{ "best_metric": null, "best_model_checkpoint": null, "epoch": 1.9978618772717553, "eval_steps": 100, "global_step": 1752, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.002280664243460908, "grad_norm": 104.30809234123367, "learning_rate": 4e-09, "logits/chosen": -1.353676199913025, "logits/rejected": -1.4426417350769043, "logps/chosen": -169.59426879882812, "logps/rejected": -212.90069580078125, "loss": 0.7312, "rewards/accuracies": 0.4375, "rewards/chosen": 0.23574542999267578, "rewards/margins": -0.06179435923695564, "rewards/rejected": 0.29753977060317993, "step": 2 }, { "epoch": 0.004561328486921816, "grad_norm": 92.96399840391902, "learning_rate": 8e-09, "logits/chosen": -1.2849147319793701, "logits/rejected": -1.3181504011154175, "logps/chosen": -156.07391357421875, "logps/rejected": -180.74073791503906, "loss": 0.6935, "rewards/accuracies": 0.59375, "rewards/chosen": 0.32024019956588745, "rewards/margins": 0.08485272526741028, "rewards/rejected": 0.23538745939731598, "step": 4 }, { "epoch": 0.006841992730382724, "grad_norm": 98.95959882761899, "learning_rate": 1.1999999999999998e-08, "logits/chosen": -1.4182077646255493, "logits/rejected": -1.4656966924667358, "logps/chosen": -162.13934326171875, "logps/rejected": -188.761474609375, "loss": 0.6766, "rewards/accuracies": 0.59375, "rewards/chosen": 0.4807170629501343, "rewards/margins": 0.1233808621764183, "rewards/rejected": 0.35733622312545776, "step": 6 }, { "epoch": 0.009122656973843632, "grad_norm": 100.15592083854503, "learning_rate": 1.6e-08, "logits/chosen": -1.3926836252212524, "logits/rejected": -1.401774287223816, "logps/chosen": -170.7425079345703, "logps/rejected": -190.57833862304688, "loss": 0.7039, "rewards/accuracies": 0.71875, "rewards/chosen": 0.4466077983379364, "rewards/margins": 0.06900040060281754, "rewards/rejected": 0.37760740518569946, "step": 8 }, { "epoch": 0.01140332121730454, "grad_norm": 97.0889776355223, "learning_rate": 2e-08, "logits/chosen": -1.376230239868164, "logits/rejected": -1.409171223640442, "logps/chosen": -169.57582092285156, "logps/rejected": -185.82571411132812, "loss": 0.7113, "rewards/accuracies": 0.6875, "rewards/chosen": 0.39193734526634216, "rewards/margins": 0.3205062747001648, "rewards/rejected": 0.07143110036849976, "step": 10 }, { "epoch": 0.013683985460765448, "grad_norm": 92.66648488760384, "learning_rate": 2.3999999999999997e-08, "logits/chosen": -1.3613837957382202, "logits/rejected": -1.4228042364120483, "logps/chosen": -142.47850036621094, "logps/rejected": -195.29649353027344, "loss": 0.6867, "rewards/accuracies": 0.5, "rewards/chosen": 0.19295203685760498, "rewards/margins": -0.01336541399359703, "rewards/rejected": 0.20631742477416992, "step": 12 }, { "epoch": 0.015964649704226355, "grad_norm": 88.1654009314199, "learning_rate": 2.8000000000000003e-08, "logits/chosen": -1.5071678161621094, "logits/rejected": -1.5190939903259277, "logps/chosen": -128.34288024902344, "logps/rejected": -131.07789611816406, "loss": 0.6954, "rewards/accuracies": 0.5625, "rewards/chosen": 0.23906439542770386, "rewards/margins": 0.07392804324626923, "rewards/rejected": 0.16513636708259583, "step": 14 }, { "epoch": 0.018245313947687265, "grad_norm": 83.0913005177034, "learning_rate": 3.2e-08, "logits/chosen": -1.2882771492004395, "logits/rejected": -1.3209519386291504, "logps/chosen": -125.15751647949219, "logps/rejected": -132.21255493164062, "loss": 0.6652, "rewards/accuracies": 0.46875, "rewards/chosen": 0.18403516709804535, "rewards/margins": -0.032971642911434174, "rewards/rejected": 0.21700681746006012, "step": 16 }, { "epoch": 0.02052597819114817, "grad_norm": 92.77604284067941, "learning_rate": 3.6e-08, "logits/chosen": -1.3992947340011597, "logits/rejected": -1.473150610923767, "logps/chosen": -173.55206298828125, "logps/rejected": -202.7044677734375, "loss": 0.7212, "rewards/accuracies": 0.34375, "rewards/chosen": 0.4242628216743469, "rewards/margins": -0.0976889505982399, "rewards/rejected": 0.5219517350196838, "step": 18 }, { "epoch": 0.02280664243460908, "grad_norm": 88.66990299818134, "learning_rate": 4e-08, "logits/chosen": -1.432981014251709, "logits/rejected": -1.5539088249206543, "logps/chosen": -182.9879913330078, "logps/rejected": -203.99020385742188, "loss": 0.7184, "rewards/accuracies": 0.5625, "rewards/chosen": 0.4513806104660034, "rewards/margins": 0.20164981484413147, "rewards/rejected": 0.24973079562187195, "step": 20 }, { "epoch": 0.02508730667806999, "grad_norm": 77.91111396247051, "learning_rate": 4.4e-08, "logits/chosen": -1.3197808265686035, "logits/rejected": -1.3996787071228027, "logps/chosen": -154.55003356933594, "logps/rejected": -232.81744384765625, "loss": 0.6479, "rewards/accuracies": 0.65625, "rewards/chosen": 0.3307289481163025, "rewards/margins": 0.18726389110088348, "rewards/rejected": 0.14346502721309662, "step": 22 }, { "epoch": 0.027367970921530895, "grad_norm": 99.08871355118758, "learning_rate": 4.799999999999999e-08, "logits/chosen": -1.3024228811264038, "logits/rejected": -1.3258510828018188, "logps/chosen": -131.18356323242188, "logps/rejected": -148.7147216796875, "loss": 0.7046, "rewards/accuracies": 0.625, "rewards/chosen": 0.421053409576416, "rewards/margins": 0.3122374713420868, "rewards/rejected": 0.10881592333316803, "step": 24 }, { "epoch": 0.029648635164991805, "grad_norm": 89.36423052851278, "learning_rate": 5.2e-08, "logits/chosen": -1.329531192779541, "logits/rejected": -1.4354490041732788, "logps/chosen": -168.1116180419922, "logps/rejected": -207.8105010986328, "loss": 0.6619, "rewards/accuracies": 0.6875, "rewards/chosen": 0.40395674109458923, "rewards/margins": 0.26189538836479187, "rewards/rejected": 0.14206132292747498, "step": 26 }, { "epoch": 0.03192929940845271, "grad_norm": 99.33427508045129, "learning_rate": 5.6000000000000005e-08, "logits/chosen": -1.3796099424362183, "logits/rejected": -1.3620996475219727, "logps/chosen": -131.89183044433594, "logps/rejected": -133.18844604492188, "loss": 0.6782, "rewards/accuracies": 0.6875, "rewards/chosen": 0.37585678696632385, "rewards/margins": 0.27094465494155884, "rewards/rejected": 0.10491211712360382, "step": 28 }, { "epoch": 0.03420996365191362, "grad_norm": 98.1725487266389, "learning_rate": 6e-08, "logits/chosen": -1.4092419147491455, "logits/rejected": -1.4259543418884277, "logps/chosen": -228.17083740234375, "logps/rejected": -249.2948455810547, "loss": 0.6923, "rewards/accuracies": 0.59375, "rewards/chosen": 0.4407646358013153, "rewards/margins": 0.335997998714447, "rewards/rejected": 0.10476663708686829, "step": 30 }, { "epoch": 0.03649062789537453, "grad_norm": 75.77879619863296, "learning_rate": 6.4e-08, "logits/chosen": -1.364654541015625, "logits/rejected": -1.4574649333953857, "logps/chosen": -134.63072204589844, "logps/rejected": -159.65719604492188, "loss": 0.6454, "rewards/accuracies": 0.5625, "rewards/chosen": 0.28305765986442566, "rewards/margins": 0.14592355489730835, "rewards/rejected": 0.1371341049671173, "step": 32 }, { "epoch": 0.038771292138835435, "grad_norm": 100.55666520099189, "learning_rate": 6.8e-08, "logits/chosen": -1.3500841856002808, "logits/rejected": -1.4460691213607788, "logps/chosen": -134.4729461669922, "logps/rejected": -159.3612823486328, "loss": 0.716, "rewards/accuracies": 0.625, "rewards/chosen": 0.44891929626464844, "rewards/margins": 0.12445079535245895, "rewards/rejected": 0.3244684934616089, "step": 34 }, { "epoch": 0.04105195638229634, "grad_norm": 104.71963688579538, "learning_rate": 7.2e-08, "logits/chosen": -1.3853086233139038, "logits/rejected": -1.434361457824707, "logps/chosen": -139.8603057861328, "logps/rejected": -171.86366271972656, "loss": 0.6988, "rewards/accuracies": 0.625, "rewards/chosen": 0.3810410499572754, "rewards/margins": 0.2534639537334442, "rewards/rejected": 0.12757712602615356, "step": 36 }, { "epoch": 0.043332620625757254, "grad_norm": 77.00258209889141, "learning_rate": 7.599999999999999e-08, "logits/chosen": -1.3689879179000854, "logits/rejected": -1.3514069318771362, "logps/chosen": -208.98460388183594, "logps/rejected": -216.9095458984375, "loss": 0.6691, "rewards/accuracies": 0.65625, "rewards/chosen": 0.5943199396133423, "rewards/margins": 0.26101234555244446, "rewards/rejected": 0.33330756425857544, "step": 38 }, { "epoch": 0.04561328486921816, "grad_norm": 83.25872912559916, "learning_rate": 8e-08, "logits/chosen": -1.3037664890289307, "logits/rejected": -1.3465715646743774, "logps/chosen": -160.97340393066406, "logps/rejected": -178.60267639160156, "loss": 0.6319, "rewards/accuracies": 0.625, "rewards/chosen": 0.4052881896495819, "rewards/margins": 0.16212955117225647, "rewards/rejected": 0.24315865337848663, "step": 40 }, { "epoch": 0.047893949112679066, "grad_norm": 113.55879280714245, "learning_rate": 8.4e-08, "logits/chosen": -1.3739519119262695, "logits/rejected": -1.442929744720459, "logps/chosen": -163.52279663085938, "logps/rejected": -217.51824951171875, "loss": 0.7171, "rewards/accuracies": 0.59375, "rewards/chosen": 0.13579119741916656, "rewards/margins": 0.14747555553913116, "rewards/rejected": -0.011684387922286987, "step": 42 }, { "epoch": 0.05017461335613998, "grad_norm": 81.14122988211874, "learning_rate": 8.8e-08, "logits/chosen": -1.373199701309204, "logits/rejected": -1.4750025272369385, "logps/chosen": -132.52114868164062, "logps/rejected": -154.50648498535156, "loss": 0.6947, "rewards/accuracies": 0.59375, "rewards/chosen": 0.4790118932723999, "rewards/margins": 0.16499680280685425, "rewards/rejected": 0.31401512026786804, "step": 44 }, { "epoch": 0.052455277599600884, "grad_norm": 89.9603968710281, "learning_rate": 9.2e-08, "logits/chosen": -1.3388586044311523, "logits/rejected": -1.4220290184020996, "logps/chosen": -181.95082092285156, "logps/rejected": -226.30923461914062, "loss": 0.6941, "rewards/accuracies": 0.46875, "rewards/chosen": 0.4439522325992584, "rewards/margins": 0.17398414015769958, "rewards/rejected": 0.26996806263923645, "step": 46 }, { "epoch": 0.05473594184306179, "grad_norm": 104.88400068525443, "learning_rate": 9.599999999999999e-08, "logits/chosen": -1.456312894821167, "logits/rejected": -1.451519250869751, "logps/chosen": -199.40493774414062, "logps/rejected": -216.68386840820312, "loss": 0.7537, "rewards/accuracies": 0.59375, "rewards/chosen": 0.4792614281177521, "rewards/margins": 0.112982377409935, "rewards/rejected": 0.36627906560897827, "step": 48 }, { "epoch": 0.0570166060865227, "grad_norm": 73.96994315968016, "learning_rate": 1e-07, "logits/chosen": -1.3086615800857544, "logits/rejected": -1.3628607988357544, "logps/chosen": -151.69989013671875, "logps/rejected": -171.46029663085938, "loss": 0.6314, "rewards/accuracies": 0.65625, "rewards/chosen": 0.1983395218849182, "rewards/margins": 0.22184070944786072, "rewards/rejected": -0.023501206189393997, "step": 50 }, { "epoch": 0.05929727032998361, "grad_norm": 100.64321021539473, "learning_rate": 1.04e-07, "logits/chosen": -1.5143202543258667, "logits/rejected": -1.5794754028320312, "logps/chosen": -118.7998046875, "logps/rejected": -149.1954803466797, "loss": 0.6704, "rewards/accuracies": 0.59375, "rewards/chosen": 0.34698718786239624, "rewards/margins": 0.06618239730596542, "rewards/rejected": 0.2808048129081726, "step": 52 }, { "epoch": 0.061577934573444515, "grad_norm": 96.37635530030595, "learning_rate": 1.08e-07, "logits/chosen": -1.254115104675293, "logits/rejected": -1.3123385906219482, "logps/chosen": -152.57516479492188, "logps/rejected": -172.66200256347656, "loss": 0.6966, "rewards/accuracies": 0.46875, "rewards/chosen": 0.28736555576324463, "rewards/margins": 0.0115619245916605, "rewards/rejected": 0.2758035957813263, "step": 54 }, { "epoch": 0.06385859881690542, "grad_norm": 99.11153586277212, "learning_rate": 1.1200000000000001e-07, "logits/chosen": -1.3405977487564087, "logits/rejected": -1.4054450988769531, "logps/chosen": -194.3892822265625, "logps/rejected": -219.24171447753906, "loss": 0.6625, "rewards/accuracies": 0.6875, "rewards/chosen": 0.382589727640152, "rewards/margins": 0.24182234704494476, "rewards/rejected": 0.14076738059520721, "step": 56 }, { "epoch": 0.06613926306036633, "grad_norm": 82.95051663248452, "learning_rate": 1.1599999999999999e-07, "logits/chosen": -1.4561753273010254, "logits/rejected": -1.4183000326156616, "logps/chosen": -189.18923950195312, "logps/rejected": -186.99264526367188, "loss": 0.6206, "rewards/accuracies": 0.59375, "rewards/chosen": 0.4602745771408081, "rewards/margins": 0.18983443081378937, "rewards/rejected": 0.27044013142585754, "step": 58 }, { "epoch": 0.06841992730382725, "grad_norm": 85.63947077517764, "learning_rate": 1.2e-07, "logits/chosen": -1.4239516258239746, "logits/rejected": -1.5685184001922607, "logps/chosen": -140.7595977783203, "logps/rejected": -184.33067321777344, "loss": 0.7255, "rewards/accuracies": 0.53125, "rewards/chosen": 0.2823754847049713, "rewards/margins": 0.02492373436689377, "rewards/rejected": 0.25745177268981934, "step": 60 }, { "epoch": 0.07070059154728815, "grad_norm": 97.67365684595, "learning_rate": 1.24e-07, "logits/chosen": -1.350022554397583, "logits/rejected": -1.3478128910064697, "logps/chosen": -171.09677124023438, "logps/rejected": -165.71694946289062, "loss": 0.6819, "rewards/accuracies": 0.5625, "rewards/chosen": 0.3082874119281769, "rewards/margins": 0.11261623352766037, "rewards/rejected": 0.1956711858510971, "step": 62 }, { "epoch": 0.07298125579074906, "grad_norm": 99.3804737289846, "learning_rate": 1.28e-07, "logits/chosen": -1.4815456867218018, "logits/rejected": -1.4830272197723389, "logps/chosen": -107.2784194946289, "logps/rejected": -121.1039047241211, "loss": 0.7165, "rewards/accuracies": 0.6875, "rewards/chosen": 0.31323567032814026, "rewards/margins": 0.11968746036291122, "rewards/rejected": 0.19354820251464844, "step": 64 }, { "epoch": 0.07526192003420996, "grad_norm": 94.44141503559203, "learning_rate": 1.32e-07, "logits/chosen": -1.493153691291809, "logits/rejected": -1.4850858449935913, "logps/chosen": -164.4472198486328, "logps/rejected": -169.9815673828125, "loss": 0.6558, "rewards/accuracies": 0.6875, "rewards/chosen": 0.5161038637161255, "rewards/margins": 0.30025073885917664, "rewards/rejected": 0.21585312485694885, "step": 66 }, { "epoch": 0.07754258427767087, "grad_norm": 86.21425469616315, "learning_rate": 1.36e-07, "logits/chosen": -1.333409070968628, "logits/rejected": -1.3404844999313354, "logps/chosen": -153.38101196289062, "logps/rejected": -159.85690307617188, "loss": 0.6781, "rewards/accuracies": 0.65625, "rewards/chosen": 0.3244006931781769, "rewards/margins": 0.22184959053993225, "rewards/rejected": 0.10255111753940582, "step": 68 }, { "epoch": 0.07982324852113178, "grad_norm": 90.03565465587579, "learning_rate": 1.3999999999999998e-07, "logits/chosen": -1.5341051816940308, "logits/rejected": -1.5601752996444702, "logps/chosen": -149.2415008544922, "logps/rejected": -156.4012908935547, "loss": 0.6936, "rewards/accuracies": 0.53125, "rewards/chosen": 0.27590540051460266, "rewards/margins": 0.12058861553668976, "rewards/rejected": 0.1553167849779129, "step": 70 }, { "epoch": 0.08210391276459268, "grad_norm": 80.65697099385187, "learning_rate": 1.44e-07, "logits/chosen": -1.3111834526062012, "logits/rejected": -1.3080787658691406, "logps/chosen": -152.25611877441406, "logps/rejected": -169.93919372558594, "loss": 0.6471, "rewards/accuracies": 0.71875, "rewards/chosen": 0.34636569023132324, "rewards/margins": 0.3154309093952179, "rewards/rejected": 0.03093479573726654, "step": 72 }, { "epoch": 0.0843845770080536, "grad_norm": 116.72716177672243, "learning_rate": 1.48e-07, "logits/chosen": -1.3597153425216675, "logits/rejected": -1.4166897535324097, "logps/chosen": -174.33534240722656, "logps/rejected": -200.9926300048828, "loss": 0.6831, "rewards/accuracies": 0.59375, "rewards/chosen": 0.2202744334936142, "rewards/margins": 0.13681186735630035, "rewards/rejected": 0.08346255123615265, "step": 74 }, { "epoch": 0.08666524125151451, "grad_norm": 99.62367819711024, "learning_rate": 1.5199999999999998e-07, "logits/chosen": -1.3973196744918823, "logits/rejected": -1.4653477668762207, "logps/chosen": -189.6744384765625, "logps/rejected": -224.8593292236328, "loss": 0.7086, "rewards/accuracies": 0.5, "rewards/chosen": 0.33062875270843506, "rewards/margins": 0.11676961928606033, "rewards/rejected": 0.21385914087295532, "step": 76 }, { "epoch": 0.08894590549497541, "grad_norm": 86.47188266565705, "learning_rate": 1.56e-07, "logits/chosen": -1.3268086910247803, "logits/rejected": -1.302788257598877, "logps/chosen": -135.62147521972656, "logps/rejected": -153.5860137939453, "loss": 0.6835, "rewards/accuracies": 0.5625, "rewards/chosen": 0.2682960629463196, "rewards/margins": 0.05024400353431702, "rewards/rejected": 0.21805202960968018, "step": 78 }, { "epoch": 0.09122656973843632, "grad_norm": 125.16786139574279, "learning_rate": 1.6e-07, "logits/chosen": -1.2104482650756836, "logits/rejected": -1.3274182081222534, "logps/chosen": -161.164794921875, "logps/rejected": -210.53590393066406, "loss": 0.7174, "rewards/accuracies": 0.53125, "rewards/chosen": 0.2355586439371109, "rewards/margins": 0.07326290011405945, "rewards/rejected": 0.16229577362537384, "step": 80 }, { "epoch": 0.09350723398189723, "grad_norm": 98.85832957481998, "learning_rate": 1.6399999999999999e-07, "logits/chosen": -1.3579671382904053, "logits/rejected": -1.4671956300735474, "logps/chosen": -153.21591186523438, "logps/rejected": -173.60589599609375, "loss": 0.6862, "rewards/accuracies": 0.46875, "rewards/chosen": 0.2675905227661133, "rewards/margins": 0.046474047005176544, "rewards/rejected": 0.22111651301383972, "step": 82 }, { "epoch": 0.09578789822535813, "grad_norm": 83.11445245790541, "learning_rate": 1.68e-07, "logits/chosen": -1.2888550758361816, "logits/rejected": -1.3688700199127197, "logps/chosen": -145.24754333496094, "logps/rejected": -174.91700744628906, "loss": 0.637, "rewards/accuracies": 0.625, "rewards/chosen": 0.36274462938308716, "rewards/margins": 0.15029877424240112, "rewards/rejected": 0.21244585514068604, "step": 84 }, { "epoch": 0.09806856246881904, "grad_norm": 82.49239750970325, "learning_rate": 1.7199999999999998e-07, "logits/chosen": -1.3739551305770874, "logits/rejected": -1.3799883127212524, "logps/chosen": -178.80386352539062, "logps/rejected": -215.7256317138672, "loss": 0.6907, "rewards/accuracies": 0.625, "rewards/chosen": 0.32459211349487305, "rewards/margins": 0.16312535107135773, "rewards/rejected": 0.1614667773246765, "step": 86 }, { "epoch": 0.10034922671227996, "grad_norm": 105.64550946139919, "learning_rate": 1.76e-07, "logits/chosen": -1.4653171300888062, "logits/rejected": -1.508003830909729, "logps/chosen": -192.55154418945312, "logps/rejected": -214.61546325683594, "loss": 0.7401, "rewards/accuracies": 0.6875, "rewards/chosen": 0.43679386377334595, "rewards/margins": 0.1878117173910141, "rewards/rejected": 0.24898216128349304, "step": 88 }, { "epoch": 0.10262989095574086, "grad_norm": 106.99677192741333, "learning_rate": 1.8e-07, "logits/chosen": -1.3307772874832153, "logits/rejected": -1.4872570037841797, "logps/chosen": -226.3917236328125, "logps/rejected": -265.8712158203125, "loss": 0.6758, "rewards/accuracies": 0.53125, "rewards/chosen": 0.42067086696624756, "rewards/margins": 0.07930833846330643, "rewards/rejected": 0.34136250615119934, "step": 90 }, { "epoch": 0.10491055519920177, "grad_norm": 104.67741510645097, "learning_rate": 1.84e-07, "logits/chosen": -1.3470810651779175, "logits/rejected": -1.3835241794586182, "logps/chosen": -162.22958374023438, "logps/rejected": -184.17697143554688, "loss": 0.6993, "rewards/accuracies": 0.5, "rewards/chosen": 0.25471583008766174, "rewards/margins": 0.08957283943891525, "rewards/rejected": 0.1651429831981659, "step": 92 }, { "epoch": 0.10719121944266267, "grad_norm": 82.2596269922547, "learning_rate": 1.88e-07, "logits/chosen": -1.3317389488220215, "logits/rejected": -1.367332935333252, "logps/chosen": -138.1045684814453, "logps/rejected": -213.2273712158203, "loss": 0.5933, "rewards/accuracies": 0.5625, "rewards/chosen": 0.27782779932022095, "rewards/margins": 0.2610364258289337, "rewards/rejected": 0.01679137721657753, "step": 94 }, { "epoch": 0.10947188368612358, "grad_norm": 106.46518450026352, "learning_rate": 1.9199999999999997e-07, "logits/chosen": -1.3722081184387207, "logits/rejected": -1.4072837829589844, "logps/chosen": -185.40281677246094, "logps/rejected": -205.63037109375, "loss": 0.7079, "rewards/accuracies": 0.71875, "rewards/chosen": 0.4706501066684723, "rewards/margins": 0.29465728998184204, "rewards/rejected": 0.17599281668663025, "step": 96 }, { "epoch": 0.11175254792958449, "grad_norm": 79.05608727628055, "learning_rate": 1.9599999999999998e-07, "logits/chosen": -1.2723050117492676, "logits/rejected": -1.2712621688842773, "logps/chosen": -170.6747589111328, "logps/rejected": -172.70376586914062, "loss": 0.6713, "rewards/accuracies": 0.625, "rewards/chosen": 0.4158933460712433, "rewards/margins": 0.22599810361862183, "rewards/rejected": 0.18989527225494385, "step": 98 }, { "epoch": 0.1140332121730454, "grad_norm": 102.50537690642588, "learning_rate": 2e-07, "logits/chosen": -1.2609540224075317, "logits/rejected": -1.3507779836654663, "logps/chosen": -126.81893157958984, "logps/rejected": -172.69830322265625, "loss": 0.7344, "rewards/accuracies": 0.59375, "rewards/chosen": 0.2087049037218094, "rewards/margins": 0.05775710940361023, "rewards/rejected": 0.15094780921936035, "step": 100 }, { "epoch": 0.1140332121730454, "eval_logits/chosen": -1.465891718864441, "eval_logits/rejected": -1.452298879623413, "eval_logps/chosen": -126.79114532470703, "eval_logps/rejected": -128.49049377441406, "eval_loss": 0.7120834589004517, "eval_rewards/accuracies": 0.4000000059604645, "eval_rewards/chosen": 0.3106829524040222, "eval_rewards/margins": -0.04654194042086601, "eval_rewards/rejected": 0.3572249114513397, "eval_runtime": 22.2628, "eval_samples_per_second": 4.492, "eval_steps_per_second": 1.123, "step": 100 }, { "epoch": 0.11631387641650631, "grad_norm": 115.65869747987765, "learning_rate": 1.9999927671554446e-07, "logits/chosen": -1.4759191274642944, "logits/rejected": -1.458261251449585, "logps/chosen": -200.50059509277344, "logps/rejected": -213.1103973388672, "loss": 0.6618, "rewards/accuracies": 0.53125, "rewards/chosen": 0.3923037648200989, "rewards/margins": 0.1664755940437317, "rewards/rejected": 0.2258281409740448, "step": 102 }, { "epoch": 0.11859454065996722, "grad_norm": 93.12582037555201, "learning_rate": 1.9999710687264071e-07, "logits/chosen": -1.311631441116333, "logits/rejected": -1.3278183937072754, "logps/chosen": -132.20401000976562, "logps/rejected": -157.69627380371094, "loss": 0.7135, "rewards/accuracies": 0.625, "rewards/chosen": 0.19503216445446014, "rewards/margins": -0.017576567828655243, "rewards/rejected": 0.21260873973369598, "step": 104 }, { "epoch": 0.12087520490342812, "grad_norm": 76.40018414480076, "learning_rate": 1.9999349050267698e-07, "logits/chosen": -1.291077733039856, "logits/rejected": -1.3958327770233154, "logps/chosen": -158.81573486328125, "logps/rejected": -199.284423828125, "loss": 0.6252, "rewards/accuracies": 0.65625, "rewards/chosen": 0.25081124901771545, "rewards/margins": 0.25090575218200684, "rewards/rejected": -9.45068895816803e-05, "step": 106 }, { "epoch": 0.12315586914688903, "grad_norm": 90.49004280394047, "learning_rate": 1.9998842765796658e-07, "logits/chosen": -1.4403332471847534, "logits/rejected": -1.4392634630203247, "logps/chosen": -143.04052734375, "logps/rejected": -156.4816436767578, "loss": 0.6816, "rewards/accuracies": 0.625, "rewards/chosen": 0.10938511788845062, "rewards/margins": 0.17133575677871704, "rewards/rejected": -0.06195063889026642, "step": 108 }, { "epoch": 0.12543653339034994, "grad_norm": 81.5018944914688, "learning_rate": 1.9998191841174702e-07, "logits/chosen": -1.4619545936584473, "logits/rejected": -1.483961582183838, "logps/chosen": -120.11573028564453, "logps/rejected": -121.2934341430664, "loss": 0.6783, "rewards/accuracies": 0.65625, "rewards/chosen": 0.4725874662399292, "rewards/margins": 0.37433862686157227, "rewards/rejected": 0.09824882447719574, "step": 110 }, { "epoch": 0.12771719763381084, "grad_norm": 91.38630396004429, "learning_rate": 1.9997396285817904e-07, "logits/chosen": -1.152282953262329, "logits/rejected": -1.252787470817566, "logps/chosen": -145.3370361328125, "logps/rejected": -174.6603240966797, "loss": 0.6269, "rewards/accuracies": 0.5625, "rewards/chosen": 0.40825897455215454, "rewards/margins": 0.25122010707855225, "rewards/rejected": 0.1570388376712799, "step": 112 }, { "epoch": 0.12999786187727175, "grad_norm": 97.62629025517847, "learning_rate": 1.9996456111234526e-07, "logits/chosen": -1.393836259841919, "logits/rejected": -1.5118399858474731, "logps/chosen": -173.96847534179688, "logps/rejected": -229.94232177734375, "loss": 0.6697, "rewards/accuracies": 0.59375, "rewards/chosen": 0.21259565651416779, "rewards/margins": 0.18360112607479095, "rewards/rejected": 0.028994524851441383, "step": 114 }, { "epoch": 0.13227852612073265, "grad_norm": 96.19706483741392, "learning_rate": 1.9995371331024833e-07, "logits/chosen": -1.299769401550293, "logits/rejected": -1.3910255432128906, "logps/chosen": -112.2242660522461, "logps/rejected": -130.05230712890625, "loss": 0.7283, "rewards/accuracies": 0.4375, "rewards/chosen": 0.14800208806991577, "rewards/margins": -0.08275800198316574, "rewards/rejected": 0.23076008260250092, "step": 116 }, { "epoch": 0.13455919036419356, "grad_norm": 114.37970019595372, "learning_rate": 1.999414196088092e-07, "logits/chosen": -1.3472720384597778, "logits/rejected": -1.4912211894989014, "logps/chosen": -202.3058319091797, "logps/rejected": -244.92904663085938, "loss": 0.7194, "rewards/accuracies": 0.46875, "rewards/chosen": 0.3503240644931793, "rewards/margins": 0.024903126060962677, "rewards/rejected": 0.32542091608047485, "step": 118 }, { "epoch": 0.1368398546076545, "grad_norm": 87.71566928589093, "learning_rate": 1.9992768018586478e-07, "logits/chosen": -1.3490660190582275, "logits/rejected": -1.3772616386413574, "logps/chosen": -146.59048461914062, "logps/rejected": -159.0959930419922, "loss": 0.6736, "rewards/accuracies": 0.5625, "rewards/chosen": 0.3947565257549286, "rewards/margins": 0.15041925013065338, "rewards/rejected": 0.244337260723114, "step": 120 }, { "epoch": 0.1391205188511154, "grad_norm": 109.98286067474862, "learning_rate": 1.9991249524016528e-07, "logits/chosen": -1.2736543416976929, "logits/rejected": -1.4115746021270752, "logps/chosen": -150.3526611328125, "logps/rejected": -189.20545959472656, "loss": 0.761, "rewards/accuracies": 0.46875, "rewards/chosen": 0.28761598467826843, "rewards/margins": -0.15278732776641846, "rewards/rejected": 0.44040337204933167, "step": 122 }, { "epoch": 0.1414011830945763, "grad_norm": 108.93431507211488, "learning_rate": 1.9989586499137135e-07, "logits/chosen": -1.2978495359420776, "logits/rejected": -1.3472024202346802, "logps/chosen": -179.5664825439453, "logps/rejected": -203.53968811035156, "loss": 0.7215, "rewards/accuracies": 0.5625, "rewards/chosen": 0.39709562063217163, "rewards/margins": 0.20268934965133667, "rewards/rejected": 0.19440627098083496, "step": 124 }, { "epoch": 0.1436818473380372, "grad_norm": 99.31026397357329, "learning_rate": 1.9987778968005106e-07, "logits/chosen": -1.2714462280273438, "logits/rejected": -1.4020087718963623, "logps/chosen": -142.49758911132812, "logps/rejected": -262.75775146484375, "loss": 0.7111, "rewards/accuracies": 0.5, "rewards/chosen": 0.1395024061203003, "rewards/margins": -0.11353214085102081, "rewards/rejected": 0.2530345320701599, "step": 126 }, { "epoch": 0.14596251158149812, "grad_norm": 92.0460328359884, "learning_rate": 1.9985826956767617e-07, "logits/chosen": -1.474685788154602, "logits/rejected": -1.5739442110061646, "logps/chosen": -135.28616333007812, "logps/rejected": -165.1414031982422, "loss": 0.6874, "rewards/accuracies": 0.5625, "rewards/chosen": 0.14916732907295227, "rewards/margins": 0.06242816895246506, "rewards/rejected": 0.0867391973733902, "step": 128 }, { "epoch": 0.14824317582495902, "grad_norm": 96.22054918626448, "learning_rate": 1.9983730493661865e-07, "logits/chosen": -1.276041030883789, "logits/rejected": -1.3194851875305176, "logps/chosen": -221.15992736816406, "logps/rejected": -242.77647399902344, "loss": 0.6457, "rewards/accuracies": 0.6875, "rewards/chosen": 0.260553777217865, "rewards/margins": 0.26679596304893494, "rewards/rejected": -0.006242218893021345, "step": 130 }, { "epoch": 0.15052384006841993, "grad_norm": 78.01858280488834, "learning_rate": 1.998148960901463e-07, "logits/chosen": -1.393154263496399, "logits/rejected": -1.423519492149353, "logps/chosen": -201.9635467529297, "logps/rejected": -235.25816345214844, "loss": 0.6893, "rewards/accuracies": 0.625, "rewards/chosen": 0.382313072681427, "rewards/margins": 0.3356338441371918, "rewards/rejected": 0.04667920991778374, "step": 132 }, { "epoch": 0.15280450431188083, "grad_norm": 103.00857915327929, "learning_rate": 1.997910433524185e-07, "logits/chosen": -1.3205957412719727, "logits/rejected": -1.3766120672225952, "logps/chosen": -189.32142639160156, "logps/rejected": -217.68544006347656, "loss": 0.724, "rewards/accuracies": 0.65625, "rewards/chosen": 0.25681746006011963, "rewards/margins": 0.15590888261795044, "rewards/rejected": 0.10090853273868561, "step": 134 }, { "epoch": 0.15508516855534174, "grad_norm": 90.52690719042177, "learning_rate": 1.9976574706848153e-07, "logits/chosen": -1.4535361528396606, "logits/rejected": -1.4828790426254272, "logps/chosen": -171.7893829345703, "logps/rejected": -213.03846740722656, "loss": 0.6788, "rewards/accuracies": 0.53125, "rewards/chosen": 0.33188748359680176, "rewards/margins": 0.07647443562746048, "rewards/rejected": 0.2554129958152771, "step": 136 }, { "epoch": 0.15736583279880265, "grad_norm": 101.69492728925366, "learning_rate": 1.9973900760426363e-07, "logits/chosen": -1.4226858615875244, "logits/rejected": -1.4269222021102905, "logps/chosen": -166.29103088378906, "logps/rejected": -268.772705078125, "loss": 0.678, "rewards/accuracies": 0.53125, "rewards/chosen": 0.3539034128189087, "rewards/margins": 0.052625708281993866, "rewards/rejected": 0.3012777268886566, "step": 138 }, { "epoch": 0.15964649704226355, "grad_norm": 83.07835169069685, "learning_rate": 1.9971082534656955e-07, "logits/chosen": -1.37977933883667, "logits/rejected": -1.466927409172058, "logps/chosen": -113.37686157226562, "logps/rejected": -142.82138061523438, "loss": 0.6442, "rewards/accuracies": 0.59375, "rewards/chosen": 0.15869390964508057, "rewards/margins": 0.20419231057167053, "rewards/rejected": -0.045498400926589966, "step": 140 }, { "epoch": 0.16192716128572446, "grad_norm": 103.54665420816978, "learning_rate": 1.99681200703075e-07, "logits/chosen": -1.4178937673568726, "logits/rejected": -1.4681955575942993, "logps/chosen": -158.346435546875, "logps/rejected": -180.1523895263672, "loss": 0.7127, "rewards/accuracies": 0.5625, "rewards/chosen": 0.2817228436470032, "rewards/margins": 0.11103209853172302, "rewards/rejected": 0.17069074511528015, "step": 142 }, { "epoch": 0.16420782552918536, "grad_norm": 77.71797198863521, "learning_rate": 1.9965013410232097e-07, "logits/chosen": -1.40109384059906, "logits/rejected": -1.4553627967834473, "logps/chosen": -172.07899475097656, "logps/rejected": -222.21438598632812, "loss": 0.6728, "rewards/accuracies": 0.5625, "rewards/chosen": 0.27610713243484497, "rewards/margins": 0.35402336716651917, "rewards/rejected": -0.077916219830513, "step": 144 }, { "epoch": 0.16648848977264627, "grad_norm": 91.27829192980697, "learning_rate": 1.996176259937072e-07, "logits/chosen": -1.3127285242080688, "logits/rejected": -1.3604404926300049, "logps/chosen": -168.33004760742188, "logps/rejected": -193.3489227294922, "loss": 0.679, "rewards/accuracies": 0.59375, "rewards/chosen": 0.18943265080451965, "rewards/margins": 0.11781211942434311, "rewards/rejected": 0.07162053138017654, "step": 146 }, { "epoch": 0.1687691540161072, "grad_norm": 105.10949942036537, "learning_rate": 1.9958367684748586e-07, "logits/chosen": -1.301593542098999, "logits/rejected": -1.3123453855514526, "logps/chosen": -153.8624267578125, "logps/rejected": -172.3568115234375, "loss": 0.6824, "rewards/accuracies": 0.53125, "rewards/chosen": 0.022907603532075882, "rewards/margins": 0.08112253248691559, "rewards/rejected": -0.058214932680130005, "step": 148 }, { "epoch": 0.1710498182595681, "grad_norm": 99.17231684606652, "learning_rate": 1.995482871547548e-07, "logits/chosen": -1.2876176834106445, "logits/rejected": -1.3324649333953857, "logps/chosen": -127.41693115234375, "logps/rejected": -158.8002166748047, "loss": 0.6995, "rewards/accuracies": 0.71875, "rewards/chosen": 0.21076691150665283, "rewards/margins": 0.24259139597415924, "rewards/rejected": -0.031824491918087006, "step": 150 }, { "epoch": 0.17333048250302902, "grad_norm": 91.41364046696481, "learning_rate": 1.9951145742745024e-07, "logits/chosen": -1.2649379968643188, "logits/rejected": -1.4087891578674316, "logps/chosen": -130.08468627929688, "logps/rejected": -158.69525146484375, "loss": 0.7016, "rewards/accuracies": 0.5, "rewards/chosen": 0.23989444971084595, "rewards/margins": 0.030172180384397507, "rewards/rejected": 0.20972229540348053, "step": 152 }, { "epoch": 0.17561114674648992, "grad_norm": 75.7202165191751, "learning_rate": 1.9947318819833962e-07, "logits/chosen": -1.3358944654464722, "logits/rejected": -1.3563066720962524, "logps/chosen": -106.91460418701172, "logps/rejected": -113.2139892578125, "loss": 0.6607, "rewards/accuracies": 0.46875, "rewards/chosen": 0.14014360308647156, "rewards/margins": -0.031036917120218277, "rewards/rejected": 0.17118054628372192, "step": 154 }, { "epoch": 0.17789181098995083, "grad_norm": 109.3316094106793, "learning_rate": 1.994334800210137e-07, "logits/chosen": -1.3152189254760742, "logits/rejected": -1.3776334524154663, "logps/chosen": -160.3906707763672, "logps/rejected": -232.65725708007812, "loss": 0.7105, "rewards/accuracies": 0.4375, "rewards/chosen": 0.10319292545318604, "rewards/margins": -0.037254467606544495, "rewards/rejected": 0.14044740796089172, "step": 156 }, { "epoch": 0.18017247523341173, "grad_norm": 84.0665837850057, "learning_rate": 1.9939233346987862e-07, "logits/chosen": -1.4214388132095337, "logits/rejected": -1.437930703163147, "logps/chosen": -172.41867065429688, "logps/rejected": -182.8797607421875, "loss": 