + "loss": 0.0786, + "rewards/accuracies": 0.8500000238418579, + "rewards/chosen": -0.1811259537935257, + "rewards/margins": 0.3043223023414612, + "rewards/rejected": -0.48544827103614807, + "step": 1440 + }, + { + "epoch": 0.48, + "learning_rate": 3.0765396768561005e-06, + "logits/chosen": -1.8524019718170166, + "logits/rejected": -0.9115484356880188, + "logps/chosen": -643.4202270507812, + "logps/rejected": -1369.478271484375, + "loss": 0.0631, + "rewards/accuracies": 0.8999999761581421, + "rewards/chosen": -0.1561841517686844, + "rewards/margins": 0.35857564210891724, + "rewards/rejected": -0.5147597193717957, + "step": 1450 + }, + { + "epoch": 0.49, + "learning_rate": 3.0481965604697582e-06, + "logits/chosen": -1.8506492376327515, + "logits/rejected": -1.3686341047286987, + "logps/chosen": -625.1241455078125, + "logps/rejected": -1145.7611083984375, + "loss": 0.0856, + "rewards/accuracies": 0.7749999761581421, + "rewards/chosen": -0.14623615145683289, + "rewards/margins": 0.25150930881500244, + "rewards/rejected": -0.39774543046951294, + "step": 1460 + }, + { + "epoch": 0.49, + "learning_rate": 3.019779227044398e-06, + "logits/chosen": -1.5452914237976074, + "logits/rejected": -1.0970354080200195, + "logps/chosen": -576.4892578125, + "logps/rejected": -1214.838623046875, + "loss": 0.054, + "rewards/accuracies": 0.800000011920929, + "rewards/chosen": -0.15811112523078918, + "rewards/margins": 0.29070132970809937, + "rewards/rejected": -0.44881248474121094, + "step": 1470 + }, + { + "epoch": 0.49, + "learning_rate": 2.9912915238320755e-06, + "logits/chosen": -1.5780893564224243, + "logits/rejected": -1.1927622556686401, + "logps/chosen": -574.8433227539062, + "logps/rejected": -1284.197998046875, + "loss": 0.0639, + "rewards/accuracies": 0.8500000238418579, + "rewards/chosen": -0.14699247479438782, + "rewards/margins": 0.30907437205314636, + "rewards/rejected": -0.4560668468475342, + "step": 1480 + }, + { + "epoch": 0.5, + "learning_rate": 2.9627373076117864e-06, + "logits/chosen": -1.6259567737579346, + "logits/rejected": -1.0762062072753906, + "logps/chosen": -532.4714965820312, + "logps/rejected": -1246.502685546875, + "loss": 0.0554, + "rewards/accuracies": 0.800000011920929, + "rewards/chosen": -0.15746064484119415, + "rewards/margins": 0.3116622865200043, + "rewards/rejected": -0.4691229462623596, + "step": 1490 + }, + { + "epoch": 0.5, + "learning_rate": 2.9341204441673267e-06, + "logits/chosen": -1.4439929723739624, + "logits/rejected": -1.0250179767608643, + "logps/chosen": -622.338134765625, + "logps/rejected": -1203.972412109375, + "loss": 0.0955, + "rewards/accuracies": 0.824999988079071, + "rewards/chosen": -0.17773118615150452, + "rewards/margins": 0.2580625116825104, + "rewards/rejected": -0.4357937276363373, + "step": 1500 + }, + { + "epoch": 0.5, + "learning_rate": 2.9054448077639193e-06, + "logits/chosen": -1.5762206315994263, + "logits/rejected": -0.9138490557670593, + "logps/chosen": -586.79345703125, + "logps/rejected": -1128.914794921875, + "loss": 0.1417, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.16046829521656036, + "rewards/margins": 0.23752140998840332, + "rewards/rejected": -0.3979896903038025, + "step": 1510 + }, + { + "epoch": 0.51, + "learning_rate": 2.876714280623708e-06, + "logits/chosen": -1.5804052352905273, + "logits/rejected": -1.3289867639541626, + "logps/chosen": -565.5819091796875, + "logps/rejected": -1220.959228515625, + "loss": 0.0819, + "rewards/accuracies": 0.824999988079071, + "rewards/chosen": -0.16751018166542053, + "rewards/margins": 0.2462599277496338, + "rewards/rejected": -0.4137701392173767, + "step": 1520 + }, + { + "epoch": 0.51, + "learning_rate": 2.847932752400164e-06, + "logits/chosen": -1.6598694324493408, + "logits/rejected": -1.195133090019226, + "logps/chosen": -587.2601318359375, + "logps/rejected": -1277.15283203125, + "loss": 0.0684, + "rewards/accuracies": 0.875, + "rewards/chosen": -0.12616164982318878, + "rewards/margins": 0.29237666726112366, + "rewards/rejected": -0.41853827238082886, + "step": 1530 + }, + { + "epoch": 0.51, + "learning_rate": 2.8191041196514874e-06, + "logits/chosen": -1.658347725868225, + "logits/rejected": -0.909757137298584, + "logps/chosen": -614.3128662109375, + "logps/rejected": -1271.3431396484375, + "loss": 0.0714, + "rewards/accuracies": 0.8999999761581421, + "rewards/chosen": -0.14127101004123688, + "rewards/margins": 0.2985311448574066, + "rewards/rejected": -0.43980222940444946, + "step": 1540 + }, + { + "epoch": 0.52, + "learning_rate": 2.7902322853130758e-06, + "logits/chosen": -1.7248843908309937, + "logits/rejected": -1.4133408069610596, + "logps/chosen": -530.6253662109375, + "logps/rejected": -1091.8160400390625, + "loss": 0.0698, + "rewards/accuracies": 0.824999988079071, + "rewards/chosen": -0.12663719058036804, + "rewards/margins": 0.252753883600235, + "rewards/rejected": -0.3793911337852478, + "step": 1550 + }, + { + "epoch": 0.52, + "learning_rate": 2.761321158169134e-06, + "logits/chosen": -1.8168293237686157, + "logits/rejected": -1.3112378120422363, + "logps/chosen": -532.3922119140625, + "logps/rejected": -1184.728515625, + "loss": 0.065, + "rewards/accuracies": 0.824999988079071, + "rewards/chosen": -0.11639068275690079, + "rewards/margins": 0.30524441599845886, + "rewards/rejected": -0.42163509130477905, + "step": 1560 + }, + { + "epoch": 0.52, + "learning_rate": 2.7323746523234813e-06, + "logits/chosen": -1.6596038341522217, + "logits/rejected": -1.0099501609802246, + "logps/chosen": -620.3221435546875, + "logps/rejected": -1408.0771484375, + "loss": 0.035, + "rewards/accuracies": 0.949999988079071, + "rewards/chosen": -0.11354754865169525, + "rewards/margins": 0.3631085455417633, + "rewards/rejected": -0.47665613889694214, + "step": 1570 + }, + { + "epoch": 0.53, + "learning_rate": 2.703396686669646e-06, + "logits/chosen": -1.677464246749878, + "logits/rejected": -1.2843691110610962, + "logps/chosen": -601.0008544921875, + "logps/rejected": -1275.202880859375, + "loss": 0.0639, + "rewards/accuracies": 0.800000011920929, + "rewards/chosen": -0.11773763597011566, + "rewards/margins": 0.30820992588996887, + "rewards/rejected": -0.4259475767612457, + "step": 1580 + }, + { + "epoch": 0.53, + "learning_rate": 2.6743911843603134e-06, + "logits/chosen": -1.915787696838379, + "logits/rejected": -0.9031229019165039, + "logps/chosen": -596.1331176757812, + "logps/rejected": -1234.94482421875, + "loss": 0.0562, + "rewards/accuracies": 0.875, + "rewards/chosen": -0.11484452337026596, + "rewards/margins": 0.32129010558128357, + "rewards/rejected": -0.4361346364021301, + "step": 1590 + }, + { + "epoch": 0.53, + "learning_rate": 2.6453620722761897e-06, + "logits/chosen": -1.800848364830017, + "logits/rejected": -1.0119121074676514, + "logps/chosen": -777.0930786132812, + "logps/rejected": -1372.353271484375, + "loss": 0.0545, + "rewards/accuracies": 0.949999988079071, + "rewards/chosen": -0.13788454234600067, + "rewards/margins": 0.31275519728660583, + "rewards/rejected": -0.4506397843360901, + "step": 1600 + }, + { + "epoch": 0.54, + "learning_rate": 2.6163132804943675e-06, + "logits/chosen": -1.7606815099716187, + "logits/rejected": -1.1106458902359009, + "logps/chosen": -619.3434448242188, + "logps/rejected": -1193.679931640625, + "loss": 0.0785, + "rewards/accuracies": 0.824999988079071, + "rewards/chosen": -0.1508094221353531, + "rewards/margins": 0.26593416929244995, + "rewards/rejected": -0.41674357652664185, + "step": 1610 + }, + { + "epoch": 0.54, + "learning_rate": 2.587248741756253e-06, + "logits/chosen": -1.8017308712005615, + "logits/rejected": -1.0231897830963135, + "logps/chosen": -573.2968139648438, + "logps/rejected": -1028.872802734375, + "loss": 0.0863, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.14274278283119202, + "rewards/margins": 0.22178173065185547, + "rewards/rejected": -0.3645245134830475, + "step": 1620 + }, + { + "epoch": 0.54, + "learning_rate": 2.558172390935141e-06, + "logits/chosen": -1.8495060205459595, + "logits/rejected": -0.9550938606262207, + "logps/chosen": -701.3528442382812, + "logps/rejected": -1321.211181640625, + "loss": 0.0576, + "rewards/accuracies": 0.8999999761581421, + "rewards/chosen": -0.20430326461791992, + "rewards/margins": 0.292900025844574, + "rewards/rejected": -0.4972032606601715, + "step": 1630 + }, + { + "epoch": 0.55, + "learning_rate": 