0.6807, "rewards/accuracies": 0.59375, "rewards/chosen": 0.16843558847904205, "rewards/margins": 0.16182081401348114, "rewards/rejected": 0.0066147902980446815, "step": 158 }, { "epoch": 0.18245313947687264, "grad_norm": 93.91811474012694, "learning_rate": 1.9934974914014762e-07, "logits/chosen": -1.3304110765457153, "logits/rejected": -1.386664628982544, "logps/chosen": -112.3897933959961, "logps/rejected": -133.53810119628906, "loss": 0.7031, "rewards/accuracies": 0.40625, "rewards/chosen": 0.19350233674049377, "rewards/margins": -0.08095596730709076, "rewards/rejected": 0.27445828914642334, "step": 160 }, { "epoch": 0.18473380372033354, "grad_norm": 90.17492690496479, "learning_rate": 1.9930572764783236e-07, "logits/chosen": -1.2432975769042969, "logits/rejected": -1.2546998262405396, "logps/chosen": -167.08847045898438, "logps/rejected": -183.3931884765625, "loss": 0.6979, "rewards/accuracies": 0.6875, "rewards/chosen": 0.05589159578084946, "rewards/margins": 0.17001007497310638, "rewards/rejected": -0.11411847919225693, "step": 162 }, { "epoch": 0.18701446796379445, "grad_norm": 95.44656254255437, "learning_rate": 1.9926026962973403e-07, "logits/chosen": -1.326186180114746, "logits/rejected": -1.4297109842300415, "logps/chosen": -183.224853515625, "logps/rejected": -239.97793579101562, "loss": 0.6436, "rewards/accuracies": 0.625, "rewards/chosen": 0.16740243136882782, "rewards/margins": 0.23012831807136536, "rewards/rejected": -0.06272587180137634, "step": 164 }, { "epoch": 0.18929513220725536, "grad_norm": 92.54409107962749, "learning_rate": 1.992133757434342e-07, "logits/chosen": -1.3230640888214111, "logits/rejected": -1.31871497631073, "logps/chosen": -226.2008819580078, "logps/rejected": -243.1304931640625, "loss": 0.6795, "rewards/accuracies": 0.65625, "rewards/chosen": 0.1608559638261795, "rewards/margins": 0.35546940565109253, "rewards/rejected": -0.1946134865283966, "step": 166 }, { "epoch": 0.19157579645071626, "grad_norm": 93.78690688865437, "learning_rate": 1.991650466672853e-07, "logits/chosen": -1.2681585550308228, "logits/rejected": -1.27677321434021, "logps/chosen": -151.71714782714844, "logps/rejected": -171.88832092285156, "loss": 0.6772, "rewards/accuracies": 0.625, "rewards/chosen": 0.21696476638317108, "rewards/margins": 0.30003517866134644, "rewards/rejected": -0.08307043462991714, "step": 168 }, { "epoch": 0.19385646069417717, "grad_norm": 79.03039382866096, "learning_rate": 1.991152831004007e-07, "logits/chosen": -1.4594199657440186, "logits/rejected": -1.456141710281372, "logps/chosen": -169.4336700439453, "logps/rejected": -180.23756408691406, "loss": 0.695, "rewards/accuracies": 0.59375, "rewards/chosen": 0.091593436896801, "rewards/margins": 0.008565463125705719, "rewards/rejected": 0.08302795141935349, "step": 170 }, { "epoch": 0.19613712493763807, "grad_norm": 84.01165418355862, "learning_rate": 1.9906408576264467e-07, "logits/chosen": -1.272233247756958, "logits/rejected": -1.309181809425354, "logps/chosen": -182.87887573242188, "logps/rejected": -218.7847442626953, "loss": 0.6868, "rewards/accuracies": 0.625, "rewards/chosen": 0.13641594350337982, "rewards/margins": 0.20986737310886383, "rewards/rejected": -0.07345142215490341, "step": 172 }, { "epoch": 0.198417789181099, "grad_norm": 95.73538074987611, "learning_rate": 1.9901145539462197e-07, "logits/chosen": -1.2831114530563354, "logits/rejected": -1.302902340888977, "logps/chosen": -147.5471954345703, "logps/rejected": -178.70321655273438, "loss": 0.6013, "rewards/accuracies": 0.6875, "rewards/chosen": 0.2684887647628784, "rewards/margins": 0.47333818674087524, "rewards/rejected": -0.20484942197799683, "step": 174 }, { "epoch": 0.20069845342455991, "grad_norm": 102.90991917593935, "learning_rate": 1.9895739275766715e-07, "logits/chosen": -1.3818506002426147, "logits/rejected": -1.4092711210250854, "logps/chosen": -227.61679077148438, "logps/rejected": -226.88558959960938, "loss": 0.737, "rewards/accuracies": 0.6875, "rewards/chosen": 0.13025416433811188, "rewards/margins": 0.26329731941223145, "rewards/rejected": -0.13304319977760315, "step": 176 }, { "epoch": 0.20297911766802082, "grad_norm": 105.89850456514735, "learning_rate": 1.9890189863383353e-07, "logits/chosen": -1.4667214155197144, "logits/rejected": -1.5101617574691772, "logps/chosen": -185.08575439453125, "logps/rejected": -237.65597534179688, "loss": 0.7025, "rewards/accuracies": 0.625, "rewards/chosen": 0.2359275072813034, "rewards/margins": 0.08144809305667877, "rewards/rejected": 0.15447941422462463, "step": 178 }, { "epoch": 0.20525978191148173, "grad_norm": 82.46406340773291, "learning_rate": 1.9884497382588183e-07, "logits/chosen": -1.2853760719299316, "logits/rejected": -1.3631393909454346, "logps/chosen": -140.35952758789062, "logps/rejected": -153.50978088378906, "loss": 0.6683, "rewards/accuracies": 0.5, "rewards/chosen": 0.1511276513338089, "rewards/margins": 0.06823855638504028, "rewards/rejected": 0.08288908749818802, "step": 180 }, { "epoch": 0.20754044615494263, "grad_norm": 90.32390874613019, "learning_rate": 1.9878661915726865e-07, "logits/chosen": -1.4392482042312622, "logits/rejected": -1.4409726858139038, "logps/chosen": -171.09303283691406, "logps/rejected": -182.13232421875, "loss": 0.6691, "rewards/accuracies": 0.53125, "rewards/chosen": 0.20515748858451843, "rewards/margins": 0.10885492712259293, "rewards/rejected": 0.0963025838136673, "step": 182 }, { "epoch": 0.20982111039840354, "grad_norm": 75.10429123148643, "learning_rate": 1.9872683547213446e-07, "logits/chosen": -1.3081107139587402, "logits/rejected": -1.317389965057373, "logps/chosen": -174.84463500976562, "logps/rejected": -210.18130493164062, "loss": 0.6158, "rewards/accuracies": 0.65625, "rewards/chosen": 0.18892019987106323, "rewards/margins": 0.4061851501464844, "rewards/rejected": -0.21726495027542114, "step": 184 }, { "epoch": 0.21210177464186444, "grad_norm": 107.2216966034497, "learning_rate": 1.9866562363529144e-07, "logits/chosen": -1.2954977750778198, "logits/rejected": -1.3127189874649048, "logps/chosen": -197.97561645507812, "logps/rejected": -204.29013061523438, "loss": 0.7154, "rewards/accuracies": 0.4375, "rewards/chosen": 0.17145520448684692, "rewards/margins": -0.03724336624145508, "rewards/rejected": 0.208698570728302, "step": 186 }, { "epoch": 0.21438243888532535, "grad_norm": 96.60657963150389, "learning_rate": 1.9860298453221107e-07, "logits/chosen": -1.3616856336593628, "logits/rejected": -1.3934649229049683, "logps/chosen": -179.2027130126953, "logps/rejected": -187.0948028564453, "loss": 0.6701, "rewards/accuracies": 0.5625, "rewards/chosen": 0.0954272449016571, "rewards/margins": -0.021092543378472328, "rewards/rejected": 0.11651977896690369, "step": 188 }, { "epoch": 0.21666310312878626, "grad_norm": 87.8142315385882, "learning_rate": 1.9853891906901108e-07, "logits/chosen": -1.3980423212051392, "logits/rejected": -1.4947185516357422, "logps/chosen": -142.8896942138672, "logps/rejected": -180.8528594970703, "loss": 0.6415, "rewards/accuracies": 0.59375, "rewards/chosen": 0.13048064708709717, "rewards/margins": 0.3490259051322937, "rewards/rejected": -0.21854525804519653, "step": 190 }, { "epoch": 0.21894376737224716, "grad_norm": 91.22189567457308, "learning_rate": 1.9847342817244254e-07, "logits/chosen": -1.3708425760269165, "logits/rejected": -1.3748908042907715, "logps/chosen": -155.70945739746094, "logps/rejected": -175.20013427734375, "loss": 0.6507, "rewards/accuracies": 0.5625, "rewards/chosen": 0.3499588370323181, "rewards/margins": 0.12689019739627838, "rewards/rejected": 0.22306862473487854, "step": 192 }, { "epoch": 0.22122443161570807, "grad_norm": 103.90331869438803, "learning_rate": 1.9840651278987642e-07, "logits/chosen": -1.3758065700531006, "logits/rejected": -1.3824987411499023, "logps/chosen": -181.644775390625, "logps/rejected": -178.72109985351562, "loss": 0.6563, "rewards/accuracies": 0.71875, "rewards/chosen": 0.29924964904785156, "rewards/margins": 0.3290708065032959, "rewards/rejected": -0.02982119843363762, "step": 194 }, { "epoch": 0.22350509585916897, "grad_norm": 103.8705512922574, "learning_rate": 1.9833817388928984e-07, "logits/chosen": -1.4562551975250244, "logits/rejected": -1.511305809020996, "logps/chosen": -168.09715270996094, "logps/rejected": -189.35787963867188, "loss": 0.6926, "rewards/accuracies": 0.5625, "rewards/chosen": 0.31778958439826965, "rewards/margins": 0.04199263080954552, "rewards/rejected": 0.27579694986343384, "step": 196 }, { "epoch": 0.22578576010262988, "grad_norm": 103.40331674623164, "learning_rate": 1.982684124592521e-07, "logits/chosen": -1.3754979372024536, "logits/rejected": -1.347690463066101, "logps/chosen": -168.61148071289062, "logps/rejected": -186.8193359375, "loss": 0.7315, "rewards/accuracies": 0.5625, "rewards/chosen": 0.14702042937278748, "rewards/margins": 0.059913113713264465, "rewards/rejected": 0.08710730820894241, "step": 198 }, { "epoch": 0.2280664243460908, "grad_norm": 102.65405433185128, "learning_rate": 1.981972295089103e-07, "logits/chosen": -1.4264771938323975, "logits/rejected": -1.5102254152297974, "logps/chosen": -160.4220428466797, "logps/rejected": -173.62269592285156, "loss": 0.6553, "rewards/accuracies": 0.59375, "rewards/chosen": 0.3021972179412842, "rewards/margins": 0.10177014023065567, "rewards/rejected": 0.2004270851612091, "step": 200 }, { "epoch": 0.2280664243460908, "eval_logits/chosen": -1.4570631980895996, "eval_logits/rejected": -1.442795753479004, "eval_logps/chosen": -127.66089630126953, "eval_logps/rejected": -129.5493621826172, "eval_loss": 0.6965904235839844, "eval_rewards/accuracies": 0.5199999809265137, "eval_rewards/chosen": 0.22370710968971252, "eval_rewards/margins": -0.02763114497065544, "eval_rewards/rejected": 0.2513382136821747, "eval_runtime": 22.1102, "eval_samples_per_second": 4.523, "eval_steps_per_second": 1.131, "step": 200 }, { "epoch": 0.23034708858955172, "grad_norm": 70.85461032661505, "learning_rate": 1.9812462606797498e-07, "logits/chosen": -1.1571446657180786, "logits/rejected": -1.2567036151885986, "logps/chosen": -170.15008544921875, "logps/rejected": -181.82806396484375, "loss": 0.6236, "rewards/accuracies": 0.6875, "rewards/chosen": 0.13917526602745056, "rewards/margins": 0.30568423867225647, "rewards/rejected": -0.1665089875459671, "step": 202 }, { "epoch": 0.23262775283301262, "grad_norm": 101.20833615282186, "learning_rate": 1.9805060318670485e-07, "logits/chosen": -1.3708992004394531, "logits/rejected": -1.3186278343200684, "logps/chosen": -154.29226684570312, "logps/rejected": -148.5332794189453, "loss": 0.6319, "rewards/accuracies": 0.5625, "rewards/chosen": 0.09488646686077118, "rewards/margins": 0.11530336737632751, "rewards/rejected": -0.020416898652911186, "step": 204 }, { "epoch": 0.23490841707647353, "grad_norm": 96.32887019136864, "learning_rate": 1.9797516193589192e-07, "logits/chosen": -1.3597060441970825, "logits/rejected": -1.3876497745513916, "logps/chosen": -101.66006469726562, "logps/rejected": -131.19041442871094, "loss": 0.7312, "rewards/accuracies": 0.53125, "rewards/chosen": 0.20645485818386078, "rewards/margins": 0.06082191318273544, "rewards/rejected": 0.14563296735286713, "step": 206 }, { "epoch": 0.23718908131993444, "grad_norm": 91.31654055574913, "learning_rate": 1.978983034068459e-07, "logits/chosen": -1.3642208576202393, "logits/rejected": -1.405060052871704, "logps/chosen": -188.59420776367188, "logps/rejected": -199.42637634277344, "loss": 0.6922, "rewards/accuracies": 0.59375, "rewards/chosen": 0.13877716660499573, "rewards/margins": 0.17159023880958557, "rewards/rejected": -0.03281306475400925, "step": 208 }, { "epoch": 0.23946974556339534, "grad_norm": 99.31042679769261, "learning_rate": 1.9782002871137832e-07, "logits/chosen": -1.3214318752288818, "logits/rejected": -1.3617138862609863, "logps/chosen": -195.82009887695312, "logps/rejected": -214.06614685058594, "loss": 0.6673, "rewards/accuracies": 0.65625, "rewards/chosen": 0.16799718141555786, "rewards/margins": 0.25821763277053833, "rewards/rejected": -0.09022042155265808, "step": 210 }, { "epoch": 0.24175040980685625, "grad_norm": 103.3617201342953, "learning_rate": 1.9774033898178666e-07, "logits/chosen": -1.4174686670303345, "logits/rejected": -1.420326590538025, "logps/chosen": -129.22848510742188, "logps/rejected": -145.9871826171875, "loss": 0.6716, "rewards/accuracies": 0.65625, "rewards/chosen": 0.3314517140388489, "rewards/margins": 0.1635245978832245, "rewards/rejected": 0.167927086353302, "step": 212 }, { "epoch": 0.24403107405031715, "grad_norm": 85.24728778878888, "learning_rate": 1.9765923537083774e-07, "logits/chosen": -1.243302583694458, "logits/rejected": -1.2585264444351196, "logps/chosen": -181.84434509277344, "logps/rejected": -238.97323608398438, "loss": 0.6159, "rewards/accuracies": 0.59375, "rewards/chosen": 0.0770760029554367, "rewards/margins": 0.2519676089286804, "rewards/rejected": -0.17489157617092133, "step": 214 }, { "epoch": 0.24631173829377806, "grad_norm": 96.37551956943565, "learning_rate": 1.9757671905175115e-07, "logits/chosen": -1.3951995372772217, "logits/rejected": -1.4096182584762573, "logps/chosen": -119.3514404296875, "logps/rejected": -131.9920196533203, "loss": 0.6021, "rewards/accuracies": 0.6875, "rewards/chosen": 0.2738375961780548, "rewards/margins": 0.2932060658931732, "rewards/rejected": -0.01936846226453781, "step": 216 }, { "epoch": 0.24859240253723897, "grad_norm": 90.01951430126202, "learning_rate": 1.9749279121818234e-07, "logits/chosen": -1.3084944486618042, "logits/rejected": -1.3836894035339355, "logps/chosen": -155.07765197753906, "logps/rejected": -191.67471313476562, "loss": 0.6953, "rewards/accuracies": 0.65625, "rewards/chosen": 0.12750835716724396, "rewards/margins": 0.31851378083229065, "rewards/rejected": -0.1910054087638855, "step": 218 }, { "epoch": 0.25087306678069987, "grad_norm": 107.56947817547281, "learning_rate": 1.9740745308420528e-07, "logits/chosen": -1.4428541660308838, "logits/rejected": -1.4522430896759033, "logps/chosen": -159.34136962890625, "logps/rejected": -201.03256225585938, "loss": 0.7085, "rewards/accuracies": 0.4375, "rewards/chosen": 0.08158425986766815, "rewards/margins": 0.062202394008636475, "rewards/rejected": 0.019381869584321976, "step": 220 }, { "epoch": 0.2531537310241608, "grad_norm": 83.84650346626758, "learning_rate": 1.9732070588429487e-07, "logits/chosen": -1.4249211549758911, "logits/rejected": -1.4040560722351074, "logps/chosen": -139.1720733642578, "logps/rejected": -150.09683227539062, "loss": 0.6528, "rewards/accuracies": 0.65625, "rewards/chosen": 0.25594449043273926, "rewards/margins": 0.27077823877334595, "rewards/rejected": -0.014833783730864525, "step": 222 }, { "epoch": 0.2554343952676217, "grad_norm": 95.84521399355921, "learning_rate": 1.972325508733091e-07, "logits/chosen": -1.4068944454193115, "logits/rejected": -1.4884459972381592, "logps/chosen": -178.6918182373047, "logps/rejected": -178.89633178710938, "loss": 0.6725, "rewards/accuracies": 0.6875, "rewards/chosen": 0.17113548517227173, "rewards/margins": 0.35012179613113403, "rewards/rejected": -0.1789863258600235, "step": 224 }, { "epoch": 0.2577150595110826, "grad_norm": 90.48418130257966, "learning_rate": 1.9714298932647098e-07, "logits/chosen": -1.3818295001983643, "logits/rejected": -1.4686387777328491, "logps/chosen": -203.4779510498047, "logps/rejected": -231.25755310058594, "loss": 0.6145, "rewards/accuracies": 0.75, "rewards/chosen": 0.2646425664424896, "rewards/margins": 0.30655670166015625, "rewards/rejected": -0.04191412031650543, "step": 226 }, { "epoch": 0.2599957237545435, "grad_norm": 96.02415026722944, "learning_rate": 1.9705202253935004e-07, "logits/chosen": -1.459298849105835, "logits/rejected": -1.4874234199523926, "logps/chosen": -190.43370056152344, "logps/rejected": -208.5654296875, "loss": 0.6259, "rewards/accuracies": 0.6875, "rewards/chosen": 0.2924637198448181, "rewards/margins": 0.3736059367656708, "rewards/rejected": -0.08114223182201385, "step": 228 }, { "epoch": 0.2622763879980044, "grad_norm": 94.75504400212658, "learning_rate": 1.9695965182784343e-07, "logits/chosen": -1.37540864944458, "logits/rejected": -1.4327397346496582, "logps/chosen": -158.9178924560547, "logps/rejected": -187.00572204589844, "loss": 0.6413, "rewards/accuracies": 0.65625, "rewards/chosen": 0.031393878161907196, "rewards/margins": 0.2354445904493332, "rewards/rejected": -0.2040507048368454, "step": 230 }, { "epoch": 0.2645570522414653, "grad_norm": 88.68651480780905, "learning_rate": 1.968658785281573e-07, "logits/chosen": -1.4091265201568604, "logits/rejected": -1.3750542402267456, "logps/chosen": -159.5924530029297, "logps/rejected": -157.25701904296875, "loss": 0.6956, "rewards/accuracies": 0.6875, "rewards/chosen": 0.05787750333547592, "rewards/margins": 0.07036175578832626, "rewards/rejected": -0.012484237551689148, "step": 232 }, { "epoch": 0.2668377164849262, "grad_norm": 87.80878700459785, "learning_rate": 1.9677070399678693e-07, "logits/chosen": -1.3906813859939575, "logits/rejected": -1.4185580015182495, "logps/chosen": -170.7882843017578, "logps/rejected": -192.6103515625, "loss": 0.6498, "rewards/accuracies": 0.625, "rewards/chosen": 0.026271583512425423, "rewards/margins": 0.08472327888011932, "rewards/rejected": -0.058451712131500244, "step": 234 }, { "epoch": 0.2691183807283871, "grad_norm": 95.5482100990242, "learning_rate": 1.9667412961049754e-07, "logits/chosen": -1.4015161991119385, "logits/rejected": -1.4669052362442017, "logps/chosen": -180.34584045410156, "logps/rejected": -205.79164123535156, "loss": 0.6991, "rewards/accuracies": 0.71875, "rewards/chosen": 0.12791365385055542, "rewards/margins": 0.28303825855255127, "rewards/rejected": -0.15512457489967346, "step": 236 }, { "epoch": 0.271399044971848, "grad_norm": 88.45630335400043, "learning_rate": 1.9657615676630418e-07, "logits/chosen": -1.2741804122924805, "logits/rejected": -1.2581013441085815, "logps/chosen": -184.23312377929688, "logps/rejected": -201.00103759765625, "loss": 0.6514, "rewards/accuracies": 0.75, "rewards/chosen": 0.036131352186203, "rewards/margins": 0.508301317691803, "rewards/rejected": -0.4721699059009552, "step": 238 }, { "epoch": 0.273679709215309, "grad_norm": 99.3871661426225, "learning_rate": 1.9647678688145159e-07, "logits/chosen": -1.327433466911316, "logits/rejected": -1.4025075435638428, "logps/chosen": -167.82005310058594, "logps/rejected": -194.77706909179688, "loss": 0.6544, "rewards/accuracies": 0.59375, "rewards/chosen": -0.01046083215624094, "rewards/margins": 0.2960438132286072, "rewards/rejected": -0.30650466680526733, "step": 240 }, { "epoch": 0.2759603734587699, "grad_norm": 129.18874275194366, "learning_rate": 1.9637602139339355e-07, "logits/chosen": -1.1904761791229248, "logits/rejected": -1.2964377403259277, "logps/chosen": -198.83749389648438, "logps/rejected": -213.1676025390625, "loss": 0.7065, "rewards/accuracies": 0.59375, "rewards/chosen": 0.018888656049966812, "rewards/margins": 0.0817704051733017, "rewards/rejected": -0.06288175284862518, "step": 242 }, { "epoch": 0.2782410377022308, "grad_norm": 78.46019369342993, "learning_rate": 1.9627386175977238e-07, "logits/chosen": -1.363703966140747, "logits/rejected": -1.385565161705017, "logps/chosen": -158.67543029785156, "logps/rejected": -171.18373107910156, "loss": 0.6328, "rewards/accuracies": 0.75, "rewards/chosen": 0.0427667461335659, "rewards/margins": 0.24535968899726868, "rewards/rejected": -0.20259293913841248, "step": 244 }, { "epoch": 0.2805217019456917, "grad_norm": 93.42637414011723, "learning_rate": 1.9617030945839747e-07, "logits/chosen": -1.4135775566101074, "logits/rejected": -1.482433557510376, "logps/chosen": -201.14869689941406, "logps/rejected": -240.91909790039062, "loss": 0.6572, "rewards/accuracies": 0.5625, "rewards/chosen": 0.14201030135154724, "rewards/margins": 0.3642757833003998, "rewards/rejected": -0.22226551175117493, "step": 246 }, { "epoch": 0.2828023661891526, "grad_norm": 94.8034192613356, "learning_rate": 1.9606536598722431e-07, "logits/chosen": -1.3897498846054077, "logits/rejected": -1.4480516910552979, "logps/chosen": -211.97976684570312, "logps/rejected": -213.59344482421875, "loss": 0.6947, "rewards/accuracies": 0.59375, "rewards/chosen": 0.1269560605287552, "rewards/margins": 0.18549999594688416, "rewards/rejected": -0.05854398384690285, "step": 248 }, { "epoch": 0.2850830304326135, "grad_norm": 107.48956854367718, "learning_rate": 1.9595903286433253e-07, "logits/chosen": -1.3192163705825806, "logits/rejected": -1.3453290462493896, "logps/chosen": -166.43765258789062, "logps/rejected": -185.8678436279297, "loss": 0.6446, "rewards/accuracies": 0.6875, "rewards/chosen": -0.22617535293102264, "rewards/margins": 0.19618834555149078, "rewards/rejected": -0.4223636984825134, "step": 250 }, { "epoch": 0.2873636946760744, "grad_norm": 97.18344743194635, "learning_rate": 1.9585131162790395e-07, "logits/chosen": -1.2871501445770264, "logits/rejected": -1.3303256034851074, "logps/chosen": -169.388671875, "logps/rejected": -222.89968872070312, "loss": 0.6597, "rewards/accuracies": 0.65625, "rewards/chosen": 0.016080807894468307, "rewards/margins": 0.22609557211399078, "rewards/rejected": -0.21001477539539337, "step": 252 }, { "epoch": 0.2896443589195353, "grad_norm": 102.3447949736702, "learning_rate": 1.9574220383620054e-07, "logits/chosen": -1.332809329032898, "logits/rejected": -1.3781144618988037, "logps/chosen": -192.14796447753906, "logps/rejected": -229.9279327392578, "loss": 0.6849, "rewards/accuracies": 0.4375, "rewards/chosen": -0.12542986869812012, "rewards/margins": 0.2946632206439972, "rewards/rejected": -0.4200930893421173, "step": 254 }, { "epoch": 0.29192502316299623, "grad_norm": 95.89945385250834, "learning_rate": 1.956317110675417e-07, "logits/chosen": -1.3515238761901855, "logits/rejected": -1.3907783031463623, "logps/chosen": -135.85316467285156, "logps/rejected": -168.83544921875, "loss": 0.6599, "rewards/accuracies": 0.625, "rewards/chosen": 0.2081318199634552, "rewards/margins": 0.3642672896385193, "rewards/rejected": -0.15613549947738647, "step": 256 }, { "epoch": 0.29420568740645714, "grad_norm": 104.38674521939737, "learning_rate": 1.9551983492028144e-07, "logits/chosen": -1.3251550197601318, "logits/rejected": -1.418371319770813, "logps/chosen": -172.90769958496094, "logps/rejected": -200.890380859375, "loss": 0.6987, "rewards/accuracies": 0.6875, "rewards/chosen": -0.051881127059459686, "rewards/margins": 0.18907570838928223, "rewards/rejected": -0.24095678329467773, "step": 258 }, { "epoch": 0.29648635164991805, "grad_norm": 110.83873967012946, "learning_rate": 1.9540657701278533e-07, "logits/chosen": -1.4099152088165283, "logits/rejected": -1.4348920583724976, "logps/chosen": -196.08592224121094, "logps/rejected": -237.6608428955078, "loss": 0.6704, "rewards/accuracies": 0.5, "rewards/chosen": -0.22441606223583221, "rewards/margins": 0.2391555905342102, "rewards/rejected": -0.4635716676712036, "step": 260 }, { "epoch": 0.29876701589337895, "grad_norm": 103.49844878423605, "learning_rate": 1.952919389834071e-07, "logits/chosen": -1.2992289066314697, "logits/rejected": -1.3971199989318848, "logps/chosen": -173.28128051757812, "logps/rejected": -213.29893493652344, "loss": 0.6847, "rewards/accuracies": 0.65625, "rewards/chosen": 0.19963645935058594, "rewards/margins": 0.11320970952510834, "rewards/rejected": 0.0864267572760582, "step": 262 }, { "epoch": 0.30104768013683986, "grad_norm": 99.13994519459824, "learning_rate": 1.9517592249046475e-07, "logits/chosen": -1.2953108549118042, "logits/rejected": -1.3212617635726929, "logps/chosen": -133.38534545898438, "logps/rejected": -153.64303588867188, "loss": 0.6807, "rewards/accuracies": 0.71875, "rewards/chosen": 0.2558751106262207, "rewards/margins": 0.31515762209892273, "rewards/rejected": -0.059282511472702026, "step": 264 }, { "epoch": 0.30332834438030076, "grad_norm": 105.16807067574956, "learning_rate": 1.950585292122169e-07, "logits/chosen": -1.380258560180664, "logits/rejected": -1.381796956062317, "logps/chosen": -265.73822021484375, "logps/rejected": -279.2653503417969, "loss": 0.6065, "rewards/accuracies": 0.65625, "rewards/chosen": 0.020402366295456886, "rewards/margins": 0.3811405599117279, "rewards/rejected": -0.36073821783065796, "step": 266 }, { "epoch": 0.30560900862376167, "grad_norm": 83.84465669976083, "learning_rate": 1.9493976084683813e-07, "logits/chosen": -1.3013784885406494, "logits/rejected": -1.3473682403564453, "logps/chosen": -143.68994140625, "logps/rejected": -161.75987243652344, "loss": 0.6565, "rewards/accuracies": 0.65625, "rewards/chosen": -0.09173958003520966, "rewards/margins": 0.3339024484157562, "rewards/rejected": -0.4256420433521271, "step": 268 }, { "epoch": 0.3078896728672226, "grad_norm": 97.4032150220662, "learning_rate": 1.9481961911239475e-07, "logits/chosen": -1.3531267642974854, "logits/rejected": -1.3918657302856445, "logps/chosen": -158.61968994140625, "logps/rejected": -196.80203247070312, "loss": 0.6872, "rewards/accuracies": 0.65625, "rewards/chosen": -0.018977202475070953, "rewards/margins": 0.31147533655166626, "rewards/rejected": -0.330452561378479, "step": 270 }, { "epoch": 0.3101703371106835, "grad_norm": 102.05098535340969, "learning_rate": 1.9469810574681968e-07, "logits/chosen": -1.2518314123153687, "logits/rejected": -1.373193383216858, "logps/chosen": -208.13992309570312, "logps/rejected": -265.2911376953125, "loss": 0.6599, "rewards/accuracies": 0.78125, "rewards/chosen": -0.10957180708646774, "rewards/margins": 0.4111281931400299, "rewards/rejected": -0.5207000374794006, "step": 272 }, { "epoch": 0.3124510013541444, "grad_norm": 101.2119365968094, "learning_rate": 1.9457522250788756e-07, "logits/chosen": -1.4733508825302124, "logits/rejected": -1.4335081577301025, "logps/chosen": -152.27394104003906, "logps/rejected": -149.08934020996094, "loss": 0.6512, "rewards/accuracies": 0.53125, "rewards/chosen": 0.0012259185314178467, "rewards/margins": 0.06475882232189178, "rewards/rejected": -0.06353290379047394, "step": 274 }, { "epoch": 0.3147316655976053, "grad_norm": 102.70187300991479, "learning_rate": 1.9445097117318912e-07, "logits/chosen": -1.2339739799499512, "logits/rejected": -1.2878998517990112, "logps/chosen": -154.39866638183594, "logps/rejected": -169.81304931640625, "loss": 0.6069, "rewards/accuracies": 0.59375, "rewards/chosen": -0.030073221772909164, "rewards/margins": 0.06755417585372925, "rewards/rejected": -0.09762738645076752, "step": 276 }, { "epoch": 0.3170123298410662, "grad_norm": 92.89132088481786, "learning_rate": 1.9432535354010545e-07, "logits/chosen": -1.381090521812439, "logits/rejected": -1.408646583557129, "logps/chosen": -193.48159790039062, "logps/rejected": -218.6549530029297, "loss": 0.721, "rewards/accuracies": 0.53125, "rewards/chosen": -0.018713481724262238, "rewards/margins": 0.14033427834510803, "rewards/rejected": -0.15904778242111206, "step": 278 }, { "epoch": 0.3192929940845271, "grad_norm": 92.57611013262112, "learning_rate": 1.9419837142578226e-07, "logits/chosen": -1.3027143478393555, "logits/rejected": -1.2925523519515991, "logps/chosen": -159.863525390625, "logps/rejected": -182.41249084472656, "loss": 0.6453, "rewards/accuracies": 0.71875, "rewards/chosen": 0.16124297678470612, "rewards/margins": 0.37421250343322754, "rewards/rejected": -0.21296954154968262, "step": 280 }, { "epoch": 0.321573658327988, "grad_norm": 88.34893118769534, "learning_rate": 1.9407002666710333e-07, "logits/chosen": -1.2514938116073608, "logits/rejected": -1.2840971946716309, "logps/chosen": -167.23568725585938, "logps/rejected": -185.19515991210938, "loss": 0.6776, "rewards/accuracies": 0.65625, "rewards/chosen": 0.0749707818031311, "rewards/margins": 0.2606170177459717, "rewards/rejected": -0.18564626574516296, "step": 282 }, { "epoch": 0.3238543225714489, "grad_norm": 87.77224347782207, "learning_rate": 1.93940321120664e-07, "logits/chosen": -1.234694242477417, "logits/rejected": -1.245900273323059, "logps/chosen": -181.2189178466797, "logps/rejected": -201.54295349121094, "loss": 0.656, "rewards/accuracies": 0.59375, "rewards/chosen": -0.03384766727685928, "rewards/margins": 0.15053534507751465, "rewards/rejected": -0.18438303470611572, "step": 284 }, { "epoch": 0.3261349868149098, "grad_norm": 102.48592883969225, "learning_rate": 1.9380925666274443e-07, "logits/chosen": -1.29789400100708, "logits/rejected": -1.3402178287506104, "logps/chosen": -200.9523468017578, "logps/rejected": -274.8287658691406, "loss": 0.6817, "rewards/accuracies": 0.625, "rewards/chosen": -0.21383914351463318, "rewards/margins": 0.0331435352563858, "rewards/rejected": -0.24698270857334137, "step": 286 }, { "epoch": 0.32841565105837073, "grad_norm": 98.03735612419584, "learning_rate": 1.9367683518928225e-07, "logits/chosen": -1.3940753936767578, "logits/rejected": -1.464486837387085, "logps/chosen": -124.9957504272461, "logps/rejected": -153.5186767578125, "loss": 0.6812, "rewards/accuracies": 0.53125, "rewards/chosen": -0.005818367004394531, "rewards/margins": 0.20061178505420685, "rewards/rejected": -0.20643013715744019, "step": 288 }, { "epoch": 0.33069631530183163, "grad_norm": 103.78616077769101, "learning_rate": 1.935430586158454e-07, "logits/chosen": -1.1263610124588013, "logits/rejected": -1.1101529598236084, "logps/chosen": -117.66938018798828, "logps/rejected": -164.22933959960938, "loss": 0.6884, "rewards/accuracies": 0.65625, "rewards/chosen": -0.0009657144546508789, "rewards/margins": 0.10708783566951752, "rewards/rejected": -0.1080535426735878, "step": 290 }, { "epoch": 0.33297697954529254, "grad_norm": 110.5063902175955, "learning_rate": 1.9340792887760412e-07, "logits/chosen": -1.324884295463562, "logits/rejected": -1.3702059984207153, "logps/chosen": -194.0547332763672, "logps/rejected": -223.9933624267578, "loss": 0.6334, "rewards/accuracies": 0.5625, "rewards/chosen": -0.0599634051322937, "rewards/margins": 0.17343786358833313, "rewards/rejected": -0.23340126872062683, "step": 292 }, { "epoch": 0.3352576437887535, "grad_norm": 85.87924361401205, "learning_rate": 1.932714479293033e-07, "logits/chosen": -1.2896180152893066, "logits/rejected": -1.3805060386657715, "logps/chosen": -182.65460205078125, "logps/rejected": -212.5992889404297, "loss": 0.6151, "rewards/accuracies": 0.65625, "rewards/chosen": -0.10547913610935211, "rewards/margins": 0.20131757855415344, "rewards/rejected": -0.30679669976234436, "step": 294 }, { "epoch": 0.3375383080322144, "grad_norm": 108.47177643694172, "learning_rate": 1.9313361774523386e-07, "logits/chosen": -1.4138604402542114, "logits/rejected": -1.402896523475647, "logps/chosen": -147.80795288085938, "logps/rejected": -165.24566650390625, "loss": 0.7377, "rewards/accuracies": 0.5, "rewards/chosen": -0.21476852893829346, "rewards/margins": 0.08000420778989792, "rewards/rejected": -0.2947727143764496, "step": 296 }, { "epoch": 0.3398189722756753, "grad_norm": 111.8896748854085, "learning_rate": 1.9299444031920436e-07, "logits/chosen": -1.2892051935195923, "logits/rejected": -1.380386233329773, "logps/chosen": -190.3002471923828, "logps/rejected": -246.29263305664062, "loss": 0.6347, "rewards/accuracies": 0.625, "rewards/chosen": -0.08067283034324646, "rewards/margins": 0.4462183713912964, "rewards/rejected": -0.5268911719322205, "step": 298 }, { "epoch": 0.3420996365191362, "grad_norm": 96.80869969762502, "learning_rate": 1.9285391766451217e-07, "logits/chosen": -1.3364348411560059, "logits/rejected": -1.3174083232879639, "logps/chosen": -161.74612426757812, "logps/rejected": -179.8140106201172, "loss": 0.669, "rewards/accuracies": 0.59375, "rewards/chosen": -0.1118321567773819, "rewards/margins": 0.23199215531349182, "rewards/rejected": -0.34382426738739014, "step": 300 }, { "epoch": 0.3420996365191362, "eval_logits/chosen": -1.4281266927719116, "eval_logits/rejected": -1.4122414588928223, "eval_logps/chosen": -128.86663818359375, "eval_logps/rejected": -131.4814910888672, "eval_loss": 0.6797733306884766, "eval_rewards/accuracies": 0.5199999809265137, "eval_rewards/chosen": 0.10313291847705841, "eval_rewards/margins": 0.045006848871707916, "eval_rewards/rejected": 0.05812607705593109, "eval_runtime": 21.8566, "eval_samples_per_second": 4.575, "eval_steps_per_second": 1.144, "step": 300 }, { "epoch": 0.3443803007625971, "grad_norm": 84.25578372291146, "learning_rate": 1.927120518139144e-07, "logits/chosen": -1.3597999811172485, "logits/rejected": -1.4241136312484741, "logps/chosen": -180.1940460205078, "logps/rejected": -199.74412536621094, "loss": 0.6421, "rewards/accuracies": 0.59375, "rewards/chosen": -0.04787690192461014, "rewards/margins": 0.13789615035057068, "rewards/rejected": -0.18577302992343903, "step": 302 }, { "epoch": 0.34666096500605803, "grad_norm": 89.24528063714898, "learning_rate": 1.925688448195983e-07, "logits/chosen": -1.1900302171707153, "logits/rejected": -1.3024637699127197, "logps/chosen": -130.7202606201172, "logps/rejected": -170.3651123046875, "loss": 0.6326, "rewards/accuracies": 0.71875, "rewards/chosen": 0.08976805210113525, "rewards/margins": 0.30762431025505066, "rewards/rejected": -0.2178562879562378, "step": 304 }, { "epoch": 0.34894162924951894, "grad_norm": 86.12361588594867, "learning_rate": 1.924242987531517e-07, "logits/chosen": -1.3129158020019531, "logits/rejected": -1.2686495780944824, "logps/chosen": -170.01808166503906, "logps/rejected": -172.82835388183594, "loss": 0.6621, "rewards/accuracies": 0.59375, "rewards/chosen": -0.010694226250052452, "rewards/margins": 0.25143933296203613, "rewards/rejected": -0.26213353872299194, "step": 306 }, { "epoch": 0.35122229349297984, "grad_norm": 80.23753458776699, "learning_rate": 1.922784157055331e-07, "logits/chosen": -1.3715269565582275, "logits/rejected": -1.3273383378982544, "logps/chosen": -190.35610961914062, "logps/rejected": -180.68020629882812, "loss": 0.6522, "rewards/accuracies": 0.625, "rewards/chosen": -0.09160695225000381, "rewards/margins": 0.13991902768611908, "rewards/rejected": -0.2315259724855423, "step": 308 }, { "epoch": 0.35350295773644075, "grad_norm": 98.71267017165292, "learning_rate": 1.9213119778704127e-07, "logits/chosen": -1.3962262868881226, "logits/rejected": -1.4428459405899048, "logps/chosen": -222.09481811523438, "logps/rejected": -248.44686889648438, "loss": 0.6602, "rewards/accuracies": 0.78125, "rewards/chosen": -0.001766495406627655, "rewards/margins": 0.569767415523529, "rewards/rejected": -0.5715339183807373, "step": 310 }, { "epoch": 0.35578362197990165, "grad_norm": 88.65548046490639, "learning_rate": 1.919826471272849e-07, "logits/chosen": -1.3152942657470703, "logits/rejected": -1.365240216255188, "logps/chosen": -128.6586151123047, "logps/rejected": -141.93685913085938, "loss": 0.6252, "rewards/accuracies": 0.59375, "rewards/chosen": -0.05175452306866646, "rewards/margins": 0.1226281225681305, "rewards/rejected": -0.17438265681266785, "step": 312 }, { "epoch": 0.35806428622336256, "grad_norm": 84.35369248301531, "learning_rate": 1.9183276587515165e-07, "logits/chosen": -1.2944692373275757, "logits/rejected": -1.3323063850402832, "logps/chosen": -165.85975646972656, "logps/rejected": -191.24041748046875, "loss": 0.6487, "rewards/accuracies": 0.53125, "rewards/chosen": -0.07458823919296265, "rewards/margins": 0.3784411549568176, "rewards/rejected": -0.4530293941497803, "step": 314 }, { "epoch": 0.36034495046682347, "grad_norm": 86.49341842550122, "learning_rate": 1.9168155619877707e-07, "logits/chosen": -1.2946043014526367, "logits/rejected": -1.4176356792449951, "logps/chosen": -130.90985107421875, "logps/rejected": -162.9954071044922, "loss": 0.6467, "rewards/accuracies": 0.78125, "rewards/chosen": 0.011317873373627663, "rewards/margins": 0.4069710373878479, "rewards/rejected": -0.39565321803092957, "step": 316 }, { "epoch": 0.36262561471028437, "grad_norm": 102.96541053849566, "learning_rate": 1.9152902028551335e-07, "logits/chosen": -1.3063163757324219, "logits/rejected": -1.3172858953475952, "logps/chosen": -157.0022735595703, "logps/rejected": -165.20347595214844, "loss": 0.6669, "rewards/accuracies": 0.59375, "rewards/chosen": -0.07298893481492996, "rewards/margins": 0.17434003949165344, "rewards/rejected": -0.2473289668560028, "step": 318 }, { "epoch": 0.3649062789537453, "grad_norm": 76.39528252287091, "learning_rate": 1.9137516034189765e-07, "logits/chosen": -1.3794448375701904, "logits/rejected": -1.4233444929122925, "logps/chosen": -164.51431274414062, "logps/rejected": -191.4024658203125, "loss": 0.629, "rewards/accuracies": 0.71875, "rewards/chosen": 0.07598446309566498, "rewards/margins": 0.355991005897522, "rewards/rejected": -0.2800065577030182, "step": 320 }, { "epoch": 0.3671869431972062, "grad_norm": 90.58917607125439, "learning_rate": 1.9121997859361997e-07, "logits/chosen": -1.3726584911346436, "logits/rejected": -1.4218438863754272, "logps/chosen": -161.8759765625, "logps/rejected": -191.62745666503906, "loss": 0.6777, "rewards/accuracies": 0.59375, "rewards/chosen": -0.03798535838723183, "rewards/margins": 0.2353454828262329, "rewards/rejected": -0.27333083748817444, "step": 322 }, { "epoch": 0.3694676074406671, "grad_norm": 88.70829690205912, "learning_rate": 1.9106347728549132e-07, "logits/chosen": -1.4393202066421509, "logits/rejected": -1.4652281999588013, "logps/chosen": -159.0114288330078, "logps/rejected": -169.2324981689453, "loss": 0.6457, "rewards/accuracies": 0.59375, "rewards/chosen": -0.03970104455947876, "rewards/margins": 0.19649550318717957, "rewards/rejected": -0.23619654774665833, "step": 324 }, { "epoch": 0.371748271684128, "grad_norm": 97.10274345850128, "learning_rate": 1.9090565868141093e-07, "logits/chosen": -1.415880560874939, "logits/rejected": -1.4168412685394287, "logps/chosen": -200.85858154296875, "logps/rejected": -214.74472045898438, "loss": 0.6936, "rewards/accuracies": 0.5, "rewards/chosen": -0.0973915159702301, "rewards/margins": -0.0053062885999679565, "rewards/rejected": -0.09208524972200394, "step": 326 }, { "epoch": 0.3740289359275889, "grad_norm": 79.53686855879208, "learning_rate": 1.9074652506433367e-07, "logits/chosen": -1.2665050029754639, "logits/rejected": -1.2990330457687378, "logps/chosen": -188.47377014160156, "logps/rejected": -196.28854370117188, "loss": 0.6894, "rewards/accuracies": 0.53125, "rewards/chosen": -0.2655767500400543, "rewards/margins": 0.0786442756652832, "rewards/rejected": -0.3442210257053375, "step": 328 }, { "epoch": 0.3763096001710498, "grad_norm": 92.25716890573669, "learning_rate": 1.9058607873623696e-07, "logits/chosen": -1.2212181091308594, "logits/rejected": -1.2750239372253418, "logps/chosen": -136.28663635253906, "logps/rejected": -182.32699584960938, "loss": 0.6309, "rewards/accuracies": 0.5, "rewards/chosen": -0.14970597624778748, "rewards/margins": 0.13012784719467163, "rewards/rejected": -0.2798338234424591, "step": 330 }, { "epoch": 0.3785902644145107, "grad_norm": 96.38565738306772, "learning_rate": 1.9042432201808753e-07, "logits/chosen": -1.3811233043670654, "logits/rejected": -1.3999402523040771, "logps/chosen": -153.0084686279297, "logps/rejected": -174.6513671875, "loss": 0.6334, "rewards/accuracies": 0.5625, "rewards/chosen": -0.06558946520090103, "rewards/margins": 0.20774412155151367, "rewards/rejected": -0.2733335494995117, "step": 332 }, { "epoch": 0.3808709286579716, "grad_norm": 101.76003883417016, "learning_rate": 1.9026125724980777e-07, "logits/chosen": -1.3316706418991089, "logits/rejected": -1.2979425191879272, "logps/chosen": -172.97792053222656, "logps/rejected": -187.61904907226562, "loss": 0.635, "rewards/accuracies": 0.65625, "rewards/chosen": 0.022555537521839142, "rewards/margins": 0.31808096170425415, "rewards/rejected": -0.2955254316329956, "step": 334 }, { "epoch": 0.3831515929014325, "grad_norm": 89.57559954819274, "learning_rate": 1.900968867902419e-07, "logits/chosen": -1.392217993736267, "logits/rejected": -1.4426984786987305, "logps/chosen": -171.40127563476562, "logps/rejected": -171.21238708496094, "loss": 0.6236, "rewards/accuracies": 0.65625, "rewards/chosen": 0.11336915194988251, "rewards/margins": 0.3153633177280426, "rewards/rejected": -0.2019941657781601, "step": 336 }, { "epoch": 0.38543225714489343, "grad_norm": 95.74770834081339, "learning_rate": 1.899312130171219e-07, "logits/chosen": -1.410531759262085, "logits/rejected": -1.4363325834274292, "logps/chosen": -176.63296508789062, "logps/rejected": -194.51480102539062, "loss": 0.6307, "rewards/accuracies": 0.65625, "rewards/chosen": -0.0749322921037674, "rewards/margins": 0.2967851161956787, "rewards/rejected": -0.3717173933982849, "step": 338 }, { "epoch": 0.38771292138835434, "grad_norm": 94.33644084218925, "learning_rate": 1.897642383270331e-07, "logits/chosen": -1.4411017894744873, "logits/rejected": -1.5141332149505615, "logps/chosen": -220.8849639892578, "logps/rejected": -249.29640197753906, "loss": 0.6396, "rewards/accuracies": 0.625, "rewards/chosen": -0.1280645728111267, "rewards/margins": 0.4015355110168457, "rewards/rejected": -0.5296000838279724, "step": 340 }, { "epoch": 0.38999358563181524, "grad_norm": 93.41769187789713, "learning_rate": 1.8959596513537937e-07, "logits/chosen": -1.4379459619522095, "logits/rejected": -1.463719367980957, "logps/chosen": -136.88717651367188, "logps/rejected": -146.65672302246094, "loss": 0.6624, "rewards/accuracies": 0.59375, "rewards/chosen": 0.024333830922842026, "rewards/margins": 0.15197159349918365, "rewards/rejected": -0.12763777375221252, "step": 342 }, { "epoch": 0.39227424987527615, "grad_norm": 110.48154958430904, "learning_rate": 1.894263958763485e-07, "logits/chosen": -1.477798581123352, "logits/rejected": -1.4197784662246704, "logps/chosen": -192.24853515625, "logps/rejected": -182.48162841796875, "loss": 0.6866, "rewards/accuracies": 0.625, "rewards/chosen": 0.029505489394068718, "rewards/margins": 0.263489693403244, "rewards/rejected": -0.23398420214653015, "step": 344 }, { "epoch": 0.39455491411873705, "grad_norm": 120.46352804291743, "learning_rate": 1.892555330028766e-07, "logits/chosen": -1.2521153688430786, "logits/rejected": -1.2826614379882812, "logps/chosen": -217.5478515625, "logps/rejected": -273.1899719238281, "loss": 0.6418, "rewards/accuracies": 0.65625, "rewards/chosen": -0.2787259817123413, "rewards/margins": 0.3908158242702484, "rewards/rejected": -0.6695418357849121, "step": 346 }, { "epoch": 0.396835578362198, "grad_norm": 94.24036369589392, "learning_rate": 1.8908337898661285e-07, "logits/chosen": -1.3746845722198486, "logits/rejected": -1.46570885181427, "logps/chosen": -179.17138671875, "logps/rejected": -217.95236206054688, "loss": 0.6201, "rewards/accuracies": 0.75, "rewards/chosen": -0.12253744155168533, "rewards/margins": 0.39310359954833984, "rewards/rejected": -0.5156410336494446, "step": 348 }, { "epoch": 0.3991162426056589, "grad_norm": 88.9093668290791, "learning_rate": 1.889099363178838e-07, "logits/chosen": -1.445816993713379, "logits/rejected": -1.4381108283996582, "logps/chosen": -132.04539489746094, "logps/rejected": -139.59835815429688, "loss": 0.6502, "rewards/accuracies": 0.625, "rewards/chosen": -0.10454928129911423, "rewards/margins": 0.15687233209609985, "rewards/rejected": -0.2614216208457947, "step": 350 }, { "epoch": 0.40139690684911983, "grad_norm": 118.96237042903994, "learning_rate": 1.8873520750565714e-07, "logits/chosen": -1.3293050527572632, "logits/rejected": -1.3804844617843628, "logps/chosen": -172.6136474609375, "logps/rejected": -184.08212280273438, "loss": 0.6857, "rewards/accuracies": 0.53125, "rewards/chosen": -0.09095679968595505, "rewards/margins": 0.1739913374185562, "rewards/rejected": -0.26494812965393066, "step": 352 }, { "epoch": 0.40367757109258073, "grad_norm": 123.21030025798191, "learning_rate": 1.8855919507750556e-07, "logits/chosen": -1.299286127090454, "logits/rejected": -1.2867302894592285, "logps/chosen": -228.6288604736328, "logps/rejected": -253.30572509765625, "loss": 0.6819, "rewards/accuracies": 0.53125, "rewards/chosen": -0.27850618958473206, "rewards/margins": 0.12168295681476593, "rewards/rejected": -0.40018919110298157, "step": 354 }, { "epoch": 0.40595823533604164, "grad_norm": 93.75103343413197, "learning_rate": 1.883819015795701e-07, "logits/chosen": -1.3721405267715454, "logits/rejected": -1.4811532497406006, "logps/chosen": -158.45986938476562, "logps/rejected": -191.99571228027344, "loss": 0.6742, "rewards/accuracies": 0.5625, "rewards/chosen": 0.06369534879922867, "rewards/margins": 0.13870595395565033, "rewards/rejected": -0.07501061260700226, "step": 356 }, { "epoch": 0.40823889957950255, "grad_norm": 102.06521296332116, "learning_rate": 1.8820332957652343e-07, "logits/chosen": -1.3636136054992676, "logits/rejected": -1.4187768697738647, "logps/chosen": -145.48468017578125, "logps/rejected": -148.94664001464844, "loss": 0.6414, "rewards/accuracies": 0.59375, "rewards/chosen": 0.0663696825504303, "rewards/margins": 0.29655539989471436, "rewards/rejected": -0.23018570244312286, "step": 358 }, { "epoch": 0.41051956382296345, "grad_norm": 80.21120922967526, "learning_rate": 1.8802348165153257e-07, "logits/chosen": -1.0907145738601685, "logits/rejected": -1.2451905012130737, "logps/chosen": -140.98895263671875, "logps/rejected": -169.56390380859375, "loss": 0.6267, "rewards/accuracies": 0.625, "rewards/chosen": 0.1324671506881714, "rewards/margins": 0.2304615080356598, "rewards/rejected": -0.0979943573474884, "step": 360 }, { "epoch": 0.41280022806642436, "grad_norm": 90.74971301077287, "learning_rate": 1.8784236040622173e-07, "logits/chosen": -1.4046244621276855, "logits/rejected": -1.5294733047485352, "logps/chosen": -220.828125, "logps/rejected": -252.51620483398438, "loss": 0.6485, "rewards/accuracies": 0.5, "rewards/chosen": -0.029995199292898178, "rewards/margins": 0.10216841101646423, "rewards/rejected": -0.1321636140346527, "step": 362 }, { "epoch": 0.41508089230988526, "grad_norm": 116.10619062475209, "learning_rate": 1.8765996846063453e-07, "logits/chosen": -1.3053789138793945, "logits/rejected": -1.331100344657898, "logps/chosen": -145.552978515625, "logps/rejected": -175.96580505371094, "loss": 0.7355, "rewards/accuracies": 0.46875, "rewards/chosen": -0.08064378052949905, "rewards/margins": 0.18256865441799164, "rewards/rejected": -0.2632124125957489, "step": 364 }, { "epoch": 0.41736155655334617, "grad_norm": 85.92321632812764, "learning_rate": 1.8747630845319612e-07, "logits/chosen": -1.3430732488632202, "logits/rejected": -1.321002721786499, "logps/chosen": -149.8716278076172, "logps/rejected": -168.82972717285156, "loss": 0.6414, "rewards/accuracies": 0.6875, "rewards/chosen": 0.08440528810024261, "rewards/margins": 0.3168274164199829, "rewards/rejected": -0.23242215812206268, "step": 366 }, { "epoch": 0.4196422207968071, "grad_norm": 89.28728181459165, "learning_rate": 1.8729138304067512e-07, "logits/chosen": -1.464853286743164, "logits/rejected": -1.478898286819458, "logps/chosen": -157.5703887939453, "logps/rejected": -171.20423889160156, "loss": 0.6186, "rewards/accuracies": 0.625, "rewards/chosen": 0.0049234069883823395, "rewards/margins": 0.2588205635547638, "rewards/rejected": -0.25389716029167175, "step": 368 }, { "epoch": 0.421922885040268, "grad_norm": 86.83502819065235, "learning_rate": 1.8710519489814503e-07, "logits/chosen": -1.2088496685028076, "logits/rejected": -1.238593578338623, "logps/chosen": -156.05690002441406, "logps/rejected": -176.14474487304688, "loss": 0.7105, "rewards/accuracies": 0.5625, "rewards/chosen": -0.01236521452665329, "rewards/margins": 0.17280086874961853, "rewards/rejected": -0.18516604602336884, "step": 370 }, { "epoch": 0.4242035492837289, "grad_norm": 101.40852416620743, "learning_rate": 1.869177467189456e-07, "logits/chosen": -1.4426904916763306, "logits/rejected": -1.409334659576416, "logps/chosen": -182.2246551513672, "logps/rejected": -194.63980102539062, "loss": 0.6937, "rewards/accuracies": 0.5625, "rewards/chosen": -0.13504472374916077, "rewards/margins": 0.12622104585170746, "rewards/rejected": -0.2612657845020294, "step": 372 }, { "epoch": 0.4264842135271898, "grad_norm": 78.92812775165392, "learning_rate": 1.8672904121464402e-07, "logits/chosen": -1.3653035163879395, "logits/rejected": -1.3275575637817383, "logps/chosen": -180.02243041992188, "logps/rejected": -195.43106079101562, "loss": 0.6208, "rewards/accuracies": 0.6875, "rewards/chosen": -0.13813042640686035, "rewards/margins": 0.25349733233451843, "rewards/rejected": -0.3916277587413788, "step": 374 }, { "epoch": 0.4287648777706507, "grad_norm": 85.95171696022348, "learning_rate": 1.8653908111499531e-07, "logits/chosen": -1.4379856586456299, "logits/rejected": -1.5062627792358398, "logps/chosen": -123.3829345703125, "logps/rejected": -155.55897521972656, "loss": 0.6752, "rewards/accuracies": 0.59375, "rewards/chosen": -0.02849392406642437, "rewards/margins": 0.1816612333059311, "rewards/rejected": -0.2101551592350006, "step": 376 }, { "epoch": 0.4310455420141116, "grad_norm": 74.55363107506629, "learning_rate": 1.8634786916790331e-07, "logits/chosen": -1.306863784790039, "logits/rejected": -1.4182804822921753, "logps/chosen": -185.36831665039062, "logps/rejected": -205.2194366455078, "loss": 0.602, "rewards/accuracies": 0.65625, "rewards/chosen": 0.23371680080890656, "rewards/margins": 0.4712272882461548, "rewards/rejected": -0.23751050233840942, "step": 378 }, { "epoch": 0.4333262062575725, "grad_norm": 98.09376209675088, "learning_rate": 1.861554081393806e-07, "logits/chosen": -1.298234462738037, "logits/rejected": -1.320231556892395, "logps/chosen": -170.1300506591797, "logps/rejected": -201.2412567138672, "loss": 0.6166, "rewards/accuracies": 0.625, "rewards/chosen": 0.11016674339771271, "rewards/margins": 0.3684804141521454, "rewards/rejected": -0.25831368565559387, "step": 380 }, { "epoch": 0.4356068705010334, "grad_norm": 95.75031333883096, "learning_rate": 1.8596170081350855e-07, "logits/chosen": -1.4295341968536377, "logits/rejected": -1.4480311870574951, "logps/chosen": -158.36959838867188, "logps/rejected": -185.86700439453125, "loss": 0.6506, "rewards/accuracies": 0.5625, "rewards/chosen": 0.11881276965141296, "rewards/margins": 0.2408534288406372, "rewards/rejected": -0.12204068899154663, "step": 382 }, { "epoch": 0.4378875347444943, "grad_norm": 89.97741935605185, "learning_rate": 1.8576674999239713e-07, "logits/chosen": -1.2293694019317627, "logits/rejected": -1.319061040878296, "logps/chosen": -168.48960876464844, "logps/rejected": -213.40200805664062, "loss": 0.6379, "rewards/accuracies": 0.71875, "rewards/chosen": 0.07465538382530212, "rewards/margins": 0.39287707209587097, "rewards/rejected": -0.31822171807289124, "step": 384 }, { "epoch": 0.44016819898795523, "grad_norm": 84.57348322808303, "learning_rate": 1.8557055849614428e-07, "logits/chosen": -1.2857288122177124, "logits/rejected": -1.317692518234253, "logps/chosen": -195.7303009033203, "logps/rejected": -239.7130126953125, "loss": 0.5716, "rewards/accuracies": 0.65625, "rewards/chosen": -0.07659934461116791, "rewards/margins": 0.5603799819946289, "rewards/rejected": -0.636979341506958, "step": 386 }, { "epoch": 0.44244886323141613, "grad_norm": 85.15633806373397, "learning_rate": 1.8537312916279523e-07, "logits/chosen": -1.3817092180252075, "logits/rejected": -1.501309871673584, "logps/chosen": -146.41033935546875, "logps/rejected": -171.60218811035156, "loss": 0.6673, "rewards/accuracies": 0.59375, "rewards/chosen": -0.07926470041275024, "rewards/margins": 0.26870349049568176, "rewards/rejected": -0.347968190908432, "step": 388 }, { "epoch": 0.44472952747487704, "grad_norm": 89.75323963715459, "learning_rate": 1.8517446484830136e-07, "logits/chosen": -1.4567725658416748, "logits/rejected": -1.5358428955078125, "logps/chosen": -172.92111206054688, "logps/rejected": -192.05184936523438, "loss": 0.6553, "rewards/accuracies": 0.65625, "rewards/chosen": -0.1497884839773178, "rewards/margins": 0.06301924586296082, "rewards/rejected": -0.21280772984027863, "step": 390 }, { "epoch": 0.44701019171833795, "grad_norm": 105.51876340320463, "learning_rate": 1.8497456842647878e-07, "logits/chosen": -1.2944555282592773, "logits/rejected": -1.3529748916625977, "logps/chosen": -157.77642822265625, "logps/rejected": -176.93093872070312, "loss": 0.682, "rewards/accuracies": 0.5625, "rewards/chosen": -0.07403340190649033, "rewards/margins": 0.10720521211624146, "rewards/rejected": -0.18123860657215118, "step": 392 }, { "epoch": 0.44929085596179885, "grad_norm": 85.63092920252892, "learning_rate": 1.8477344278896706e-07, "logits/chosen": -1.4767736196517944, "logits/rejected": -1.5019617080688477, "logps/chosen": -173.24639892578125, "logps/rejected": -193.271240234375, "loss": 0.6935, "rewards/accuracies": 0.65625, "rewards/chosen": 0.13190394639968872, "rewards/margins": 0.48726946115493774, "rewards/rejected": -0.355365514755249, "step": 394 }, { "epoch": 0.45157152020525976, "grad_norm": 100.4356452029139, "learning_rate": 1.8457109084518718e-07, "logits/chosen": -1.4190278053283691, "logits/rejected": -1.3654460906982422, "logps/chosen": -227.8990020751953, "logps/rejected": -215.39312744140625, "loss": 0.6697, "rewards/accuracies": 0.6875, "rewards/chosen": -0.15163551270961761, "rewards/margins": 0.19186191260814667, "rewards/rejected": -0.3434974253177643, "step": 396 }, { "epoch": 0.45385218444872066, "grad_norm": 89.70522326638444, "learning_rate": 1.8436751552229938e-07, "logits/chosen": -1.3089743852615356, "logits/rejected": -1.340443730354309, "logps/chosen": -145.11398315429688, "logps/rejected": -172.60635375976562, "loss": 0.6571, "rewards/accuracies": 0.75, "rewards/chosen": -0.20591212809085846, "rewards/margins": 0.34871169924736023, "rewards/rejected": -0.5546237230300903, "step": 398 }, { "epoch": 0.4561328486921816, "grad_norm": 84.89862446594006, "learning_rate": 1.84162719765161e-07, "logits/chosen": -1.2627918720245361, "logits/rejected": -1.2893693447113037, "logps/chosen": -130.3240966796875, "logps/rejected": -160.49227905273438, "loss": 0.6402, "rewards/accuracies": 0.78125, "rewards/chosen": 0.0029626190662384033, "rewards/margins": 0.4357815384864807, "rewards/rejected": -0.4328188896179199, "step": 400 }, { "epoch": 0.4561328486921816, "eval_logits/chosen": -1.4406086206436157, "eval_logits/rejected": -1.4253712892532349, "eval_logps/chosen": -129.20408630371094, "eval_logps/rejected": -132.1772003173828, "eval_loss": 0.659533679485321, "eval_rewards/accuracies": 0.6399999856948853, "eval_rewards/chosen": 0.06938672810792923, "eval_rewards/margins": 0.08083352446556091, "eval_rewards/rejected": -0.011446798220276833, "eval_runtime": 21.3693, "eval_samples_per_second": 4.68, "eval_steps_per_second": 1.17, "step": 400 }, { "epoch": 0.45841351293564253, "grad_norm": 86.51322437777803, "learning_rate": 1.839567065362838e-07, "logits/chosen": -1.368638038635254, "logits/rejected": -1.3851830959320068, "logps/chosen": -169.3848876953125, "logps/rejected": -175.4376678466797, "loss": 0.635, "rewards/accuracies": 0.625, "rewards/chosen": -0.0501415841281414, "rewards/margins": 0.1427716761827469, "rewards/rejected": -0.1929132342338562, "step": 402 }, { "epoch": 0.46069417717910344, "grad_norm": 101.82366291792587, "learning_rate": 1.8374947881579112e-07, "logits/chosen": -1.201027512550354, "logits/rejected": -1.2725883722305298, "logps/chosen": -191.0504150390625, "logps/rejected": -233.1793212890625, "loss": 0.6195, "rewards/accuracies": 0.625, "rewards/chosen": -0.2396811544895172, "rewards/margins": 0.2062177062034607, "rewards/rejected": -0.4458988606929779, "step": 404 }, { "epoch": 0.46297484142256434, "grad_norm": 108.6063244513385, "learning_rate": 1.8354103960137473e-07, "logits/chosen": -1.4036638736724854, "logits/rejected": -1.4488118886947632, "logps/chosen": -230.92767333984375, "logps/rejected": -239.439208984375, "loss": 0.6795, "rewards/accuracies": 0.6875, "rewards/chosen": -0.32087260484695435, "rewards/margins": 0.10698382556438446, "rewards/rejected": -0.4278564453125, "step": 406 }, { "epoch": 0.46525550566602525, "grad_norm": 86.59174830548812, "learning_rate": 1.833313919082515e-07, "logits/chosen": -1.1800165176391602, "logits/rejected": -1.3113179206848145, "logps/chosen": -161.62985229492188, "logps/rejected": -202.1416473388672, "loss": 0.6378, "rewards/accuracies": 0.65625, "rewards/chosen": -0.14624057710170746, "rewards/margins": 0.3915923237800598, "rewards/rejected": -0.5378329157829285, "step": 408 }, { "epoch": 0.46753616990948615, "grad_norm": 92.36447587751275, "learning_rate": 1.8312053876911977e-07, "logits/chosen": -1.360296368598938, "logits/rejected": -1.3766343593597412, "logps/chosen": -203.7003173828125, "logps/rejected": -233.6885223388672, "loss": 0.6525, "rewards/accuracies": 0.5, "rewards/chosen": -0.13416332006454468, "rewards/margins": 0.36563166975975037, "rewards/rejected": -0.49979501962661743, "step": 410 }, { "epoch": 0.46981683415294706, "grad_norm": 79.86005120772512, "learning_rate": 1.8290848323411553e-07, "logits/chosen": -1.4597082138061523, "logits/rejected": -1.4648618698120117, "logps/chosen": -151.07449340820312, "logps/rejected": -151.00494384765625, "loss": 0.7101, "rewards/accuracies": 0.65625, "rewards/chosen": 0.05570954829454422, "rewards/margins": 0.08499263972043991, "rewards/rejected": -0.02928308956325054, "step": 412 }, { "epoch": 0.47209749839640797, "grad_norm": 70.72545574881394, "learning_rate": 1.8269522837076817e-07, "logits/chosen": -1.4345017671585083, "logits/rejected": -1.498835802078247, "logps/chosen": -180.71951293945312, "logps/rejected": -195.18490600585938, "loss": 0.6126, "rewards/accuracies": 0.5625, "rewards/chosen": -0.060558006167411804, "rewards/margins": 0.23918625712394714, "rewards/rejected": -0.29974424839019775, "step": 414 }, { "epoch": 0.47437816263986887, "grad_norm": 91.45944584413142, "learning_rate": 1.8248077726395631e-07, "logits/chosen": -1.381393313407898, "logits/rejected": -1.3962711095809937, "logps/chosen": -139.09548950195312, "logps/rejected": -154.32943725585938, "loss": 0.644, "rewards/accuracies": 0.78125, "rewards/chosen": 0.08523928374052048, "rewards/margins": 0.4487174153327942, "rewards/rejected": -0.3634781539440155, "step": 416 }, { "epoch": 0.4766588268833298, "grad_norm": 118.15308317787861, "learning_rate": 1.8226513301586297e-07, "logits/chosen": -1.1643140316009521, "logits/rejected": -1.1523492336273193, "logps/chosen": -181.3363494873047, "logps/rejected": -208.14474487304688, "loss": 0.6631, "rewards/accuracies": 0.5, "rewards/chosen": -0.18151137232780457, "rewards/margins": 0.11719075590372086, "rewards/rejected": -0.2987021207809448, "step": 418 }, { "epoch": 0.4789394911267907, "grad_norm": 87.18607024470172, "learning_rate": 1.820482987459308e-07, "logits/chosen": -1.2820664644241333, "logits/rejected": -1.340864658355713, "logps/chosen": -152.3217315673828, "logps/rejected": -166.40199279785156, "loss": 0.6306, "rewards/accuracies": 0.5, "rewards/chosen": -0.20804628729820251, "rewards/margins": 0.08607995510101318, "rewards/rejected": -0.2941262423992157, "step": 420 }, { "epoch": 0.4812201553702516, "grad_norm": 101.2197264458815, "learning_rate": 1.818302775908169e-07, "logits/chosen": -1.3642746210098267, "logits/rejected": -1.4530210494995117, "logps/chosen": -177.27783203125, "logps/rejected": -209.4615478515625, "loss": 0.6002, "rewards/accuracies": 0.65625, "rewards/chosen": -0.06609566509723663, "rewards/margins": 0.24875135719776154, "rewards/rejected": -0.31484702229499817, "step": 422 }, { "epoch": 0.4835008196137125, "grad_norm": 102.24009472626234, "learning_rate": 1.8161107270434757e-07, "logits/chosen": -1.322772741317749, "logits/rejected": -1.3549699783325195, "logps/chosen": -175.10890197753906, "logps/rejected": -189.1114959716797, "loss": 0.7052, "rewards/accuracies": 0.53125, "rewards/chosen": -0.2034468799829483, "rewards/margins": -0.003699667751789093, "rewards/rejected": -0.199747234582901, "step": 424 }, { "epoch": 0.4857814838571734, "grad_norm": 88.40359207083654, "learning_rate": 1.8139068725747251e-07, "logits/chosen": -1.2980220317840576, "logits/rejected": -1.328129768371582, "logps/chosen": -127.48295593261719, "logps/rejected": -145.63137817382812, "loss": 0.6353, "rewards/accuracies": 0.5625, "rewards/chosen": -0.17486554384231567, "rewards/margins": 0.16773542761802673, "rewards/rejected": -0.3426010012626648, "step": 426 }, { "epoch": 0.4880621481006343, "grad_norm": 85.01546232524993, "learning_rate": 1.811691244382191e-07, "logits/chosen": -1.240834355354309, "logits/rejected": -1.315454125404358, "logps/chosen": -171.84417724609375, "logps/rejected": -196.60919189453125, "loss": 0.6332, "rewards/accuracies": 0.625, "rewards/chosen": -0.043825387954711914, "rewards/margins": 0.35121482610702515, "rewards/rejected": -0.39504021406173706, "step": 428 }, { "epoch": 0.4903428123440952, "grad_norm": 92.50180646153169, "learning_rate": 1.8094638745164619e-07, "logits/chosen": -1.368233561515808, "logits/rejected": -1.4737908840179443, "logps/chosen": -231.7982177734375, "logps/rejected": -257.540771484375, "loss": 0.6393, "rewards/accuracies": 0.5, "rewards/chosen": -0.41820961236953735, "rewards/margins": 0.03227778524160385, "rewards/rejected": -0.4504873752593994, "step": 430 }, { "epoch": 0.4926234765875561, "grad_norm": 93.17182791936958, "learning_rate": 1.8072247951979782e-07, "logits/chosen": -1.218642234802246, "logits/rejected": -1.3522933721542358, "logps/chosen": -161.9176025390625, "logps/rejected": -191.59205627441406, "loss": 0.6532, "rewards/accuracies": 0.46875, "rewards/chosen": -0.2410704791545868, "rewards/margins": 0.09214940667152405, "rewards/rejected": -0.33321988582611084, "step": 432 }, { "epoch": 0.494904140831017, "grad_norm": 85.22671128905367, "learning_rate": 1.8049740388165646e-07, "logits/chosen": -1.3610548973083496, "logits/rejected": -1.369373083114624, "logps/chosen": -170.4285888671875, "logps/rejected": -203.44435119628906, "loss": 0.6788, "rewards/accuracies": 0.53125, "rewards/chosen": -0.12429852783679962, "rewards/margins": 0.06445370614528656, "rewards/rejected": -0.18875223398208618, "step": 434 }, { "epoch": 0.49718480507447793, "grad_norm": 106.2899671913355, "learning_rate": 1.8027116379309635e-07, "logits/chosen": -1.3293408155441284, "logits/rejected": -1.3227120637893677, "logps/chosen": -189.62835693359375, "logps/rejected": -196.80776977539062, "loss": 0.6658, "rewards/accuracies": 0.4375, "rewards/chosen": -0.36924922466278076, "rewards/margins": 0.04584325850009918, "rewards/rejected": -0.41509246826171875, "step": 436 }, { "epoch": 0.49946546931793884, "grad_norm": 97.4627918918024, "learning_rate": 1.8004376252683629e-07, "logits/chosen": -1.3211669921875, "logits/rejected": -1.318382740020752, "logps/chosen": -170.44007873535156, "logps/rejected": -177.15567016601562, "loss": 0.652, "rewards/accuracies": 0.65625, "rewards/chosen": -0.0650164932012558, "rewards/margins": 0.2474796026945114, "rewards/rejected": -0.3124960958957672, "step": 438 }, { "epoch": 0.5017461335613997, "grad_norm": 94.35984756124336, "learning_rate": 1.7981520337239229e-07, "logits/chosen": -1.4195128679275513, "logits/rejected": -1.517584204673767, "logps/chosen": -124.39069366455078, "logps/rejected": -148.39462280273438, "loss": 0.6071, "rewards/accuracies": 0.65625, "rewards/chosen": 0.07836849987506866, "rewards/margins": 0.2909252941608429, "rewards/rejected": -0.21255679428577423, "step": 440 }, { "epoch": 0.5040267978048607, "grad_norm": 105.60849785169918, "learning_rate": 1.7958548963603e-07, "logits/chosen": -1.3688621520996094, "logits/rejected": -1.3791152238845825, "logps/chosen": -193.5460968017578, "logps/rejected": -198.70162963867188, "loss": 0.6155, "rewards/accuracies": 0.8125, "rewards/chosen": 0.006075154058635235, "rewards/margins": 0.4186677634716034, "rewards/rejected": -0.41259267926216125, "step": 442 }, { "epoch": 0.5063074620483216, "grad_norm": 105.19505214300226, "learning_rate": 1.7935462464071694e-07, "logits/chosen": -1.4063529968261719, "logits/rejected": -1.4609131813049316, "logps/chosen": -136.6985321044922, "logps/rejected": -142.0517120361328, "loss": 0.6638, "rewards/accuracies": 0.71875, "rewards/chosen": -0.021208282560110092, "rewards/margins": 0.23011060059070587, "rewards/rejected": -0.25131887197494507, "step": 444 }, { "epoch": 0.5085881262917825, "grad_norm": 80.98126219811707, "learning_rate": 1.7912261172607434e-07, "logits/chosen": -1.4103684425354004, "logits/rejected": -1.4659305810928345, "logps/chosen": -181.66671752929688, "logps/rejected": -211.0421142578125, "loss": 0.667, "rewards/accuracies": 0.6875, "rewards/chosen": -0.0669156014919281, "rewards/margins": 0.23970045149326324, "rewards/rejected": -0.30661600828170776, "step": 446 }, { "epoch": 0.5108687905352434, "grad_norm": 86.7187808268374, "learning_rate": 1.7888945424832892e-07, "logits/chosen": -1.2415688037872314, "logits/rejected": -1.2903611660003662, "logps/chosen": -195.07174682617188, "logps/rejected": -223.18804931640625, "loss": 0.6344, "rewards/accuracies": 0.625, "rewards/chosen": -0.26416096091270447, "rewards/margins": 0.23690298199653625, "rewards/rejected": -0.5010639429092407, "step": 448 }, { "epoch": 0.5131494547787043, "grad_norm": 91.51916447980646, "learning_rate": 1.7865515558026425e-07, "logits/chosen": -1.2545722723007202, "logits/rejected": -1.320673942565918, "logps/chosen": -190.4950714111328, "logps/rejected": -196.8434600830078, "loss": 0.645, "rewards/accuracies": 0.6875, "rewards/chosen": -0.12865591049194336, "rewards/margins": 0.2827560007572174, "rewards/rejected": -0.41141194105148315, "step": 450 }, { "epoch": 0.5154301190221652, "grad_norm": 78.95115222899102, "learning_rate": 1.78419719111172e-07, "logits/chosen": -1.2161290645599365, "logits/rejected": -1.2955366373062134, "logps/chosen": -133.3023223876953, "logps/rejected": -173.38735961914062, "loss": 0.6202, "rewards/accuracies": 0.625, "rewards/chosen": -0.031168239191174507, "rewards/margins": 0.3425666093826294, "rewards/rejected": -0.37373483180999756, "step": 452 }, { "epoch": 0.5177107832656261, "grad_norm": 85.03550287122602, "learning_rate": 1.78183148246803e-07, "logits/chosen": -1.3149534463882446, "logits/rejected": -1.3734480142593384, "logps/chosen": -181.44749450683594, "logps/rejected": -190.66098022460938, "loss": 0.6579, "rewards/accuracies": 0.4375, "rewards/chosen": -0.02150813303887844, "rewards/margins": 0.09824002534151077, "rewards/rejected": -0.11974816769361496, "step": 454 }, { "epoch": 0.519991447509087, "grad_norm": 101.82433211173363, "learning_rate": 1.779454464093177e-07, "logits/chosen": -1.2327746152877808, "logits/rejected": -1.2479290962219238, "logps/chosen": -123.62596130371094, "logps/rejected": -145.822021484375, "loss": 0.71, "rewards/accuracies": 0.65625, "rewards/chosen": 0.008450210094451904, "rewards/margins": 0.1642961949110031, "rewards/rejected": -0.1558459848165512, "step": 456 }, { "epoch": 0.522272111752548, "grad_norm": 86.25249319253086, "learning_rate": 1.7770661703723716e-07, "logits/chosen": -1.4179208278656006, "logits/rejected": -1.4704089164733887, "logps/chosen": -155.395263671875, "logps/rejected": -241.91619873046875, "loss": 0.6544, "rewards/accuracies": 0.625, "rewards/chosen": -0.0192805714905262, "rewards/margins": 0.4996611773967743, "rewards/rejected": -0.5189418792724609, "step": 458 }, { "epoch": 0.5245527759960088, "grad_norm": 95.73185772448069, "learning_rate": 1.7746666358539268e-07, "logits/chosen": -1.324285864830017, "logits/rejected": -1.3773202896118164, "logps/chosen": -138.01214599609375, "logps/rejected": -187.76605224609375, "loss": 0.6781, "rewards/accuracies": 0.5, "rewards/chosen": -0.11806146800518036, "rewards/margins": 0.12408170104026794, "rewards/rejected": -0.2421431541442871, "step": 460 }, { "epoch": 0.5268334402394698, "grad_norm": 83.04802100256194, "learning_rate": 1.7722558952487637e-07, "logits/chosen": -1.2252494096755981, "logits/rejected": -1.2402747869491577, "logps/chosen": -145.02134704589844, "logps/rejected": -157.798095703125, "loss": 0.6658, "rewards/accuracies": 0.59375, "rewards/chosen": -0.16228270530700684, "rewards/margins": -0.006807304918766022, "rewards/rejected": -0.1554754078388214, "step": 462 }, { "epoch": 0.5291141044829306, "grad_norm": 106.4784679324614, "learning_rate": 1.7698339834299062e-07, "logits/chosen": -1.2572526931762695, "logits/rejected": -1.2763482332229614, "logps/chosen": -205.0850067138672, "logps/rejected": -216.61383056640625, "loss": 0.6758, "rewards/accuracies": 0.71875, "rewards/chosen": -0.10254046320915222, "rewards/margins": 0.47703590989112854, "rewards/rejected": -0.5795763731002808, "step": 464 }, { "epoch": 0.5313947687263916, "grad_norm": 88.23039919938233, "learning_rate": 1.7674009354319776e-07, "logits/chosen": -1.2906126976013184, "logits/rejected": -1.3812601566314697, "logps/chosen": -182.90606689453125, "logps/rejected": -192.94520568847656, "loss": 0.5967, "rewards/accuracies": 0.59375, "rewards/chosen": -0.12573322653770447, "rewards/margins": 