2.5290881645034932e-06, + "logits/chosen": -1.613519310951233, + "logits/rejected": -1.0299922227859497, + "logps/chosen": -682.4573974609375, + "logps/rejected": -1327.379150390625, + "loss": 0.0725, + "rewards/accuracies": 0.875, + "rewards/chosen": -0.2211301326751709, + "rewards/margins": 0.2834627032279968, + "rewards/rejected": -0.5045928359031677, + "step": 1640 + }, + { + "epoch": 0.55, + "learning_rate": 2.5e-06, + "logits/chosen": -1.8198554515838623, + "logits/rejected": -1.2243192195892334, + "logps/chosen": -673.7959594726562, + "logps/rejected": -1209.7906494140625, + "loss": 0.0879, + "rewards/accuracies": 0.8500000238418579, + "rewards/chosen": -0.15559658408164978, + "rewards/margins": 0.2655668258666992, + "rewards/rejected": -0.4211633801460266, + "step": 1650 + }, + { + "epoch": 0.55, + "learning_rate": 2.470911835496508e-06, + "logits/chosen": -1.6637722253799438, + "logits/rejected": -1.15647292137146, + "logps/chosen": -611.1282348632812, + "logps/rejected": -1223.1854248046875, + "loss": 0.0801, + "rewards/accuracies": 0.824999988079071, + "rewards/chosen": -0.1450941264629364, + "rewards/margins": 0.2773655652999878, + "rewards/rejected": -0.4224596917629242, + "step": 1660 + }, + { + "epoch": 0.56, + "learning_rate": 2.4418276090648596e-06, + "logits/chosen": -1.8464033603668213, + "logits/rejected": -1.052440881729126, + "logps/chosen": -644.9616088867188, + "logps/rejected": -1258.923095703125, + "loss": 0.0874, + "rewards/accuracies": 0.8500000238418579, + "rewards/chosen": -0.1209816113114357, + "rewards/margins": 0.2808608412742615, + "rewards/rejected": -0.40184250473976135, + "step": 1670 + }, + { + "epoch": 0.56, + "learning_rate": 2.4127512582437486e-06, + "logits/chosen": -1.685725212097168, + "logits/rejected": -1.0185719728469849, + "logps/chosen": -620.1656494140625, + "logps/rejected": -1276.2811279296875, + "loss": 0.0583, + "rewards/accuracies": 0.875, + "rewards/chosen": -0.14368347823619843, + "rewards/margins": 0.2920317053794861, + "rewards/rejected": -0.4357151985168457, + "step": 1680 + }, + { + "epoch": 0.56, + "learning_rate": 2.3836867195056333e-06, + "logits/chosen": -1.6978428363800049, + "logits/rejected": -1.209702730178833, + "logps/chosen": -633.7965698242188, + "logps/rejected": -1311.2330322265625, + "loss": 0.0559, + "rewards/accuracies": 0.875, + "rewards/chosen": -0.15191814303398132, + "rewards/margins": 0.2970329523086548, + "rewards/rejected": -0.4489511549472809, + "step": 1690 + }, + { + "epoch": 0.57, + "learning_rate": 2.3546379277238107e-06, + "logits/chosen": -1.4833507537841797, + "logits/rejected": -1.0119761228561401, + "logps/chosen": -680.2658081054688, + "logps/rejected": -1296.93212890625, + "loss": 0.0718, + "rewards/accuracies": 0.8500000238418579, + "rewards/chosen": -0.23375864326953888, + "rewards/margins": 0.2688060402870178, + "rewards/rejected": -0.5025646686553955, + "step": 1700 + }, + { + "epoch": 0.57, + "learning_rate": 2.325608815639687e-06, + "logits/chosen": -1.6255954504013062, + "logits/rejected": -0.9894803166389465, + "logps/chosen": -608.648681640625, + "logps/rejected": -1257.4925537109375, + "loss": 0.0806, + "rewards/accuracies": 0.8500000238418579, + "rewards/chosen": -0.19161316752433777, + "rewards/margins": 0.27543801069259644, + "rewards/rejected": -0.46705111861228943, + "step": 1710 + }, + { + "epoch": 0.57, + "learning_rate": 2.296603313330355e-06, + "logits/chosen": -1.3378230333328247, + "logits/rejected": -0.8268178105354309, + "logps/chosen": -624.6492309570312, + "logps/rejected": -1211.5384521484375, + "loss": 0.1075, + "rewards/accuracies": 0.824999988079071, + "rewards/chosen": -0.17298360168933868, + "rewards/margins": 0.25485867261886597, + "rewards/rejected": -0.42784222960472107, + "step": 1720 + }, + { + "epoch": 0.58, + "learning_rate": 2.26762534767652e-06, + "logits/chosen": -1.7436068058013916, + "logits/rejected": -1.0120248794555664, + "logps/chosen": -667.2508544921875, + "logps/rejected": -1316.541015625, + "loss": 0.0723, + "rewards/accuracies": 0.949999988079071, + "rewards/chosen": -0.18062946200370789, + "rewards/margins": 0.2949804961681366, + "rewards/rejected": -0.47561001777648926, + "step": 1730 + }, + { + "epoch": 0.58, + "learning_rate": 2.238678841830867e-06, + "logits/chosen": -1.743772268295288, + "logits/rejected": -0.9547621607780457, + "logps/chosen": -719.8886108398438, + "logps/rejected": -1386.740478515625, + "loss": 0.0533, + "rewards/accuracies": 0.925000011920929, + "rewards/chosen": -0.1676359474658966, + "rewards/margins": 0.31254929304122925, + "rewards/rejected": -0.48018521070480347, + "step": 1740 + }, + { + "epoch": 0.58, + "learning_rate": 2.2097677146869242e-06, + "logits/chosen": -1.7221221923828125, + "logits/rejected": -1.1789166927337646, + "logps/chosen": -515.1943359375, + "logps/rejected": -1109.064453125, + "loss": 0.0696, + "rewards/accuracies": 0.8500000238418579, + "rewards/chosen": -0.11262670904397964, + "rewards/margins": 0.2698899805545807, + "rewards/rejected": -0.38251665234565735, + "step": 1750 + }, + { + "epoch": 0.59, + "learning_rate": 2.1808958803485134e-06, + "logits/chosen": -1.7244141101837158, + "logits/rejected": -1.0081803798675537, + "logps/chosen": -617.8436889648438, + "logps/rejected": -1301.739990234375, + "loss": 0.0449, + "rewards/accuracies": 0.949999988079071, + "rewards/chosen": -0.13489536941051483, + "rewards/margins": 0.30581116676330566, + "rewards/rejected": -0.4407065510749817, + "step": 1760 + }, + { + "epoch": 0.59, + "learning_rate": 2.1520672475998374e-06, + "logits/chosen": -1.5150662660598755, + "logits/rejected": -1.2830677032470703, + "logps/chosen": -469.53497314453125, + "logps/rejected": -1185.6640625, + "loss": 0.1188, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.12466877698898315, + "rewards/margins": 0.25860312581062317, + "rewards/rejected": -0.3832719027996063, + "step": 1770 + }, + { + "epoch": 0.59, + "learning_rate": 2.1232857193762923e-06, + "logits/chosen": -1.6666500568389893, + "logits/rejected": -1.1478335857391357, + "logps/chosen": -630.8321533203125, + "logps/rejected": -1219.052978515625, + "loss": 0.0988, + "rewards/accuracies": 0.824999988079071, + "rewards/chosen": -0.14840959012508392, + "rewards/margins": 0.2695569396018982, + "rewards/rejected": -0.4179665148258209, + "step": 1780 + }, + { + "epoch": 0.6, + "learning_rate": 2.094555192236082e-06, + "logits/chosen": -1.6702663898468018, + "logits/rejected": -1.1331311464309692, + "logps/chosen": -639.830322265625, + "logps/rejected": -1440.0772705078125, + "loss": 0.0343, + "rewards/accuracies": 0.9750000238418579, + "rewards/chosen": -0.16856852173805237, + "rewards/margins": 0.34757930040359497, + "rewards/rejected": -0.5161478519439697, + "step": 1790 + }, + { + "epoch": 0.6, + "learning_rate": 2.0658795558326745e-06, + "logits/chosen": -1.3389908075332642, + "logits/rejected": -1.1332718133926392, + "logps/chosen": -560.32763671875, + "logps/rejected": -1140.04541015625, + "loss": 0.0962, + "rewards/accuracies": 0.7749999761581421, + "rewards/chosen": -0.18182942271232605, + "rewards/margins": 0.22485823929309845, + "rewards/rejected": -0.4066876769065857, + "step": 1800 + }, + { + "epoch": 0.6, + "learning_rate": 2.037262692388214e-06, + "logits/chosen": -1.5946425199508667, + "logits/rejected": -1.050821304321289, + "logps/chosen": -601.1361083984375, + "logps/rejected": -1288.375244140625, + "loss": 0.0656, + "rewards/accuracies": 0.8999999761581421, + "rewards/chosen": -0.13252010941505432, + "rewards/margins": 0.3152475953102112, + "rewards/rejected": -0.4477676749229431, + "step": 1810 + }, + { + "epoch": 0.61, + "learning_rate": 2.0087084761679245e-06, + "logits/chosen": -1.9240398406982422, + "logits/rejected": -1.0551602840423584, + "logps/chosen": -725.8641967773438, + "logps/rejected": -1185.0716552734375, + "loss": 0.0752, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.1920987367630005, + "rewards/margins": 0.24843771755695343, + "rewards/rejected": -0.4405364394187927, + "step": 1820 + }, + { + "epoch": 0.61, + "learning_rate": 1.9802207729556023e-06, + "logits/chosen": -1.827103853225708, + "logits/rejected": -1.2141703367233276, + "logps/chosen": -591.169677734375, + "logps/rejected": -1252.523193359375, + "loss": 0.0783, + "rewards/accuracies": 0.8500000238418579, + "rewards/chosen": -0.18946315348148346, + "rewards/margins": 0.2804568111896515, + "rewards/rejected": -0.46991997957229614, + "step": 1830 + }, + { + "epoch": 0.61, + "learning_rate": 1.9518034395302413e-06, + "logits/chosen": -1.4736969470977783, + "logits/rejected": -0.9016133546829224, + "logps/chosen": -606.1239013671875, + "logps/rejected": -1232.34912109375, + "loss": 0.0613, + "rewards/accuracies": 0.8500000238418579, + "rewards/chosen": -0.16971653699874878, + "rewards/margins": 0.3104923367500305, + "rewards/rejected": -0.4802088141441345, + "step": 1840 + }, + { + "epoch": 0.62, + "learning_rate": 1.9234603231439e-06, + "logits/chosen": -1.7311079502105713, + "logits/rejected": -1.0143775939941406, + "logps/chosen": -625.5157470703125, + "logps/rejected": -1218.10693359375, + "loss": 0.0795, + "rewards/accuracies": 0.824999988079071, + "rewards/chosen": -0.12972071766853333, + "rewards/margins": 0.30200275778770447, + "rewards/rejected": -0.4317234456539154, + "step": 1850 + }, + { + "epoch": 0.62, + "learning_rate": 1.895195261000831e-06, + "logits/chosen": -1.702005386352539, + "logits/rejected": -1.2821683883666992, + "logps/chosen": -580.2069091796875, + "logps/rejected": -1129.827880859375, + "loss": 0.092, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.12076668441295624, + "rewards/margins": 0.2600669264793396, + "rewards/rejected": -0.38083356618881226, + "step": 1860 + }, + { + "epoch": 0.62, + "learning_rate": 1.8670120797379958e-06, + "logits/chosen": -1.5771738290786743, + "logits/rejected": -1.2320917844772339, + "logps/chosen": -590.98046875, + "logps/rejected": -1251.115234375, + "loss": 0.1923, + "rewards/accuracies": 0.800000011920929, + "rewards/chosen": -0.14493119716644287, + "rewards/margins": 0.3147684633731842, + "rewards/rejected": -0.4596996307373047, + "step": 1870 + }, + { + "epoch": 0.63, + "learning_rate": 1.8389145949069953e-06, + "logits/chosen": -1.7696335315704346, + "logits/rejected": -1.096189260482788, + "logps/chosen": -578.0641479492188, + "logps/rejected": -1174.998291015625, + "loss": 0.0784, + "rewards/accuracies": 0.800000011920929, + "rewards/chosen": -0.13082124292850494, + "rewards/margins": 0.29184988141059875, + "rewards/rejected": -0.4226710796356201, + "step": 1880 + }, + { + "epoch": 0.63, + "learning_rate": 1.8109066104575023e-06, + "logits/chosen": -1.5320732593536377, + "logits/rejected": -0.7170606851577759, + "logps/chosen": -602.5262451171875, + "logps/rejected": -1216.552978515625, + "loss": 0.0593, + "rewards/accuracies": 0.875, + "rewards/chosen": -0.13688640296459198, + "rewards/margins": 0.3167448043823242, + "rewards/rejected": -0.4536312222480774, + "step": 1890 + }, + { + "epoch": 0.63, + "learning_rate": 1.7829919182222752e-06, + "logits/chosen": -1.6282037496566772, + "logits/rejected": -0.7399900555610657, + "logps/chosen": -665.1109008789062, + "logps/rejected": -1210.0506591796875, + "loss": 0.0832, + "rewards/accuracies": 0.824999988079071, + "rewards/chosen": -0.17592889070510864, + "rewards/margins": 0.27498775720596313, + "rewards/rejected": -0.4509166181087494, + "step": 1900 + }, + { + "epoch": 0.64, + "learning_rate": 1.755174297403795e-06, + "logits/chosen": -1.5313676595687866, + "logits/rejected": -1.0872125625610352, + "logps/chosen": -673.0841064453125, + "logps/rejected": -1284.4078369140625, + "loss": 0.0509, + "rewards/accuracies": 0.875, + "rewards/chosen": -0.22001402080059052, + "rewards/margins": 0.2925686240196228, + "rewards/rejected": -0.5125826597213745, + "step": 1910 + }, + { + "epoch": 0.64, + "learning_rate": 1.7274575140626318e-06, + "logits/chosen": -1.5627118349075317, + "logits/rejected": -0.9476549029350281, + "logps/chosen": -889.22509765625, + "logps/rejected": -1250.838134765625, + "loss": 0.1224, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.24946339428424835, + "rewards/margins": 0.240512415766716, + "rewards/rejected": -0.4899757504463196, + "step": 1920 + }, + { + "epoch": 0.64, + "learning_rate": 1.699845320607571e-06, + "logits/chosen": -1.3387973308563232, + "logits/rejected": -0.737518310546875, + "logps/chosen": -649.2557373046875, + "logps/rejected": -1256.31103515625, + "loss": 0.0787, + "rewards/accuracies": 0.8999999761581421, + "rewards/chosen": -0.2079259604215622, + "rewards/margins": 0.2849615216255188, + "rewards/rejected": -0.4928874373435974, + "step": 1930 + }, + { + "epoch": 0.65, + "learning_rate": 1.6723414552876052e-06, + "logits/chosen": -1.6167535781860352, + "logits/rejected": -1.154432773590088, + "logps/chosen": -748.0882568359375, + "logps/rejected": -1344.591064453125, + "loss": 0.0877, + "rewards/accuracies": 0.7749999761581421, + "rewards/chosen": -0.2506067454814911, + "rewards/margins": 0.25507742166519165, + "rewards/rejected": -0.5056841969490051, + "step": 1940 + }, + { + "epoch": 0.65, + "learning_rate": 1.6449496416858285e-06, + "logits/chosen": -1.4003807306289673, + "logits/rejected": -0.9338014721870422, + "logps/chosen": -809.3727416992188, + "logps/rejected": -1325.3411865234375, + "loss": 0.0916, + "rewards/accuracies": 0.8500000238418579, + "rewards/chosen": -0.31113409996032715, + "rewards/margins": 0.2267119586467743, + "rewards/rejected": -0.5378460884094238, + "step": 1950 + }, + { + "epoch": 0.65, + "learning_rate": 1.6176735882153284e-06, + "logits/chosen": -1.6452484130859375, + "logits/rejected": -0.8376502990722656, + "logps/chosen": -785.1629028320312, + "logps/rejected": -1453.1751708984375, + "loss": 0.0648, + "rewards/accuracies": 0.8500000238418579, + "rewards/chosen": -0.24854576587677002, + "rewards/margins": 0.31561970710754395, + "rewards/rejected": -0.564165472984314, + "step": 1960 + }, + { + "epoch": 0.66, + "learning_rate": 1.5905169876171223e-06, + "logits/chosen": -1.3292005062103271, + "logits/rejected": -1.1126399040222168, + "logps/chosen": -715.500244140625, + "logps/rejected": -1252.599365234375, + "loss": 0.0795, + "rewards/accuracies": 0.800000011920929, + "rewards/chosen": -0.2204395830631256, + "rewards/margins": 0.24764451384544373, + "rewards/rejected": -0.46808409690856934, + "step": 1970 + }, + { + "epoch": 0.66, + "learning_rate": 1.56348351646022e-06, + "logits/chosen": -1.7485458850860596, + "logits/rejected": -1.010534405708313, + "logps/chosen": -690.407958984375, + "logps/rejected": -1369.098876953125, + "loss": 0.068, + "rewards/accuracies": 0.875, + "rewards/chosen": -0.21612036228179932, + "rewards/margins": 0.3184736371040344, + "rewards/rejected": -0.534593939781189, + "step": 1980 + }, + { + "epoch": 0.66, + "learning_rate": 1.5365768346438798e-06, + "logits/chosen": -1.5913209915161133, + "logits/rejected": -0.9691961407661438, + "logps/chosen": -642.9549560546875, + "logps/rejected": -1401.8271484375, + "loss": 0.0482, + "rewards/accuracies": 0.8999999761581421, + "rewards/chosen": -0.17725911736488342, + "rewards/margins": 0.3471994996070862, + "rewards/rejected": -0.5244585871696472, + "step": 1990 + }, + { + "epoch": 0.67, + "learning_rate": 1.509800584902108e-06, + "logits/chosen": -1.489768147468567, + "logits/rejected": -0.9495828747749329, + "logps/chosen": -701.9700317382812, + "logps/rejected": -1363.816650390625, + "loss": 0.1071, + "rewards/accuracies": 0.875, + "rewards/chosen": -0.21182182431221008, + "rewards/margins": 0.28239232301712036, + "rewards/rejected": -0.4942142069339752, + "step": 2000 + }, + { + "epoch": 0.67, + "learning_rate": 1.4831583923105e-06, + "logits/chosen": -1.5869534015655518, + "logits/rejected": -0.7756304740905762, + "logps/chosen": -702.1505737304688, + "logps/rejected": -1240.6885986328125, + "loss": 0.0858, + "rewards/accuracies": 0.8500000238418579, + "rewards/chosen": -0.17973639070987701, + "rewards/margins": 0.26760077476501465, + "rewards/rejected": -0.44733715057373047, + "step": 2010 + }, + { + "epoch": 0.67, + "learning_rate": 1.4566538637954556e-06, + "logits/chosen": -1.632197618484497, + "logits/rejected": -1.0302485227584839, + "logps/chosen": -545.2433471679688, + "logps/rejected": -1228.961669921875, + "loss": 0.0726, + "rewards/accuracies": 0.800000011920929, + "rewards/chosen": -0.15146581828594208, + "rewards/margins": 0.30513542890548706, + "rewards/rejected": -0.45660123229026794, + "step": 2020 + }, + { + "epoch": 0.68, + "learning_rate": 1.430290587645865e-06, + "logits/chosen": -1.6993017196655273, + "logits/rejected": -1.0094877481460571, + "logps/chosen": -642.3917236328125, + "logps/rejected": -1230.881103515625, + "loss": 