0.15527519583702087, "rewards/rejected": -0.28100842237472534, "step": 466 }, { "epoch": 0.5336754329698524, "grad_norm": 96.24054074341065, "learning_rate": 1.764956786450694e-07, "logits/chosen": -1.2798047065734863, "logits/rejected": -1.3678182363510132, "logps/chosen": -160.85719299316406, "logps/rejected": -189.01951599121094, "loss": 0.6321, "rewards/accuracies": 0.78125, "rewards/chosen": -0.01916433498263359, "rewards/margins": 0.328861802816391, "rewards/rejected": -0.3480261266231537, "step": 468 }, { "epoch": 0.5359560972133134, "grad_norm": 87.51604676838089, "learning_rate": 1.7625015718423548e-07, "logits/chosen": -1.3382031917572021, "logits/rejected": -1.4446802139282227, "logps/chosen": -152.4910125732422, "logps/rejected": -192.94979858398438, "loss": 0.6364, "rewards/accuracies": 0.65625, "rewards/chosen": -0.021129202097654343, "rewards/margins": 0.3612152636051178, "rewards/rejected": -0.38234445452690125, "step": 470 }, { "epoch": 0.5382367614567742, "grad_norm": 118.40842425272356, "learning_rate": 1.7600353271233312e-07, "logits/chosen": -1.319710373878479, "logits/rejected": -1.284549355506897, "logps/chosen": -152.95933532714844, "logps/rejected": -156.8072052001953, "loss": 0.7236, "rewards/accuracies": 0.53125, "rewards/chosen": -0.10047149658203125, "rewards/margins": 0.010637946426868439, "rewards/rejected": -0.11110943555831909, "step": 472 }, { "epoch": 0.5405174257002352, "grad_norm": 91.83509455780727, "learning_rate": 1.7575580879695522e-07, "logits/chosen": -1.445102572441101, "logits/rejected": -1.4954876899719238, "logps/chosen": -196.539306640625, "logps/rejected": -211.00616455078125, "loss": 0.6409, "rewards/accuracies": 0.625, "rewards/chosen": -0.04536067694425583, "rewards/margins": 0.1365879625082016, "rewards/rejected": -0.18194863200187683, "step": 474 }, { "epoch": 0.542798089943696, "grad_norm": 90.74830800218245, "learning_rate": 1.7550698902159894e-07, "logits/chosen": -1.3519353866577148, "logits/rejected": -1.4368261098861694, "logps/chosen": -228.4932098388672, "logps/rejected": -261.87109375, "loss": 0.6373, "rewards/accuracies": 0.46875, "rewards/chosen": -0.1739276647567749, "rewards/margins": 0.011357128620147705, "rewards/rejected": -0.1852847784757614, "step": 476 }, { "epoch": 0.545078754187157, "grad_norm": 114.14083612276099, "learning_rate": 1.7525707698561384e-07, "logits/chosen": -1.296964168548584, "logits/rejected": -1.3777580261230469, "logps/chosen": -178.81484985351562, "logps/rejected": -220.63436889648438, "loss": 0.6515, "rewards/accuracies": 0.75, "rewards/chosen": -0.10972025990486145, "rewards/margins": 0.45839735865592957, "rewards/rejected": -0.5681176781654358, "step": 478 }, { "epoch": 0.547359418430618, "grad_norm": 106.70099483754956, "learning_rate": 1.750060763041497e-07, "logits/chosen": -1.3242031335830688, "logits/rejected": -1.3594613075256348, "logps/chosen": -139.518310546875, "logps/rejected": -171.3057403564453, "loss": 0.6932, "rewards/accuracies": 0.59375, "rewards/chosen": -0.09948215633630753, "rewards/margins": 0.13577090203762054, "rewards/rejected": -0.23525305092334747, "step": 480 }, { "epoch": 0.5496400826740788, "grad_norm": 93.76871490458784, "learning_rate": 1.7475399060810435e-07, "logits/chosen": -1.5048158168792725, "logits/rejected": -1.567376971244812, "logps/chosen": -175.3671112060547, "logps/rejected": -195.3438262939453, "loss": 0.6284, "rewards/accuracies": 0.625, "rewards/chosen": 0.08378006517887115, "rewards/margins": 0.32175135612487793, "rewards/rejected": -0.23797132074832916, "step": 482 }, { "epoch": 0.5519207469175398, "grad_norm": 87.30478432077047, "learning_rate": 1.7450082354407107e-07, "logits/chosen": -1.3454846143722534, "logits/rejected": -1.3264262676239014, "logps/chosen": -150.41058349609375, "logps/rejected": -158.28414916992188, "loss": 0.6392, "rewards/accuracies": 0.53125, "rewards/chosen": 0.06661003828048706, "rewards/margins": 0.05708365514874458, "rewards/rejected": 0.009526383131742477, "step": 484 }, { "epoch": 0.5542014111610006, "grad_norm": 97.70119982177076, "learning_rate": 1.7424657877428594e-07, "logits/chosen": -1.4438061714172363, "logits/rejected": -1.5046687126159668, "logps/chosen": -210.83221435546875, "logps/rejected": -220.98056030273438, "loss": 0.6711, "rewards/accuracies": 0.59375, "rewards/chosen": 0.02250128984451294, "rewards/margins": 0.25783565640449524, "rewards/rejected": -0.23533441126346588, "step": 486 }, { "epoch": 0.5564820754044616, "grad_norm": 92.7470076491593, "learning_rate": 1.7399125997657475e-07, "logits/chosen": -1.4578030109405518, "logits/rejected": -1.5047590732574463, "logps/chosen": -173.8478546142578, "logps/rejected": -218.70567321777344, "loss": 0.6464, "rewards/accuracies": 0.75, "rewards/chosen": -0.005423933267593384, "rewards/margins": 0.22333881258964539, "rewards/rejected": -0.22876277565956116, "step": 488 }, { "epoch": 0.5587627396479224, "grad_norm": 80.43868813641923, "learning_rate": 1.7373487084429986e-07, "logits/chosen": -1.4293080568313599, "logits/rejected": -1.3778735399246216, "logps/chosen": -137.72915649414062, "logps/rejected": -144.8123779296875, "loss": 0.6175, "rewards/accuracies": 0.65625, "rewards/chosen": -0.04381662234663963, "rewards/margins": 0.17386358976364136, "rewards/rejected": -0.2176802158355713, "step": 490 }, { "epoch": 0.5610434038913834, "grad_norm": 88.32928179136898, "learning_rate": 1.734774150863067e-07, "logits/chosen": -1.4234795570373535, "logits/rejected": -1.4454480409622192, "logps/chosen": -168.4951171875, "logps/rejected": -183.8735809326172, "loss": 0.6522, "rewards/accuracies": 0.625, "rewards/chosen": 0.008489616215229034, "rewards/margins": 0.20173315703868866, "rewards/rejected": -0.19324351847171783, "step": 492 }, { "epoch": 0.5633240681348443, "grad_norm": 93.76675043932717, "learning_rate": 1.732188964268703e-07, "logits/chosen": -1.2950657606124878, "logits/rejected": -1.3514856100082397, "logps/chosen": -174.49368286132812, "logps/rejected": -198.4128875732422, "loss": 0.6429, "rewards/accuracies": 0.5, "rewards/chosen": -0.17520295083522797, "rewards/margins": 0.17245244979858398, "rewards/rejected": -0.34765538573265076, "step": 494 }, { "epoch": 0.5656047323783052, "grad_norm": 124.66844580791373, "learning_rate": 1.7295931860564118e-07, "logits/chosen": -1.3973047733306885, "logits/rejected": -1.3848158121109009, "logps/chosen": -187.01031494140625, "logps/rejected": -201.88232421875, "loss": 0.661, "rewards/accuracies": 0.4375, "rewards/chosen": -0.4809916317462921, "rewards/margins": -0.037112899124622345, "rewards/rejected": -0.443878710269928, "step": 496 }, { "epoch": 0.5678853966217661, "grad_norm": 98.27416452810739, "learning_rate": 1.7269868537759138e-07, "logits/chosen": -1.4414607286453247, "logits/rejected": -1.4525971412658691, "logps/chosen": -186.87220764160156, "logps/rejected": -201.6339874267578, "loss": 0.6551, "rewards/accuracies": 0.75, "rewards/chosen": 0.048535339534282684, "rewards/margins": 0.2541671097278595, "rewards/rejected": -0.20563176274299622, "step": 498 }, { "epoch": 0.570166060865227, "grad_norm": 81.20755107084783, "learning_rate": 1.7243700051296016e-07, "logits/chosen": -1.4566954374313354, "logits/rejected": -1.458418369293213, "logps/chosen": -186.07911682128906, "logps/rejected": -181.54051208496094, "loss": 0.6716, "rewards/accuracies": 0.4375, "rewards/chosen": -0.08042742311954498, "rewards/margins": 0.003591485321521759, "rewards/rejected": -0.08401892334222794, "step": 500 }, { "epoch": 0.570166060865227, "eval_logits/chosen": -1.4689228534698486, "eval_logits/rejected": -1.4550352096557617, "eval_logps/chosen": -128.87640380859375, "eval_logps/rejected": -132.28384399414062, "eval_loss": 0.6351009011268616, "eval_rewards/accuracies": 0.6399999856948853, "eval_rewards/chosen": 0.10215496271848679, "eval_rewards/margins": 0.12426460534334183, "eval_rewards/rejected": -0.022109635174274445, "eval_runtime": 20.6735, "eval_samples_per_second": 4.837, "eval_steps_per_second": 1.209, "step": 500 }, { "epoch": 0.5724467251086879, "grad_norm": 87.15380241514247, "learning_rate": 1.7217426779719944e-07, "logits/chosen": -1.461560606956482, "logits/rejected": -1.4978594779968262, "logps/chosen": -145.8336639404297, "logps/rejected": -164.410888671875, "loss": 0.623, "rewards/accuracies": 0.65625, "rewards/chosen": -0.030693676322698593, "rewards/margins": 0.31371259689331055, "rewards/rejected": -0.34440624713897705, "step": 502 }, { "epoch": 0.5747273893521488, "grad_norm": 109.6734083472702, "learning_rate": 1.71910491030919e-07, "logits/chosen": -1.4200553894042969, "logits/rejected": -1.4383351802825928, "logps/chosen": -185.2432861328125, "logps/rejected": -198.3214569091797, "loss": 0.6592, "rewards/accuracies": 0.5625, "rewards/chosen": -0.09178508818149567, "rewards/margins": 0.1750696301460266, "rewards/rejected": -0.2668547034263611, "step": 504 }, { "epoch": 0.5770080535956097, "grad_norm": 95.33984559143799, "learning_rate": 1.716456740298315e-07, "logits/chosen": -1.356040120124817, "logits/rejected": -1.3635413646697998, "logps/chosen": -221.06320190429688, "logps/rejected": -246.34091186523438, "loss": 0.6466, "rewards/accuracies": 0.59375, "rewards/chosen": -0.224991112947464, "rewards/margins": 0.12949411571025848, "rewards/rejected": -0.35448524355888367, "step": 506 }, { "epoch": 0.5792887178390707, "grad_norm": 104.51344570968739, "learning_rate": 1.7137982062469739e-07, "logits/chosen": -1.199120044708252, "logits/rejected": -1.2254160642623901, "logps/chosen": -182.35678100585938, "logps/rejected": -205.9598388671875, "loss": 0.6302, "rewards/accuracies": 0.6875, "rewards/chosen": -0.01045935694128275, "rewards/margins": 0.4567835330963135, "rewards/rejected": -0.46724286675453186, "step": 508 }, { "epoch": 0.5815693820825315, "grad_norm": 80.61166146639357, "learning_rate": 1.7111293466126936e-07, "logits/chosen": -1.4290285110473633, "logits/rejected": -1.465309977531433, "logps/chosen": -183.41546630859375, "logps/rejected": -190.38209533691406, "loss": 0.6303, "rewards/accuracies": 0.625, "rewards/chosen": -0.04882081598043442, "rewards/margins": 0.2510834336280823, "rewards/rejected": -0.2999042272567749, "step": 510 }, { "epoch": 0.5838500463259925, "grad_norm": 81.58673928063743, "learning_rate": 1.7084502000023678e-07, "logits/chosen": -1.3187546730041504, "logits/rejected": -1.3721165657043457, "logps/chosen": -166.35317993164062, "logps/rejected": -205.14459228515625, "loss": 0.6609, "rewards/accuracies": 0.6875, "rewards/chosen": 0.06919605284929276, "rewards/margins": 0.25993865728378296, "rewards/rejected": -0.1907426118850708, "step": 512 }, { "epoch": 0.5861307105694533, "grad_norm": 107.99959199152349, "learning_rate": 1.7057608051716987e-07, "logits/chosen": -1.2958874702453613, "logits/rejected": -1.3409713506698608, "logps/chosen": -189.3512420654297, "logps/rejected": -229.98101806640625, "loss": 0.6805, "rewards/accuracies": 0.625, "rewards/chosen": -0.1412532776594162, "rewards/margins": 0.1916051059961319, "rewards/rejected": -0.3328583836555481, "step": 514 }, { "epoch": 0.5884113748129143, "grad_norm": 103.31085588981476, "learning_rate": 1.7030612010246357e-07, "logits/chosen": -1.4542144536972046, "logits/rejected": -1.4537440538406372, "logps/chosen": -187.4440155029297, "logps/rejected": -197.0413818359375, "loss": 0.6901, "rewards/accuracies": 0.5625, "rewards/chosen": -0.20216982066631317, "rewards/margins": -0.05732645466923714, "rewards/rejected": -0.14484335482120514, "step": 516 }, { "epoch": 0.5906920390563751, "grad_norm": 93.7004047268541, "learning_rate": 1.7003514266128128e-07, "logits/chosen": -1.3705557584762573, "logits/rejected": -1.4125196933746338, "logps/chosen": -176.84820556640625, "logps/rejected": -218.4021453857422, "loss": 0.688, "rewards/accuracies": 0.5625, "rewards/chosen": -0.13926447927951813, "rewards/margins": 0.1602565199136734, "rewards/rejected": -0.2995210289955139, "step": 518 }, { "epoch": 0.5929727032998361, "grad_norm": 76.7210943847488, "learning_rate": 1.6976315211349848e-07, "logits/chosen": -1.402063250541687, "logits/rejected": -1.5004804134368896, "logps/chosen": -183.45787048339844, "logps/rejected": -226.31532287597656, "loss": 0.6405, "rewards/accuracies": 0.71875, "rewards/chosen": 0.08890336751937866, "rewards/margins": 0.2809264659881592, "rewards/rejected": -0.19202306866645813, "step": 520 }, { "epoch": 0.5952533675432969, "grad_norm": 79.99952128240997, "learning_rate": 1.694901523936458e-07, "logits/chosen": -1.3303985595703125, "logits/rejected": -1.4111953973770142, "logps/chosen": -161.44390869140625, "logps/rejected": -201.76593017578125, "loss": 0.601, "rewards/accuracies": 0.78125, "rewards/chosen": 0.12671524286270142, "rewards/margins": 0.4966307282447815, "rewards/rejected": -0.3699154853820801, "step": 522 }, { "epoch": 0.5975340317867579, "grad_norm": 89.60709297255305, "learning_rate": 1.6921614745085235e-07, "logits/chosen": -1.4100148677825928, "logits/rejected": -1.4378656148910522, "logps/chosen": -193.54835510253906, "logps/rejected": -202.9269256591797, "loss": 0.6182, "rewards/accuracies": 0.71875, "rewards/chosen": -0.035577308386564255, "rewards/margins": 0.3006408214569092, "rewards/rejected": -0.33621811866760254, "step": 524 }, { "epoch": 0.5998146960302188, "grad_norm": 104.27911734761277, "learning_rate": 1.689411412487885e-07, "logits/chosen": -1.3611271381378174, "logits/rejected": -1.3725131750106812, "logps/chosen": -166.821533203125, "logps/rejected": -197.65394592285156, "loss": 0.6274, "rewards/accuracies": 0.65625, "rewards/chosen": 0.09458069503307343, "rewards/margins": 0.400374174118042, "rewards/rejected": -0.30579352378845215, "step": 526 }, { "epoch": 0.6020953602736797, "grad_norm": 77.9183795766502, "learning_rate": 1.6866513776560841e-07, "logits/chosen": -1.2477551698684692, "logits/rejected": -1.2647314071655273, "logps/chosen": -143.03448486328125, "logps/rejected": -149.70594787597656, "loss": 0.6035, "rewards/accuracies": 0.71875, "rewards/chosen": 0.14375856518745422, "rewards/margins": 0.17092323303222656, "rewards/rejected": -0.02716466784477234, "step": 528 }, { "epoch": 0.6043760245171406, "grad_norm": 79.08140854679833, "learning_rate": 1.6838814099389265e-07, "logits/chosen": -1.379310965538025, "logits/rejected": -1.49821138381958, "logps/chosen": -176.78890991210938, "logps/rejected": -228.0948028564453, "loss": 0.6114, "rewards/accuracies": 0.59375, "rewards/chosen": -0.18548306822776794, "rewards/margins": 0.3985109329223633, "rewards/rejected": -0.5839939117431641, "step": 530 }, { "epoch": 0.6066566887606015, "grad_norm": 86.55652744872312, "learning_rate": 1.6811015494059045e-07, "logits/chosen": -1.340354323387146, "logits/rejected": -1.372084617614746, "logps/chosen": -139.92453002929688, "logps/rejected": -170.04629516601562, "loss": 0.5944, "rewards/accuracies": 0.625, "rewards/chosen": 0.09279327094554901, "rewards/margins": 0.36332762241363525, "rewards/rejected": -0.27053436636924744, "step": 532 }, { "epoch": 0.6089373530040625, "grad_norm": 86.16548448796566, "learning_rate": 1.678311836269616e-07, "logits/chosen": -1.2494535446166992, "logits/rejected": -1.3522768020629883, "logps/chosen": -133.36306762695312, "logps/rejected": -140.9103240966797, "loss": 0.626, "rewards/accuracies": 0.625, "rewards/chosen": -0.22486190497875214, "rewards/margins": 0.10757801681756973, "rewards/rejected": -0.3324398994445801, "step": 534 }, { "epoch": 0.6112180172475233, "grad_norm": 78.5036790731728, "learning_rate": 1.6755123108851842e-07, "logits/chosen": -1.2597969770431519, "logits/rejected": -1.3842471837997437, "logps/chosen": -102.91565704345703, "logps/rejected": -149.77996826171875, "loss": 0.6558, "rewards/accuracies": 0.59375, "rewards/chosen": 0.049101147800683975, "rewards/margins": 0.2616952359676361, "rewards/rejected": -0.21259410679340363, "step": 536 }, { "epoch": 0.6134986814909843, "grad_norm": 97.00034287566903, "learning_rate": 1.6727030137496727e-07, "logits/chosen": -1.333500862121582, "logits/rejected": -1.407701015472412, "logps/chosen": -168.34678649902344, "logps/rejected": -212.6487579345703, "loss": 0.6333, "rewards/accuracies": 0.65625, "rewards/chosen": -0.23587258160114288, "rewards/margins": 0.40141719579696655, "rewards/rejected": -0.637289822101593, "step": 538 }, { "epoch": 0.6157793457344451, "grad_norm": 93.14382617410926, "learning_rate": 1.6698839855015007e-07, "logits/chosen": -1.3826831579208374, "logits/rejected": -1.3780534267425537, "logps/chosen": -197.9063262939453, "logps/rejected": -201.8406982421875, "loss": 0.653, "rewards/accuracies": 0.46875, "rewards/chosen": -0.19411635398864746, "rewards/margins": 0.06094612553715706, "rewards/rejected": -0.2550624907016754, "step": 540 }, { "epoch": 0.6180600099779061, "grad_norm": 84.06073173280129, "learning_rate": 1.6670552669198546e-07, "logits/chosen": -1.459622859954834, "logits/rejected": -1.5475270748138428, "logps/chosen": -149.38450622558594, "logps/rejected": -174.00827026367188, "loss": 0.6878, "rewards/accuracies": 0.625, "rewards/chosen": -0.22387176752090454, "rewards/margins": 0.04707152396440506, "rewards/rejected": -0.2709433138370514, "step": 542 }, { "epoch": 0.620340674221367, "grad_norm": 75.86401555746195, "learning_rate": 1.6642168989240974e-07, "logits/chosen": -1.3617055416107178, "logits/rejected": -1.3725388050079346, "logps/chosen": -177.85269165039062, "logps/rejected": -204.47271728515625, "loss": 0.6393, "rewards/accuracies": 0.75, "rewards/chosen": -0.2369232326745987, "rewards/margins": 0.4420955181121826, "rewards/rejected": -0.6790187358856201, "step": 544 }, { "epoch": 0.6226213384648279, "grad_norm": 99.08946862124856, "learning_rate": 1.6613689225731787e-07, "logits/chosen": -1.437178611755371, "logits/rejected": -1.4497863054275513, "logps/chosen": -152.61231994628906, "logps/rejected": -163.36196899414062, "loss": 0.6514, "rewards/accuracies": 0.5625, "rewards/chosen": -0.24256393313407898, "rewards/margins": 0.14335207641124725, "rewards/rejected": -0.3859160244464874, "step": 546 }, { "epoch": 0.6249020027082888, "grad_norm": 82.17381709045306, "learning_rate": 1.6585113790650386e-07, "logits/chosen": -1.3418258428573608, "logits/rejected": -1.357533574104309, "logps/chosen": -150.5135955810547, "logps/rejected": -168.92135620117188, "loss": 0.6404, "rewards/accuracies": 0.625, "rewards/chosen": -0.1530638188123703, "rewards/margins": 0.16317017376422882, "rewards/rejected": -0.3162340223789215, "step": 548 }, { "epoch": 0.6271826669517497, "grad_norm": 98.6032079855375, "learning_rate": 1.6556443097360133e-07, "logits/chosen": -1.305914282798767, "logits/rejected": -1.3782196044921875, "logps/chosen": -187.4833526611328, "logps/rejected": -213.8814697265625, "loss": 0.6415, "rewards/accuracies": 0.75, "rewards/chosen": -0.03956926241517067, "rewards/margins": 0.45714423060417175, "rewards/rejected": -0.4967134892940521, "step": 550 }, { "epoch": 0.6294633311952106, "grad_norm": 89.5606898436777, "learning_rate": 1.6527677560602363e-07, "logits/chosen": -1.4830090999603271, "logits/rejected": -1.4891180992126465, "logps/chosen": -178.29112243652344, "logps/rejected": -182.18019104003906, "loss": 0.658, "rewards/accuracies": 0.6875, "rewards/chosen": -0.10583019256591797, "rewards/margins": 0.16508930921554565, "rewards/rejected": -0.270919531583786, "step": 552 }, { "epoch": 0.6317439954386715, "grad_norm": 94.29213072448552, "learning_rate": 1.6498817596490388e-07, "logits/chosen": -1.3679401874542236, "logits/rejected": -1.397881031036377, "logps/chosen": -170.22218322753906, "logps/rejected": -190.22608947753906, "loss": 0.6632, "rewards/accuracies": 0.625, "rewards/chosen": -0.07908223569393158, "rewards/margins": 0.2489246428012848, "rewards/rejected": -0.32800689339637756, "step": 554 }, { "epoch": 0.6340246596821324, "grad_norm": 97.6468147380264, "learning_rate": 1.6469863622503475e-07, "logits/chosen": -1.3809480667114258, "logits/rejected": -1.3862886428833008, "logps/chosen": -161.6693878173828, "logps/rejected": -186.37608337402344, "loss": 0.6762, "rewards/accuracies": 0.6875, "rewards/chosen": -0.08833756297826767, "rewards/margins": 0.3514275848865509, "rewards/rejected": -0.4397651255130768, "step": 556 }, { "epoch": 0.6363053239255934, "grad_norm": 90.63822694085678, "learning_rate": 1.6440816057480812e-07, "logits/chosen": -1.4811036586761475, "logits/rejected": -1.554297924041748, "logps/chosen": -214.304931640625, "logps/rejected": -267.9236145019531, "loss": 0.638, "rewards/accuracies": 0.625, "rewards/chosen": 0.08694766461849213, "rewards/margins": 0.4463149309158325, "rewards/rejected": -0.359367311000824, "step": 558 }, { "epoch": 0.6385859881690542, "grad_norm": 109.09749536253133, "learning_rate": 1.641167532161545e-07, "logits/chosen": -1.3816839456558228, "logits/rejected": -1.3854830265045166, "logps/chosen": -146.0878448486328, "logps/rejected": -151.40838623046875, "loss": 0.6199, "rewards/accuracies": 0.625, "rewards/chosen": -0.2251240313053131, "rewards/margins": 0.2835414409637451, "rewards/rejected": -0.5086655020713806, "step": 560 }, { "epoch": 0.6408666524125152, "grad_norm": 85.56418756561474, "learning_rate": 1.63824418364482e-07, "logits/chosen": -1.3435323238372803, "logits/rejected": -1.4011439085006714, "logps/chosen": -171.78982543945312, "logps/rejected": -188.12899780273438, "loss": 0.6342, "rewards/accuracies": 0.5625, "rewards/chosen": -0.03885346278548241, "rewards/margins": 0.2555912137031555, "rewards/rejected": -0.29444462060928345, "step": 562 }, { "epoch": 0.643147316655976, "grad_norm": 91.83203136203487, "learning_rate": 1.6353116024861583e-07, "logits/chosen": -1.3896362781524658, "logits/rejected": -1.3640623092651367, "logps/chosen": -186.40646362304688, "logps/rejected": -194.1944580078125, "loss": 0.6166, "rewards/accuracies": 0.71875, "rewards/chosen": -0.14515984058380127, "rewards/margins": 0.4773363471031189, "rewards/rejected": -0.6224961876869202, "step": 564 }, { "epoch": 0.645427980899437, "grad_norm": 81.53572929971928, "learning_rate": 1.6323698311073665e-07, "logits/chosen": -1.2487475872039795, "logits/rejected": -1.298531413078308, "logps/chosen": -157.927734375, "logps/rejected": -172.3856658935547, "loss": 0.6697, "rewards/accuracies": 0.65625, "rewards/chosen": -0.2098311483860016, "rewards/margins": 0.2927185297012329, "rewards/rejected": -0.5025496482849121, "step": 566 }, { "epoch": 0.6477086451428978, "grad_norm": 114.13313398743495, "learning_rate": 1.6294189120631955e-07, "logits/chosen": -1.4274886846542358, "logits/rejected": -1.5344736576080322, "logps/chosen": -182.54502868652344, "logps/rejected": -216.51434326171875, "loss": 0.6837, "rewards/accuracies": 0.71875, "rewards/chosen": -0.11003254354000092, "rewards/margins": 0.4595148265361786, "rewards/rejected": -0.5695474147796631, "step": 568 }, { "epoch": 0.6499893093863588, "grad_norm": 91.72506895488603, "learning_rate": 1.6264588880407218e-07, "logits/chosen": -1.3987553119659424, "logits/rejected": -1.375367283821106, "logps/chosen": -150.62136840820312, "logps/rejected": -147.3133087158203, "loss": 0.655, "rewards/accuracies": 0.59375, "rewards/chosen": 0.012118123471736908, "rewards/margins": 0.23711130023002625, "rewards/rejected": -0.22499315440654755, "step": 570 }, { "epoch": 0.6522699736298196, "grad_norm": 107.88986883450896, "learning_rate": 1.6234898018587336e-07, "logits/chosen": -1.2692500352859497, "logits/rejected": -1.369061827659607, "logps/chosen": -139.50912475585938, "logps/rejected": -160.3131561279297, "loss": 0.6783, "rewards/accuracies": 0.46875, "rewards/chosen": -0.21677826344966888, "rewards/margins": 0.14994385838508606, "rewards/rejected": -0.36672213673591614, "step": 572 }, { "epoch": 0.6545506378732806, "grad_norm": 78.69743235560887, "learning_rate": 1.620511696467108e-07, "logits/chosen": -1.4847999811172485, "logits/rejected": -1.4455361366271973, "logps/chosen": -143.21315002441406, "logps/rejected": -142.0272674560547, "loss": 0.5931, "rewards/accuracies": 0.59375, "rewards/chosen": 0.023255709558725357, "rewards/margins": 0.20145481824874878, "rewards/rejected": -0.17819911241531372, "step": 574 }, { "epoch": 0.6568313021167415, "grad_norm": 88.95142783400384, "learning_rate": 1.6175246149461918e-07, "logits/chosen": -1.2740530967712402, "logits/rejected": -1.454386830329895, "logps/chosen": -151.84759521484375, "logps/rejected": -200.50332641601562, "loss": 0.6686, "rewards/accuracies": 0.5, "rewards/chosen": -0.14912785589694977, "rewards/margins": 0.19120272994041443, "rewards/rejected": -0.3403306007385254, "step": 576 }, { "epoch": 0.6591119663602024, "grad_norm": 106.08082954445281, "learning_rate": 1.614528600506178e-07, "logits/chosen": -1.3865227699279785, "logits/rejected": -1.375576138496399, "logps/chosen": -190.82093811035156, "logps/rejected": -221.74673461914062, "loss": 0.6078, "rewards/accuracies": 0.625, "rewards/chosen": 0.02210266701877117, "rewards/margins": 0.2864355444908142, "rewards/rejected": -0.2643328607082367, "step": 578 }, { "epoch": 0.6613926306036633, "grad_norm": 87.3395501088531, "learning_rate": 1.6115236964864796e-07, "logits/chosen": -1.4399878978729248, "logits/rejected": -1.4395971298217773, "logps/chosen": -173.22372436523438, "logps/rejected": -176.63436889648438, "loss": 0.6855, "rewards/accuracies": 0.59375, "rewards/chosen": -0.08596701920032501, "rewards/margins": 0.2504124939441681, "rewards/rejected": -0.3363794982433319, "step": 580 }, { "epoch": 0.6636732948471242, "grad_norm": 106.76668867973154, "learning_rate": 1.6085099463551042e-07, "logits/chosen": -1.4452544450759888, "logits/rejected": -1.5641038417816162, "logps/chosen": -168.6453094482422, "logps/rejected": -173.80926513671875, "loss": 0.6725, "rewards/accuracies": 0.625, "rewards/chosen": -0.173700749874115, "rewards/margins": 0.12216023355722427, "rewards/rejected": -0.29586100578308105, "step": 582 }, { "epoch": 0.6659539590905851, "grad_norm": 87.05178703169928, "learning_rate": 1.6054873937080243e-07, "logits/chosen": -1.304762601852417, "logits/rejected": -1.368096113204956, "logps/chosen": -172.9827423095703, "logps/rejected": -219.77606201171875, "loss": 0.6104, "rewards/accuracies": 0.6875, "rewards/chosen": -0.15153777599334717, "rewards/margins": 0.4469776451587677, "rewards/rejected": -0.5985154509544373, "step": 584 }, { "epoch": 0.668234623334046, "grad_norm": 110.06770845486187, "learning_rate": 1.6024560822685464e-07, "logits/chosen": -1.2189353704452515, "logits/rejected": -1.2109813690185547, "logps/chosen": -124.88150024414062, "logps/rejected": -134.57229614257812, "loss": 0.6398, "rewards/accuracies": 0.59375, "rewards/chosen": -0.04388846457004547, "rewards/margins": 0.09341102838516235, "rewards/rejected": -0.13729947805404663, "step": 586 }, { "epoch": 0.670515287577507, "grad_norm": 96.86048631567245, "learning_rate": 1.5994160558866803e-07, "logits/chosen": -1.4283928871154785, "logits/rejected": -1.454673171043396, "logps/chosen": -141.67201232910156, "logps/rejected": -165.86553955078125, "loss": 0.6099, "rewards/accuracies": 0.75, "rewards/chosen": -0.005323350429534912, "rewards/margins": 0.37266457080841064, "rewards/rejected": -0.37798795104026794, "step": 588 }, { "epoch": 0.6727959518209679, "grad_norm": 87.33194066270343, "learning_rate": 1.5963673585385014e-07, "logits/chosen": -1.3119065761566162, "logits/rejected": -1.3611897230148315, "logps/chosen": -149.3885498046875, "logps/rejected": -233.38589477539062, "loss": 0.6465, "rewards/accuracies": 0.71875, "rewards/chosen": -0.1451117843389511, "rewards/margins": 0.5241594314575195, "rewards/rejected": -0.6692711710929871, "step": 590 }, { "epoch": 0.6750766160644288, "grad_norm": 89.80632782962597, "learning_rate": 1.5933100343255182e-07, "logits/chosen": -1.478175401687622, "logits/rejected": -1.5088427066802979, "logps/chosen": -160.82077026367188, "logps/rejected": -175.51266479492188, "loss": 0.6521, "rewards/accuracies": 0.53125, "rewards/chosen": -0.026576366275548935, "rewards/margins": 0.050725750625133514, "rewards/rejected": -0.07730211317539215, "step": 592 }, { "epoch": 0.6773572803078897, "grad_norm": 74.35318626181845, "learning_rate": 1.5902441274740328e-07, "logits/chosen": -1.3951764106750488, "logits/rejected": -1.5363534688949585, "logps/chosen": -199.1707000732422, "logps/rejected": -223.73233032226562, "loss": 0.6397, "rewards/accuracies": 0.71875, "rewards/chosen": -0.12182176113128662, "rewards/margins": 0.23543262481689453, "rewards/rejected": -0.35725438594818115, "step": 594 }, { "epoch": 0.6796379445513506, "grad_norm": 94.79924897563647, "learning_rate": 1.5871696823344998e-07, "logits/chosen": -1.296017050743103, "logits/rejected": -1.3113356828689575, "logps/chosen": -217.41751098632812, "logps/rejected": -248.5911865234375, "loss": 0.6955, "rewards/accuracies": 0.5, "rewards/chosen": -0.08776041865348816, "rewards/margins": -0.076215460896492, "rewards/rejected": -0.011544965207576752, "step": 596 }, { "epoch": 0.6819186087948115, "grad_norm": 92.59481105091255, "learning_rate": 1.584086743380887e-07, "logits/chosen": -1.4293127059936523, "logits/rejected": -1.4433321952819824, "logps/chosen": -118.54342651367188, "logps/rejected": -134.85447692871094, "loss": 0.6368, "rewards/accuracies": 0.65625, "rewards/chosen": -0.17175720632076263, "rewards/margins": 0.14412283897399902, "rewards/rejected": -0.31588006019592285, "step": 598 }, { "epoch": 0.6841992730382724, "grad_norm": 89.66618993893586, "learning_rate": 1.580995355210031e-07, "logits/chosen": -1.2930192947387695, "logits/rejected": -1.368417501449585, "logps/chosen": -156.90586853027344, "logps/rejected": -204.53976440429688, "loss": 0.655, "rewards/accuracies": 0.59375, "rewards/chosen": -0.10334792733192444, "rewards/margins": 0.27799686789512634, "rewards/rejected": -0.3813447952270508, "step": 600 }, { "epoch": 0.6841992730382724, "eval_logits/chosen": -1.4766055345535278, "eval_logits/rejected": -1.4625221490859985, "eval_logps/chosen": -128.85867309570312, "eval_logps/rejected": -132.3487091064453, "eval_loss": 0.6277530789375305, "eval_rewards/accuracies": 0.6000000238418579, "eval_rewards/chosen": 0.10392862558364868, "eval_rewards/margins": 0.13252468407154083, "eval_rewards/rejected": -0.028596054762601852, "eval_runtime": 20.497, "eval_samples_per_second": 4.879, "eval_steps_per_second": 1.22, "step": 600 }, { "epoch": 0.6864799372817333, "grad_norm": 99.81191522440201, "learning_rate": 1.577895562540992e-07, "logits/chosen": -1.5043866634368896, "logits/rejected": -1.5639445781707764, "logps/chosen": -176.97195434570312, "logps/rejected": -200.05955505371094, "loss": 0.6298, "rewards/accuracies": 0.5, "rewards/chosen": -0.08353392779827118, "rewards/margins": 0.1179167851805687, "rewards/rejected": -0.20145073533058167, "step": 602 }, { "epoch": 0.6887606015251942, "grad_norm": 76.70830865796096, "learning_rate": 1.574787410214407e-07, "logits/chosen": -1.409602165222168, "logits/rejected": -1.4615368843078613, "logps/chosen": -155.466064453125, "logps/rejected": -179.92701721191406, "loss": 0.5929, "rewards/accuracies": 0.6875, "rewards/chosen": -0.06697350740432739, "rewards/margins": 0.43057531118392944, "rewards/rejected": -0.4975488483905792, "step": 604 }, { "epoch": 0.6910412657686551, "grad_norm": 94.3888285857028, "learning_rate": 1.571670943191841e-07, "logits/chosen": -1.4215627908706665, "logits/rejected": -1.4542145729064941, "logps/chosen": -179.4984130859375, "logps/rejected": -194.54330444335938, "loss": 0.6238, "rewards/accuracies": 0.59375, "rewards/chosen": -0.06045013293623924, "rewards/margins": 0.22754909098148346, "rewards/rejected": -0.2879992127418518, "step": 606 }, { "epoch": 0.6933219300121161, "grad_norm": 85.0198174251867, "learning_rate": 1.5685462065551372e-07, "logits/chosen": -1.3846269845962524, "logits/rejected": -1.403570532798767, "logps/chosen": -119.62281799316406, "logps/rejected": -122.60442352294922, "loss": 0.6716, "rewards/accuracies": 0.5625, "rewards/chosen": 0.05273966118693352, "rewards/margins": 0.23132431507110596, "rewards/rejected": -0.17858465015888214, "step": 608 }, { "epoch": 0.6956025942555769, "grad_norm": 106.39373760782738, "learning_rate": 1.5654132455057648e-07, "logits/chosen": -1.4965307712554932, "logits/rejected": -1.5246574878692627, "logps/chosen": -172.1771240234375, "logps/rejected": -178.22027587890625, "loss": 0.6898, "rewards/accuracies": 0.46875, "rewards/chosen": -0.24939650297164917, "rewards/margins": -0.02339550480246544, "rewards/rejected": -0.22600099444389343, "step": 610 }, { "epoch": 0.6978832584990379, "grad_norm": 90.59146954300502, "learning_rate": 1.562272105364164e-07, "logits/chosen": -1.2977298498153687, "logits/rejected": -1.356859803199768, "logps/chosen": -184.52252197265625, "logps/rejected": -211.69007873535156, "loss": 0.6426, "rewards/accuracies": 0.59375, "rewards/chosen": -0.14194564521312714, "rewards/margins": 0.10268077999353409, "rewards/rejected": -0.24462641775608063, "step": 612 }, { "epoch": 0.7001639227424987, "grad_norm": 132.7396200312291, "learning_rate": 1.5591228315690912e-07, "logits/chosen": -1.312849760055542, "logits/rejected": -1.3643529415130615, "logps/chosen": -183.94964599609375, "logps/rejected": -215.9215087890625, "loss": 0.6904, "rewards/accuracies": 0.65625, "rewards/chosen": -0.048665039241313934, "rewards/margins": 0.34853097796440125, "rewards/rejected": -0.3971959948539734, "step": 614 }, { "epoch": 0.7024445869859597, "grad_norm": 85.32152495878141, "learning_rate": 1.5559654696769627e-07, "logits/chosen": -1.3156163692474365, "logits/rejected": -1.3534281253814697, "logps/chosen": -135.82110595703125, "logps/rejected": -146.7247314453125, "loss": 0.6404, "rewards/accuracies": 0.625, "rewards/chosen": -0.19835428893566132, "rewards/margins": 0.1589125692844391, "rewards/rejected": -0.3572668731212616, "step": 616 }, { "epoch": 0.7047252512294205, "grad_norm": 97.72254578285961, "learning_rate": 1.5528000653611933e-07, "logits/chosen": -1.3361620903015137, "logits/rejected": -1.3171348571777344, "logps/chosen": -133.0448760986328, "logps/rejected": -157.213623046875, "loss": 0.6296, "rewards/accuracies": 0.5625, "rewards/chosen": -0.2093026041984558, "rewards/margins": 0.04571903124451637, "rewards/rejected": -0.2550216317176819, "step": 618 }, { "epoch": 0.7070059154728815, "grad_norm": 81.04476758820772, "learning_rate": 