0.075, + "rewards/accuracies": 0.8999999761581421, + "rewards/chosen": -0.17171506583690643, + "rewards/margins": 0.2653944194316864, + "rewards/rejected": -0.43710947036743164, + "step": 2030 + }, + { + "epoch": 0.68, + "learning_rate": 1.4040721330273063e-06, + "logits/chosen": -1.5952852964401245, + "logits/rejected": -1.0270891189575195, + "logps/chosen": -582.1309814453125, + "logps/rejected": -1269.764892578125, + "loss": 0.0746, + "rewards/accuracies": 0.925000011920929, + "rewards/chosen": -0.1761622130870819, + "rewards/margins": 0.2915706932544708, + "rewards/rejected": -0.46773290634155273, + "step": 2040 + }, + { + "epoch": 0.68, + "learning_rate": 1.3780020494988447e-06, + "logits/chosen": -1.3955180644989014, + "logits/rejected": -0.8177375793457031, + "logps/chosen": -722.6862182617188, + "logps/rejected": -1415.1123046875, + "loss": 0.0664, + "rewards/accuracies": 0.875, + "rewards/chosen": -0.24629633128643036, + "rewards/margins": 0.30118346214294434, + "rewards/rejected": -0.5474798083305359, + "step": 2050 + }, + { + "epoch": 0.69, + "learning_rate": 1.3520838665324704e-06, + "logits/chosen": -1.5909086465835571, + "logits/rejected": -1.0436404943466187, + "logps/chosen": -642.248291015625, + "logps/rejected": -1155.293212890625, + "loss": 0.1028, + "rewards/accuracies": 0.800000011920929, + "rewards/chosen": -0.233381986618042, + "rewards/margins": 0.2285003662109375, + "rewards/rejected": -0.4618823528289795, + "step": 2060 + }, + { + "epoch": 0.69, + "learning_rate": 1.3263210930352737e-06, + "logits/chosen": -1.6532557010650635, + "logits/rejected": -1.1368399858474731, + "logps/chosen": -737.0447387695312, + "logps/rejected": -1343.158203125, + "loss": 0.076, + "rewards/accuracies": 0.8500000238418579, + "rewards/chosen": -0.2329857349395752, + "rewards/margins": 0.27646780014038086, + "rewards/rejected": -0.5094535946846008, + "step": 2070 + }, + { + "epoch": 0.69, + "learning_rate": 1.3007172168743854e-06, + "logits/chosen": -1.4727094173431396, + "logits/rejected": -0.934604287147522, + "logps/chosen": -640.7996826171875, + "logps/rejected": -1427.772216796875, + "loss": 0.0472, + "rewards/accuracies": 0.925000011920929, + "rewards/chosen": -0.19586428999900818, + "rewards/margins": 0.3154672384262085, + "rewards/rejected": -0.5113314986228943, + "step": 2080 + }, + { + "epoch": 0.7, + "learning_rate": 1.275275704404783e-06, + "logits/chosen": -1.6787872314453125, + "logits/rejected": -1.2128015756607056, + "logps/chosen": -691.6954345703125, + "logps/rejected": -1384.380859375, + "loss": 0.0654, + "rewards/accuracies": 0.824999988079071, + "rewards/chosen": -0.19925867021083832, + "rewards/margins": 0.3383576273918152, + "rewards/rejected": -0.5376163721084595, + "step": 2090 + }, + { + "epoch": 0.7, + "learning_rate": 1.2500000000000007e-06, + "logits/chosen": -1.4302475452423096, + "logits/rejected": -0.8775968551635742, + "logps/chosen": -651.3941040039062, + "logps/rejected": -1211.107177734375, + "loss": 0.0971, + "rewards/accuracies": 0.824999988079071, + "rewards/chosen": -0.20647192001342773, + "rewards/margins": 0.2682201862335205, + "rewards/rejected": -0.4746921956539154, + "step": 2100 + }, + { + "epoch": 0.7, + "learning_rate": 1.2248935255858118e-06, + "logits/chosen": -1.6478124856948853, + "logits/rejected": -0.8647529482841492, + "logps/chosen": -657.9767456054688, + "logps/rejected": -1445.98291015625, + "loss": 0.0464, + "rewards/accuracies": 0.925000011920929, + "rewards/chosen": -0.1874004602432251, + "rewards/margins": 0.3412622809410095, + "rewards/rejected": -0.5286628007888794, + "step": 2110 + }, + { + "epoch": 0.71, + "learning_rate": 1.1999596801769617e-06, + "logits/chosen": -1.3120667934417725, + "logits/rejected": -0.8835352063179016, + "logps/chosen": -623.9490356445312, + "logps/rejected": -1219.7760009765625, + "loss": 0.0814, + "rewards/accuracies": 0.7749999761581421, + "rewards/chosen": -0.20036788284778595, + "rewards/margins": 0.27707821130752563, + "rewards/rejected": -0.4774460792541504, + "step": 2120 + }, + { + "epoch": 0.71, + "learning_rate": 1.1752018394169882e-06, + "logits/chosen": -1.7081232070922852, + "logits/rejected": -0.9557887315750122, + "logps/chosen": -739.8932495117188, + "logps/rejected": -1354.7886962890625, + "loss": 0.063, + "rewards/accuracies": 0.8500000238418579, + "rewards/chosen": -0.1747320294380188, + "rewards/margins": 0.3161642551422119, + "rewards/rejected": -0.4908962845802307, + "step": 2130 + }, + { + "epoch": 0.71, + "learning_rate": 1.1506233551212186e-06, + "logits/chosen": -1.655374526977539, + "logits/rejected": -0.7405561208724976, + "logps/chosen": -661.6314697265625, + "logps/rejected": -1389.5576171875, + "loss": 0.0551, + "rewards/accuracies": 0.8999999761581421, + "rewards/chosen": -0.17144644260406494, + "rewards/margins": 0.3735453486442566, + "rewards/rejected": -0.5449918508529663, + "step": 2140 + }, + { + "epoch": 0.72, + "learning_rate": 1.1262275548229852e-06, + "logits/chosen": -1.257212519645691, + "logits/rejected": -0.9685500860214233, + "logps/chosen": -522.0648193359375, + "logps/rejected": -1224.8739013671875, + "loss": 0.0756, + "rewards/accuracies": 0.800000011920929, + "rewards/chosen": -0.16989552974700928, + "rewards/margins": 0.29921796917915344, + "rewards/rejected": -0.4691134989261627, + "step": 2150 + }, + { + "epoch": 0.72, + "learning_rate": 1.1020177413231334e-06, + "logits/chosen": -1.2228362560272217, + "logits/rejected": -0.7848671674728394, + "logps/chosen": -745.6207275390625, + "logps/rejected": -1389.057373046875, + "loss": 0.0749, + "rewards/accuracies": 0.875, + "rewards/chosen": -0.23939558863639832, + "rewards/margins": 0.28681597113609314, + "rewards/rejected": -0.5262116193771362, + "step": 2160 + }, + { + "epoch": 0.72, + "learning_rate": 1.077997192242871e-06, + "logits/chosen": -1.4660097360610962, + "logits/rejected": -0.974969208240509, + "logps/chosen": -637.7769165039062, + "logps/rejected": -1316.93017578125, + "loss": 0.065, + "rewards/accuracies": 0.925000011920929, + "rewards/chosen": -0.15726247429847717, + "rewards/margins": 0.31098154187202454, + "rewards/rejected": -0.4682440757751465, + "step": 2170 + }, + { + "epoch": 0.73, + "learning_rate": 1.0541691595800338e-06, + "logits/chosen": -1.7434594631195068, + "logits/rejected": -0.9831005930900574, + "logps/chosen": -688.5543212890625, + "logps/rejected": -1286.400634765625, + "loss": 0.0618, + "rewards/accuracies": 0.875, + "rewards/chosen": -0.19961665570735931, + "rewards/margins": 0.28607600927352905, + "rewards/rejected": -0.48569273948669434, + "step": 2180 + }, + { + "epoch": 0.73, + "learning_rate": 1.0305368692688175e-06, + "logits/chosen": -1.7225663661956787, + "logits/rejected": -1.016959309577942, + "logps/chosen": -655.6326904296875, + "logps/rejected": -1282.568115234375, + "loss": 0.0711, + "rewards/accuracies": 0.875, + "rewards/chosen": -0.19480985403060913, + "rewards/margins": 0.30676570534706116, + "rewards/rejected": -0.5015754699707031, + "step": 2190 + }, + { + "epoch": 0.73, + "learning_rate": 1.0071035207430352e-06, + "logits/chosen": -1.6399486064910889, + "logits/rejected": -1.1826612949371338, + "logps/chosen": -634.7337646484375, + "logps/rejected": -1298.8203125, + "loss": 0.0641, + "rewards/accuracies": 0.824999988079071, + "rewards/chosen": -0.1779237538576126, + "rewards/margins": 0.31172335147857666, + "rewards/rejected": -0.4896470904350281, + "step": 2200 + }, + { + "epoch": 0.74, + "learning_rate": 9.838722865029674e-07, + "logits/chosen": -1.6983495950698853, + "logits/rejected": -0.968538761138916, + "logps/chosen": -577.1983642578125, + "logps/rejected": -1137.4327392578125, + "loss": 0.0539, + "rewards/accuracies": 0.7749999761581421, + "rewards/chosen": -0.150478333234787, + "rewards/margins": 0.2877087891101837, + "rewards/rejected": -0.4381870627403259, + "step": 2210 + }, + { + "epoch": 0.74, + "learning_rate": 9.608463116858544e-07, + "logits/chosen": -1.341863989830017, + "logits/rejected": -0.9842678904533386, + "logps/chosen": -667.66015625, + "logps/rejected": -1293.984130859375, + "loss": 0.0759, + "rewards/accuracies": 0.824999988079071, + "rewards/chosen": -0.21305403113365173, + "rewards/margins": 0.2678312361240387, + "rewards/rejected": -0.48088526725769043, + "step": 2220 + }, + { + "epoch": 0.74, + "learning_rate": 