1.5496266644115383e-07, "logits/chosen": -1.335296869277954, "logits/rejected": -1.4067720174789429, "logps/chosen": -166.6826934814453, "logps/rejected": -201.6751251220703, "loss": 0.5956, "rewards/accuracies": 0.6875, "rewards/chosen": 0.02005193382501602, "rewards/margins": 0.4108864665031433, "rewards/rejected": -0.3908345103263855, "step": 620 }, { "epoch": 0.7092865797163423, "grad_norm": 117.48862378162204, "learning_rate": 1.5464453127334292e-07, "logits/chosen": -1.3234319686889648, "logits/rejected": -1.3931816816329956, "logps/chosen": -211.87327575683594, "logps/rejected": -244.9843292236328, "loss": 0.6435, "rewards/accuracies": 0.5625, "rewards/chosen": -0.23849377036094666, "rewards/margins": 0.13237711787223816, "rewards/rejected": -0.3708708584308624, "step": 622 }, { "epoch": 0.7115672439598033, "grad_norm": 97.28790849798257, "learning_rate": 1.5432560563473102e-07, "logits/chosen": -1.3715243339538574, "logits/rejected": -1.4115769863128662, "logps/chosen": -167.96871948242188, "logps/rejected": -195.65823364257812, "loss": 0.6435, "rewards/accuracies": 0.5625, "rewards/chosen": -0.18149994313716888, "rewards/margins": 0.21998891234397888, "rewards/rejected": -0.4014888405799866, "step": 624 }, { "epoch": 0.7138479082032642, "grad_norm": 74.77329414265908, "learning_rate": 1.5400589413879728e-07, "logits/chosen": -1.3941603899002075, "logits/rejected": -1.4356334209442139, "logps/chosen": -146.8162841796875, "logps/rejected": -166.09603881835938, "loss": 0.6115, "rewards/accuracies": 0.71875, "rewards/chosen": 0.0008982233703136444, "rewards/margins": 0.2602536678314209, "rewards/rejected": -0.25935542583465576, "step": 626 }, { "epoch": 0.7161285724467251, "grad_norm": 102.40110691278403, "learning_rate": 1.5368540141038877e-07, "logits/chosen": -1.33980393409729, "logits/rejected": -1.3720303773880005, "logps/chosen": -177.0899658203125, "logps/rejected": -177.4815216064453, "loss": 0.6489, "rewards/accuracies": 0.625, "rewards/chosen": -0.17625108361244202, "rewards/margins": 0.22537095844745636, "rewards/rejected": -0.40162205696105957, "step": 628 }, { "epoch": 0.718409236690186, "grad_norm": 77.60752469183791, "learning_rate": 1.5336413208565372e-07, "logits/chosen": -1.424604058265686, "logits/rejected": -1.4951881170272827, "logps/chosen": -217.2845458984375, "logps/rejected": -298.6929016113281, "loss": 0.5522, "rewards/accuracies": 0.625, "rewards/chosen": -0.31031134724617004, "rewards/margins": 0.769706130027771, "rewards/rejected": -1.0800174474716187, "step": 630 }, { "epoch": 0.7206899009336469, "grad_norm": 86.60572317721935, "learning_rate": 1.5304209081197424e-07, "logits/chosen": -1.2321975231170654, "logits/rejected": -1.1972179412841797, "logps/chosen": -162.22976684570312, "logps/rejected": -173.07118225097656, "loss": 0.651, "rewards/accuracies": 0.6875, "rewards/chosen": -0.0614384189248085, "rewards/margins": 0.27045127749443054, "rewards/rejected": -0.33188968896865845, "step": 632 }, { "epoch": 0.7229705651771078, "grad_norm": 105.40040114011227, "learning_rate": 1.5271928224789933e-07, "logits/chosen": -1.3975963592529297, "logits/rejected": -1.4231470823287964, "logps/chosen": -202.03836059570312, "logps/rejected": -236.25344848632812, "loss": 0.6755, "rewards/accuracies": 0.59375, "rewards/chosen": -0.32155218720436096, "rewards/margins": 0.17959336936473846, "rewards/rejected": -0.5011455416679382, "step": 634 }, { "epoch": 0.7252512294205687, "grad_norm": 81.36310946045379, "learning_rate": 1.5239571106307727e-07, "logits/chosen": -1.3148503303527832, "logits/rejected": -1.3296858072280884, "logps/chosen": -97.38423919677734, "logps/rejected": -119.96273803710938, "loss": 0.632, "rewards/accuracies": 0.6875, "rewards/chosen": 0.0010671375785022974, "rewards/margins": 0.2470298409461975, "rewards/rejected": -0.2459626942873001, "step": 636 }, { "epoch": 0.7275318936640296, "grad_norm": 96.83510386866747, "learning_rate": 1.5207138193818824e-07, "logits/chosen": -1.4544883966445923, "logits/rejected": -1.4232516288757324, "logps/chosen": -173.56967163085938, "logps/rejected": -202.63575744628906, "loss": 0.618, "rewards/accuracies": 0.59375, "rewards/chosen": -0.11890554428100586, "rewards/margins": 0.4641531705856323, "rewards/rejected": -0.5830587148666382, "step": 638 }, { "epoch": 0.7298125579074906, "grad_norm": 98.36905363836429, "learning_rate": 1.5174629956487657e-07, "logits/chosen": -1.3621355295181274, "logits/rejected": -1.470801591873169, "logps/chosen": -152.83348083496094, "logps/rejected": -180.09095764160156, "loss": 0.6283, "rewards/accuracies": 0.5625, "rewards/chosen": -0.1331956833600998, "rewards/margins": 0.12644681334495544, "rewards/rejected": -0.25964248180389404, "step": 640 }, { "epoch": 0.7320932221509515, "grad_norm": 99.21748796643946, "learning_rate": 1.514204686456828e-07, "logits/chosen": -1.4174845218658447, "logits/rejected": -1.5134220123291016, "logps/chosen": -199.45867919921875, "logps/rejected": -233.53509521484375, "loss": 0.6424, "rewards/accuracies": 0.71875, "rewards/chosen": 0.029536306858062744, "rewards/margins": 0.40678656101226807, "rewards/rejected": -0.3772502541542053, "step": 642 }, { "epoch": 0.7343738863944124, "grad_norm": 104.85775879115107, "learning_rate": 1.5109389389397564e-07, "logits/chosen": -1.309433102607727, "logits/rejected": -1.3806891441345215, "logps/chosen": -198.72012329101562, "logps/rejected": -229.50213623046875, "loss": 0.636, "rewards/accuracies": 0.6875, "rewards/chosen": -0.2979825735092163, "rewards/margins": 0.2780360281467438, "rewards/rejected": -0.5760185718536377, "step": 644 }, { "epoch": 0.7366545506378733, "grad_norm": 83.70012784480859, "learning_rate": 1.50766580033884e-07, "logits/chosen": -1.3813220262527466, "logits/rejected": -1.4268507957458496, "logps/chosen": -144.8968048095703, "logps/rejected": -168.29891967773438, "loss": 0.6301, "rewards/accuracies": 0.65625, "rewards/chosen": -0.2506921887397766, "rewards/margins": 0.17526662349700928, "rewards/rejected": -0.4259588122367859, "step": 646 }, { "epoch": 0.7389352148813342, "grad_norm": 76.15531374501609, "learning_rate": 1.5043853180022838e-07, "logits/chosen": -1.3109767436981201, "logits/rejected": -1.393604040145874, "logps/chosen": -123.48320770263672, "logps/rejected": -146.57994079589844, "loss": 0.627, "rewards/accuracies": 0.6875, "rewards/chosen": -0.030930444598197937, "rewards/margins": 0.237480029463768, "rewards/rejected": -0.26841047406196594, "step": 648 }, { "epoch": 0.7412158791247951, "grad_norm": 80.96947358910683, "learning_rate": 1.5010975393845257e-07, "logits/chosen": -1.249889612197876, "logits/rejected": -1.2592819929122925, "logps/chosen": -210.37823486328125, "logps/rejected": -225.4598388671875, "loss": 0.6471, "rewards/accuracies": 0.625, "rewards/chosen": -0.39817529916763306, "rewards/margins": 0.08833743631839752, "rewards/rejected": -0.4865127205848694, "step": 650 }, { "epoch": 0.743496543368256, "grad_norm": 87.1921118015952, "learning_rate": 1.4978025120455482e-07, "logits/chosen": -1.3601963520050049, "logits/rejected": -1.376826286315918, "logps/chosen": -167.55612182617188, "logps/rejected": -175.45620727539062, "loss": 0.6323, "rewards/accuracies": 0.75, "rewards/chosen": 0.03152412921190262, "rewards/margins": 0.33548736572265625, "rewards/rejected": -0.30396321415901184, "step": 652 }, { "epoch": 0.745777207611717, "grad_norm": 101.45108780703592, "learning_rate": 1.4945002836501935e-07, "logits/chosen": -1.4448864459991455, "logits/rejected": -1.5169280767440796, "logps/chosen": -251.17156982421875, "logps/rejected": -303.9085693359375, "loss": 0.6505, "rewards/accuracies": 0.5, "rewards/chosen": -0.29235783219337463, "rewards/margins": 0.28852301836013794, "rewards/rejected": -0.580880880355835, "step": 654 }, { "epoch": 0.7480578718551778, "grad_norm": 78.1814307088951, "learning_rate": 1.4911909019674702e-07, "logits/chosen": -1.3950117826461792, "logits/rejected": -1.3936371803283691, "logps/chosen": -221.95086669921875, "logps/rejected": -246.12425231933594, "loss": 0.7095, "rewards/accuracies": 0.625, "rewards/chosen": -0.40981170535087585, "rewards/margins": 0.04004380851984024, "rewards/rejected": -0.4498555362224579, "step": 656 }, { "epoch": 0.7503385360986388, "grad_norm": 86.3290682570179, "learning_rate": 1.4878744148698654e-07, "logits/chosen": -1.4538379907608032, "logits/rejected": -1.4509309530258179, "logps/chosen": -240.70957946777344, "logps/rejected": -277.9938049316406, "loss": 0.6274, "rewards/accuracies": 0.71875, "rewards/chosen": -0.10867351293563843, "rewards/margins": 0.7040268182754517, "rewards/rejected": -0.8127003908157349, "step": 658 }, { "epoch": 0.7526192003420996, "grad_norm": 76.663224249533, "learning_rate": 1.48455087033265e-07, "logits/chosen": -1.347740888595581, "logits/rejected": -1.4315268993377686, "logps/chosen": -175.90298461914062, "logps/rejected": -204.00741577148438, "loss": 0.589, "rewards/accuracies": 0.8125, "rewards/chosen": 0.0198584645986557, "rewards/margins": 0.3997357189655304, "rewards/rejected": -0.3798772692680359, "step": 660 }, { "epoch": 0.7548998645855606, "grad_norm": 89.45613833312, "learning_rate": 1.4812203164331865e-07, "logits/chosen": -1.3782517910003662, "logits/rejected": -1.3579304218292236, "logps/chosen": -106.40101623535156, "logps/rejected": -115.47900390625, "loss": 0.6276, "rewards/accuracies": 0.6875, "rewards/chosen": 0.002452205866575241, "rewards/margins": 0.2672392427921295, "rewards/rejected": -0.2647870182991028, "step": 662 }, { "epoch": 0.7571805288290214, "grad_norm": 90.62698292022132, "learning_rate": 1.4778828013502315e-07, "logits/chosen": -1.3181474208831787, "logits/rejected": -1.2984850406646729, "logps/chosen": -121.110595703125, "logps/rejected": -144.5844268798828, "loss": 0.6345, "rewards/accuracies": 0.46875, "rewards/chosen": -0.14049722254276276, "rewards/margins": 0.16425339877605438, "rewards/rejected": -0.30475062131881714, "step": 664 }, { "epoch": 0.7594611930724824, "grad_norm": 90.46387262398376, "learning_rate": 1.474538373363241e-07, "logits/chosen": -1.2038509845733643, "logits/rejected": -1.1771519184112549, "logps/chosen": -123.54405212402344, "logps/rejected": -128.04415893554688, "loss": 0.6513, "rewards/accuracies": 0.6875, "rewards/chosen": -0.0023752544075250626, "rewards/margins": 0.20540934801101685, "rewards/rejected": -0.20778462290763855, "step": 666 }, { "epoch": 0.7617418573159432, "grad_norm": 98.8814238604518, "learning_rate": 1.4711870808516705e-07, "logits/chosen": -1.3800256252288818, "logits/rejected": -1.4212613105773926, "logps/chosen": -143.5724639892578, "logps/rejected": -149.91879272460938, "loss": 0.6367, "rewards/accuracies": 0.71875, "rewards/chosen": -0.25999924540519714, "rewards/margins": 0.30095675587654114, "rewards/rejected": -0.5609559416770935, "step": 668 }, { "epoch": 0.7640225215594042, "grad_norm": 91.14237505783925, "learning_rate": 1.4678289722942755e-07, "logits/chosen": -1.4362088441848755, "logits/rejected": -1.4652812480926514, "logps/chosen": -238.3679962158203, "logps/rejected": -245.81390380859375, "loss": 0.6248, "rewards/accuracies": 0.625, "rewards/chosen": -0.2839330732822418, "rewards/margins": 0.23081044852733612, "rewards/rejected": -0.5147435069084167, "step": 670 }, { "epoch": 0.766303185802865, "grad_norm": 92.40909256739194, "learning_rate": 1.4644640962684106e-07, "logits/chosen": -1.2180765867233276, "logits/rejected": -1.2128483057022095, "logps/chosen": -139.85572814941406, "logps/rejected": -155.68014526367188, "loss": 0.6594, "rewards/accuracies": 0.65625, "rewards/chosen": -0.31525713205337524, "rewards/margins": 0.2499125599861145, "rewards/rejected": -0.5651696920394897, "step": 672 }, { "epoch": 0.768583850046326, "grad_norm": 85.52183536423232, "learning_rate": 1.461092501449326e-07, "logits/chosen": -1.432080864906311, "logits/rejected": -1.4694151878356934, "logps/chosen": -192.5015411376953, "logps/rejected": -206.3730926513672, "loss": 0.6026, "rewards/accuracies": 0.6875, "rewards/chosen": -0.06770830601453781, "rewards/margins": 0.42051053047180176, "rewards/rejected": -0.48821884393692017, "step": 674 }, { "epoch": 0.7708645142897869, "grad_norm": 90.10321239819011, "learning_rate": 1.4577142366094641e-07, "logits/chosen": -1.357291579246521, "logits/rejected": -1.4703514575958252, "logps/chosen": -159.84365844726562, "logps/rejected": -179.19932556152344, "loss": 0.6417, "rewards/accuracies": 0.6875, "rewards/chosen": -0.26335883140563965, "rewards/margins": 0.19603171944618225, "rewards/rejected": -0.4593905210494995, "step": 676 }, { "epoch": 0.7731451785332478, "grad_norm": 89.71495801422127, "learning_rate": 1.454329350617754e-07, "logits/chosen": -1.2440226078033447, "logits/rejected": -1.2795155048370361, "logps/chosen": -142.0009765625, "logps/rejected": -174.45877075195312, "loss": 0.8425, "rewards/accuracies": 0.75, "rewards/chosen": -0.18753905594348907, "rewards/margins": 0.41140902042388916, "rewards/rejected": -0.598948061466217, "step": 678 }, { "epoch": 0.7754258427767087, "grad_norm": 86.78570750881197, "learning_rate": 1.4509378924389042e-07, "logits/chosen": -1.3585668802261353, "logits/rejected": -1.4263851642608643, "logps/chosen": -138.0133056640625, "logps/rejected": -149.3664093017578, "loss": 0.6293, "rewards/accuracies": 0.5, "rewards/chosen": -0.08084426820278168, "rewards/margins": 0.16473475098609924, "rewards/rejected": -0.24557898938655853, "step": 680 }, { "epoch": 0.7777065070201696, "grad_norm": 107.27075654983396, "learning_rate": 1.4475399111326942e-07, "logits/chosen": -1.3274388313293457, "logits/rejected": -1.4092543125152588, "logps/chosen": -205.84066772460938, "logps/rejected": -229.88511657714844, "loss": 0.6449, "rewards/accuracies": 0.65625, "rewards/chosen": -0.28960123658180237, "rewards/margins": 0.23577138781547546, "rewards/rejected": -0.5253726243972778, "step": 682 }, { "epoch": 0.7799871712636305, "grad_norm": 78.77386252056206, "learning_rate": 1.4441354558532653e-07, "logits/chosen": -1.3599066734313965, "logits/rejected": -1.4156947135925293, "logps/chosen": -178.33787536621094, "logps/rejected": -178.77232360839844, "loss": 0.644, "rewards/accuracies": 0.6875, "rewards/chosen": -0.20014306902885437, "rewards/margins": 0.3175792396068573, "rewards/rejected": -0.5177222490310669, "step": 684 }, { "epoch": 0.7822678355070914, "grad_norm": 96.56956615564326, "learning_rate": 1.4407245758484092e-07, "logits/chosen": -1.4600579738616943, "logits/rejected": -1.4812790155410767, "logps/chosen": -200.6736602783203, "logps/rejected": -215.11302185058594, "loss": 0.6408, "rewards/accuracies": 0.625, "rewards/chosen": -0.31072649359703064, "rewards/margins": 0.2427929937839508, "rewards/rejected": -0.5535194873809814, "step": 686 }, { "epoch": 0.7845484997505523, "grad_norm": 93.7282766427954, "learning_rate": 1.4373073204588556e-07, "logits/chosen": -1.392540454864502, "logits/rejected": -1.5576982498168945, "logps/chosen": -173.74102783203125, "logps/rejected": -209.58255004882812, "loss": 0.6396, "rewards/accuracies": 0.625, "rewards/chosen": -0.07411661744117737, "rewards/margins": 0.2528524100780487, "rewards/rejected": -0.32696905732154846, "step": 688 }, { "epoch": 0.7868291639940133, "grad_norm": 78.03441039978401, "learning_rate": 1.433883739117558e-07, "logits/chosen": -1.4191378355026245, "logits/rejected": -1.376889705657959, "logps/chosen": -167.5023651123047, "logps/rejected": -175.2780303955078, "loss": 0.6232, "rewards/accuracies": 0.6875, "rewards/chosen": 0.10266311466693878, "rewards/margins": 0.33145037293434143, "rewards/rejected": -0.22878730297088623, "step": 690 }, { "epoch": 0.7891098282374741, "grad_norm": 102.68741637989442, "learning_rate": 1.4304538813489807e-07, "logits/chosen": -1.365789771080017, "logits/rejected": -1.4465529918670654, "logps/chosen": -241.91314697265625, "logps/rejected": -287.3153381347656, "loss": 0.626, "rewards/accuracies": 0.71875, "rewards/chosen": -0.22250030934810638, "rewards/margins": 0.3011859655380249, "rewards/rejected": -0.5236862301826477, "step": 692 }, { "epoch": 0.7913904924809351, "grad_norm": 126.01863649654494, "learning_rate": 1.4270177967683795e-07, "logits/chosen": -1.1870087385177612, "logits/rejected": -1.2576615810394287, "logps/chosen": -193.8482666015625, "logps/rejected": -214.00135803222656, "loss": 0.683, "rewards/accuracies": 0.5625, "rewards/chosen": -0.6029391884803772, "rewards/margins": 0.08865418285131454, "rewards/rejected": -0.69159334897995, "step": 694 }, { "epoch": 0.793671156724396, "grad_norm": 105.78445666178517, "learning_rate": 1.4235755350810852e-07, "logits/chosen": -1.4039561748504639, "logits/rejected": -1.483564853668213, "logps/chosen": -131.8105926513672, "logps/rejected": -161.2996368408203, "loss": 0.5968, "rewards/accuracies": 0.71875, "rewards/chosen": -0.06916500627994537, "rewards/margins": 0.4792707562446594, "rewards/rejected": -0.5484358072280884, "step": 696 }, { "epoch": 0.7959518209678569, "grad_norm": 111.82639934813994, "learning_rate": 1.420127146081786e-07, "logits/chosen": -1.182243824005127, "logits/rejected": -1.2272214889526367, "logps/chosen": -142.89569091796875, "logps/rejected": -167.72943115234375, "loss": 0.7077, "rewards/accuracies": 0.4375, "rewards/chosen": -0.33521801233291626, "rewards/margins": 0.031463623046875, "rewards/rejected": -0.36668163537979126, "step": 698 }, { "epoch": 0.7982324852113178, "grad_norm": 93.0448822586256, "learning_rate": 1.4166726796538043e-07, "logits/chosen": -1.337456464767456, "logits/rejected": -1.38818359375, "logps/chosen": -161.9310760498047, "logps/rejected": -177.00717163085938, "loss": 0.5943, "rewards/accuracies": 0.6875, "rewards/chosen": -0.11526203155517578, "rewards/margins": 0.27307793498039246, "rewards/rejected": -0.38833993673324585, "step": 700 }, { "epoch": 0.7982324852113178, "eval_logits/chosen": -1.4622485637664795, "eval_logits/rejected": -1.4484994411468506, "eval_logps/chosen": -129.2548065185547, "eval_logps/rejected": -133.13601684570312, "eval_loss": 0.6083559989929199, "eval_rewards/accuracies": 0.6399999856948853, "eval_rewards/chosen": 0.06431641429662704, "eval_rewards/margins": 0.17164374887943268, "eval_rewards/rejected": -0.10732734948396683, "eval_runtime": 19.822, "eval_samples_per_second": 5.045, "eval_steps_per_second": 1.261, "step": 700 }, { "epoch": 0.8005131494547787, "grad_norm": 81.54591894736863, "learning_rate": 1.413212185768378e-07, "logits/chosen": -1.3647618293762207, "logits/rejected": -1.4832388162612915, "logps/chosen": -163.0651092529297, "logps/rejected": -199.4841766357422, "loss": 0.6247, "rewards/accuracies": 0.8125, "rewards/chosen": 0.14452999830245972, "rewards/margins": 0.4548623561859131, "rewards/rejected": -0.31033238768577576, "step": 702 }, { "epoch": 0.8027938136982397, "grad_norm": 88.3455225094938, "learning_rate": 1.409745714483936e-07, "logits/chosen": -1.3221628665924072, "logits/rejected": -1.379399299621582, "logps/chosen": -124.75774383544922, "logps/rejected": -142.49942016601562, "loss": 0.6477, "rewards/accuracies": 0.65625, "rewards/chosen": -0.004875157028436661, "rewards/margins": 0.23325148224830627, "rewards/rejected": -0.23812663555145264, "step": 704 }, { "epoch": 0.8050744779417005, "grad_norm": 104.14495646431855, "learning_rate": 1.406273315945374e-07, "logits/chosen": -1.4262161254882812, "logits/rejected": -1.460605502128601, "logps/chosen": -188.75218200683594, "logps/rejected": -196.2736358642578, "loss": 0.6307, "rewards/accuracies": 0.625, "rewards/chosen": -0.09237577021121979, "rewards/margins": 0.26734021306037903, "rewards/rejected": -0.35971593856811523, "step": 706 }, { "epoch": 0.8073551421851615, "grad_norm": 88.27883050256283, "learning_rate": 1.4027950403833294e-07, "logits/chosen": -1.3447182178497314, "logits/rejected": -1.3849260807037354, "logps/chosen": -194.97628784179688, "logps/rejected": -231.31488037109375, "loss": 0.608, "rewards/accuracies": 0.65625, "rewards/chosen": -0.018238365650177002, "rewards/margins": 0.4262167513370514, "rewards/rejected": -0.444455087184906, "step": 708 }, { "epoch": 0.8096358064286223, "grad_norm": 83.81720367834012, "learning_rate": 1.3993109381134552e-07, "logits/chosen": -1.458938717842102, "logits/rejected": -1.4975433349609375, "logps/chosen": -217.40234375, "logps/rejected": -230.56849670410156, "loss": 0.5997, "rewards/accuracies": 0.6875, "rewards/chosen": -0.06927414983510971, "rewards/margins": 0.34377092123031616, "rewards/rejected": -0.41304510831832886, "step": 710 }, { "epoch": 0.8119164706720833, "grad_norm": 83.92843092918012, "learning_rate": 1.3958210595356923e-07, "logits/chosen": -1.295701265335083, "logits/rejected": -1.3689717054367065, "logps/chosen": -184.71035766601562, "logps/rejected": -231.5858154296875, "loss": 0.6496, "rewards/accuracies": 0.75, "rewards/chosen": -0.07217526435852051, "rewards/margins": 0.5205077528953552, "rewards/rejected": -0.5926830768585205, "step": 712 }, { "epoch": 0.8141971349155441, "grad_norm": 64.10905806570788, "learning_rate": 1.3923254551335385e-07, "logits/chosen": -1.3552354574203491, "logits/rejected": -1.5121649503707886, "logps/chosen": -162.65443420410156, "logps/rejected": -202.2783203125, "loss": 0.6038, "rewards/accuracies": 0.65625, "rewards/chosen": -0.21346963942050934, "rewards/margins": 0.2602553367614746, "rewards/rejected": -0.47372499108314514, "step": 714 }, { "epoch": 0.8164777991590051, "grad_norm": 96.65072480066111, "learning_rate": 1.388824175473321e-07, "logits/chosen": -1.4826998710632324, "logits/rejected": -1.525536060333252, "logps/chosen": -157.91253662109375, "logps/rejected": -189.42970275878906, "loss": 0.6418, "rewards/accuracies": 0.5625, "rewards/chosen": -0.15659984946250916, "rewards/margins": 0.5801843404769897, "rewards/rejected": -0.7367842197418213, "step": 716 }, { "epoch": 0.8187584634024659, "grad_norm": 133.38988345223416, "learning_rate": 1.3853172712034618e-07, "logits/chosen": -1.453382968902588, "logits/rejected": -1.4554516077041626, "logps/chosen": -123.51729583740234, "logps/rejected": -133.98687744140625, "loss": 0.6316, "rewards/accuracies": 0.5625, "rewards/chosen": -0.17259877920150757, "rewards/margins": 0.09528855979442596, "rewards/rejected": -0.26788732409477234, "step": 718 }, { "epoch": 0.8210391276459269, "grad_norm": 103.6833810750344, "learning_rate": 1.3818047930537489e-07, "logits/chosen": -1.4238344430923462, "logits/rejected": -1.4430863857269287, "logps/chosen": -185.29090881347656, "logps/rejected": -190.6168212890625, "loss": 0.6817, "rewards/accuracies": 0.5, "rewards/chosen": -0.1492576152086258, "rewards/margins": 0.1548936814069748, "rewards/rejected": -0.304151326417923, "step": 720 }, { "epoch": 0.8233197918893878, "grad_norm": 104.78480311225195, "learning_rate": 1.3782867918345986e-07, "logits/chosen": -1.4426988363265991, "logits/rejected": -1.474928855895996, "logps/chosen": -182.13693237304688, "logps/rejected": -207.8540802001953, "loss": 0.688, "rewards/accuracies": 0.5625, "rewards/chosen": -0.28544071316719055, "rewards/margins": 0.010512780398130417, "rewards/rejected": -0.2959534525871277, "step": 722 }, { "epoch": 0.8256004561328487, "grad_norm": 83.99721091366573, "learning_rate": 1.374763318436323e-07, "logits/chosen": -1.3846015930175781, "logits/rejected": -1.374580979347229, "logps/chosen": -188.639892578125, "logps/rejected": -203.4746856689453, "loss": 0.6455, "rewards/accuracies": 0.59375, "rewards/chosen": -0.09084916859865189, "rewards/margins": 0.13614203035831451, "rewards/rejected": -0.2269912213087082, "step": 724 }, { "epoch": 0.8278811203763096, "grad_norm": 100.80568473474456, "learning_rate": 1.371234423828393e-07, "logits/chosen": -1.3973044157028198, "logits/rejected": -1.4167506694793701, "logps/chosen": -221.90057373046875, "logps/rejected": -223.55625915527344, "loss": 0.6358, "rewards/accuracies": 0.5, "rewards/chosen": -0.5376750826835632, "rewards/margins": 0.1100195124745369, "rewards/rejected": -0.6476945281028748, "step": 726 }, { "epoch": 0.8301617846197705, "grad_norm": 95.2695276072168, "learning_rate": 1.367700159058701e-07, "logits/chosen": -1.304479718208313, "logits/rejected": -1.3024879693984985, "logps/chosen": -132.629150390625, "logps/rejected": -157.943115234375, "loss": 0.6385, "rewards/accuracies": 0.625, "rewards/chosen": -0.17815637588500977, "rewards/margins": 0.16010738909244537, "rewards/rejected": -0.33826377987861633, "step": 728 }, { "epoch": 0.8324424488632314, "grad_norm": 84.57675708069675, "learning_rate": 1.3641605752528223e-07, "logits/chosen": -1.345677137374878, "logits/rejected": -1.369272232055664, "logps/chosen": -174.67979431152344, "logps/rejected": -177.95265197753906, "loss": 0.6631, "rewards/accuracies": 0.375, "rewards/chosen": -0.33565065264701843, "rewards/margins": -0.10611479729413986, "rewards/rejected": -0.22953587770462036, "step": 730 }, { "epoch": 0.8347231131066923, "grad_norm": 95.98437315557469, "learning_rate": 1.3606157236132753e-07, "logits/chosen": -1.3388769626617432, "logits/rejected": -1.4168953895568848, "logps/chosen": -128.86917114257812, "logps/rejected": -155.26756286621094, "loss": 0.5995, "rewards/accuracies": 0.65625, "rewards/chosen": -0.12645135819911957, "rewards/margins": 0.4350703954696655, "rewards/rejected": -0.5615217685699463, "step": 732 }, { "epoch": 0.8370037773501532, "grad_norm": 98.72975210848782, "learning_rate": 1.3570656554187823e-07, "logits/chosen": -1.293025255203247, "logits/rejected": -1.3117451667785645, "logps/chosen": -198.599609375, "logps/rejected": -195.0568084716797, "loss": 0.6664, "rewards/accuracies": 0.53125, "rewards/chosen": -0.3894595205783844, "rewards/margins": 0.01842033676803112, "rewards/rejected": -0.40787985920906067, "step": 734 }, { "epoch": 0.8392844415936142, "grad_norm": 83.44989797339352, "learning_rate": 1.353510422023526e-07, "logits/chosen": -1.3290454149246216, "logits/rejected": -1.3762670755386353, "logps/chosen": -108.33480834960938, "logps/rejected": -132.43887329101562, "loss": 0.6699, "rewards/accuracies": 0.625, "rewards/chosen": -0.07590903341770172, "rewards/margins": 0.31655895709991455, "rewards/rejected": -0.3924679458141327, "step": 736 }, { "epoch": 0.841565105837075, "grad_norm": 94.8310467886239, "learning_rate": 1.3499500748564075e-07, "logits/chosen": -1.4289618730545044, "logits/rejected": -1.4254275560379028, "logps/chosen": -111.20790100097656, "logps/rejected": -120.1327896118164, "loss": 0.6651, "rewards/accuracies": 0.5, "rewards/chosen": -0.06255226582288742, "rewards/margins": 0.16382303833961487, "rewards/rejected": -0.2263752967119217, "step": 738 }, { "epoch": 0.843845770080536, "grad_norm": 106.55945054665669, "learning_rate": 1.346384665420302e-07, "logits/chosen": -1.2947108745574951, "logits/rejected": -1.3571364879608154, "logps/chosen": -157.76895141601562, "logps/rejected": -187.6517333984375, "loss": 0.6321, "rewards/accuracies": 0.75, "rewards/chosen": -0.14570750296115875, "rewards/margins": 0.39257413148880005, "rewards/rejected": -0.5382816791534424, "step": 740 }, { "epoch": 0.8461264343239968, "grad_norm": 112.76083171183551, "learning_rate": 1.3428142452913133e-07, "logits/chosen": -1.4128669500350952, "logits/rejected": -1.4613770246505737, "logps/chosen": -202.3125457763672, "logps/rejected": -242.2345428466797, "loss": 0.6542, "rewards/accuracies": 0.6875, "rewards/chosen": -0.27778810262680054, "rewards/margins": 0.3144941031932831, "rewards/rejected": -0.592282235622406, "step": 742 }, { "epoch": 0.8484070985674578, "grad_norm": 100.11503458436644, "learning_rate": 1.3392388661180302e-07, "logits/chosen": -1.3573894500732422, "logits/rejected": -1.344806432723999, "logps/chosen": -186.35108947753906, "logps/rejected": -188.49490356445312, "loss": 0.6701, "rewards/accuracies": 0.5625, "rewards/chosen": -0.16072505712509155, "rewards/margins": 0.19017082452774048, "rewards/rejected": -0.3508959114551544, "step": 744 }, { "epoch": 0.8506877628109187, "grad_norm": 91.60219394383267, "learning_rate": 1.3356585796207755e-07, "logits/chosen": -1.3965177536010742, "logits/rejected": -1.4244239330291748, "logps/chosen": -201.50318908691406, "logps/rejected": -217.1870574951172, "loss": 0.6803, "rewards/accuracies": 0.6875, "rewards/chosen": 0.007203985005617142, "rewards/margins": 0.4475135803222656, "rewards/rejected": -0.44030967354774475, "step": 746 }, { "epoch": 0.8529684270543796, "grad_norm": 93.27202711374441, "learning_rate": 1.3320734375908606e-07, "logits/chosen": -1.4266613721847534, "logits/rejected": -1.4168901443481445, "logps/chosen": -208.13624572753906, "logps/rejected": -238.91502380371094, "loss": 0.6581, "rewards/accuracies": 0.65625, "rewards/chosen": -0.2992842197418213, "rewards/margins": 0.2575336992740631, "rewards/rejected": -0.556817889213562, "step": 748 }, { "epoch": 0.8552490912978405, "grad_norm": 99.8766408873739, "learning_rate": 1.328483491889836e-07, "logits/chosen": -1.427262306213379, "logits/rejected": -1.4984745979309082, "logps/chosen": -215.9974365234375, "logps/rejected": -236.14759826660156, "loss": 0.6592, "rewards/accuracies": 0.59375, "rewards/chosen": -0.048569947481155396, "rewards/margins": 0.15008097887039185, "rewards/rejected": -0.19865092635154724, "step": 750 }, { "epoch": 0.8575297555413014, "grad_norm": 82.7550556411054, "learning_rate": 1.3248887944487393e-07, "logits/chosen": -1.410501480102539, "logits/rejected": -1.452683687210083, "logps/chosen": -122.24147033691406, "logps/rejected": -137.63116455078125, "loss": 0.6134, "rewards/accuracies": 0.6875, "rewards/chosen": 0.11725883185863495, "rewards/margins": 0.1699836403131485, "rewards/rejected": -0.05272480472922325, "step": 752 }, { "epoch": 0.8598104197847624, "grad_norm": 88.9952584850963, "learning_rate": 1.321289397267347e-07, "logits/chosen": -1.4281740188598633, "logits/rejected": -1.445003628730774, "logps/chosen": -177.8570556640625, "logps/rejected": -181.70663452148438, "loss": 0.6919, "rewards/accuracies": 0.5625, "rewards/chosen": -0.2182515561580658, "rewards/margins": 0.0669536367058754, "rewards/rejected": -0.28520524501800537, "step": 754 }, { "epoch": 0.8620910840282232, "grad_norm": 100.22335380917765, "learning_rate": 1.3176853524134198e-07, "logits/chosen": -1.391309142112732, "logits/rejected": -1.4018808603286743, "logps/chosen": -164.9562530517578, "logps/rejected": -178.225341796875, "loss": 0.6706, "rewards/accuracies": 0.59375, "rewards/chosen": -0.04197298735380173, "rewards/margins": 0.15188831090927124, "rewards/rejected": -0.19386129081249237, "step": 756 }, { "epoch": 0.8643717482716842, "grad_norm": 100.1727346446802, "learning_rate": 1.314076712021949e-07, "logits/chosen": -1.35850989818573, "logits/rejected": -1.420215129852295, "logps/chosen": -157.33843994140625, "logps/rejected": -192.61769104003906, "loss": 0.6228, "rewards/accuracies": 0.625, "rewards/chosen": -0.10978934168815613, "rewards/margins": 0.24940092861652374, "rewards/rejected": -0.3591902256011963, "step": 758 }, { "epoch": 0.866652412515145, "grad_norm": 89.59203114100742, "learning_rate": 1.3104635282944052e-07, "logits/chosen": -1.4790459871292114, "logits/rejected": -1.4666205644607544, "logps/chosen": -161.86483764648438, "logps/rejected": -211.35423278808594, "loss": 0.6404, "rewards/accuracies": 0.71875, "rewards/chosen": 0.004301717504858971, "rewards/margins": 0.35839876532554626, "rewards/rejected": -0.35409703850746155, "step": 760 }, { "epoch": 0.868933076758606, "grad_norm": 94.6581914253334, "learning_rate": 1.3068458534979812e-07, "logits/chosen": -1.3811688423156738, "logits/rejected": -1.4190725088119507, "logps/chosen": -80.00721740722656, "logps/rejected": -101.73347473144531, "loss": 0.6688, "rewards/accuracies": 0.625, "rewards/chosen": -0.02035554125905037, "rewards/margins": 0.07091270387172699, "rewards/rejected": -0.09126824140548706, "step": 762 }, { "epoch": 0.8712137410020668, "grad_norm": 84.05822092329534, "learning_rate": 1.3032237399648357e-07, "logits/chosen": -1.3661377429962158, "logits/rejected": -1.3815851211547852, "logps/chosen": -106.82656860351562, "logps/rejected": -128.75656127929688, "loss": 0.6309, "rewards/accuracies": 0.5625, "rewards/chosen": -0.03877873718738556, "rewards/margins": 0.08257640898227692, "rewards/rejected": -0.12135513871908188, "step": 764 }, { "epoch": 0.8734944052455278, "grad_norm": 99.58060667623182, "learning_rate": 1.2995972400913367e-07, "logits/chosen": -1.4144562482833862, "logits/rejected": -1.4412004947662354, "logps/chosen": -136.15481567382812, "logps/rejected": -141.8118438720703, "loss": 0.6589, "rewards/accuracies": 0.65625, "rewards/chosen": -0.0782175287604332, "rewards/margins": 0.17598725855350494, "rewards/rejected": -0.25420477986335754, "step": 766 }, { "epoch": 0.8757750694889886, "grad_norm": 80.17753876582776, "learning_rate": 1.2959664063373042e-07, "logits/chosen": -1.3012042045593262, "logits/rejected": -1.2482383251190186, "logps/chosen": -100.65060424804688, "logps/rejected": -112.65821838378906, "loss": 0.5368, "rewards/accuracies": 0.78125, "rewards/chosen": 0.04776890203356743, "rewards/margins": 0.4462125897407532, "rewards/rejected": -0.39844369888305664, "step": 768 }, { "epoch": 0.8780557337324496, "grad_norm": 88.16104383702894, "learning_rate": 1.2923312912252506e-07, "logits/chosen": -1.4704458713531494, "logits/rejected": -1.5337769985198975, "logps/chosen": -118.58712768554688, "logps/rejected": -128.90980529785156, "loss": 0.6971, "rewards/accuracies": 0.53125, "rewards/chosen": 0.005523152183741331, "rewards/margins": 0.031648553907871246, "rewards/rejected": -0.02612539939582348, "step": 770 }, { "epoch": 0.8803363979759105, "grad_norm": 88.57684727206627, "learning_rate": 1.288691947339621e-07, "logits/chosen": -1.4641995429992676, "logits/rejected": -1.4404255151748657, "logps/chosen": -188.07818603515625, "logps/rejected": -189.4072265625, "loss": 0.6435, "rewards/accuracies": 0.65625, "rewards/chosen": -0.11661610007286072, "rewards/margins": 0.22115452587604523, "rewards/rejected": -0.33777061104774475, "step": 772 }, { "epoch": 0.8826170622193714, "grad_norm": 88.03404600894119, "learning_rate": 1.2850484273260325e-07, "logits/chosen": -1.2959342002868652, "logits/rejected": -1.3783100843429565, "logps/chosen": -205.9984130859375, "logps/rejected": -227.6925506591797, "loss": 0.6664, "rewards/accuracies": 0.59375, "rewards/chosen": -0.16467887163162231, "rewards/margins": 0.33329781889915466, "rewards/rejected": -0.497976690530777, "step": 774 }, { "epoch": 0.8848977264628323, "grad_norm": 