9.380287136401001e-07, + "logits/chosen": -1.5788919925689697, + "logits/rejected": -0.9648571014404297, + "logps/chosen": -630.4158325195312, + "logps/rejected": -1360.427001953125, + "loss": 0.07, + "rewards/accuracies": 0.8500000238418579, + "rewards/chosen": -0.1882987916469574, + "rewards/margins": 0.3067074418067932, + "rewards/rejected": -0.495006263256073, + "step": 2230 + }, + { + "epoch": 0.75, + "learning_rate": 9.154225815032242e-07, + "logits/chosen": -1.4030269384384155, + "logits/rejected": -0.967923641204834, + "logps/chosen": -672.2672119140625, + "logps/rejected": -1179.1063232421875, + "loss": 0.0872, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.20492562651634216, + "rewards/margins": 0.24575969576835632, + "rewards/rejected": -0.4506853520870209, + "step": 2240 + }, + { + "epoch": 0.75, + "learning_rate": 8.930309757836517e-07, + "logits/chosen": -1.5156638622283936, + "logits/rejected": -1.0182697772979736, + "logps/chosen": -763.292724609375, + "logps/rejected": -1367.725341796875, + "loss": 0.094, + "rewards/accuracies": 0.824999988079071, + "rewards/chosen": -0.2593090534210205, + "rewards/margins": 0.27271029353141785, + "rewards/rejected": -0.532019317150116, + "step": 2250 + }, + { + "epoch": 0.75, + "learning_rate": 8.708569279463622e-07, + "logits/chosen": -1.521528959274292, + "logits/rejected": -0.7515259981155396, + "logps/chosen": -701.5262451171875, + "logps/rejected": -1299.10693359375, + "loss": 0.0501, + "rewards/accuracies": 0.949999988079071, + "rewards/chosen": -0.2057270109653473, + "rewards/margins": 0.3127078413963318, + "rewards/rejected": -0.5184348821640015, + "step": 2260 + }, + { + "epoch": 0.76, + "learning_rate": 8.489034400024812e-07, + "logits/chosen": -1.5848454236984253, + "logits/rejected": -1.0163071155548096, + "logps/chosen": -675.6258544921875, + "logps/rejected": -1327.443603515625, + "loss": 0.0724, + "rewards/accuracies": 0.824999988079071, + "rewards/chosen": -0.2220022976398468, + "rewards/margins": 0.297424852848053, + "rewards/rejected": -0.5194271802902222, + "step": 2270 + }, + { + "epoch": 0.76, + "learning_rate": 8.271734841028553e-07, + "logits/chosen": -1.6320521831512451, + "logits/rejected": -0.905319094657898, + "logps/chosen": -691.1963500976562, + "logps/rejected": -1361.6270751953125, + "loss": 0.0634, + "rewards/accuracies": 0.8500000238418579, + "rewards/chosen": -0.1820472925901413, + "rewards/margins": 0.337322473526001, + "rewards/rejected": -0.5193697810173035, + "step": 2280 + }, + { + "epoch": 0.76, + "learning_rate": 8.056700021356695e-07, + "logits/chosen": -1.7158523797988892, + "logits/rejected": -1.1264632940292358, + "logps/chosen": -621.7196044921875, + "logps/rejected": -1224.8692626953125, + "loss": 0.0612, + "rewards/accuracies": 0.875, + "rewards/chosen": -0.17123040556907654, + "rewards/margins": 0.28949856758117676, + "rewards/rejected": -0.4607289731502533, + "step": 2290 + }, + { + "epoch": 0.77, + "learning_rate": 7.843959053281663e-07, + "logits/chosen": -1.6252124309539795, + "logits/rejected": -1.1049168109893799, + "logps/chosen": -624.00390625, + "logps/rejected": -1279.696044921875, + "loss": 0.0827, + "rewards/accuracies": 0.824999988079071, + "rewards/chosen": -0.18267154693603516, + "rewards/margins": 0.31361207365989685, + "rewards/rejected": -0.496283620595932, + "step": 2300 + }, + { + "epoch": 0.77, + "learning_rate": 7.633540738525066e-07, + "logits/chosen": -1.5565481185913086, + "logits/rejected": -1.1449733972549438, + "logps/chosen": -699.10400390625, + "logps/rejected": -1397.216796875, + "loss": 0.0733, + "rewards/accuracies": 0.8500000238418579, + "rewards/chosen": -0.18508002161979675, + "rewards/margins": 0.33612412214279175, + "rewards/rejected": -0.5212041735649109, + "step": 2310 + }, + { + "epoch": 0.77, + "learning_rate": 7.425473564358457e-07, + "logits/chosen": -1.5639346837997437, + "logits/rejected": -1.0611830949783325, + "logps/chosen": -636.5604248046875, + "logps/rejected": -1440.516357421875, + "loss": 0.0714, + "rewards/accuracies": 0.824999988079071, + "rewards/chosen": -0.18250158429145813, + "rewards/margins": 0.3447728157043457, + "rewards/rejected": -0.5272743701934814, + "step": 2320 + }, + { + "epoch": 0.78, + "learning_rate": 7.219785699746573e-07, + "logits/chosen": -1.436949610710144, + "logits/rejected": -1.1259959936141968, + "logps/chosen": -717.715576171875, + "logps/rejected": -1378.193359375, + "loss": 0.0757, + "rewards/accuracies": 0.875, + "rewards/chosen": -0.22643479704856873, + "rewards/margins": 0.29645177721977234, + "rewards/rejected": -0.5228865742683411, + "step": 2330 + }, + { + "epoch": 0.78, + "learning_rate": 7.016504991533727e-07, + "logits/chosen": -1.5969784259796143, + "logits/rejected": -0.9250253438949585, + "logps/chosen": -714.7103881835938, + "logps/rejected": -1392.815185546875, + "loss": 0.0606, + "rewards/accuracies": 0.875, + "rewards/chosen": -0.20075106620788574, + "rewards/margins": 0.321119487285614, + "rewards/rejected": -0.5218705534934998, + "step": 2340 + }, + { + "epoch": 0.78, + "learning_rate": 6.815658960673782e-07, + "logits/chosen": -1.3801023960113525, + "logits/rejected": -0.8244778513908386, + "logps/chosen": -762.7855834960938, + "logps/rejected": -1422.857177734375, + "loss": 0.0722, + "rewards/accuracies": 0.875, + "rewards/chosen": -0.22916767001152039, + "rewards/margins": 0.2947383224964142, + "rewards/rejected": -0.5239059925079346, + "step": 2350 + }, + { + "epoch": 0.79, + "learning_rate": 6.617274798504286e-07, + "logits/chosen": -1.501314640045166, + "logits/rejected": -1.1214609146118164, + "logps/chosen": -670.3170776367188, + "logps/rejected": -1361.3216552734375, + "loss": 0.0577, + "rewards/accuracies": 0.8500000238418579, + "rewards/chosen": -0.1791990101337433, + "rewards/margins": 0.3246581256389618, + "rewards/rejected": -0.5038571357727051, + "step": 2360 + }, + { + "epoch": 0.79, + "learning_rate": 6.421379363065142e-07, + "logits/chosen": -1.3665945529937744, + "logits/rejected": -0.8138443231582642, + "logps/chosen": -703.3013916015625, + "logps/rejected": -1420.8404541015625, + "loss": 0.0468, + "rewards/accuracies": 0.925000011920929, + "rewards/chosen": -0.20666304230690002, + "rewards/margins": 0.3045525550842285, + "rewards/rejected": -0.5112155675888062, + "step": 2370 + }, + { + "epoch": 0.79, + "learning_rate": 6.227999175462521e-07, + "logits/chosen": -1.4125187397003174, + "logits/rejected": -1.0406558513641357, + "logps/chosen": -726.2548828125, + "logps/rejected": -1322.2655029296875, + "loss": 0.0807, + "rewards/accuracies": 0.875, + "rewards/chosen": -0.1995895951986313, + "rewards/margins": 0.27303341031074524, + "rewards/rejected": -0.47262296080589294, + "step": 2380 + }, + { + "epoch": 0.8, + "learning_rate": 6.037160416278279e-07, + "logits/chosen": -1.5233482122421265, + "logits/rejected": -0.8162555694580078, + "logps/chosen": -642.1959228515625, + "logps/rejected": -1301.4527587890625, + "loss": 0.0589, + "rewards/accuracies": 0.8500000238418579, + "rewards/chosen": -0.18815341591835022, + "rewards/margins": 0.29811039566993713, + "rewards/rejected": -0.48626384139060974, + "step": 2390 + }, + { + "epoch": 0.8, + "learning_rate": 5.848888922025553e-07, + "logits/chosen": -1.6081314086914062, + "logits/rejected": -1.1109412908554077, + "logps/chosen": -642.6141357421875, + "logps/rejected": -1297.754150390625, + "loss": 0.0596, + "rewards/accuracies": 0.8999999761581421, + "rewards/chosen": -0.16165341436862946, + "rewards/margins": 0.29734665155410767, + "rewards/rejected": -0.45900002121925354, + "step": 2400 + }, + { + "epoch": 0.8, + "learning_rate": 5.663210181650905e-07, + "logits/chosen": -1.4959220886230469, + "logits/rejected": -1.2034730911254883, + "logps/chosen": -575.3321533203125, + "logps/rejected": -1197.032470703125, + "loss": 0.0979, + "rewards/accuracies": 0.800000011920929, + "rewards/chosen": -0.14985838532447815, + "rewards/margins": 0.26243895292282104, + "rewards/rejected": -0.4122973084449768, + "step": 2410 + }, + { + "epoch": 0.81, + "learning_rate": 5.48014933308352e-07, + "logits/chosen": -1.6470867395401, + "logits/rejected": -1.3305919170379639, + "logps/chosen": -586.1883544921875, + "logps/rejected": -1270.7503662109375, + "loss": 0.0629, + "rewards/accuracies": 0.824999988079071, + "rewards/chosen": -0.17088322341442108, + "rewards/margins": 0.29427218437194824, + "rewards/rejected": -0.4651554226875305, + "step": 