85.2906054415725, "learning_rate": 1.2814007838905129e-07, "logits/chosen": -1.3777484893798828, "logits/rejected": -1.4851555824279785, "logps/chosen": -169.7268524169922, "logps/rejected": -183.84872436523438, "loss": 0.5976, "rewards/accuracies": 0.65625, "rewards/chosen": -0.050482071936130524, "rewards/margins": 0.21894827485084534, "rewards/rejected": -0.26943036913871765, "step": 776 }, { "epoch": 0.8871783907062932, "grad_norm": 113.67510179823725, "learning_rate": 1.2777490697987375e-07, "logits/chosen": -1.361803412437439, "logits/rejected": -1.4689048528671265, "logps/chosen": -188.6908416748047, "logps/rejected": -239.35243225097656, "loss": 0.6593, "rewards/accuracies": 0.71875, "rewards/chosen": 0.024178601801395416, "rewards/margins": 0.5260721445083618, "rewards/rejected": -0.5018935203552246, "step": 778 }, { "epoch": 0.8894590549497541, "grad_norm": 77.93754881822308, "learning_rate": 1.2740933378752683e-07, "logits/chosen": -1.4177353382110596, "logits/rejected": -1.4752013683319092, "logps/chosen": -196.52670288085938, "logps/rejected": -216.70066833496094, "loss": 0.5885, "rewards/accuracies": 0.75, "rewards/chosen": -0.18388278782367706, "rewards/margins": 0.34562817215919495, "rewards/rejected": -0.5295109748840332, "step": 780 }, { "epoch": 0.891739719193215, "grad_norm": 85.24959090845118, "learning_rate": 1.2704336410027862e-07, "logits/chosen": -1.4634852409362793, "logits/rejected": -1.5205847024917603, "logps/chosen": -250.54759216308594, "logps/rejected": -285.6510009765625, "loss": 0.6189, "rewards/accuracies": 0.53125, "rewards/chosen": -0.32170242071151733, "rewards/margins": 0.21656639873981476, "rewards/rejected": -0.5382688045501709, "step": 782 }, { "epoch": 0.8940203834366759, "grad_norm": 94.62957856064564, "learning_rate": 1.2667700321213279e-07, "logits/chosen": -1.216140627861023, "logits/rejected": -1.3199231624603271, "logps/chosen": -163.63653564453125, "logps/rejected": -182.559814453125, "loss": 0.6221, "rewards/accuracies": 0.53125, "rewards/chosen": -0.21417541801929474, "rewards/margins": 0.07622166723012924, "rewards/rejected": -0.29039713740348816, "step": 784 }, { "epoch": 0.8963010476801369, "grad_norm": 84.13982267155843, "learning_rate": 1.263102564227521e-07, "logits/chosen": -1.3924446105957031, "logits/rejected": -1.3973028659820557, "logps/chosen": -164.72738647460938, "logps/rejected": -185.91358947753906, "loss": 0.6378, "rewards/accuracies": 0.6875, "rewards/chosen": -0.017229687422513962, "rewards/margins": 0.29563748836517334, "rewards/rejected": -0.3128671646118164, "step": 786 }, { "epoch": 0.8985817119235977, "grad_norm": 92.91613247214089, "learning_rate": 1.2594312903738162e-07, "logits/chosen": -1.4232394695281982, "logits/rejected": -1.5436615943908691, "logps/chosen": -137.60520935058594, "logps/rejected": -159.5936737060547, "loss": 0.6099, "rewards/accuracies": 0.6875, "rewards/chosen": -0.14493076503276825, "rewards/margins": 0.23380112648010254, "rewards/rejected": -0.3787318468093872, "step": 788 }, { "epoch": 0.9008623761670587, "grad_norm": 109.24475281169839, "learning_rate": 1.2557562636677192e-07, "logits/chosen": -1.3665200471878052, "logits/rejected": -1.4350488185882568, "logps/chosen": -191.39637756347656, "logps/rejected": -255.76321411132812, "loss": 0.6655, "rewards/accuracies": 0.6875, "rewards/chosen": -0.16933086514472961, "rewards/margins": 0.5223888158798218, "rewards/rejected": -0.691719651222229, "step": 790 }, { "epoch": 0.9031430404105195, "grad_norm": 66.209268910187, "learning_rate": 1.252077537271024e-07, "logits/chosen": -1.318701982498169, "logits/rejected": -1.3234184980392456, "logps/chosen": -180.39767456054688, "logps/rejected": -182.42808532714844, "loss": 0.5986, "rewards/accuracies": 0.71875, "rewards/chosen": -0.012874465435743332, "rewards/margins": 0.4246137738227844, "rewards/rejected": -0.43748825788497925, "step": 792 }, { "epoch": 0.9054237046539805, "grad_norm": 110.64800895876068, "learning_rate": 1.2483951643990425e-07, "logits/chosen": -1.427907943725586, "logits/rejected": -1.4352999925613403, "logps/chosen": -234.02430725097656, "logps/rejected": -252.70291137695312, "loss": 0.6882, "rewards/accuracies": 0.59375, "rewards/chosen": -0.23474670946598053, "rewards/margins": 0.3372371196746826, "rewards/rejected": -0.571983814239502, "step": 794 }, { "epoch": 0.9077043688974413, "grad_norm": 71.22024881783034, "learning_rate": 1.2447091983198366e-07, "logits/chosen": -1.4595139026641846, "logits/rejected": -1.494476556777954, "logps/chosen": -177.576904296875, "logps/rejected": -200.62152099609375, "loss": 0.622, "rewards/accuracies": 0.6875, "rewards/chosen": -0.10567592829465866, "rewards/margins": 0.4641418755054474, "rewards/rejected": -0.5698177814483643, "step": 796 }, { "epoch": 0.9099850331409023, "grad_norm": 77.66730114746076, "learning_rate": 1.2410196923534453e-07, "logits/chosen": -1.3751342296600342, "logits/rejected": -1.4557826519012451, "logps/chosen": -156.71981811523438, "logps/rejected": -175.78079223632812, "loss": 0.6241, "rewards/accuracies": 0.5625, "rewards/chosen": -0.17370270192623138, "rewards/margins": 0.5142669081687927, "rewards/rejected": -0.6879696249961853, "step": 798 }, { "epoch": 0.9122656973843632, "grad_norm": 79.01703720689792, "learning_rate": 1.237326699871115e-07, "logits/chosen": -1.3648015260696411, "logits/rejected": -1.4242901802062988, "logps/chosen": -139.64234924316406, "logps/rejected": -162.11544799804688, "loss": 0.6048, "rewards/accuracies": 0.65625, "rewards/chosen": -0.2178276777267456, "rewards/margins": 0.2655281722545624, "rewards/rejected": -0.48335587978363037, "step": 800 }, { "epoch": 0.9122656973843632, "eval_logits/chosen": -1.4735329151153564, "eval_logits/rejected": -1.460737705230713, "eval_logps/chosen": -128.9962158203125, "eval_logps/rejected": -133.23785400390625, "eval_loss": 0.6001502871513367, "eval_rewards/accuracies": 0.6800000071525574, "eval_rewards/chosen": 0.09017454832792282, "eval_rewards/margins": 0.20768418908119202, "eval_rewards/rejected": -0.11750967055559158, "eval_runtime": 19.4284, "eval_samples_per_second": 5.147, "eval_steps_per_second": 1.287, "step": 800 }, { "epoch": 0.9145463616278241, "grad_norm": 114.49291402716334, "learning_rate": 1.2336302742945265e-07, "logits/chosen": -1.1926456689834595, "logits/rejected": -1.1694424152374268, "logps/chosen": -151.49020385742188, "logps/rejected": -221.17051696777344, "loss": 0.6541, "rewards/accuracies": 0.4375, "rewards/chosen": -0.3034975230693817, "rewards/margins": 0.10761985182762146, "rewards/rejected": -0.41111740469932556, "step": 802 }, { "epoch": 0.9168270258712851, "grad_norm": 87.25759889144325, "learning_rate": 1.2299304690950234e-07, "logits/chosen": -1.314978837966919, "logits/rejected": -1.3282618522644043, "logps/chosen": -130.7468719482422, "logps/rejected": -169.4188690185547, "loss": 0.6349, "rewards/accuracies": 0.5625, "rewards/chosen": -0.07509980350732803, "rewards/margins": 0.23689687252044678, "rewards/rejected": -0.3119966685771942, "step": 804 }, { "epoch": 0.9191076901147459, "grad_norm": 90.2755897487053, "learning_rate": 1.2262273377928374e-07, "logits/chosen": -1.4546884298324585, "logits/rejected": -1.5022200345993042, "logps/chosen": -155.28553771972656, "logps/rejected": -186.28756713867188, "loss": 0.6376, "rewards/accuracies": 0.5, "rewards/chosen": -0.20369568467140198, "rewards/margins": 0.15442761778831482, "rewards/rejected": -0.3581232726573944, "step": 806 }, { "epoch": 0.9213883543582069, "grad_norm": 87.61957852766155, "learning_rate": 1.2225209339563143e-07, "logits/chosen": -1.1901319026947021, "logits/rejected": -1.2732369899749756, "logps/chosen": -126.2178726196289, "logps/rejected": -131.4507598876953, "loss": 0.6119, "rewards/accuracies": 0.71875, "rewards/chosen": -0.06662334501743317, "rewards/margins": 0.1810164749622345, "rewards/rejected": -0.24763983488082886, "step": 808 }, { "epoch": 0.9236690186016677, "grad_norm": 123.86797719697702, "learning_rate": 1.2188113112011404e-07, "logits/chosen": -1.260641098022461, "logits/rejected": -1.3186612129211426, "logps/chosen": -247.8528594970703, "logps/rejected": -279.6898498535156, "loss": 0.6643, "rewards/accuracies": 0.6875, "rewards/chosen": -0.3332889676094055, "rewards/margins": 0.2615528106689453, "rewards/rejected": -0.594841718673706, "step": 810 }, { "epoch": 0.9259496828451287, "grad_norm": 92.05739920896082, "learning_rate": 1.2150985231895645e-07, "logits/chosen": -1.5352187156677246, "logits/rejected": -1.5089805126190186, "logps/chosen": -168.19351196289062, "logps/rejected": -155.74728393554688, "loss": 0.6464, "rewards/accuracies": 0.65625, "rewards/chosen": -0.05093672126531601, "rewards/margins": 0.23030051589012146, "rewards/rejected": -0.2812372148036957, "step": 812 }, { "epoch": 0.9282303470885895, "grad_norm": 93.01772334051867, "learning_rate": 1.2113826236296244e-07, "logits/chosen": -1.4838998317718506, "logits/rejected": -1.4913954734802246, "logps/chosen": -145.08023071289062, "logps/rejected": -237.1237030029297, "loss": 0.63, "rewards/accuracies": 0.5625, "rewards/chosen": -0.10176398605108261, "rewards/margins": 0.37156644463539124, "rewards/rejected": -0.47333040833473206, "step": 814 }, { "epoch": 0.9305110113320505, "grad_norm": 104.77223934596387, "learning_rate": 1.207663666274367e-07, "logits/chosen": -1.3763738870620728, "logits/rejected": -1.3552130460739136, "logps/chosen": -177.25376892089844, "logps/rejected": -221.85498046875, "loss": 0.6857, "rewards/accuracies": 0.5, "rewards/chosen": -0.18679024279117584, "rewards/margins": 0.05987125262618065, "rewards/rejected": -0.2466614842414856, "step": 816 }, { "epoch": 0.9327916755755113, "grad_norm": 103.0413802693114, "learning_rate": 1.2039417049210742e-07, "logits/chosen": -1.3318116664886475, "logits/rejected": -1.3942524194717407, "logps/chosen": -142.11920166015625, "logps/rejected": -182.98092651367188, "loss": 0.6606, "rewards/accuracies": 0.65625, "rewards/chosen": -0.08502069115638733, "rewards/margins": 0.2950421869754791, "rewards/rejected": -0.38006284832954407, "step": 818 }, { "epoch": 0.9350723398189723, "grad_norm": 85.38211010208585, "learning_rate": 1.2002167934104814e-07, "logits/chosen": -1.3738641738891602, "logits/rejected": -1.3845763206481934, "logps/chosen": -120.14466857910156, "logps/rejected": -148.59951782226562, "loss": 0.6025, "rewards/accuracies": 0.65625, "rewards/chosen": -0.09485463052988052, "rewards/margins": 0.2216353416442871, "rewards/rejected": -0.31648993492126465, "step": 820 }, { "epoch": 0.9373530040624332, "grad_norm": 78.59698161262021, "learning_rate": 1.1964889856260001e-07, "logits/chosen": -1.405772089958191, "logits/rejected": -1.4993829727172852, "logps/chosen": -135.0908203125, "logps/rejected": -158.73516845703125, "loss": 0.6449, "rewards/accuracies": 0.59375, "rewards/chosen": 0.057033490389585495, "rewards/margins": 0.19890281558036804, "rewards/rejected": -0.14186930656433105, "step": 822 }, { "epoch": 0.9396336683058941, "grad_norm": 77.09101408818802, "learning_rate": 1.1927583354929392e-07, "logits/chosen": -1.2669503688812256, "logits/rejected": -1.3981753587722778, "logps/chosen": -185.62281799316406, "logps/rejected": -229.72349548339844, "loss": 0.5898, "rewards/accuracies": 0.78125, "rewards/chosen": -0.05814466252923012, "rewards/margins": 0.5641728043556213, "rewards/rejected": -0.6223174333572388, "step": 824 }, { "epoch": 0.941914332549355, "grad_norm": 80.62233528509246, "learning_rate": 1.1890248969777239e-07, "logits/chosen": -1.4591572284698486, "logits/rejected": -1.4886600971221924, "logps/chosen": -133.4588165283203, "logps/rejected": -161.96484375, "loss": 0.6355, "rewards/accuracies": 0.71875, "rewards/chosen": -0.1525450050830841, "rewards/margins": 0.35945460200309753, "rewards/rejected": -0.5119996666908264, "step": 826 }, { "epoch": 0.9441949967928159, "grad_norm": 82.91104665388441, "learning_rate": 1.1852887240871144e-07, "logits/chosen": -1.3768235445022583, "logits/rejected": -1.4797062873840332, "logps/chosen": -168.19825744628906, "logps/rejected": -195.83607482910156, "loss": 0.6328, "rewards/accuracies": 0.75, "rewards/chosen": 0.021572627127170563, "rewards/margins": 0.4793768525123596, "rewards/rejected": -0.45780429244041443, "step": 828 }, { "epoch": 0.9464756610362768, "grad_norm": 99.28432995782116, "learning_rate": 1.1815498708674265e-07, "logits/chosen": -1.3641839027404785, "logits/rejected": -1.5057623386383057, "logps/chosen": -188.45806884765625, "logps/rejected": -205.60987854003906, "loss": 0.6396, "rewards/accuracies": 0.5, "rewards/chosen": -0.27040964365005493, "rewards/margins": 0.16016271710395813, "rewards/rejected": -0.4305723309516907, "step": 830 }, { "epoch": 0.9487563252797377, "grad_norm": 91.59735539302851, "learning_rate": 1.1778083914037489e-07, "logits/chosen": -1.4061131477355957, "logits/rejected": -1.425171136856079, "logps/chosen": -127.95852661132812, "logps/rejected": -161.05148315429688, "loss": 0.6552, "rewards/accuracies": 0.65625, "rewards/chosen": 0.016702737659215927, "rewards/margins": 0.2845785915851593, "rewards/rejected": -0.2678758502006531, "step": 832 }, { "epoch": 0.9510369895231986, "grad_norm": 86.74397487981402, "learning_rate": 1.17406433981916e-07, "logits/chosen": -1.3160350322723389, "logits/rejected": -1.3871701955795288, "logps/chosen": -176.73638916015625, "logps/rejected": -211.8169403076172, "loss": 0.5804, "rewards/accuracies": 0.75, "rewards/chosen": -0.43320897221565247, "rewards/margins": 0.319485068321228, "rewards/rejected": -0.7526940107345581, "step": 834 }, { "epoch": 0.9533176537666596, "grad_norm": 95.1172952955302, "learning_rate": 1.1703177702739459e-07, "logits/chosen": -1.4391237497329712, "logits/rejected": -1.4496339559555054, "logps/chosen": -170.07687377929688, "logps/rejected": -187.2700958251953, "loss": 0.6198, "rewards/accuracies": 0.59375, "rewards/chosen": -0.08456218987703323, "rewards/margins": 0.2940000295639038, "rewards/rejected": -0.37856221199035645, "step": 836 }, { "epoch": 0.9555983180101204, "grad_norm": 88.76345502432684, "learning_rate": 1.1665687369648172e-07, "logits/chosen": -1.323454737663269, "logits/rejected": -1.330315351486206, "logps/chosen": -124.01346588134766, "logps/rejected": -141.37887573242188, "loss": 0.6466, "rewards/accuracies": 0.5, "rewards/chosen": -0.2537275552749634, "rewards/margins": -0.010971667245030403, "rewards/rejected": -0.24275588989257812, "step": 838 }, { "epoch": 0.9578789822535814, "grad_norm": 88.84157969865495, "learning_rate": 1.1628172941241239e-07, "logits/chosen": -1.2515794038772583, "logits/rejected": -1.2858678102493286, "logps/chosen": -161.89578247070312, "logps/rejected": -192.5193328857422, "loss": 0.5897, "rewards/accuracies": 0.65625, "rewards/chosen": -0.062142424285411835, "rewards/margins": 0.38304537534713745, "rewards/rejected": -0.44518783688545227, "step": 840 }, { "epoch": 0.9601596464970422, "grad_norm": 77.39513907682372, "learning_rate": 1.159063496019072e-07, "logits/chosen": -1.3906288146972656, "logits/rejected": -1.529634952545166, "logps/chosen": -149.05487060546875, "logps/rejected": -175.32589721679688, "loss": 0.607, "rewards/accuracies": 0.46875, "rewards/chosen": -0.14833964407444, "rewards/margins": 0.27842655777931213, "rewards/rejected": -0.42676615715026855, "step": 842 }, { "epoch": 0.9624403107405032, "grad_norm": 95.00453640521121, "learning_rate": 1.1553073969509382e-07, "logits/chosen": -1.3770264387130737, "logits/rejected": -1.4574774503707886, "logps/chosen": -154.23617553710938, "logps/rejected": -165.87744140625, "loss": 0.6221, "rewards/accuracies": 0.65625, "rewards/chosen": -0.23750467598438263, "rewards/margins": 0.23638057708740234, "rewards/rejected": -0.47388529777526855, "step": 844 }, { "epoch": 0.964720974983964, "grad_norm": 106.04741673208737, "learning_rate": 1.1515490512542831e-07, "logits/chosen": -1.467265248298645, "logits/rejected": -1.4767037630081177, "logps/chosen": -154.89556884765625, "logps/rejected": -169.7006072998047, "loss": 0.6862, "rewards/accuracies": 0.6875, "rewards/chosen": -0.292231023311615, "rewards/margins": 0.32214629650115967, "rewards/rejected": -0.6143773198127747, "step": 846 }, { "epoch": 0.967001639227425, "grad_norm": 105.71509066302269, "learning_rate": 1.1477885132961677e-07, "logits/chosen": -1.3954505920410156, "logits/rejected": -1.4496076107025146, "logps/chosen": -135.10621643066406, "logps/rejected": -155.632080078125, "loss": 0.6695, "rewards/accuracies": 0.71875, "rewards/chosen": -0.11768642067909241, "rewards/margins": 0.2924567759037018, "rewards/rejected": -0.4101432263851166, "step": 848 }, { "epoch": 0.9692823034708858, "grad_norm": 83.14192239326704, "learning_rate": 1.1440258374753649e-07, "logits/chosen": -1.2167763710021973, "logits/rejected": -1.2719136476516724, "logps/chosen": -161.74093627929688, "logps/rejected": -186.17892456054688, "loss": 0.6215, "rewards/accuracies": 0.65625, "rewards/chosen": -0.2962597608566284, "rewards/margins": 0.19979730248451233, "rewards/rejected": -0.49605703353881836, "step": 850 }, { "epoch": 0.9715629677143468, "grad_norm": 91.00056492305369, "learning_rate": 1.1402610782215733e-07, "logits/chosen": -1.338948369026184, "logits/rejected": -1.4312294721603394, "logps/chosen": -184.9549560546875, "logps/rejected": -219.67489624023438, "loss": 0.6323, "rewards/accuracies": 0.5625, "rewards/chosen": -0.3477857708930969, "rewards/margins": 0.11345378309488297, "rewards/rejected": -0.46123960614204407, "step": 852 }, { "epoch": 0.9738436319578078, "grad_norm": 83.43376640938686, "learning_rate": 1.1364942899946299e-07, "logits/chosen": -1.4227657318115234, "logits/rejected": -1.4756063222885132, "logps/chosen": -176.06297302246094, "logps/rejected": -190.72410583496094, "loss": 0.6332, "rewards/accuracies": 0.65625, "rewards/chosen": -0.2725405991077423, "rewards/margins": 0.23061612248420715, "rewards/rejected": -0.5031567215919495, "step": 854 }, { "epoch": 0.9761242962012686, "grad_norm": 80.98108181650579, "learning_rate": 1.132725527283722e-07, "logits/chosen": -1.390743613243103, "logits/rejected": -1.4082766771316528, "logps/chosen": -148.54055786132812, "logps/rejected": -182.93699645996094, "loss": 0.6133, "rewards/accuracies": 0.75, "rewards/chosen": -0.07731722295284271, "rewards/margins": 0.25885114073753357, "rewards/rejected": -0.3361683785915375, "step": 856 }, { "epoch": 0.9784049604447296, "grad_norm": 84.02405944942352, "learning_rate": 1.1289548446065993e-07, "logits/chosen": -1.3017364740371704, "logits/rejected": -1.3097370862960815, "logps/chosen": -162.28753662109375, "logps/rejected": -187.27439880371094, "loss": 0.6506, "rewards/accuracies": 0.6875, "rewards/chosen": -0.13367247581481934, "rewards/margins": 0.2926619052886963, "rewards/rejected": -0.426334410905838, "step": 858 }, { "epoch": 0.9806856246881904, "grad_norm": 92.76881750126093, "learning_rate": 1.1251822965087854e-07, "logits/chosen": -1.2643218040466309, "logits/rejected": -1.4101805686950684, "logps/chosen": -162.29969787597656, "logps/rejected": -204.03713989257812, "loss": 0.6025, "rewards/accuracies": 0.6875, "rewards/chosen": -0.28711238503456116, "rewards/margins": 0.36070847511291504, "rewards/rejected": -0.6478208899497986, "step": 860 }, { "epoch": 0.9829662889316514, "grad_norm": 118.85388390167415, "learning_rate": 1.1214079375627883e-07, "logits/chosen": -1.4487006664276123, "logits/rejected": -1.474176049232483, "logps/chosen": -271.73126220703125, "logps/rejected": -296.800537109375, "loss": 0.6742, "rewards/accuracies": 0.59375, "rewards/chosen": -0.5793529152870178, "rewards/margins": 0.24643933773040771, "rewards/rejected": -0.8257922530174255, "step": 862 }, { "epoch": 0.9852469531751122, "grad_norm": 86.67252914799577, "learning_rate": 1.1176318223673105e-07, "logits/chosen": -1.4337862730026245, "logits/rejected": -1.4527971744537354, "logps/chosen": -200.96485900878906, "logps/rejected": -221.7783660888672, "loss": 0.6498, "rewards/accuracies": 0.625, "rewards/chosen": -0.2869877815246582, "rewards/margins": 0.2341936230659485, "rewards/rejected": -0.5211814045906067, "step": 864 }, { "epoch": 0.9875276174185732, "grad_norm": 113.98036843821767, "learning_rate": 1.1138540055464609e-07, "logits/chosen": -1.3671294450759888, "logits/rejected": -1.368064522743225, "logps/chosen": -163.43055725097656, "logps/rejected": -175.2020721435547, "loss": 0.5793, "rewards/accuracies": 0.65625, "rewards/chosen": -0.022854965180158615, "rewards/margins": 0.4233308434486389, "rewards/rejected": -0.44618579745292664, "step": 866 }, { "epoch": 0.989808281662034, "grad_norm": 80.29058110520482, "learning_rate": 1.110074541748963e-07, "logits/chosen": -1.346540093421936, "logits/rejected": -1.3472204208374023, "logps/chosen": -186.08084106445312, "logps/rejected": -207.60470581054688, "loss": 0.6321, "rewards/accuracies": 0.6875, "rewards/chosen": -0.25492382049560547, "rewards/margins": 0.35832715034484863, "rewards/rejected": -0.6132509708404541, "step": 868 }, { "epoch": 0.992088945905495, "grad_norm": 90.95115779695014, "learning_rate": 1.1062934856473653e-07, "logits/chosen": -1.4387091398239136, "logits/rejected": -1.455345630645752, "logps/chosen": -98.49878692626953, "logps/rejected": -172.24810791015625, "loss": 0.6444, "rewards/accuracies": 0.625, "rewards/chosen": -0.03821319341659546, "rewards/margins": 0.8005204796791077, "rewards/rejected": -0.8387336730957031, "step": 870 }, { "epoch": 0.9943696101489559, "grad_norm": 100.52505053286734, "learning_rate": 1.1025108919372499e-07, "logits/chosen": -1.3714473247528076, "logits/rejected": -1.3371658325195312, "logps/chosen": -167.58447265625, "logps/rejected": -186.04751586914062, "loss": 0.6675, "rewards/accuracies": 0.65625, "rewards/chosen": -0.16858822107315063, "rewards/margins": 0.19196297228336334, "rewards/rejected": -0.36055123805999756, "step": 872 }, { "epoch": 0.9966502743924168, "grad_norm": 110.94600706114623, "learning_rate": 1.0987268153364411e-07, "logits/chosen": -1.3899271488189697, "logits/rejected": -1.3658580780029297, "logps/chosen": -130.26220703125, "logps/rejected": -134.71063232421875, "loss": 0.66, "rewards/accuracies": 0.625, "rewards/chosen": -0.10100418329238892, "rewards/margins": 0.2500309348106384, "rewards/rejected": -0.35103511810302734, "step": 874 }, { "epoch": 0.9989309386358777, "grad_norm": 105.88720383741304, "learning_rate": 1.0949413105842146e-07, "logits/chosen": -1.4607656002044678, "logits/rejected": -1.5963387489318848, "logps/chosen": -196.40066528320312, "logps/rejected": -225.38638305664062, "loss": 0.6348, "rewards/accuracies": 0.625, "rewards/chosen": -0.15517286956310272, "rewards/margins": 0.15176478028297424, "rewards/rejected": -0.30693763494491577, "step": 876 }, { "epoch": 1.0012116028793385, "grad_norm": 56.3694410581267, "learning_rate": 1.091154432440506e-07, "logits/chosen": -1.389705777168274, "logits/rejected": -1.4425851106643677, "logps/chosen": -184.08709716796875, "logps/rejected": -215.4456024169922, "loss": 0.5323, "rewards/accuracies": 0.71875, "rewards/chosen": 0.004449841566383839, "rewards/margins": 0.635086178779602, "rewards/rejected": -0.6306363344192505, "step": 878 }, { "epoch": 1.0034922671227995, "grad_norm": 59.14871364307595, "learning_rate": 1.0873662356851164e-07, "logits/chosen": -1.2264068126678467, "logits/rejected": -1.2039250135421753, "logps/chosen": -143.2238311767578, "logps/rejected": -148.1532745361328, "loss": 0.4487, "rewards/accuracies": 0.71875, "rewards/chosen": 0.14279066026210785, "rewards/margins": 0.6712049841880798, "rewards/rejected": -0.528414249420166, "step": 880 }, { "epoch": 1.0057729313662604, "grad_norm": 56.08399590075636, "learning_rate": 1.0835767751169225e-07, "logits/chosen": -1.3238468170166016, "logits/rejected": -1.4208531379699707, "logps/chosen": -172.80247497558594, "logps/rejected": -221.4132843017578, "loss": 0.4898, "rewards/accuracies": 0.75, "rewards/chosen": -0.04028752073645592, "rewards/margins": 0.7049790024757385, "rewards/rejected": -0.7452664375305176, "step": 882 }, { "epoch": 1.0080535956097214, "grad_norm": 52.855939708340124, "learning_rate": 1.0797861055530831e-07, "logits/chosen": -1.3633054494857788, "logits/rejected": -1.3413807153701782, "logps/chosen": -146.97352600097656, "logps/rejected": -183.86334228515625, "loss": 0.4712, "rewards/accuracies": 0.875, "rewards/chosen": -0.06403888016939163, "rewards/margins": 0.8986248970031738, "rewards/rejected": -0.9626636505126953, "step": 884 }, { "epoch": 1.0103342598531821, "grad_norm": 66.20782299721449, "learning_rate": 1.0759942818282453e-07, "logits/chosen": -1.2454776763916016, "logits/rejected": -1.3409074544906616, "logps/chosen": -165.26669311523438, "logps/rejected": -199.02630615234375, "loss": 0.4379, "rewards/accuracies": 0.90625, "rewards/chosen": -0.18418747186660767, "rewards/margins": 0.8542241454124451, "rewards/rejected": -1.0384116172790527, "step": 886 }, { "epoch": 1.012614924096643, "grad_norm": 61.538116919835716, "learning_rate": 1.0722013587937526e-07, "logits/chosen": -1.345977783203125, "logits/rejected": -1.3574477434158325, "logps/chosen": -140.93460083007812, "logps/rejected": -170.11268615722656, "loss": 0.4964, "rewards/accuracies": 0.8125, "rewards/chosen": 0.001187225803732872, "rewards/margins": 0.6773791313171387, "rewards/rejected": -0.6761919260025024, "step": 888 }, { "epoch": 1.014895588340104, "grad_norm": 66.18054031071868, "learning_rate": 1.0684073913168501e-07, "logits/chosen": -1.506268858909607, "logits/rejected": -1.5550968647003174, "logps/chosen": -227.16659545898438, "logps/rejected": -231.74356079101562, "loss": 0.4649, "rewards/accuracies": 0.875, "rewards/chosen": -0.04133976250886917, "rewards/margins": 0.7655045986175537, "rewards/rejected": -0.8068442940711975, "step": 890 }, { "epoch": 1.017176252583565, "grad_norm": 63.53563886574213, "learning_rate": 1.0646124342798919e-07, "logits/chosen": -1.2807819843292236, "logits/rejected": -1.3626489639282227, "logps/chosen": -149.43206787109375, "logps/rejected": -194.97207641601562, "loss": 0.4679, "rewards/accuracies": 0.875, "rewards/chosen": 0.026193831115961075, "rewards/margins": 0.7646716237068176, "rewards/rejected": -0.7384778261184692, "step": 892 }, { "epoch": 1.0194569168270258, "grad_norm": 63.988279055249954, "learning_rate": 1.0608165425795468e-07, "logits/chosen": -1.5072122812271118, "logits/rejected": -1.5509088039398193, "logps/chosen": -185.84625244140625, "logps/rejected": -213.9545440673828, "loss": 0.5205, "rewards/accuracies": 0.84375, "rewards/chosen": -0.03688034415245056, "rewards/margins": 0.8168615102767944, "rewards/rejected": -0.8537418842315674, "step": 894 }, { "epoch": 1.0217375810704867, "grad_norm": 58.69101964281229, "learning_rate": 1.0570197711260038e-07, "logits/chosen": -1.388396978378296, "logits/rejected": -1.3820849657058716, "logps/chosen": -127.9742431640625, "logps/rejected": -137.78054809570312, "loss": 0.4494, "rewards/accuracies": 0.875, "rewards/chosen": 0.10387776046991348, "rewards/margins": 0.5781891942024231, "rewards/rejected": -0.47431135177612305, "step": 896 }, { "epoch": 1.0240182453139477, "grad_norm": 74.70102903392454, "learning_rate": 1.0532221748421786e-07, "logits/chosen": -1.3617676496505737, "logits/rejected": -1.4110198020935059, "logps/chosen": -160.6496124267578, "logps/rejected": -182.05860900878906, "loss": 0.4881, "rewards/accuracies": 0.78125, "rewards/chosen": -0.21052153408527374, "rewards/margins": 0.5095985531806946, "rewards/rejected": -0.7201201319694519, "step": 898 }, { "epoch": 1.0262989095574087, "grad_norm": 61.26905190831476, "learning_rate": 1.0494238086629183e-07, "logits/chosen": -1.4342904090881348, "logits/rejected": -1.4664863348007202, "logps/chosen": -173.26043701171875, "logps/rejected": -202.07130432128906, "loss": 0.4934, "rewards/accuracies": 0.78125, "rewards/chosen": -0.21440905332565308, "rewards/margins": 0.7449226379394531, "rewards/rejected": -0.9593316316604614, "step": 900 }, { "epoch": 1.0262989095574087, "eval_logits/chosen": -1.4491240978240967, "eval_logits/rejected": -1.4348580837249756, "eval_logps/chosen": -129.59962463378906, "eval_logps/rejected": -134.80783081054688, "eval_loss": 0.5797997713088989, "eval_rewards/accuracies": 0.7200000286102295, "eval_rewards/chosen": 0.029834765940904617, "eval_rewards/margins": 0.3043439984321594, "eval_rewards/rejected": -0.2745092511177063, "eval_runtime": 20.081, "eval_samples_per_second": 4.98, "eval_steps_per_second": 1.245, "step": 900 }, { "epoch": 1.0285795738008694, "grad_norm": 58.67281714259608, "learning_rate": 1.0456247275342074e-07, "logits/chosen": -1.3800104856491089, "logits/rejected": -1.4262235164642334, "logps/chosen": -148.66639709472656, "logps/rejected": -190.34461975097656, "loss": 0.4164, "rewards/accuracies": 0.75, "rewards/chosen": -0.14912372827529907, "rewards/margins": 0.7551414370536804, "rewards/rejected": -0.9042652249336243, "step": 902 }, { "epoch": 1.0308602380443304, "grad_norm": 51.15662895087173, "learning_rate": 1.0418249864123723e-07, "logits/chosen": -1.312926173210144, "logits/rejected": -1.3743692636489868, "logps/chosen": -210.05198669433594, "logps/rejected": -243.07321166992188, "loss": 0.446, "rewards/accuracies": 0.90625, "rewards/chosen": -0.018690135329961777, "rewards/margins": 0.9038247466087341, "rewards/rejected": -0.9225149154663086, "step": 904 }, { "epoch": 1.0331409022877913, "grad_norm": 60.88958208699774, "learning_rate": 1.0380246402632869e-07, "logits/chosen": -1.2510696649551392, "logits/rejected": -1.2662408351898193, "logps/chosen": -101.8481674194336, "logps/rejected": -151.2491912841797, "loss": 0.4442, "rewards/accuracies": 0.71875, "rewards/chosen": 0.025427617132663727, "rewards/margins": 0.740162193775177, "rewards/rejected": -0.7147345542907715, "step": 906 }, { "epoch": 1.0354215665312523, "grad_norm": 54.60415707659223, "learning_rate": 1.0342237440615764e-07, "logits/chosen": -1.3037830591201782, "logits/rejected": -1.3146332502365112, "logps/chosen": -202.0905303955078, "logps/rejected": -266.09552001953125, "loss": 0.414, "rewards/accuracies": 0.875, "rewards/chosen": -0.2531195878982544, "rewards/margins": 1.4251919984817505, "rewards/rejected": -1.6783115863800049, "step": 908 }, { "epoch": 1.0377022307747132, "grad_norm": 55.23259639285888, "learning_rate": 1.0304223527898243e-07, "logits/chosen": -1.3536947965621948, "logits/rejected": -1.371093988418579, "logps/chosen": -151.1886444091797, "logps/rejected": -162.5618438720703, "loss": 0.4135, "rewards/accuracies": 0.90625, "rewards/chosen": -0.06908969581127167, "rewards/margins": 0.756403386592865, "rewards/rejected": -0.8254930973052979, "step": 910 }, { "epoch": 1.039982895018174, "grad_norm": 55.36663265349372, "learning_rate": 1.0266205214377748e-07, "logits/chosen": -1.3304215669631958, "logits/rejected": -1.412177324295044, "logps/chosen": -165.882080078125, "logps/rejected": -179.88272094726562, "loss": 0.4386, "rewards/accuracies": 0.78125, "rewards/chosen": -0.23252317309379578, "rewards/margins": 0.8055883646011353, "rewards/rejected": -1.038111686706543, "step": 912 }, { "epoch": 1.042263559261635, "grad_norm": 57.72994301158152, "learning_rate": 1.0228183050015383e-07, "logits/chosen": -1.4341506958007812, "logits/rejected": -1.416825771331787, "logps/chosen": -149.43185424804688, "logps/rejected": -155.654296875, "loss": 0.4917, "rewards/accuracies": 0.78125, "rewards/chosen": -0.058750756084918976, "rewards/margins": 0.5951379537582397, "rewards/rejected": -0.6538887619972229, "step": 914 }, { "epoch": 1.044544223505096, "grad_norm": 55.93678999952222, "learning_rate": 1.0190157584827952e-07, "logits/chosen": -1.2692397832870483, "logits/rejected": -1.3736000061035156, "logps/chosen": -187.53558349609375, "logps/rejected": -219.8994903564453, "loss": 0.4593, "rewards/accuracies": 0.8125, "rewards/chosen": -0.10293256491422653, "rewards/margins": 0.9443311095237732, "rewards/rejected": -1.047263741493225, "step": 916 }, { "epoch": 1.0468248877485569, "grad_norm": 54.86148488530044, "learning_rate": 1.0152129368880021e-07, "logits/chosen": -1.359283208847046, "logits/rejected": -1.423500657081604, "logps/chosen": -160.22647094726562, "logps/rejected": -174.9389190673828, "loss": 0.4786, "rewards/accuracies": 0.875, "rewards/chosen": -0.20612327754497528, "rewards/margins": 0.813952624797821, "rewards/rejected": -1.0200759172439575, "step": 918 }, { "epoch": 1.0491055519920176, "grad_norm": 53.53882999084853, "learning_rate": 1.0114098952275933e-07, "logits/chosen": -1.4099137783050537, "logits/rejected": -1.4628063440322876, "logps/chosen": -158.679443359375, "logps/rejected": -185.23330688476562, "loss": 0.4698, "rewards/accuracies": 0.90625, "rewards/chosen": -0.0398465096950531, "rewards/margins": 0.869282603263855, "rewards/rejected": -0.9091290831565857, "step": 920 }, { "epoch": 1.0513862162354786, "grad_norm": 63.92382605562123, "learning_rate": 1.0076066885151877e-07, "logits/chosen": -1.2206496000289917, "logits/rejected": -1.3061589002609253, "logps/chosen": -192.36740112304688, "logps/rejected": -219.992431640625, "loss": 0.4498, "rewards/accuracies": 0.9375, "rewards/chosen": -0.20084695518016815, "rewards/margins": 1.0578863620758057, "rewards/rejected": -1.2587332725524902, "step": 922 }, { "epoch": 1.0536668804789395, "grad_norm": 58.361553235329055, "learning_rate": 1.0038033717667906e-07, "logits/chosen": -1.3777412176132202, "logits/rejected": -1.4997334480285645, "logps/chosen": -144.8574676513672, "logps/rejected": -185.2276611328125, "loss": 0.4444, "rewards/accuracies": 0.875, "rewards/chosen": -0.12493613362312317, "rewards/margins": 0.8183287978172302, "rewards/rejected": -0.943264901638031, "step": 924 }, { "epoch": 1.0559475447224005, "grad_norm": 60.9980807817603, "learning_rate": 1e-07, "logits/chosen": -1.3119468688964844, "logits/rejected": -1.4459898471832275, "logps/chosen": -204.2576904296875, "logps/rejected": -243.5587921142578, "loss": 0.4801, "rewards/accuracies": 0.78125, "rewards/chosen": -0.2800896167755127, "rewards/margins": 0.8604587912559509, "rewards/rejected": -1.1405483484268188, "step": 926 }, { "epoch": 1.0582282089658612, "grad_norm": 58.9377536477793, "learning_rate": 9.961966282332094e-08, "logits/chosen": -1.4508683681488037, "logits/rejected": -1.4245975017547607, "logps/chosen": -168.61058044433594, "logps/rejected": -165.25489807128906, "loss": 0.4758, "rewards/accuracies": 