2420 + }, + { + "epoch": 0.81, + "learning_rate": 5.299731159831953e-07, + "logits/chosen": -1.416419506072998, + "logits/rejected": -0.9456356167793274, + "logps/chosen": -537.1207275390625, + "logps/rejected": -1264.95556640625, + "loss": 0.0765, + "rewards/accuracies": 0.824999988079071, + "rewards/chosen": -0.15687143802642822, + "rewards/margins": 0.3016758859157562, + "rewards/rejected": -0.45854735374450684, + "step": 2430 + }, + { + "epoch": 0.81, + "learning_rate": 5.121980087628802e-07, + "logits/chosen": -1.742026686668396, + "logits/rejected": -1.1850230693817139, + "logps/chosen": -548.540283203125, + "logps/rejected": -1200.302734375, + "loss": 0.0542, + "rewards/accuracies": 0.824999988079071, + "rewards/chosen": -0.15248605608940125, + "rewards/margins": 0.29664915800094604, + "rewards/rejected": -0.4491352438926697, + "step": 2440 + }, + { + "epoch": 0.82, + "learning_rate": 4.946920181123904e-07, + "logits/chosen": -1.5911678075790405, + "logits/rejected": -0.8707629442214966, + "logps/chosen": -719.4567260742188, + "logps/rejected": -1296.5850830078125, + "loss": 0.062, + "rewards/accuracies": 0.925000011920929, + "rewards/chosen": -0.19431941211223602, + "rewards/margins": 0.2867654860019684, + "rewards/rejected": -0.4810849130153656, + "step": 2450 + }, + { + "epoch": 0.82, + "learning_rate": 4.774575140626317e-07, + "logits/chosen": -1.5443083047866821, + "logits/rejected": -1.0547590255737305, + "logps/chosen": -627.6547241210938, + "logps/rejected": -1386.137451171875, + "loss": 0.0469, + "rewards/accuracies": 0.949999988079071, + "rewards/chosen": -0.19703541696071625, + "rewards/margins": 0.36265698075294495, + "rewards/rejected": -0.5596923828125, + "step": 2460 + }, + { + "epoch": 0.82, + "learning_rate": 4.6049682988957037e-07, + "logits/chosen": -1.7288634777069092, + "logits/rejected": -0.8852437138557434, + "logps/chosen": -762.8133544921875, + "logps/rejected": -1455.493896484375, + "loss": 0.0512, + "rewards/accuracies": 0.8999999761581421, + "rewards/chosen": -0.1930183470249176, + "rewards/margins": 0.3427968919277191, + "rewards/rejected": -0.5358152389526367, + "step": 2470 + }, + { + "epoch": 0.83, + "learning_rate": 4.438122617983442e-07, + "logits/chosen": -1.6805801391601562, + "logits/rejected": -1.0256656408309937, + "logps/chosen": -660.5001220703125, + "logps/rejected": -1275.15625, + "loss": 0.0772, + "rewards/accuracies": 0.800000011920929, + "rewards/chosen": -0.208724707365036, + "rewards/margins": 0.28110271692276, + "rewards/rejected": -0.48982739448547363, + "step": 2480 + }, + { + "epoch": 0.83, + "learning_rate": 4.27406068612396e-07, + "logits/chosen": -1.6279792785644531, + "logits/rejected": -0.9308198094367981, + "logps/chosen": -610.5615844726562, + "logps/rejected": -1186.8466796875, + "loss": 0.0817, + "rewards/accuracies": 0.7749999761581421, + "rewards/chosen": -0.15405488014221191, + "rewards/margins": 0.27832460403442383, + "rewards/rejected": -0.43237948417663574, + "step": 2490 + }, + { + "epoch": 0.83, + "learning_rate": 4.1128047146765936e-07, + "logits/chosen": -1.5780996084213257, + "logits/rejected": -0.9356195330619812, + "logps/chosen": -742.1123046875, + "logps/rejected": -1309.2537841796875, + "loss": 0.0887, + "rewards/accuracies": 0.8500000238418579, + "rewards/chosen": -0.21093162894248962, + "rewards/margins": 0.29827579855918884, + "rewards/rejected": -0.5092074275016785, + "step": 2500 + }, + { + "epoch": 0.84, + "learning_rate": 3.954376535118562e-07, + "logits/chosen": -1.451724648475647, + "logits/rejected": -1.0602259635925293, + "logps/chosen": -743.4791870117188, + "logps/rejected": -1402.3712158203125, + "loss": 0.058, + "rewards/accuracies": 0.8999999761581421, + "rewards/chosen": -0.21480798721313477, + "rewards/margins": 0.3146688640117645, + "rewards/rejected": -0.5294768214225769, + "step": 2510 + }, + { + "epoch": 0.84, + "learning_rate": 3.798797596089351e-07, + "logits/chosen": -1.3941175937652588, + "logits/rejected": -0.991183876991272, + "logps/chosen": -676.042236328125, + "logps/rejected": -1340.735595703125, + "loss": 0.0826, + "rewards/accuracies": 0.8999999761581421, + "rewards/chosen": -0.23510794341564178, + "rewards/margins": 0.2676921486854553, + "rewards/rejected": -0.5028000473976135, + "step": 2520 + }, + { + "epoch": 0.84, + "learning_rate": 3.646088960486863e-07, + "logits/chosen": -1.6253753900527954, + "logits/rejected": -1.1532971858978271, + "logps/chosen": -519.7340087890625, + "logps/rejected": -1152.824462890625, + "loss": 0.0921, + "rewards/accuracies": 0.800000011920929, + "rewards/chosen": -0.14441636204719543, + "rewards/margins": 0.2724858820438385, + "rewards/rejected": -0.41690224409103394, + "step": 2530 + }, + { + "epoch": 0.85, + "learning_rate": 3.4962713026158697e-07, + "logits/chosen": -1.607985258102417, + "logits/rejected": -1.2400459051132202, + "logps/chosen": -500.02020263671875, + "logps/rejected": -1161.1961669921875, + "loss": 0.0885, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.1489984691143036, + "rewards/margins": 0.2574361562728882, + "rewards/rejected": -0.4064346253871918, + "step": 2540 + }, + { + "epoch": 0.85, + "learning_rate": 3.3493649053890325e-07, + "logits/chosen": -1.5089218616485596, + "logits/rejected": -0.8405025601387024, + "logps/chosen": -708.8694458007812, + "logps/rejected": -1470.578857421875, + "loss": 0.0492, + "rewards/accuracies": 0.9750000238418579, + "rewards/chosen": -0.21505184471607208, + "rewards/margins": 0.34505897760391235, + "rewards/rejected": -0.5601108074188232, + "step": 2550 + }, + { + "epoch": 0.85, + "learning_rate": 3.2053896575809426e-07, + "logits/chosen": -1.639923334121704, + "logits/rejected": -1.043080449104309, + "logps/chosen": -783.112060546875, + "logps/rejected": -1343.9681396484375, + "loss": 0.0747, + "rewards/accuracies": 0.824999988079071, + "rewards/chosen": -0.24453020095825195, + "rewards/margins": 0.2654086947441101, + "rewards/rejected": -0.5099388957023621, + "step": 2560 + }, + { + "epoch": 0.86, + "learning_rate": 3.0643650511354483e-07, + "logits/chosen": -1.6855356693267822, + "logits/rejected": -0.8734441995620728, + "logps/chosen": -713.609130859375, + "logps/rejected": -1338.822998046875, + "loss": 0.0699, + "rewards/accuracies": 0.875, + "rewards/chosen": -0.2088811695575714, + "rewards/margins": 0.30321091413497925, + "rewards/rejected": -0.512092113494873, + "step": 2570 + }, + { + "epoch": 0.86, + "learning_rate": 2.9263101785268253e-07, + "logits/chosen": -1.40034818649292, + "logits/rejected": -1.074161171913147, + "logps/chosen": -614.1613159179688, + "logps/rejected": -1244.793212890625, + "loss": 0.0945, + "rewards/accuracies": 0.824999988079071, + "rewards/chosen": -0.20661716163158417, + "rewards/margins": 0.28118330240249634, + "rewards/rejected": -0.4878004193305969, + "step": 2580 + }, + { + "epoch": 0.86, + "learning_rate": 2.791243730174903e-07, + "logits/chosen": -1.6715885400772095, + "logits/rejected": -1.1529837846755981, + "logps/chosen": -547.1351928710938, + "logps/rejected": -1222.099365234375, + "loss": 0.0664, + "rewards/accuracies": 0.8500000238418579, + "rewards/chosen": -0.12953749299049377, + "rewards/margins": 0.31725236773490906, + "rewards/rejected": -0.44678983092308044, + "step": 2590 + }, + { + "epoch": 0.87, + "learning_rate": 2.6591839919146963e-07, + "logits/chosen": -1.6793832778930664, + "logits/rejected": -0.8579391241073608, + "logps/chosen": -668.2606201171875, + "logps/rejected": -1342.045166015625, + "loss": 0.0454, + "rewards/accuracies": 0.949999988079071, + "rewards/chosen": -0.16217480599880219, + "rewards/margins": 0.35208019614219666, + "rewards/rejected": -0.5142549276351929, + "step": 2600 + }, + { + "epoch": 0.87, + "learning_rate": 2.53014884252083e-07, + "logits/chosen": -1.469300627708435, + "logits/rejected": -0.767549455165863, + "logps/chosen": -720.5267333984375, + "logps/rejected": -1461.4560546875, + "loss": 0.0383, + "rewards/accuracies": 0.925000011920929, + "rewards/chosen": -0.19582341611385345, + "rewards/margins": 0.36342793703079224, + "rewards/rejected": -0.5592513680458069, + "step": 2610 + }, + { + "epoch": 0.87, + "learning_rate": 2.404155751286988e-07, + "logits/chosen": -1.249018907546997, + "logits/rejected": -0.6379393339157104, + "logps/chosen": -634.1907958984375, + "logps/rejected": -1247.8612060546875, + "loss": 0.0579, + "rewards/accuracies": 0.824999988079071, + "rewards/chosen": -0.18129587173461914, + "rewards/margins": 0.29834333062171936, + "rewards/rejected": -0.4796391427516937, + "step": 2620 + }, + { + "epoch": 0.88, + "learning_rate": 2.281221775660894e-07, + "logits/chosen": -1.5495988130569458, + "logits/rejected": -0.8539125323295593, + "logps/chosen": -634.5662231445312, + "logps/rejected": -1174.760498046875, + "loss": 0.0756, + "rewards/accuracies": 0.875, + "rewards/chosen": -0.1546512097120285, + "rewards/margins": 0.26017943024635315, + "rewards/rejected": -0.41483062505722046, + "step": 2630 + }, + { + "epoch": 0.88, + "learning_rate": 2.1613635589349756e-07, + "logits/chosen": -1.432281494140625, + "logits/rejected": -0.9715501070022583, + "logps/chosen": -674.9788818359375, + "logps/rejected": -1332.5562744140625, + "loss": 0.0969, + "rewards/accuracies": 0.875, + "rewards/chosen": -0.2190193384885788, + "rewards/margins": 0.2877424359321594, + "rewards/rejected": -0.5067617893218994, + "step": 2640 + }, + { + "epoch": 0.88, + "learning_rate": 2.044597327993153e-07, + "logits/chosen": -1.5616180896759033, + "logits/rejected": -1.1563438177108765, + "logps/chosen": -606.47509765625, + "logps/rejected": -1302.349365234375, + "loss": 0.0886, + "rewards/accuracies": 0.8500000238418579, + "rewards/chosen": -0.17486010491847992, + "rewards/margins": 0.30272260308265686, + "rewards/rejected": -0.4775826930999756, + "step": 2650 + }, + { + "epoch": 0.89, + "learning_rate": 1.9309388911139427e-07, + "logits/chosen": -1.6554429531097412, + "logits/rejected": -0.9854539036750793, + "logps/chosen": -705.8251342773438, + "logps/rejected": -1167.0521240234375, + "loss": 0.0627, + "rewards/accuracies": 0.875, + "rewards/chosen": -0.1600712388753891, + "rewards/margins": 0.2684277892112732, + "rewards/rejected": -0.4284990429878235, + "step": 2660 + }, + { + "epoch": 0.89, + "learning_rate": 1.8204036358303173e-07, + "logits/chosen": -1.5156413316726685, + "logits/rejected": -0.9938446879386902, + "logps/chosen": -598.7017822265625, + "logps/rejected": -1183.290771484375, + "loss": 0.0615, + "rewards/accuracies": 0.8500000238418579, + "rewards/chosen": -0.16524121165275574, + "rewards/margins": 0.28775864839553833, + "rewards/rejected": -0.45299988985061646, + "step": 2670 + }, + { + "epoch": 0.89, + "learning_rate": 1.713006526846439e-07, + "logits/chosen": -1.6532678604125977, + "logits/rejected": -0.9341715574264526, + "logps/chosen": -686.2755126953125, + "logps/rejected": -1177.0118408203125, + "loss": 0.1065, + "rewards/accuracies": 0.800000011920929, + "rewards/chosen": -0.18550768494606018, + "rewards/margins": 0.24031631648540497, + "rewards/rejected": -0.42582401633262634, + "step": 2680 + }, + { + "epoch": 0.9, + "learning_rate": 1.6087621040117157e-07, + "logits/chosen": -1.7682663202285767, + "logits/rejected": -0.9737985730171204, + "logps/chosen": -617.552001953125, + "logps/rejected": -1250.820068359375, + "loss": 0.0657, + "rewards/accuracies": 0.875, + "rewards/chosen": -0.16673611104488373, + "rewards/margins": 0.3138349950313568, + "rewards/rejected": -0.48057112097740173, + "step": 2690 + }, + { + "epoch": 0.9, + "learning_rate": 1.507684480352292e-07, + "logits/chosen": -1.4090583324432373, + "logits/rejected": -0.8811914324760437, + "logps/chosen": -614.2474975585938, + "logps/rejected": -1185.4610595703125, + "loss": 0.0782, + "rewards/accuracies": 0.824999988079071, + "rewards/chosen": -0.1610531508922577, + "rewards/margins": 0.2807305157184601, + "rewards/rejected": -0.4417836666107178, + "step": 2700 + }, + { + "epoch": 0.9, + "learning_rate": 1.4097873401604124e-07, + "logits/chosen": -1.5380343198776245, + "logits/rejected": -1.1294854879379272, + "logps/chosen": -611.0988159179688, + "logps/rejected": -1329.0465087890625, + "loss": 0.0768, + "rewards/accuracies": 0.8500000238418579, + "rewards/chosen": -0.18254908919334412, + "rewards/margins": 0.304229736328125, + "rewards/rejected": -0.4867788255214691, + "step": 2710 + }, + { + "epoch": 0.91, + "learning_rate": 1.31508393714177e-07, + "logits/chosen": -1.3804000616073608, + "logits/rejected": -0.9093223810195923, + "logps/chosen": -625.8529052734375, + "logps/rejected": -1367.1817626953125, + "loss": 0.0544, + "rewards/accuracies": 0.925000011920929, + "rewards/chosen": -0.1812657117843628, + "rewards/margins": 0.3490126132965088, + "rewards/rejected": -0.5302783250808716, + "step": 2720 + }, + { + "epoch": 0.91, + "learning_rate": 1.223587092621162e-07, + "logits/chosen": -1.5867640972137451, + "logits/rejected": -0.9399985074996948, + "logps/chosen": -676.7730102539062, + "logps/rejected": -1228.8385009765625, + "loss": 0.0731, + "rewards/accuracies": 0.8500000238418579, + "rewards/chosen": -0.17374689877033234, + "rewards/margins": 0.2911613881587982, + "rewards/rejected": -0.46490830183029175, + "step": 2730 + }, + { + "epoch": 0.91, + "learning_rate": 1.1353091938067024e-07, + "logits/chosen": -1.3089746236801147, + "logits/rejected": -1.0505130290985107, + "logps/chosen": -693.4176025390625, + "logps/rejected": -1227.46142578125, + "loss": 0.108, + "rewards/accuracies": 0.7749999761581421, + "rewards/chosen": -0.20728778839111328, + "rewards/margins": 0.26565903425216675, + "rewards/rejected": -0.47294682264328003, + "step": 2740 + }, + { + "epoch": 0.92, + "learning_rate": 1.0502621921127776e-07, + "logits/chosen": -1.6415075063705444, + "logits/rejected": -1.284096360206604, + "logps/chosen": -757.7032470703125, + "logps/rejected": -1357.983154296875, + "loss": 0.0815, + "rewards/accuracies": 0.8999999761581421, + "rewards/chosen": -0.2325299233198166, + "rewards/margins": 0.27662572264671326, + "rewards/rejected": -0.5091556906700134, + "step": 2750 + }, + { + "epoch": 0.92, + "learning_rate": 9.684576015420277e-08, + "logits/chosen": -1.3691219091415405, + "logits/rejected": -1.0339184999465942, + "logps/chosen": -582.3773193359375, + "logps/rejected": -1222.3829345703125, + "loss": 0.0626, + "rewards/accuracies": 0.824999988079071, + "rewards/chosen": -0.17146027088165283, + "rewards/margins": 0.299213707447052, + "rewards/rejected": -0.47067388892173767, + "step": 2760 + }, + { + "epoch": 0.92, + "learning_rate": 8.899064971265275e-08, + "logits/chosen": -1.581775188446045, + "logits/rejected": -1.0051343441009521, + "logps/chosen": -718.9093017578125, + "logps/rejected": -1241.8062744140625, + "loss": 0.0669, + "rewards/accuracies": 0.824999988079071, + "rewards/chosen": -0.1594964563846588, + "rewards/margins": 0.28466928005218506, + "rewards/rejected": -0.44416576623916626, + "step": 2770 + }, + { + "epoch": 0.93, + "learning_rate": 8.146195134284052e-08, + "logits/chosen": -1.4833767414093018, + "logits/rejected": -1.0147238969802856, + "logps/chosen": -611.6956176757812, + "logps/rejected": -1192.66015625, + "loss": 0.0914, + "rewards/accuracies": 0.7749999761581421, + "rewards/chosen": -0.20624539256095886, + "rewards/margins": 0.2417052984237671, + "rewards/rejected": -0.4479507505893707, + "step": 2780 + }, + { + "epoch": 0.93, + "learning_rate": 7.426068431000883e-08, + "logits/chosen": -1.4513158798217773, + "logits/rejected": -0.7021422982215881, + "logps/chosen": -749.6524658203125, + "logps/rejected": -1255.1658935546875, + "loss": 0.0931, + "rewards/accuracies": 0.8999999761581421, + "rewards/chosen": -0.2141745388507843, + "rewards/margins": 0.2607952654361725, + "rewards/rejected": -0.4749697744846344, + "step": 2790 + }, + { + "epoch": 0.93, + "learning_rate": 6.738782355044048e-08, + "logits/chosen": -1.3975991010665894, + "logits/rejected": -0.9725745320320129, + "logps/chosen": -653.4349365234375, + "logps/rejected": -1304.32666015625, + "loss": 0.0911, + "rewards/accuracies": 0.8500000238418579, + "rewards/chosen": -0.2119159698486328, + "rewards/margins": 0.2688130736351013, + "rewards/rejected": -0.48072901368141174, + "step": 2800 + }, + { + "epoch": 0.94, + "learning_rate": 6.084429953946452e-08, + "logits/chosen": -1.5564731359481812, + "logits/rejected": -1.0867516994476318, + "logps/chosen": -614.5533447265625, + "logps/rejected": -1332.2275390625, + "loss": 0.0512, + "rewards/accuracies": 0.925000011920929, + "rewards/chosen": -0.18066832423210144, + "rewards/margins": 0.32241129875183105, + "rewards/rejected": -0.5030795335769653, + "step": 2810 + }, + { + "epoch": 0.94, + "learning_rate": 