0.6875, "rewards/chosen": -0.2604736089706421, "rewards/margins": 0.6006544828414917, "rewards/rejected": -0.8611280918121338, "step": 928 }, { "epoch": 1.0605088732093222, "grad_norm": 56.86318861636952, "learning_rate": 9.923933114848123e-08, "logits/chosen": -1.1611130237579346, "logits/rejected": -1.1586978435516357, "logps/chosen": -137.10975646972656, "logps/rejected": -159.64285278320312, "loss": 0.4456, "rewards/accuracies": 0.78125, "rewards/chosen": -0.2737983167171478, "rewards/margins": 0.6588336825370789, "rewards/rejected": -0.9326320290565491, "step": 930 }, { "epoch": 1.0627895374527832, "grad_norm": 47.87292073584732, "learning_rate": 9.885901047724065e-08, "logits/chosen": -1.3931517601013184, "logits/rejected": -1.3711789846420288, "logps/chosen": -123.40711212158203, "logps/rejected": -148.42681884765625, "loss": 0.457, "rewards/accuracies": 0.9375, "rewards/chosen": 0.12486562132835388, "rewards/margins": 0.9808838963508606, "rewards/rejected": -0.8560183048248291, "step": 932 }, { "epoch": 1.0650702016962441, "grad_norm": 60.27486161031482, "learning_rate": 9.847870631119978e-08, "logits/chosen": -1.1859281063079834, "logits/rejected": -1.167513370513916, "logps/chosen": -141.37530517578125, "logps/rejected": -153.80101013183594, "loss": 0.4799, "rewards/accuracies": 0.8125, "rewards/chosen": -0.2075161635875702, "rewards/margins": 0.6786595582962036, "rewards/rejected": -0.886175811290741, "step": 934 }, { "epoch": 1.0673508659397049, "grad_norm": 52.567168828966985, "learning_rate": 9.809842415172047e-08, "logits/chosen": -1.322398066520691, "logits/rejected": -1.4248247146606445, "logps/chosen": -136.26565551757812, "logps/rejected": -173.91111755371094, "loss": 0.4799, "rewards/accuracies": 0.75, "rewards/chosen": -0.21634379029273987, "rewards/margins": 0.6690733432769775, "rewards/rejected": -0.8854170441627502, "step": 936 }, { "epoch": 1.0696315301831658, "grad_norm": 55.376994474776346, "learning_rate": 9.771816949984619e-08, "logits/chosen": -1.1317143440246582, "logits/rejected": -1.2357646226882935, "logps/chosen": -102.08535766601562, "logps/rejected": -140.366943359375, "loss": 0.4312, "rewards/accuracies": 0.875, "rewards/chosen": -0.09264838695526123, "rewards/margins": 0.8339763879776001, "rewards/rejected": -0.9266247153282166, "step": 938 }, { "epoch": 1.0719121944266268, "grad_norm": 52.064161742890775, "learning_rate": 9.733794785622253e-08, "logits/chosen": -1.2546545267105103, "logits/rejected": -1.2125529050827026, "logps/chosen": -137.27352905273438, "logps/rejected": -155.36073303222656, "loss": 0.429, "rewards/accuracies": 0.8125, "rewards/chosen": -0.31309816241264343, "rewards/margins": 0.9382134675979614, "rewards/rejected": -1.2513116598129272, "step": 940 }, { "epoch": 1.0741928586700877, "grad_norm": 54.0017692542014, "learning_rate": 9.695776472101757e-08, "logits/chosen": -1.444411039352417, "logits/rejected": -1.4067819118499756, "logps/chosen": -167.93966674804688, "logps/rejected": -159.4579315185547, "loss": 0.4581, "rewards/accuracies": 0.59375, "rewards/chosen": -0.14569707214832306, "rewards/margins": 0.46156030893325806, "rewards/rejected": -0.6072573661804199, "step": 942 }, { "epoch": 1.0764735229135485, "grad_norm": 59.11201512010301, "learning_rate": 9.657762559384235e-08, "logits/chosen": -1.2698876857757568, "logits/rejected": -1.3079754114151, "logps/chosen": -191.69493103027344, "logps/rejected": -204.89511108398438, "loss": 0.45, "rewards/accuracies": 0.78125, "rewards/chosen": -0.35640037059783936, "rewards/margins": 0.654572606086731, "rewards/rejected": -1.0109730958938599, "step": 944 }, { "epoch": 1.0787541871570094, "grad_norm": 62.61929808626616, "learning_rate": 9.619753597367133e-08, "logits/chosen": -1.3809797763824463, "logits/rejected": -1.465749979019165, "logps/chosen": -113.58399963378906, "logps/rejected": -143.52798461914062, "loss": 0.5093, "rewards/accuracies": 0.59375, "rewards/chosen": -0.23512893915176392, "rewards/margins": 0.46248409152030945, "rewards/rejected": -0.6976129412651062, "step": 946 }, { "epoch": 1.0810348514004704, "grad_norm": 62.384552822468294, "learning_rate": 9.581750135876276e-08, "logits/chosen": -1.156672477722168, "logits/rejected": -1.1879075765609741, "logps/chosen": -136.56668090820312, "logps/rejected": -164.78175354003906, "loss": 0.4693, "rewards/accuracies": 0.84375, "rewards/chosen": -0.1409253031015396, "rewards/margins": 0.768065333366394, "rewards/rejected": -0.9089905619621277, "step": 948 }, { "epoch": 1.0833155156439314, "grad_norm": 62.82871732010386, "learning_rate": 9.543752724657923e-08, "logits/chosen": -1.2612335681915283, "logits/rejected": -1.3249022960662842, "logps/chosen": -148.45864868164062, "logps/rejected": -204.80255126953125, "loss": 0.4975, "rewards/accuracies": 0.78125, "rewards/chosen": -0.337429940700531, "rewards/margins": 1.0719372034072876, "rewards/rejected": -1.4093672037124634, "step": 950 }, { "epoch": 1.0855961798873923, "grad_norm": 51.27361713332628, "learning_rate": 9.505761913370813e-08, "logits/chosen": -1.2637767791748047, "logits/rejected": -1.2903871536254883, "logps/chosen": -123.89924621582031, "logps/rejected": -151.03236389160156, "loss": 0.4743, "rewards/accuracies": 0.8125, "rewards/chosen": -0.15131911635398865, "rewards/margins": 0.7663697600364685, "rewards/rejected": -0.9176889061927795, "step": 952 }, { "epoch": 1.087876844130853, "grad_norm": 60.844362449035934, "learning_rate": 9.467778251578216e-08, "logits/chosen": -1.264873743057251, "logits/rejected": -1.308799147605896, "logps/chosen": -211.9695281982422, "logps/rejected": -249.72434997558594, "loss": 0.4411, "rewards/accuracies": 0.84375, "rewards/chosen": -0.4224779009819031, "rewards/margins": 1.1005442142486572, "rewards/rejected": -1.523022174835205, "step": 954 }, { "epoch": 1.090157508374314, "grad_norm": 62.233166596514444, "learning_rate": 9.429802288739961e-08, "logits/chosen": -1.2648420333862305, "logits/rejected": -1.3022546768188477, "logps/chosen": -204.7000732421875, "logps/rejected": -234.1602783203125, "loss": 0.5, "rewards/accuracies": 0.75, "rewards/chosen": -0.3554447889328003, "rewards/margins": 0.6820401549339294, "rewards/rejected": -1.037484884262085, "step": 956 }, { "epoch": 1.092438172617775, "grad_norm": 52.22595835402743, "learning_rate": 9.391834574204533e-08, "logits/chosen": -1.3593980073928833, "logits/rejected": -1.3903865814208984, "logps/chosen": -180.24501037597656, "logps/rejected": -211.7206268310547, "loss": 0.4345, "rewards/accuracies": 0.9375, "rewards/chosen": -0.10301375389099121, "rewards/margins": 1.223108172416687, "rewards/rejected": -1.3261220455169678, "step": 958 }, { "epoch": 1.0947188368612357, "grad_norm": 59.31222116238765, "learning_rate": 9.353875657201083e-08, "logits/chosen": -1.2963013648986816, "logits/rejected": -1.3758771419525146, "logps/chosen": -183.5509033203125, "logps/rejected": -246.8356475830078, "loss": 0.4517, "rewards/accuracies": 0.90625, "rewards/chosen": -0.12891785800457, "rewards/margins": 1.3549247980117798, "rewards/rejected": -1.4838427305221558, "step": 960 }, { "epoch": 1.0969995011046967, "grad_norm": 70.97226420908964, "learning_rate": 9.3159260868315e-08, "logits/chosen": -1.3708178997039795, "logits/rejected": -1.43119478225708, "logps/chosen": -188.970703125, "logps/rejected": -210.37744140625, "loss": 0.4582, "rewards/accuracies": 0.8125, "rewards/chosen": -0.3142687976360321, "rewards/margins": 0.9785380959510803, "rewards/rejected": -1.29280686378479, "step": 962 }, { "epoch": 1.0992801653481576, "grad_norm": 62.07148030594709, "learning_rate": 9.277986412062474e-08, "logits/chosen": -1.2510358095169067, "logits/rejected": -1.3639835119247437, "logps/chosen": -151.8282012939453, "logps/rejected": -186.52447509765625, "loss": 0.3958, "rewards/accuracies": 0.8125, "rewards/chosen": -0.29675284028053284, "rewards/margins": 0.9109016060829163, "rewards/rejected": -1.2076544761657715, "step": 964 }, { "epoch": 1.1015608295916186, "grad_norm": 68.63281362186014, "learning_rate": 9.240057181717546e-08, "logits/chosen": -1.199402093887329, "logits/rejected": -1.3185701370239258, "logps/chosen": -170.2873077392578, "logps/rejected": -201.01829528808594, "loss": 0.4737, "rewards/accuracies": 0.78125, "rewards/chosen": -0.3943040668964386, "rewards/margins": 0.6716153025627136, "rewards/rejected": -1.0659193992614746, "step": 966 }, { "epoch": 1.1038414938350796, "grad_norm": 53.596924936810275, "learning_rate": 9.202138944469167e-08, "logits/chosen": -1.3583130836486816, "logits/rejected": -1.3962550163269043, "logps/chosen": -138.78871154785156, "logps/rejected": -169.90960693359375, "loss": 0.4341, "rewards/accuracies": 0.78125, "rewards/chosen": -0.13780230283737183, "rewards/margins": 0.8456271886825562, "rewards/rejected": -0.9834293723106384, "step": 968 }, { "epoch": 1.1061221580785403, "grad_norm": 57.028026126969664, "learning_rate": 9.164232248830777e-08, "logits/chosen": -1.271955966949463, "logits/rejected": -1.3388372659683228, "logps/chosen": -176.40234375, "logps/rejected": -217.4737091064453, "loss": 0.4131, "rewards/accuracies": 0.8125, "rewards/chosen": -0.16253671050071716, "rewards/margins": 1.022351861000061, "rewards/rejected": -1.1848886013031006, "step": 970 }, { "epoch": 1.1084028223220013, "grad_norm": 69.9853718884853, "learning_rate": 9.126337643148838e-08, "logits/chosen": -1.2280056476593018, "logits/rejected": -1.2864279747009277, "logps/chosen": -188.2535858154297, "logps/rejected": -217.4619598388672, "loss": 0.4778, "rewards/accuracies": 0.8125, "rewards/chosen": -0.3671016991138458, "rewards/margins": 0.9746519923210144, "rewards/rejected": -1.3417538404464722, "step": 972 }, { "epoch": 1.1106834865654622, "grad_norm": 65.03116909989372, "learning_rate": 9.088455675594942e-08, "logits/chosen": -1.2964333295822144, "logits/rejected": -1.3419139385223389, "logps/chosen": -208.0177764892578, "logps/rejected": -250.48910522460938, "loss": 0.4301, "rewards/accuracies": 0.875, "rewards/chosen": -0.375073105096817, "rewards/margins": 1.1881136894226074, "rewards/rejected": -1.5631868839263916, "step": 974 }, { "epoch": 1.1129641508089232, "grad_norm": 53.40636924718595, "learning_rate": 9.050586894157853e-08, "logits/chosen": -1.2230535745620728, "logits/rejected": -1.3127639293670654, "logps/chosen": -152.9366455078125, "logps/rejected": -209.3880157470703, "loss": 0.4103, "rewards/accuracies": 0.75, "rewards/chosen": -0.21992121636867523, "rewards/margins": 1.1222590208053589, "rewards/rejected": -1.3421802520751953, "step": 976 }, { "epoch": 1.115244815052384, "grad_norm": 65.88010848155325, "learning_rate": 9.012731846635589e-08, "logits/chosen": -1.1757484674453735, "logits/rejected": -1.1423841714859009, "logps/chosen": -194.4547119140625, "logps/rejected": -219.962890625, "loss": 0.4572, "rewards/accuracies": 0.78125, "rewards/chosen": -0.47297030687332153, "rewards/margins": 0.9668577909469604, "rewards/rejected": -1.4398282766342163, "step": 978 }, { "epoch": 1.117525479295845, "grad_norm": 63.30481794169962, "learning_rate": 8.974891080627503e-08, "logits/chosen": -1.297542929649353, "logits/rejected": -1.387624740600586, "logps/chosen": -152.45774841308594, "logps/rejected": -171.50283813476562, "loss": 0.4435, "rewards/accuracies": 0.8125, "rewards/chosen": -0.17080621421337128, "rewards/margins": 0.6862813830375671, "rewards/rejected": -0.8570876717567444, "step": 980 }, { "epoch": 1.1198061435393059, "grad_norm": 58.37292774536043, "learning_rate": 8.937065143526346e-08, "logits/chosen": -1.2513972520828247, "logits/rejected": -1.2623106241226196, "logps/chosen": -151.0131378173828, "logps/rejected": -171.0509490966797, "loss": 0.4517, "rewards/accuracies": 0.6875, "rewards/chosen": -0.1952919065952301, "rewards/margins": 0.6836872696876526, "rewards/rejected": -0.8789791464805603, "step": 982 }, { "epoch": 1.1220868077827668, "grad_norm": 71.24214091954167, "learning_rate": 8.899254582510367e-08, "logits/chosen": -1.2960295677185059, "logits/rejected": -1.3374513387680054, "logps/chosen": -170.47344970703125, "logps/rejected": -197.69920349121094, "loss": 0.4205, "rewards/accuracies": 0.84375, "rewards/chosen": -0.2936790883541107, "rewards/margins": 1.0447689294815063, "rewards/rejected": -1.3384480476379395, "step": 984 }, { "epoch": 1.1243674720262276, "grad_norm": 58.68134074489735, "learning_rate": 8.861459944535389e-08, "logits/chosen": -1.3058708906173706, "logits/rejected": -1.3617157936096191, "logps/chosen": -149.42727661132812, "logps/rejected": -179.4463653564453, "loss": 0.4586, "rewards/accuracies": 0.75, "rewards/chosen": -0.1447882056236267, "rewards/margins": 0.7445257902145386, "rewards/rejected": -0.8893140554428101, "step": 986 }, { "epoch": 1.1266481362696885, "grad_norm": 61.951258926530045, "learning_rate": 8.823681776326898e-08, "logits/chosen": -1.3737095594406128, "logits/rejected": -1.4248648881912231, "logps/chosen": -189.31312561035156, "logps/rejected": -220.4053192138672, "loss": 0.4602, "rewards/accuracies": 0.8125, "rewards/chosen": -0.6107076406478882, "rewards/margins": 0.6425344944000244, "rewards/rejected": -1.2532421350479126, "step": 988 }, { "epoch": 1.1289288005131495, "grad_norm": 62.59059177140616, "learning_rate": 8.785920624372121e-08, "logits/chosen": -1.169434666633606, "logits/rejected": -1.1585787534713745, "logps/chosen": -154.62611389160156, "logps/rejected": -163.4484100341797, "loss": 0.4619, "rewards/accuracies": 0.75, "rewards/chosen": -0.1728251576423645, "rewards/margins": 0.8388264775276184, "rewards/rejected": -1.0116515159606934, "step": 990 }, { "epoch": 1.1312094647566104, "grad_norm": 56.29054558070934, "learning_rate": 8.748177034912147e-08, "logits/chosen": -1.4068464040756226, "logits/rejected": -1.4235914945602417, "logps/chosen": -203.57357788085938, "logps/rejected": -247.82298278808594, "loss": 0.4738, "rewards/accuracies": 0.84375, "rewards/chosen": -0.4565941393375397, "rewards/margins": 1.059171438217163, "rewards/rejected": -1.5157654285430908, "step": 992 }, { "epoch": 1.1334901290000712, "grad_norm": 51.085177964624826, "learning_rate": 8.710451553934008e-08, "logits/chosen": -1.3480877876281738, "logits/rejected": -1.3525364398956299, "logps/chosen": -155.8323974609375, "logps/rejected": -188.3721466064453, "loss": 0.4321, "rewards/accuracies": 0.8125, "rewards/chosen": -0.2992142140865326, "rewards/margins": 0.7550831437110901, "rewards/rejected": -1.0542973279953003, "step": 994 }, { "epoch": 1.1357707932435321, "grad_norm": 58.843609149415194, "learning_rate": 8.672744727162781e-08, "logits/chosen": -1.2812210321426392, "logits/rejected": -1.2667509317398071, "logps/chosen": -148.864990234375, "logps/rejected": -171.8406219482422, "loss": 0.4565, "rewards/accuracies": 0.78125, "rewards/chosen": -0.2008969783782959, "rewards/margins": 0.9470231533050537, "rewards/rejected": -1.1479202508926392, "step": 996 }, { "epoch": 1.138051457486993, "grad_norm": 55.1542603604775, "learning_rate": 8.635057100053701e-08, "logits/chosen": -1.2231156826019287, "logits/rejected": -1.2690283060073853, "logps/chosen": -134.86245727539062, "logps/rejected": -175.73106384277344, "loss": 0.4239, "rewards/accuracies": 0.9375, "rewards/chosen": -0.14157333970069885, "rewards/margins": 1.2773213386535645, "rewards/rejected": -1.4188947677612305, "step": 998 }, { "epoch": 1.140332121730454, "grad_norm": 61.40854796108108, "learning_rate": 8.597389217784267e-08, "logits/chosen": -1.2670400142669678, "logits/rejected": -1.3082315921783447, "logps/chosen": -159.52301025390625, "logps/rejected": -179.5859832763672, "loss": 0.4284, "rewards/accuracies": 0.875, "rewards/chosen": -0.2508872151374817, "rewards/margins": 0.8329470157623291, "rewards/rejected": -1.083834171295166, "step": 1000 }, { "epoch": 1.140332121730454, "eval_logits/chosen": -1.3981410264968872, "eval_logits/rejected": -1.382390022277832, "eval_logps/chosen": -131.1500701904297, "eval_logps/rejected": -136.96005249023438, "eval_loss": 0.5724203586578369, "eval_rewards/accuracies": 0.6800000071525574, "eval_rewards/chosen": -0.12521003186702728, "eval_rewards/margins": 0.36452141404151917, "eval_rewards/rejected": -0.48973140120506287, "eval_runtime": 19.6429, "eval_samples_per_second": 5.091, "eval_steps_per_second": 1.273, "step": 1000 }, { "epoch": 1.142612785973915, "grad_norm": 67.69710183071128, "learning_rate": 8.55974162524635e-08, "logits/chosen": -1.1661014556884766, "logits/rejected": -1.1639918088912964, "logps/chosen": -152.52914428710938, "logps/rejected": -175.15721130371094, "loss": 0.4415, "rewards/accuracies": 0.6875, "rewards/chosen": -0.44051098823547363, "rewards/margins": 0.7846835255622864, "rewards/rejected": -1.2251945734024048, "step": 1002 }, { "epoch": 1.1448934502173758, "grad_norm": 60.82110671671323, "learning_rate": 8.522114867038322e-08, "logits/chosen": -1.3812358379364014, "logits/rejected": -1.3619797229766846, "logps/chosen": -155.48292541503906, "logps/rejected": -154.6956024169922, "loss": 0.4675, "rewards/accuracies": 0.90625, "rewards/chosen": -0.1355004459619522, "rewards/margins": 0.798072099685669, "rewards/rejected": -0.93357253074646, "step": 1004 }, { "epoch": 1.1471741144608367, "grad_norm": 53.52801508976481, "learning_rate": 8.484509487457171e-08, "logits/chosen": -1.3384727239608765, "logits/rejected": -1.3696333169937134, "logps/chosen": -202.6947479248047, "logps/rejected": -222.9131622314453, "loss": 0.4287, "rewards/accuracies": 0.84375, "rewards/chosen": -0.21906551718711853, "rewards/margins": 1.0475835800170898, "rewards/rejected": -1.2666492462158203, "step": 1006 }, { "epoch": 1.1494547787042977, "grad_norm": 51.71321310860304, "learning_rate": 8.446926030490622e-08, "logits/chosen": -1.391283392906189, "logits/rejected": -1.4445586204528809, "logps/chosen": -123.97080993652344, "logps/rejected": -172.02490234375, "loss": 0.4404, "rewards/accuracies": 0.78125, "rewards/chosen": -0.22346752882003784, "rewards/margins": 0.7628045082092285, "rewards/rejected": -0.9862719774246216, "step": 1008 }, { "epoch": 1.1517354429477584, "grad_norm": 50.571283153367446, "learning_rate": 8.40936503980928e-08, "logits/chosen": -1.2062405347824097, "logits/rejected": -1.2253963947296143, "logps/chosen": -225.68972778320312, "logps/rejected": -253.89883422851562, "loss": 0.4194, "rewards/accuracies": 0.90625, "rewards/chosen": -0.2782267928123474, "rewards/margins": 1.586256504058838, "rewards/rejected": -1.86448335647583, "step": 1010 }, { "epoch": 1.1540161071912194, "grad_norm": 104.63385992757593, "learning_rate": 8.371827058758761e-08, "logits/chosen": -1.3567335605621338, "logits/rejected": -1.373883843421936, "logps/chosen": -173.80685424804688, "logps/rejected": -193.8018798828125, "loss": 0.4405, "rewards/accuracies": 0.875, "rewards/chosen": -0.4084673523902893, "rewards/margins": 0.7369635701179504, "rewards/rejected": -1.1454309225082397, "step": 1012 }, { "epoch": 1.1562967714346803, "grad_norm": 64.35691482136876, "learning_rate": 8.334312630351828e-08, "logits/chosen": -1.3794384002685547, "logits/rejected": -1.4179136753082275, "logps/chosen": -164.9583282470703, "logps/rejected": -195.81439208984375, "loss": 0.4849, "rewards/accuracies": 0.8125, "rewards/chosen": -0.21411646902561188, "rewards/margins": 0.8294214606285095, "rewards/rejected": -1.043537974357605, "step": 1014 }, { "epoch": 1.1585774356781413, "grad_norm": 55.14828540532963, "learning_rate": 8.29682229726054e-08, "logits/chosen": -1.2123723030090332, "logits/rejected": -1.2490003108978271, "logps/chosen": -170.7882537841797, "logps/rejected": -199.4991912841797, "loss": 0.4305, "rewards/accuracies": 0.78125, "rewards/chosen": -0.279404878616333, "rewards/margins": 0.8790395855903625, "rewards/rejected": -1.1584445238113403, "step": 1016 }, { "epoch": 1.1608580999216023, "grad_norm": 56.75948473097209, "learning_rate": 8.2593566018084e-08, "logits/chosen": -1.218788743019104, "logits/rejected": -1.3168693780899048, "logps/chosen": -198.46771240234375, "logps/rejected": -236.20420837402344, "loss": 0.3977, "rewards/accuracies": 0.875, "rewards/chosen": -0.30342423915863037, "rewards/margins": 1.051550269126892, "rewards/rejected": -1.3549745082855225, "step": 1018 }, { "epoch": 1.163138764165063, "grad_norm": 69.31406225752083, "learning_rate": 8.22191608596251e-08, "logits/chosen": -1.1068880558013916, "logits/rejected": -1.1753079891204834, "logps/chosen": -151.9117889404297, "logps/rejected": -186.4574432373047, "loss": 0.5023, "rewards/accuracies": 0.75, "rewards/chosen": -0.28111034631729126, "rewards/margins": 0.7174416184425354, "rewards/rejected": -0.9985520839691162, "step": 1020 }, { "epoch": 1.165419428408524, "grad_norm": 66.36797160504774, "learning_rate": 8.184501291325732e-08, "logits/chosen": -1.4255850315093994, "logits/rejected": -1.4551324844360352, "logps/chosen": -154.44932556152344, "logps/rejected": -180.50991821289062, "loss": 0.4782, "rewards/accuracies": 0.8125, "rewards/chosen": -0.34893348813056946, "rewards/margins": 0.6230674982070923, "rewards/rejected": -0.9720010757446289, "step": 1022 }, { "epoch": 1.167700092651985, "grad_norm": 69.9233343096583, "learning_rate": 8.147112759128858e-08, "logits/chosen": -1.2810206413269043, "logits/rejected": -1.3186777830123901, "logps/chosen": -195.02601623535156, "logps/rejected": -233.76730346679688, "loss": 0.4403, "rewards/accuracies": 0.875, "rewards/chosen": -0.3124176263809204, "rewards/margins": 1.22174870967865, "rewards/rejected": -1.5341663360595703, "step": 1024 }, { "epoch": 1.169980756895446, "grad_norm": 63.32204131506361, "learning_rate": 8.109751030222764e-08, "logits/chosen": -1.3566696643829346, "logits/rejected": -1.4292207956314087, "logps/chosen": -202.13467407226562, "logps/rejected": -226.0039825439453, "loss": 0.4815, "rewards/accuracies": 0.8125, "rewards/chosen": -0.3643057644367218, "rewards/margins": 0.7674057483673096, "rewards/rejected": -1.1317113637924194, "step": 1026 }, { "epoch": 1.1722614211389066, "grad_norm": 47.95687326599466, "learning_rate": 8.072416645070608e-08, "logits/chosen": -1.331808090209961, "logits/rejected": -1.375180959701538, "logps/chosen": -158.04380798339844, "logps/rejected": -180.04217529296875, "loss": 0.4145, "rewards/accuracies": 0.90625, "rewards/chosen": -0.10679011046886444, "rewards/margins": 0.9117978811264038, "rewards/rejected": -1.0185880661010742, "step": 1028 }, { "epoch": 1.1745420853823676, "grad_norm": 56.45833885104353, "learning_rate": 8.035110143739999e-08, "logits/chosen": -1.246973991394043, "logits/rejected": -1.2526628971099854, "logps/chosen": -149.625, "logps/rejected": -221.7806396484375, "loss": 0.458, "rewards/accuracies": 0.75, "rewards/chosen": -0.383423775434494, "rewards/margins": 0.8781226873397827, "rewards/rejected": -1.2615464925765991, "step": 1030 }, { "epoch": 1.1768227496258286, "grad_norm": 54.84274531559865, "learning_rate": 7.997832065895187e-08, "logits/chosen": -1.3046379089355469, "logits/rejected": -1.335304617881775, "logps/chosen": -184.51785278320312, "logps/rejected": -226.02542114257812, "loss": 0.4619, "rewards/accuracies": 0.78125, "rewards/chosen": -0.39717113971710205, "rewards/margins": 0.9696211218833923, "rewards/rejected": -1.3667922019958496, "step": 1032 }, { "epoch": 1.1791034138692895, "grad_norm": 70.40644247341677, "learning_rate": 7.960582950789257e-08, "logits/chosen": -1.4094276428222656, "logits/rejected": -1.3907825946807861, "logps/chosen": -177.15933227539062, "logps/rejected": -189.71926879882812, "loss": 0.466, "rewards/accuracies": 0.8125, "rewards/chosen": -0.43957361578941345, "rewards/margins": 0.5969427824020386, "rewards/rejected": -1.0365164279937744, "step": 1034 }, { "epoch": 1.1813840781127503, "grad_norm": 59.549418196573015, "learning_rate": 7.923363337256327e-08, "logits/chosen": -1.3654837608337402, "logits/rejected": -1.4075124263763428, "logps/chosen": -153.509033203125, "logps/rejected": -202.06356811523438, "loss": 0.4249, "rewards/accuracies": 0.90625, "rewards/chosen": -0.264252245426178, "rewards/margins": 1.4065436124801636, "rewards/rejected": -1.6707959175109863, "step": 1036 }, { "epoch": 1.1836647423562112, "grad_norm": 60.574371000814885, "learning_rate": 7.886173763703757e-08, "logits/chosen": -1.309372067451477, "logits/rejected": -1.3306725025177002, "logps/chosen": -147.5194091796875, "logps/rejected": -155.11155700683594, "loss": 0.4683, "rewards/accuracies": 0.625, "rewards/chosen": -0.18043869733810425, "rewards/margins": 0.39405256509780884, "rewards/rejected": -0.5744912624359131, "step": 1038 }, { "epoch": 1.1859454065996722, "grad_norm": 58.21498952759633, "learning_rate": 7.849014768104353e-08, "logits/chosen": -1.4345394372940063, "logits/rejected": -1.454176425933838, "logps/chosen": -208.239501953125, "logps/rejected": -231.63497924804688, "loss": 0.4582, "rewards/accuracies": 0.8125, "rewards/chosen": -0.41373205184936523, "rewards/margins": 0.9180522561073303, "rewards/rejected": -1.3317843675613403, "step": 1040 }, { "epoch": 1.1882260708431331, "grad_norm": 60.11382019448424, "learning_rate": 7.811886887988596e-08, "logits/chosen": -1.1493500471115112, "logits/rejected": -1.2578613758087158, "logps/chosen": -156.71392822265625, "logps/rejected": -203.9082794189453, "loss": 0.43, "rewards/accuracies": 0.78125, "rewards/chosen": -0.3507915437221527, "rewards/margins": 0.6405390501022339, "rewards/rejected": -0.991330623626709, "step": 1042 }, { "epoch": 1.1905067350865939, "grad_norm": 62.63091780254508, "learning_rate": 7.774790660436856e-08, "logits/chosen": -1.35098135471344, "logits/rejected": -1.3219125270843506, "logps/chosen": -164.2214813232422, "logps/rejected": -184.14907836914062, "loss": 0.4559, "rewards/accuracies": 0.78125, "rewards/chosen": -0.37730497121810913, "rewards/margins": 0.7604411244392395, "rewards/rejected": -1.1377460956573486, "step": 1044 }, { "epoch": 1.1927873993300548, "grad_norm": 58.393702096339915, "learning_rate": 7.737726622071628e-08, "logits/chosen": -1.2446062564849854, "logits/rejected": -1.2715435028076172, "logps/chosen": -161.71238708496094, "logps/rejected": -208.52255249023438, "loss": 0.4614, "rewards/accuracies": 0.71875, "rewards/chosen": -0.5151055455207825, "rewards/margins": 0.7842628359794617, "rewards/rejected": -1.2993682622909546, "step": 1046 }, { "epoch": 1.1950680635735158, "grad_norm": 68.88961801831904, "learning_rate": 7.700695309049766e-08, "logits/chosen": -1.3265782594680786, "logits/rejected": -1.3526149988174438, "logps/chosen": -155.804443359375, "logps/rejected": -158.39266967773438, "loss": 0.4413, "rewards/accuracies": 0.71875, "rewards/chosen": -0.2930993437767029, "rewards/margins": 0.6833476424217224, "rewards/rejected": -0.9764469861984253, "step": 1048 }, { "epoch": 1.1973487278169768, "grad_norm": 62.58520626376503, "learning_rate": 7.663697257054736e-08, "logits/chosen": -1.193183183670044, "logits/rejected": -1.3192574977874756, "logps/chosen": -146.06712341308594, "logps/rejected": -209.09132385253906, "loss": 0.4365, "rewards/accuracies": 0.90625, "rewards/chosen": -0.26093000173568726, "rewards/margins": 0.8910123705863953, "rewards/rejected": -1.151942253112793, "step": 1050 }, { "epoch": 1.1996293920604377, "grad_norm": 60.54325139288285, "learning_rate": 7.62673300128885e-08, "logits/chosen": -1.2697768211364746, "logits/rejected": -1.3025808334350586, "logps/chosen": -140.79417419433594, "logps/rejected": -190.2462158203125, "loss": 0.4533, "rewards/accuracies": 0.84375, "rewards/chosen": -0.18518134951591492, "rewards/margins": 0.8462077975273132, "rewards/rejected": -1.0313892364501953, "step": 1052 }, { "epoch": 1.2019100563038985, "grad_norm": 62.02420670554748, "learning_rate": 7.589803076465546e-08, "logits/chosen": -1.3338916301727295, "logits/rejected": -1.3496989011764526, "logps/chosen": -190.69876098632812, "logps/rejected": -227.00088500976562, "loss": 0.4498, "rewards/accuracies": 0.84375, "rewards/chosen": -0.321012020111084, "rewards/margins": 1.0334088802337646, "rewards/rejected": -1.3544210195541382, "step": 1054 }, { "epoch": 1.2041907205473594, "grad_norm": 65.72599471201299, "learning_rate": 7.552908016801632e-08, "logits/chosen": -1.3135321140289307, "logits/rejected": -1.3163644075393677, "logps/chosen": -183.17430114746094, "logps/rejected": -195.53515625, "loss": 0.4853, "rewards/accuracies": 0.78125, "rewards/chosen": -0.4703221619129181, "rewards/margins": 0.690421998500824, "rewards/rejected": -1.1607441902160645, "step": 1056 }, { "epoch": 1.2064713847908204, "grad_norm": 67.75915859922333, "learning_rate": 7.516048356009577e-08, "logits/chosen": -1.309540867805481, "logits/rejected": -1.3292198181152344, "logps/chosen": -169.94964599609375, "logps/rejected": -181.25917053222656, "loss": 0.4585, "rewards/accuracies": 0.75, "rewards/chosen": -0.44265216588974, "rewards/margins": 0.8768545985221863, "rewards/rejected": -1.3195068836212158, "step": 1058 }, { "epoch": 1.2087520490342811, "grad_norm": 60.73391494662122, "learning_rate": 7.479224627289764e-08, "logits/chosen": -1.2144145965576172, "logits/rejected": -1.2872897386550903, "logps/chosen": -129.9945831298828, "logps/rejected": -174.8819580078125, "loss": 0.4193, "rewards/accuracies": 0.84375, "rewards/chosen": -0.19057044386863708, "rewards/margins": 0.8820828199386597, "rewards/rejected": -1.0726532936096191, "step": 1060 }, { "epoch": 1.211032713277742, "grad_norm": 57.13712403831016, "learning_rate": 7.442437363322812e-08, "logits/chosen": -1.307482361793518, "logits/rejected": -1.3386861085891724, "logps/chosen": -186.2252197265625, "logps/rejected": -227.38156127929688, "loss": 0.4097, "rewards/accuracies": 0.8125, "rewards/chosen": -0.2698296904563904, "rewards/margins": 1.0196818113327026, "rewards/rejected": -1.2895114421844482, "step": 1062 }, { "epoch": 1.213313377521203, "grad_norm": 61.28403933125535, "learning_rate": 7.405687096261839e-08, "logits/chosen": -1.3965141773223877, "logits/rejected": -1.4495915174484253, "logps/chosen": -220.30531311035156, "logps/rejected": -261.9841613769531, "loss": 0.4575, "rewards/accuracies": 0.8125, "rewards/chosen": -0.45436590909957886, "rewards/margins": 1.2083145380020142, "rewards/rejected": -1.6626802682876587, "step": 1064 }, { "epoch": 1.215594041764664, "grad_norm": 66.4214015966849, "learning_rate": 7.368974357724789e-08, "logits/chosen": -1.2381188869476318, "logits/rejected": -1.2131423950195312, "logps/chosen": -164.27276611328125, "logps/rejected": -279.0275573730469, "loss": 0.3947, "rewards/accuracies": 0.875, "rewards/chosen": -0.26724374294281006, "rewards/margins": 1.546440601348877, "rewards/rejected": -1.8136844635009766, "step": 1066 }, { "epoch": 1.217874706008125, "grad_norm": 52.96231029305883, "learning_rate": 7.332299678786722e-08, "logits/chosen": -1.1868572235107422, "logits/rejected": -1.224455714225769, "logps/chosen": -165.09494018554688, "logps/rejected": -200.85031127929688, "loss": 0.4338, "rewards/accuracies": 0.78125, "rewards/chosen": -0.35685399174690247, "rewards/margins": 1.0505053997039795, "rewards/rejected": -1.4073594808578491, "step": 1068 }, { "epoch": 1.2201553702515857, "grad_norm": 65.66793449701784, "learning_rate": 7.295663589972138e-08, "logits/chosen": -1.4080750942230225, "logits/rejected": -1.40921151638031, "logps/chosen": -157.87884521484375, "logps/rejected": -179.00694274902344, "loss": 0.4529, "rewards/accuracies": 0.90625, "rewards/chosen": -0.19519281387329102, "rewards/margins": 0.7116326093673706, "rewards/rejected": -0.9068254828453064, "step": 1070 }, { "epoch": 1.2224360344950467, "grad_norm": 58.03621828629834, "learning_rate": 7.259066621247316e-08, "logits/chosen": -1.3837441205978394, "logits/rejected": -1.4137468338012695, "logps/chosen": -169.6446075439453, "logps/rejected": -189.69842529296875, "loss": 0.4533, "rewards/accuracies": 0.75, "rewards/chosen": -0.2820315361022949, "rewards/margins": 0.677888035774231, "rewards/rejected": -0.9599195718765259, "step": 1072 }, { "epoch": 1.2247166987385076, "grad_norm": 65.14721879160109, "learning_rate": 7.222509302012621e-08, "logits/chosen": -1.3673007488250732, "logits/rejected": -1.3410099744796753, "logps/chosen": -240.47653198242188, "logps/rejected": -257.4016418457031, "loss": 0.4147, "rewards/accuracies": 0.84375, "rewards/chosen": -0.28077253699302673, "rewards/margins": 0.9187532663345337, "rewards/rejected": -1.1995258331298828, "step": 1074 }, { "epoch": 1.2269973629819684, "grad_norm": 61.93965308054813, "learning_rate": 7.185992161094876e-08, "logits/chosen": -1.3297830820083618, "logits/rejected": -1.2881451845169067, "logps/chosen": -202.92349243164062, "logps/rejected": -220.54769897460938, "loss": 0.4607, "rewards/accuracies": 0.78125, "rewards/chosen": -0.2249366044998169, "rewards/margins": 1.0748928785324097, "rewards/rejected": -1.2998292446136475, "step": 1076 }, { "epoch": 1.2292780272254293, "grad_norm": 71.36489123402376, "learning_rate": 7.149515726739678e-08, "logits/chosen": -1.2684777975082397, "logits/rejected": -1.3346588611602783, "logps/chosen": -135.23745727539062, "logps/rejected": -170.05029296875, "loss": 0.5085, "rewards/accuracies": 0.84375, "rewards/chosen": -0.37416163086891174, "rewards/margins": 0.7284354567527771, "rewards/rejected": -1.1025971174240112, "step": 1078 }, { "epoch": 1.2315586914688903, "grad_norm": 63.847418083603706, "learning_rate": 7.113080526603792e-08, "logits/chosen": -1.3170560598373413, "logits/rejected": -1.3284251689910889, "logps/chosen": -226.8069305419922, "logps/rejected": -265.87310791015625, "loss": 0.3977, "rewards/accuracies": 0.78125, "rewards/chosen": -0.6735376715660095, "rewards/margins": 1.2206555604934692, "rewards/rejected": -1.894193172454834, "step": 1080 }, { "epoch": 1.2338393557123513, "grad_norm": 71.41962712931492, "learning_rate": 7.076687087747495e-08, "logits/chosen": -1.2745944261550903, "logits/rejected": -1.3628857135772705, "logps/chosen": -167.65524291992188, "logps/rejected": -192.4241180419922, "loss": 0.5152, "rewards/accuracies": 0.84375, "rewards/chosen": -0.3967045545578003, "rewards/margins": 0.7864773869514465, "rewards/rejected": -1.1831820011138916, "step": 1082 }, { "epoch": 1.2361200199558122, "grad_norm": 63.43060169338188, "learning_rate": 7.040335936626957e-08, "logits/chosen": -1.2247568368911743, "logits/rejected": -1.334959864616394, "logps/chosen": -154.92623901367188, "logps/rejected": -192.81825256347656, "loss": 0.3931, "rewards/accuracies": 0.90625, "rewards/chosen": -0.37357938289642334, "rewards/margins": 