5.463099816548578e-08, + "logits/chosen": -1.6008579730987549, + "logits/rejected": -1.274701476097107, + "logps/chosen": -693.5843505859375, + "logps/rejected": -1352.4996337890625, + "loss": 0.0756, + "rewards/accuracies": 0.824999988079071, + "rewards/chosen": -0.21343615651130676, + "rewards/margins": 0.2871994972229004, + "rewards/rejected": -0.5006356835365295, + "step": 2820 + }, + { + "epoch": 0.94, + "learning_rate": 4.8748760610051735e-08, + "logits/chosen": -1.5801304578781128, + "logits/rejected": -1.0879913568496704, + "logps/chosen": -632.2335205078125, + "logps/rejected": -1163.8685302734375, + "loss": 0.1023, + "rewards/accuracies": 0.7749999761581421, + "rewards/chosen": -0.16370609402656555, + "rewards/margins": 0.23538489639759064, + "rewards/rejected": -0.3990909457206726, + "step": 2830 + }, + { + "epoch": 0.95, + "learning_rate": 4.319838323396691e-08, + "logits/chosen": -1.6419999599456787, + "logits/rejected": -1.148740530014038, + "logps/chosen": -622.5277099609375, + "logps/rejected": -1213.3511962890625, + "loss": 0.0825, + "rewards/accuracies": 0.7749999761581421, + "rewards/chosen": -0.18520578742027283, + "rewards/margins": 0.2649041712284088, + "rewards/rejected": -0.45010989904403687, + "step": 2840 + }, + { + "epoch": 0.95, + "learning_rate": 3.798061746947995e-08, + "logits/chosen": -1.5798577070236206, + "logits/rejected": -1.0320630073547363, + "logps/chosen": -704.6697998046875, + "logps/rejected": -1352.01806640625, + "loss": 0.0633, + "rewards/accuracies": 0.8999999761581421, + "rewards/chosen": -0.21166665852069855, + "rewards/margins": 0.2847224175930023, + "rewards/rejected": -0.49638909101486206, + "step": 2850 + }, + { + "epoch": 0.95, + "learning_rate": 3.309616971855195e-08, + "logits/chosen": -1.5923948287963867, + "logits/rejected": -0.976751446723938, + "logps/chosen": -629.5615844726562, + "logps/rejected": -1377.3729248046875, + "loss": 0.0713, + "rewards/accuracies": 0.8500000238418579, + "rewards/chosen": -0.16095896065235138, + "rewards/margins": 0.3290019631385803, + "rewards/rejected": -0.4899609684944153, + "step": 2860 + }, + { + "epoch": 0.96, + "learning_rate": 2.8545701257221003e-08, + "logits/chosen": -1.5409938097000122, + "logits/rejected": -1.0359934568405151, + "logps/chosen": -694.2177124023438, + "logps/rejected": -1343.0057373046875, + "loss": 0.0705, + "rewards/accuracies": 0.875, + "rewards/chosen": -0.20338015258312225, + "rewards/margins": 0.2844991981983185, + "rewards/rejected": -0.48787933588027954, + "step": 2870 + }, + { + "epoch": 0.96, + "learning_rate": 2.4329828146074096e-08, + "logits/chosen": -1.6312742233276367, + "logits/rejected": -1.199237585067749, + "logps/chosen": -547.9122924804688, + "logps/rejected": -1200.981689453125, + "loss": 0.0883, + "rewards/accuracies": 0.824999988079071, + "rewards/chosen": -0.14529578387737274, + "rewards/margins": 0.2816771864891052, + "rewards/rejected": -0.42697301506996155, + "step": 2880 + }, + { + "epoch": 0.96, + "learning_rate": 2.0449121146845773e-08, + "logits/chosen": -1.4686208963394165, + "logits/rejected": -0.851632297039032, + "logps/chosen": -645.3411254882812, + "logps/rejected": -1267.0413818359375, + "loss": 0.0672, + "rewards/accuracies": 0.875, + "rewards/chosen": -0.20314595103263855, + "rewards/margins": 0.28285151720046997, + "rewards/rejected": -0.4859974980354309, + "step": 2890 + }, + { + "epoch": 0.97, + "learning_rate": 1.6904105645142443e-08, + "logits/chosen": -1.631974220275879, + "logits/rejected": -1.05061674118042, + "logps/chosen": -530.013671875, + "logps/rejected": -1199.45068359375, + "loss": 0.0715, + "rewards/accuracies": 0.800000011920929, + "rewards/chosen": -0.15192578732967377, + "rewards/margins": 0.31625282764434814, + "rewards/rejected": -0.4681786596775055, + "step": 2900 + }, + { + "epoch": 0.97, + "learning_rate": 1.3695261579316776e-08, + "logits/chosen": -1.4659382104873657, + "logits/rejected": -0.9017642736434937, + "logps/chosen": -520.8211669921875, + "logps/rejected": -1325.6385498046875, + "loss": 0.056, + "rewards/accuracies": 0.8999999761581421, + "rewards/chosen": -0.15825147926807404, + "rewards/margins": 0.35485899448394775, + "rewards/rejected": -0.5131104588508606, + "step": 2910 + }, + { + "epoch": 0.97, + "learning_rate": 1.0823023375489128e-08, + "logits/chosen": -1.5723521709442139, + "logits/rejected": -1.1977678537368774, + "logps/chosen": -600.8292236328125, + "logps/rejected": -1365.1005859375, + "loss": 0.0466, + "rewards/accuracies": 0.875, + "rewards/chosen": -0.16827178001403809, + "rewards/margins": 0.3262236714363098, + "rewards/rejected": -0.4944954812526703, + "step": 2920 + }, + { + "epoch": 0.98, + "learning_rate": 8.28777988873486e-09, + "logits/chosen": -1.6231420040130615, + "logits/rejected": -1.0731710195541382, + "logps/chosen": -671.5623779296875, + "logps/rejected": -1358.5367431640625, + "loss": 0.0688, + "rewards/accuracies": 0.8999999761581421, + "rewards/chosen": -0.21294662356376648, + "rewards/margins": 0.3046295642852783, + "rewards/rejected": -0.5175761580467224, + "step": 2930 + }, + { + "epoch": 0.98, + "learning_rate": 6.089874350439507e-09, + "logits/chosen": -1.4259278774261475, + "logits/rejected": -0.8798543810844421, + "logps/chosen": -694.62548828125, + "logps/rejected": -1346.7584228515625, + "loss": 0.0846, + "rewards/accuracies": 0.8500000238418579, + "rewards/chosen": -0.21204762160778046, + "rewards/margins": 0.27931392192840576, + "rewards/rejected": -0.4913616180419922, + "step": 2940 + }, + { + "epoch": 0.98, + "learning_rate": 4.229604321829561e-09, + "logits/chosen": -1.596842885017395, + "logits/rejected": -1.0910255908966064, + "logps/chosen": -653.8787231445312, + "logps/rejected": -1310.4539794921875, + "loss": 0.0672, + "rewards/accuracies": 0.875, + "rewards/chosen": -0.17626503109931946, + "rewards/margins": 0.3032272756099701, + "rewards/rejected": -0.47949227690696716, + "step": 2950 + }, + { + "epoch": 0.99, + "learning_rate": 2.7072216536885855e-09, + "logits/chosen": -1.3221776485443115, + "logits/rejected": -0.8105962872505188, + "logps/chosen": -586.7859497070312, + "logps/rejected": -1137.4781494140625, + "loss": 0.099, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.19391945004463196, + "rewards/margins": 0.2519974112510681, + "rewards/rejected": -0.4459168314933777, + "step": 2960 + }, + { + "epoch": 0.99, + "learning_rate": 1.5229324522605949e-09, + "logits/chosen": -1.553269624710083, + "logits/rejected": -1.0229682922363281, + "logps/chosen": -545.6754760742188, + "logps/rejected": -1225.473388671875, + "loss": 0.0631, + "rewards/accuracies": 0.800000011920929, + "rewards/chosen": -0.16567066311836243, + "rewards/margins": 0.2939736247062683, + "rewards/rejected": -0.45964425802230835, + "step": 2970 + }, + { + "epoch": 0.99, + "learning_rate": 6.768970513457151e-10, + "logits/chosen": -1.6384735107421875, + "logits/rejected": -0.9449461102485657, + "logps/chosen": -580.2606811523438, + "logps/rejected": -1307.0478515625, + "loss": 0.0692, + "rewards/accuracies": 0.875, + "rewards/chosen": -0.16747145354747772, + "rewards/margins": 0.31812483072280884, + "rewards/rejected": -0.48559626936912537, + "step": 2980 + }, + { + "epoch": 1.0, + "learning_rate": 1.692299905944883e-10, + "logits/chosen": -1.3800950050354004, + "logits/rejected": -0.8507388234138489, + "logps/chosen": -683.5490112304688, + "logps/rejected": -1347.037841796875, + "loss": 0.052, + "rewards/accuracies": 0.875, + "rewards/chosen": -0.19447694718837738, + "rewards/margins": 0.3061479330062866, + "rewards/rejected": -0.5006248950958252, + "step": 2990 + }, + { + "epoch": 1.0, + "learning_rate": 0.0, + "logits/chosen": -1.6093953847885132, + "logits/rejected": -0.8852132558822632, + "logps/chosen": -607.3538208007812, + "logps/rejected": -1118.9757080078125, + "loss": 0.1084, + "rewards/accuracies": 0.7749999761581421, + "rewards/chosen": -0.13724899291992188, + "rewards/margins": 0.2455356866121292, + "rewards/rejected": -0.3827846646308899, + "step": 3000 + }, + { + "epoch": 1.0, + "step": 3000, + "total_flos": 0.0, + "train_loss": 0.08193727386991183, + "train_runtime": 13050.0697, + "train_samples_per_second": 0.92, + "train_steps_per_second": 0.23 + } + ], + "logging_steps": 10, + "max_steps": 3000, + "num_input_tokens_seen": 0, + "num_train_epochs": 1, + "save_steps": 100, + "total_flos": 0.0, + "train_batch_size": 2, + "trial_name": null, + "trial_params": null +}