1.085322380065918, "rewards/rejected": -1.4589016437530518, "step": 1084 }, { "epoch": 1.238400684199273, "grad_norm": 57.073138047606875, "learning_rate": 7.004027599086634e-08, "logits/chosen": -1.2639906406402588, "logits/rejected": -1.3193964958190918, "logps/chosen": -210.07972717285156, "logps/rejected": -277.6387634277344, "loss": 0.4629, "rewards/accuracies": 0.84375, "rewards/chosen": -0.7190876007080078, "rewards/margins": 1.5234203338623047, "rewards/rejected": -2.2425079345703125, "step": 1086 }, { "epoch": 1.240681348442734, "grad_norm": 59.243249538144596, "learning_rate": 6.967762600351646e-08, "logits/chosen": -1.2877843379974365, "logits/rejected": -1.3444148302078247, "logps/chosen": -159.09422302246094, "logps/rejected": -178.9310760498047, "loss": 0.4381, "rewards/accuracies": 0.6875, "rewards/chosen": -0.4013168215751648, "rewards/margins": 0.7007678747177124, "rewards/rejected": -1.1020846366882324, "step": 1088 }, { "epoch": 1.2429620126861949, "grad_norm": 61.681333545711915, "learning_rate": 6.93154146502019e-08, "logits/chosen": -1.1838276386260986, "logits/rejected": -1.178552269935608, "logps/chosen": -132.74107360839844, "logps/rejected": -161.54119873046875, "loss": 0.453, "rewards/accuracies": 0.875, "rewards/chosen": -0.2650779187679291, "rewards/margins": 0.9995027184486389, "rewards/rejected": -1.2645806074142456, "step": 1090 }, { "epoch": 1.2452426769296558, "grad_norm": 52.74342769824309, "learning_rate": 6.895364717055948e-08, "logits/chosen": -1.2756752967834473, "logits/rejected": -1.339556336402893, "logps/chosen": -166.25335693359375, "logps/rejected": -215.28515625, "loss": 0.4548, "rewards/accuracies": 0.8125, "rewards/chosen": -0.27491295337677, "rewards/margins": 0.772299587726593, "rewards/rejected": -1.0472124814987183, "step": 1092 }, { "epoch": 1.2475233411731166, "grad_norm": 67.05885459459954, "learning_rate": 6.859232879780515e-08, "logits/chosen": -1.2031257152557373, "logits/rejected": -1.2640451192855835, "logps/chosen": -167.1966094970703, "logps/rejected": -168.45986938476562, "loss": 0.438, "rewards/accuracies": 0.75, "rewards/chosen": -0.5370901823043823, "rewards/margins": 0.6850543022155762, "rewards/rejected": -1.222144603729248, "step": 1094 }, { "epoch": 1.2498040054165775, "grad_norm": 61.68077906707433, "learning_rate": 6.823146475865807e-08, "logits/chosen": -1.3824557065963745, "logits/rejected": -1.4685417413711548, "logps/chosen": -170.927490234375, "logps/rejected": -193.9454803466797, "loss": 0.4349, "rewards/accuracies": 0.90625, "rewards/chosen": -0.21729141473770142, "rewards/margins": 0.8320575952529907, "rewards/rejected": -1.049349069595337, "step": 1096 }, { "epoch": 1.2520846696600385, "grad_norm": 60.117698245447464, "learning_rate": 6.787106027326529e-08, "logits/chosen": -1.2755475044250488, "logits/rejected": -1.3001313209533691, "logps/chosen": -176.72113037109375, "logps/rejected": -194.21383666992188, "loss": 0.4634, "rewards/accuracies": 0.71875, "rewards/chosen": -0.5146129131317139, "rewards/margins": 0.5713402628898621, "rewards/rejected": -1.0859532356262207, "step": 1098 }, { "epoch": 1.2543653339034995, "grad_norm": 60.13020970533813, "learning_rate": 6.751112055512604e-08, "logits/chosen": -1.3733175992965698, "logits/rejected": -1.4645140171051025, "logps/chosen": -153.76385498046875, "logps/rejected": -180.75132751464844, "loss": 0.4132, "rewards/accuracies": 0.71875, "rewards/chosen": -0.1185435876250267, "rewards/margins": 0.8378748297691345, "rewards/rejected": -0.9564183950424194, "step": 1100 }, { "epoch": 1.2543653339034995, "eval_logits/chosen": -1.3715007305145264, "eval_logits/rejected": -1.3544986248016357, "eval_logps/chosen": -131.8278350830078, "eval_logps/rejected": -137.9906005859375, "eval_loss": 0.5563453435897827, "eval_rewards/accuracies": 0.7599999904632568, "eval_rewards/chosen": -0.19298861920833588, "eval_rewards/margins": 0.3997974693775177, "eval_rewards/rejected": -0.5927860736846924, "eval_runtime": 19.8497, "eval_samples_per_second": 5.038, "eval_steps_per_second": 1.259, "step": 1100 }, { "epoch": 1.2566459981469604, "grad_norm": 51.569029451796766, "learning_rate": 6.715165081101641e-08, "logits/chosen": -1.2064361572265625, "logits/rejected": -1.268586277961731, "logps/chosen": -170.99136352539062, "logps/rejected": -214.76968383789062, "loss": 0.4277, "rewards/accuracies": 0.8125, "rewards/chosen": -0.3422897160053253, "rewards/margins": 1.1319069862365723, "rewards/rejected": -1.4741966724395752, "step": 1102 }, { "epoch": 1.2589266623904212, "grad_norm": 68.83152296238853, "learning_rate": 6.679265624091392e-08, "logits/chosen": -1.3187097311019897, "logits/rejected": -1.3244363069534302, "logps/chosen": -177.85157775878906, "logps/rejected": -204.6130828857422, "loss": 0.483, "rewards/accuracies": 0.75, "rewards/chosen": -0.32719144225120544, "rewards/margins": 1.074700951576233, "rewards/rejected": -1.4018925428390503, "step": 1104 }, { "epoch": 1.2612073266338821, "grad_norm": 66.02612056942255, "learning_rate": 6.643414203792243e-08, "logits/chosen": -1.229019045829773, "logits/rejected": -1.3234455585479736, "logps/chosen": -184.5186004638672, "logps/rejected": -208.52243041992188, "loss": 0.4241, "rewards/accuracies": 0.84375, "rewards/chosen": -0.3899713158607483, "rewards/margins": 1.090962290763855, "rewards/rejected": -1.4809335470199585, "step": 1106 }, { "epoch": 1.263487990877343, "grad_norm": 59.215567502448074, "learning_rate": 6.607611338819696e-08, "logits/chosen": -1.2996883392333984, "logits/rejected": -1.319692850112915, "logps/chosen": -146.01797485351562, "logps/rejected": -178.61734008789062, "loss": 0.4377, "rewards/accuracies": 0.9375, "rewards/chosen": -0.31348976492881775, "rewards/margins": 0.9113309383392334, "rewards/rejected": -1.2248207330703735, "step": 1108 }, { "epoch": 1.2657686551208038, "grad_norm": 57.07885969385745, "learning_rate": 6.571857547086863e-08, "logits/chosen": -1.3365250825881958, "logits/rejected": -1.213597297668457, "logps/chosen": -202.40350341796875, "logps/rejected": -213.51048278808594, "loss": 0.4194, "rewards/accuracies": 0.875, "rewards/chosen": -0.38011184334754944, "rewards/margins": 1.0593245029449463, "rewards/rejected": -1.4394363164901733, "step": 1110 }, { "epoch": 1.2680493193642648, "grad_norm": 52.36073610875736, "learning_rate": 6.536153345796983e-08, "logits/chosen": -1.212735652923584, "logits/rejected": -1.2983253002166748, "logps/chosen": -208.72747802734375, "logps/rejected": -259.2344665527344, "loss": 0.4074, "rewards/accuracies": 0.8125, "rewards/chosen": -0.7484298348426819, "rewards/margins": 1.3461904525756836, "rewards/rejected": -2.0946204662323, "step": 1112 }, { "epoch": 1.2703299836077258, "grad_norm": 52.44708462406558, "learning_rate": 6.500499251435926e-08, "logits/chosen": -1.2901079654693604, "logits/rejected": -1.3678314685821533, "logps/chosen": -168.54234313964844, "logps/rejected": -214.02752685546875, "loss": 0.3998, "rewards/accuracies": 0.90625, "rewards/chosen": -0.32987186312675476, "rewards/margins": 1.0558862686157227, "rewards/rejected": -1.3857581615447998, "step": 1114 }, { "epoch": 1.2726106478511867, "grad_norm": 52.22832210272109, "learning_rate": 6.464895779764741e-08, "logits/chosen": -1.1556543111801147, "logits/rejected": -1.271488070487976, "logps/chosen": -131.3715057373047, "logps/rejected": -171.28237915039062, "loss": 0.4324, "rewards/accuracies": 0.90625, "rewards/chosen": -0.349406898021698, "rewards/margins": 0.989849328994751, "rewards/rejected": -1.3392562866210938, "step": 1116 }, { "epoch": 1.2748913120946477, "grad_norm": 62.70237944614458, "learning_rate": 6.429343445812177e-08, "logits/chosen": -1.146794319152832, "logits/rejected": -1.218197226524353, "logps/chosen": -194.02381896972656, "logps/rejected": -233.72328186035156, "loss": 0.4746, "rewards/accuracies": 0.75, "rewards/chosen": -0.5142129063606262, "rewards/margins": 0.9550236463546753, "rewards/rejected": -1.4692366123199463, "step": 1118 }, { "epoch": 1.2771719763381084, "grad_norm": 62.09523040000935, "learning_rate": 6.393842763867248e-08, "logits/chosen": -1.3159422874450684, "logits/rejected": -1.3710412979125977, "logps/chosen": -217.96322631835938, "logps/rejected": -217.71226501464844, "loss": 0.4275, "rewards/accuracies": 0.8125, "rewards/chosen": -0.5582747459411621, "rewards/margins": 0.7185696959495544, "rewards/rejected": -1.2768443822860718, "step": 1120 }, { "epoch": 1.2794526405815694, "grad_norm": 63.94486693511555, "learning_rate": 6.358394247471777e-08, "logits/chosen": -1.1774568557739258, "logits/rejected": -1.2224806547164917, "logps/chosen": -190.8328857421875, "logps/rejected": -234.33628845214844, "loss": 0.4315, "rewards/accuracies": 0.875, "rewards/chosen": -0.6915558576583862, "rewards/margins": 1.1698423624038696, "rewards/rejected": -1.8613982200622559, "step": 1122 }, { "epoch": 1.2817333048250303, "grad_norm": 59.101218821423586, "learning_rate": 6.322998409412988e-08, "logits/chosen": -1.33734929561615, "logits/rejected": -1.409659504890442, "logps/chosen": -187.33851623535156, "logps/rejected": -218.93324279785156, "loss": 0.439, "rewards/accuracies": 0.90625, "rewards/chosen": -0.5945166945457458, "rewards/margins": 0.9995917081832886, "rewards/rejected": -1.5941084623336792, "step": 1124 }, { "epoch": 1.284013969068491, "grad_norm": 59.13516077796158, "learning_rate": 6.287655761716067e-08, "logits/chosen": -1.3064815998077393, "logits/rejected": -1.3387144804000854, "logps/chosen": -191.6893768310547, "logps/rejected": -218.80178833007812, "loss": 0.4099, "rewards/accuracies": 0.90625, "rewards/chosen": -0.41915464401245117, "rewards/margins": 0.9991384744644165, "rewards/rejected": -1.4182932376861572, "step": 1126 }, { "epoch": 1.286294633311952, "grad_norm": 73.05044565719012, "learning_rate": 6.252366815636767e-08, "logits/chosen": -1.2290081977844238, "logits/rejected": -1.2006083726882935, "logps/chosen": -205.61956787109375, "logps/rejected": -196.34405517578125, "loss": 0.4412, "rewards/accuracies": 0.84375, "rewards/chosen": -0.7214517593383789, "rewards/margins": 0.7542897462844849, "rewards/rejected": -1.4757413864135742, "step": 1128 }, { "epoch": 1.288575297555413, "grad_norm": 58.9381764531572, "learning_rate": 6.217132081654014e-08, "logits/chosen": -1.2354235649108887, "logits/rejected": -1.2623546123504639, "logps/chosen": -160.61277770996094, "logps/rejected": -189.03482055664062, "loss": 0.4053, "rewards/accuracies": 0.78125, "rewards/chosen": -0.610559344291687, "rewards/margins": 1.2747613191604614, "rewards/rejected": -1.8853206634521484, "step": 1130 }, { "epoch": 1.290855961798874, "grad_norm": 63.57271664339758, "learning_rate": 6.181952069462513e-08, "logits/chosen": -1.2558073997497559, "logits/rejected": -1.3187260627746582, "logps/chosen": -182.9962158203125, "logps/rejected": -224.17080688476562, "loss": 0.4249, "rewards/accuracies": 0.78125, "rewards/chosen": -0.39853131771087646, "rewards/margins": 1.0594068765640259, "rewards/rejected": -1.4579381942749023, "step": 1132 }, { "epoch": 1.293136626042335, "grad_norm": 54.23291951216934, "learning_rate": 6.146827287965381e-08, "logits/chosen": -1.3496458530426025, "logits/rejected": -1.3700135946273804, "logps/chosen": -213.7442626953125, "logps/rejected": -243.01422119140625, "loss": 0.4322, "rewards/accuracies": 0.625, "rewards/chosen": -0.5680805444717407, "rewards/margins": 1.0590269565582275, "rewards/rejected": -1.6271076202392578, "step": 1134 }, { "epoch": 1.2954172902857957, "grad_norm": 56.45605987126908, "learning_rate": 6.111758245266794e-08, "logits/chosen": -1.3360776901245117, "logits/rejected": -1.2823421955108643, "logps/chosen": -135.13218688964844, "logps/rejected": -145.7807159423828, "loss": 0.4184, "rewards/accuracies": 0.75, "rewards/chosen": -0.2994767427444458, "rewards/margins": 0.7106032967567444, "rewards/rejected": -1.010080099105835, "step": 1136 }, { "epoch": 1.2976979545292566, "grad_norm": 67.42621295544383, "learning_rate": 6.076745448664615e-08, "logits/chosen": -1.3192064762115479, "logits/rejected": -1.3712198734283447, "logps/chosen": -173.24655151367188, "logps/rejected": -196.41253662109375, "loss": 0.4958, "rewards/accuracies": 0.84375, "rewards/chosen": -0.2075835019350052, "rewards/margins": 0.8700404763221741, "rewards/rejected": -1.077623963356018, "step": 1138 }, { "epoch": 1.2999786187727176, "grad_norm": 63.99357195093472, "learning_rate": 6.041789404643078e-08, "logits/chosen": -1.245415210723877, "logits/rejected": -1.3109159469604492, "logps/chosen": -131.79176330566406, "logps/rejected": -169.92637634277344, "loss": 0.4754, "rewards/accuracies": 0.75, "rewards/chosen": -0.31556999683380127, "rewards/margins": 0.7632789015769958, "rewards/rejected": -1.0788488388061523, "step": 1140 }, { "epoch": 1.3022592830161783, "grad_norm": 67.11890446031333, "learning_rate": 6.006890618865446e-08, "logits/chosen": -1.250425100326538, "logits/rejected": -1.3222934007644653, "logps/chosen": -192.59100341796875, "logps/rejected": -223.69459533691406, "loss": 0.4643, "rewards/accuracies": 0.875, "rewards/chosen": -0.32158908247947693, "rewards/margins": 1.268847107887268, "rewards/rejected": -1.5904361009597778, "step": 1142 }, { "epoch": 1.3045399472596393, "grad_norm": 76.00247534873895, "learning_rate": 5.972049596166705e-08, "logits/chosen": -1.1002310514450073, "logits/rejected": -1.1576155424118042, "logps/chosen": -148.41189575195312, "logps/rejected": -185.99664306640625, "loss": 0.4429, "rewards/accuracies": 0.875, "rewards/chosen": -0.632346510887146, "rewards/margins": 1.0468285083770752, "rewards/rejected": -1.6791750192642212, "step": 1144 }, { "epoch": 1.3068206115031002, "grad_norm": 70.43915324421563, "learning_rate": 5.9372668405462646e-08, "logits/chosen": -1.3556523323059082, "logits/rejected": -1.4173972606658936, "logps/chosen": -212.95169067382812, "logps/rejected": -238.5531768798828, "loss": 0.4643, "rewards/accuracies": 0.9375, "rewards/chosen": -0.40602514147758484, "rewards/margins": 1.0105608701705933, "rewards/rejected": -1.4165860414505005, "step": 1146 }, { "epoch": 1.3091012757465612, "grad_norm": 63.83492445068513, "learning_rate": 5.9025428551606415e-08, "logits/chosen": -1.244753360748291, "logits/rejected": -1.273009181022644, "logps/chosen": -164.7984619140625, "logps/rejected": -216.9807586669922, "loss": 0.4013, "rewards/accuracies": 0.90625, "rewards/chosen": -0.4480228126049042, "rewards/margins": 1.1551215648651123, "rewards/rejected": -1.6031445264816284, "step": 1148 }, { "epoch": 1.3113819399900222, "grad_norm": 60.05293035275106, "learning_rate": 5.867878142316221e-08, "logits/chosen": -1.3057725429534912, "logits/rejected": -1.3353570699691772, "logps/chosen": -145.6079559326172, "logps/rejected": -162.2236328125, "loss": 0.4074, "rewards/accuracies": 0.8125, "rewards/chosen": -0.13641442358493805, "rewards/margins": 0.7820553779602051, "rewards/rejected": -0.9184697270393372, "step": 1150 }, { "epoch": 1.3136626042334831, "grad_norm": 65.80534748016166, "learning_rate": 5.833273203461959e-08, "logits/chosen": -1.2780495882034302, "logits/rejected": -1.3019449710845947, "logps/chosen": -196.4916229248047, "logps/rejected": -222.07977294921875, "loss": 0.4397, "rewards/accuracies": 0.8125, "rewards/chosen": -0.3648252487182617, "rewards/margins": 1.3595582246780396, "rewards/rejected": -1.7243834733963013, "step": 1152 }, { "epoch": 1.3159432684769439, "grad_norm": 77.90046778105933, "learning_rate": 5.798728539182143e-08, "logits/chosen": -1.3802424669265747, "logits/rejected": -1.3905105590820312, "logps/chosen": -158.01589965820312, "logps/rejected": -197.33522033691406, "loss": 0.4201, "rewards/accuracies": 0.78125, "rewards/chosen": -0.4259686768054962, "rewards/margins": 0.7603559494018555, "rewards/rejected": -1.1863245964050293, "step": 1154 }, { "epoch": 1.3182239327204048, "grad_norm": 57.72791245543785, "learning_rate": 5.764244649189149e-08, "logits/chosen": -1.2129665613174438, "logits/rejected": -1.2697325944900513, "logps/chosen": -178.7556610107422, "logps/rejected": -201.34042358398438, "loss": 0.4094, "rewards/accuracies": 0.8125, "rewards/chosen": -0.3163738548755646, "rewards/margins": 0.971676766872406, "rewards/rejected": -1.288050651550293, "step": 1156 }, { "epoch": 1.3205045969638658, "grad_norm": 70.16500681265146, "learning_rate": 5.729822032316208e-08, "logits/chosen": -1.2184062004089355, "logits/rejected": -1.3518949747085571, "logps/chosen": -127.61387634277344, "logps/rejected": -177.82791137695312, "loss": 0.4509, "rewards/accuracies": 0.84375, "rewards/chosen": -0.3493412733078003, "rewards/margins": 1.1434459686279297, "rewards/rejected": -1.49278724193573, "step": 1158 }, { "epoch": 1.3227852612073265, "grad_norm": 65.43520672265807, "learning_rate": 5.6954611865101935e-08, "logits/chosen": -1.35573410987854, "logits/rejected": -1.4151753187179565, "logps/chosen": -219.81471252441406, "logps/rejected": -242.21942138671875, "loss": 0.4384, "rewards/accuracies": 0.90625, "rewards/chosen": -0.508373498916626, "rewards/margins": 1.2028579711914062, "rewards/rejected": -1.7112314701080322, "step": 1160 }, { "epoch": 1.3250659254507875, "grad_norm": 63.06305687675219, "learning_rate": 5.661162608824419e-08, "logits/chosen": -1.2267005443572998, "logits/rejected": -1.3211195468902588, "logps/chosen": -124.29898071289062, "logps/rejected": -159.11904907226562, "loss": 0.448, "rewards/accuracies": 0.90625, "rewards/chosen": -0.21557267010211945, "rewards/margins": 0.8949982523918152, "rewards/rejected": -1.110571026802063, "step": 1162 }, { "epoch": 1.3273465896942485, "grad_norm": 50.82426863386907, "learning_rate": 5.6269267954114464e-08, "logits/chosen": -1.232271671295166, "logits/rejected": -1.1851279735565186, "logps/chosen": -157.5489501953125, "logps/rejected": -258.4756774902344, "loss": 0.4687, "rewards/accuracies": 0.84375, "rewards/chosen": -0.4816133379936218, "rewards/margins": 1.42643141746521, "rewards/rejected": -1.9080448150634766, "step": 1164 }, { "epoch": 1.3296272539377094, "grad_norm": 59.70507538342753, "learning_rate": 5.592754241515908e-08, "logits/chosen": -1.3426597118377686, "logits/rejected": -1.379440426826477, "logps/chosen": -213.9489288330078, "logps/rejected": -245.739501953125, "loss": 0.4244, "rewards/accuracies": 0.8125, "rewards/chosen": -0.4688984155654907, "rewards/margins": 1.3086878061294556, "rewards/rejected": -1.7775862216949463, "step": 1166 }, { "epoch": 1.3319079181811704, "grad_norm": 68.73916859324237, "learning_rate": 5.5586454414673454e-08, "logits/chosen": -1.3552615642547607, "logits/rejected": -1.3137784004211426, "logps/chosen": -172.09751892089844, "logps/rejected": -183.2836456298828, "loss": 0.4544, "rewards/accuracies": 0.84375, "rewards/chosen": -0.30267298221588135, "rewards/margins": 1.0140584707260132, "rewards/rejected": -1.3167314529418945, "step": 1168 }, { "epoch": 1.3341885824246311, "grad_norm": 60.83087482532621, "learning_rate": 5.524600888673057e-08, "logits/chosen": -1.2173874378204346, "logits/rejected": -1.325149655342102, "logps/chosen": -156.27577209472656, "logps/rejected": -196.09017944335938, "loss": 0.4569, "rewards/accuracies": 0.9375, "rewards/chosen": -0.2902172803878784, "rewards/margins": 1.0630940198898315, "rewards/rejected": -1.3533114194869995, "step": 1170 }, { "epoch": 1.336469246668092, "grad_norm": 65.80542191040618, "learning_rate": 5.4906210756109575e-08, "logits/chosen": -1.223040223121643, "logits/rejected": -1.1960320472717285, "logps/chosen": -200.8917694091797, "logps/rejected": -229.284423828125, "loss": 0.453, "rewards/accuracies": 0.75, "rewards/chosen": -0.5325464606285095, "rewards/margins": 1.22454833984375, "rewards/rejected": -1.7570948600769043, "step": 1172 }, { "epoch": 1.338749910911553, "grad_norm": 60.664933897307115, "learning_rate": 5.456706493822458e-08, "logits/chosen": -1.3014591932296753, "logits/rejected": -1.277079463005066, "logps/chosen": -167.12942504882812, "logps/rejected": -180.93209838867188, "loss": 0.4185, "rewards/accuracies": 0.84375, "rewards/chosen": -0.23974554240703583, "rewards/margins": 0.8305579423904419, "rewards/rejected": -1.0703034400939941, "step": 1174 }, { "epoch": 1.3410305751550138, "grad_norm": 67.3008532319244, "learning_rate": 5.4228576339053577e-08, "logits/chosen": -1.2223035097122192, "logits/rejected": -1.2677267789840698, "logps/chosen": -170.18605041503906, "logps/rejected": -188.31619262695312, "loss": 0.4476, "rewards/accuracies": 0.75, "rewards/chosen": -0.4260440766811371, "rewards/margins": 0.6966915130615234, "rewards/rejected": -1.1227355003356934, "step": 1176 }, { "epoch": 1.3433112393984747, "grad_norm": 58.74035582903428, "learning_rate": 5.38907498550674e-08, "logits/chosen": -1.263763666152954, "logits/rejected": -1.3518104553222656, "logps/chosen": -218.0510711669922, "logps/rejected": -249.43292236328125, "loss": 0.4739, "rewards/accuracies": 0.96875, "rewards/chosen": -0.3573762774467468, "rewards/margins": 1.326561450958252, "rewards/rejected": -1.6839375495910645, "step": 1178 }, { "epoch": 1.3455919036419357, "grad_norm": 57.92848976074756, "learning_rate": 5.355359037315893e-08, "logits/chosen": -1.2013936042785645, "logits/rejected": -1.3312106132507324, "logps/chosen": -142.94082641601562, "logps/rejected": -221.46524047851562, "loss": 0.4267, "rewards/accuracies": 0.875, "rewards/chosen": -0.3215872645378113, "rewards/margins": 1.4045162200927734, "rewards/rejected": -1.726103663444519, "step": 1180 }, { "epoch": 1.3478725678853967, "grad_norm": 55.43645182737105, "learning_rate": 5.3217102770572466e-08, "logits/chosen": -1.3156689405441284, "logits/rejected": -1.2861976623535156, "logps/chosen": -195.7543182373047, "logps/rejected": -248.62838745117188, "loss": 0.4405, "rewards/accuracies": 0.84375, "rewards/chosen": -0.4373939037322998, "rewards/margins": 1.2866684198379517, "rewards/rejected": -1.7240623235702515, "step": 1182 }, { "epoch": 1.3501532321288576, "grad_norm": 57.579451083148605, "learning_rate": 5.288129191483296e-08, "logits/chosen": -1.2084459066390991, "logits/rejected": -1.1476478576660156, "logps/chosen": -136.50244140625, "logps/rejected": -144.5672149658203, "loss": 0.4725, "rewards/accuracies": 0.6875, "rewards/chosen": -0.35504642128944397, "rewards/margins": 0.5072875022888184, "rewards/rejected": -0.8623339533805847, "step": 1184 }, { "epoch": 1.3524338963723184, "grad_norm": 51.29532599637972, "learning_rate": 5.254616266367591e-08, "logits/chosen": -1.2815008163452148, "logits/rejected": -1.3232269287109375, "logps/chosen": -141.40736389160156, "logps/rejected": -170.84877014160156, "loss": 0.4018, "rewards/accuracies": 0.78125, "rewards/chosen": -0.3718317449092865, "rewards/margins": 0.9131482839584351, "rewards/rejected": -1.284980058670044, "step": 1186 }, { "epoch": 1.3547145606157793, "grad_norm": 68.71395143271354, "learning_rate": 5.2211719864976855e-08, "logits/chosen": -1.3111282587051392, "logits/rejected": -1.2713509798049927, "logps/chosen": -168.83644104003906, "logps/rejected": -189.7281951904297, "loss": 0.4716, "rewards/accuracies": 0.71875, "rewards/chosen": -0.8040088415145874, "rewards/margins": 0.7993437647819519, "rewards/rejected": -1.603352665901184, "step": 1188 }, { "epoch": 1.3569952248592403, "grad_norm": 47.5011022235632, "learning_rate": 5.1877968356681365e-08, "logits/chosen": -1.3059942722320557, "logits/rejected": -1.3891113996505737, "logps/chosen": -139.248046875, "logps/rejected": -177.55447387695312, "loss": 0.4136, "rewards/accuracies": 0.96875, "rewards/chosen": -0.3202630281448364, "rewards/margins": 1.2098579406738281, "rewards/rejected": -1.530120849609375, "step": 1190 }, { "epoch": 1.359275889102701, "grad_norm": 61.28282316729743, "learning_rate": 5.154491296673499e-08, "logits/chosen": -1.3505187034606934, "logits/rejected": -1.3525949716567993, "logps/chosen": -151.95126342773438, "logps/rejected": -170.53799438476562, "loss": 0.4111, "rewards/accuracies": 0.78125, "rewards/chosen": -0.29294249415397644, "rewards/margins": 0.8566387295722961, "rewards/rejected": -1.1495811939239502, "step": 1192 }, { "epoch": 1.361556553346162, "grad_norm": 50.10110632099787, "learning_rate": 5.121255851301345e-08, "logits/chosen": -1.182908058166504, "logits/rejected": -1.268027901649475, "logps/chosen": -184.67160034179688, "logps/rejected": -275.9751892089844, "loss": 0.4072, "rewards/accuracies": 0.96875, "rewards/chosen": -0.4911373257637024, "rewards/margins": 1.55191171169281, "rewards/rejected": -2.043048858642578, "step": 1194 }, { "epoch": 1.363837217589623, "grad_norm": 54.80752104995682, "learning_rate": 5.088090980325297e-08, "logits/chosen": -1.2852065563201904, "logits/rejected": -1.3752135038375854, "logps/chosen": -202.42112731933594, "logps/rejected": -243.62234497070312, "loss": 0.4245, "rewards/accuracies": 0.875, "rewards/chosen": -0.540120542049408, "rewards/margins": 1.2409517765045166, "rewards/rejected": -1.7810722589492798, "step": 1196 }, { "epoch": 1.366117881833084, "grad_norm": 57.983612303856106, "learning_rate": 5.054997163498065e-08, "logits/chosen": -1.4842596054077148, "logits/rejected": -1.454995036125183, "logps/chosen": -213.9556884765625, "logps/rejected": -208.71621704101562, "loss": 0.4446, "rewards/accuracies": 0.8125, "rewards/chosen": -0.512903094291687, "rewards/margins": 0.7354660034179688, "rewards/rejected": -1.2483689785003662, "step": 1198 }, { "epoch": 1.3683985460765449, "grad_norm": 56.17410591987438, "learning_rate": 5.0219748795445214e-08, "logits/chosen": -1.39608895778656, "logits/rejected": -1.4197230339050293, "logps/chosen": -151.9307403564453, "logps/rejected": -177.78945922851562, "loss": 0.3957, "rewards/accuracies": 0.90625, "rewards/chosen": -0.07908481359481812, "rewards/margins": 1.1620317697525024, "rewards/rejected": -1.2411167621612549, "step": 1200 }, { "epoch": 1.3683985460765449, "eval_logits/chosen": -1.3583488464355469, "eval_logits/rejected": -1.341153860092163, "eval_logps/chosen": -132.06036376953125, "eval_logps/rejected": -138.48941040039062, "eval_loss": 0.5543201565742493, "eval_rewards/accuracies": 0.7599999904632568, "eval_rewards/chosen": -0.2162403166294098, "eval_rewards/margins": 0.42642509937286377, "eval_rewards/rejected": -0.6426653861999512, "eval_runtime": 20.2372, "eval_samples_per_second": 4.941, "eval_steps_per_second": 1.235, "step": 1200 }, { "epoch": 1.3706792103200058, "grad_norm": 64.03519542961854, "learning_rate": 4.98902460615475e-08, "logits/chosen": -1.3389551639556885, "logits/rejected": -1.3822637796401978, "logps/chosen": -175.14820861816406, "logps/rejected": -222.5068359375, "loss": 0.4712, "rewards/accuracies": 0.75, "rewards/chosen": -0.35669347643852234, "rewards/margins": 1.1921041011810303, "rewards/rejected": -1.548797607421875, "step": 1202 }, { "epoch": 1.3729598745634666, "grad_norm": 51.71910925101713, "learning_rate": 4.956146819977166e-08, "logits/chosen": -1.253232717514038, "logits/rejected": -1.2863636016845703, "logps/chosen": -176.71702575683594, "logps/rejected": -210.63230895996094, "loss": 0.4178, "rewards/accuracies": 0.84375, "rewards/chosen": -0.32385319471359253, "rewards/margins": 1.246777057647705, "rewards/rejected": -1.570630431175232, "step": 1204 }, { "epoch": 1.3752405388069275, "grad_norm": 56.79692940311852, "learning_rate": 4.923341996611603e-08, "logits/chosen": -1.1557482481002808, "logits/rejected": -1.180600881576538, "logps/chosen": -150.3299102783203, "logps/rejected": -167.61911010742188, "loss": 0.4255, "rewards/accuracies": 0.6875, "rewards/chosen": -0.3654107451438904, "rewards/margins": 0.7367621064186096, "rewards/rejected": -1.1021727323532104, "step": 1206 }, { "epoch": 1.3775212030503885, "grad_norm": 57.06474376403373, "learning_rate": 4.890610610602437e-08, "logits/chosen": -1.297890067100525, "logits/rejected": -1.347840666770935, "logps/chosen": -209.23716735839844, "logps/rejected": -259.1905517578125, "loss": 0.3961, "rewards/accuracies": 0.875, "rewards/chosen": -0.45185887813568115, "rewards/margins": 1.451210856437683, "rewards/rejected": -1.9030694961547852, "step": 1208 }, { "epoch": 1.3798018672938492, "grad_norm": 60.29263036985363, "learning_rate": 4.8579531354317225e-08, "logits/chosen": -1.397212266921997, "logits/rejected": -1.3925597667694092, "logps/chosen": -183.3704833984375, "logps/rejected": -195.10496520996094, "loss": 0.4579, "rewards/accuracies": 0.75, "rewards/chosen": -0.7833220362663269, "rewards/margins": 0.7912861108779907, "rewards/rejected": -1.574608325958252, "step": 1210 }, { "epoch": 1.3820825315373102, "grad_norm": 52.32294118146283, "learning_rate": 4.825370043512339e-08, "logits/chosen": -1.3067998886108398, "logits/rejected": -1.3849916458129883, "logps/chosen": -162.16232299804688, "logps/rejected": -195.49609375, "loss": 0.4108, "rewards/accuracies": 0.875, "rewards/chosen": -0.33422791957855225, "rewards/margins": 1.0588434934616089, "rewards/rejected": -1.3930714130401611, "step": 1212 }, { "epoch": 1.3843631957807712, "grad_norm": 59.66742993964118, "learning_rate": 4.792861806181171e-08, "logits/chosen": -1.30001962184906, "logits/rejected": -1.386979341506958, "logps/chosen": -170.89529418945312, "logps/rejected": -197.6066131591797, "loss": 0.4525, "rewards/accuracies": 0.8125, "rewards/chosen": -0.2925351560115814, "rewards/margins": 0.9441651105880737, "rewards/rejected": -1.2367002964019775, "step": 1214 }, { "epoch": 1.3866438600242321, "grad_norm": 65.32724813467337, "learning_rate": 4.760428893692273e-08, "logits/chosen": -1.2875810861587524, "logits/rejected": -1.3601633310317993, "logps/chosen": -160.88156127929688, "logps/rejected": -190.6903076171875, "loss": 0.4288, "rewards/accuracies": 0.71875, "rewards/chosen": -0.6599161624908447, "rewards/margins": 0.7817404270172119, "rewards/rejected": -1.4416565895080566, "step": 1216 }, { "epoch": 1.388924524267693, "grad_norm": 52.68799788674585, "learning_rate": 4.728071775210069e-08, "logits/chosen": -1.3100471496582031, "logits/rejected": -1.295668125152588, "logps/chosen": -157.36289978027344, "logps/rejected": -179.96127319335938, "loss": 0.4265, "rewards/accuracies": 0.75, "rewards/chosen": -0.36468327045440674, "rewards/margins": 0.9460710883140564, "rewards/rejected": -1.3107542991638184, "step": 1218 }, { "epoch": 1.3912051885111538, "grad_norm": 59.812629327754166, "learning_rate": 4.695790918802576e-08, "logits/chosen": -1.4171504974365234, "logits/rejected": -1.3959426879882812, "logps/chosen": -226.54220581054688, "logps/rejected": -242.77142333984375, "loss": 0.4516, "rewards/accuracies": 0.71875, "rewards/chosen": -0.6130508184432983, "rewards/margins": 1.1557624340057373, "rewards/rejected": -1.768813133239746, "step": 1220 }, { "epoch": 1.3934858527546148, "grad_norm": 66.91529385196273, "learning_rate": 4.663586791434628e-08, "logits/chosen": -1.1362406015396118, "logits/rejected": -1.2372556924819946, "logps/chosen": -182.88442993164062, "logps/rejected": -217.1130828857422, "loss": 0.4164, "rewards/accuracies": 0.78125, "rewards/chosen": -0.5246292352676392, "rewards/margins": 1.0273668766021729, "rewards/rejected": -1.551996111869812, "step": 1222 }, { "epoch": 1.3957665169980757, "grad_norm": 57.29717940934624, "learning_rate": 4.631459858961122e-08, "logits/chosen": -1.219807744026184, "logits/rejected": -1.2460401058197021, "logps/chosen": -153.93907165527344, "logps/rejected": -200.4837188720703, "loss": 0.4358, "rewards/accuracies": 0.84375, "rewards/chosen": -0.2589726746082306, "rewards/margins": 1.4164376258850098, "rewards/rejected": -1.675410509109497, "step": 1224 }, { "epoch": 1.3980471812415365, "grad_norm": 61.08923102786796, "learning_rate": 4.5994105861202715e-08, "logits/chosen": -1.3403871059417725, "logits/rejected": -1.3382725715637207, "logps/chosen": -147.61839294433594, "logps/rejected": -150.13453674316406, "loss": 0.4522, "rewards/accuracies": 0.875, "rewards/chosen": -0.403720498085022, "rewards/margins": 0.7644599080085754, "rewards/rejected": -1.1681804656982422, "step": 1226 }, { "epoch": 1.4003278454849974, "grad_norm": 55.589268418781224, "learning_rate": 4.5674394365268965e-08, "logits/chosen": -1.1604863405227661, "logits/rejected": -1.2054895162582397, "logps/chosen": -169.14215087890625, "logps/rejected": -203.62513732910156, "loss": 0.4333, "rewards/accuracies": 0.78125, "rewards/chosen": -0.32466423511505127, "rewards/margins": 1.318765640258789, "rewards/rejected": -1.6434298753738403, "step": 1228 }, { "epoch": 1.4026085097284584, "grad_norm": 50.39174380587635, "learning_rate": 4.535546872665707e-08, "logits/chosen": -1.2403908967971802, "logits/rejected": -1.3284348249435425, "logps/chosen": -154.95285034179688, "logps/rejected": -183.48558044433594, "loss": 0.5428, "rewards/accuracies": 0.75, "rewards/chosen": -0.2970479428768158, "rewards/margins": 0.813206672668457, "rewards/rejected": -1.1102546453475952, "step": 1230 }, { "epoch": 1.4048891739719194, "grad_norm": 64.62636544867223, "learning_rate": 4.5037333558846145e-08, "logits/chosen": -1.1855900287628174, "logits/rejected": -1.1975244283676147, "logps/chosen": -102.4107437133789, "logps/rejected": -124.4788589477539, "loss": 0.4184, "rewards/accuracies": 0.84375, "rewards/chosen": -0.20104435086250305, "rewards/margins": 0.7573148608207703, "rewards/rejected": -0.9583592414855957, "step": 1232 }, { "epoch": 1.4071698382153803, "grad_norm": 65.48395677041327, "learning_rate": 4.471999346388069e-08, "logits/chosen": -1.2656984329223633, "logits/rejected": -1.2520796060562134, "logps/chosen": -145.40838623046875, "logps/rejected": -172.25161743164062, "loss": 0.4563, "rewards/accuracies": 0.875, "rewards/chosen": -0.48840075731277466, "rewards/margins": 0.857879102230072, "rewards/rejected": -1.3462798595428467, "step": 1234 }, { "epoch": 1.409450502458841, "grad_norm": 53.15387042970869, "learning_rate": 4.4403453032303765e-08, "logits/chosen": -1.393466591835022, "logits/rejected": -1.4557361602783203, "logps/chosen": -160.5977020263672, "logps/rejected": -189.77520751953125, "loss": 0.4181, "rewards/accuracies": 0.90625, "rewards/chosen": -0.379285603761673, "rewards/margins": 1.075732946395874, "rewards/rejected": -1.4550185203552246, "step": 1236 }, { "epoch": 1.411731166702302, "grad_norm": 63.15445502776712, "learning_rate": 4.4087716843090895e-08, "logits/chosen": -1.3475301265716553, "logits/rejected": -1.3905658721923828, "logps/chosen": -166.34042358398438, "logps/rejected": -213.61553955078125, "loss": 0.461, "rewards/accuracies": 0.75, "rewards/chosen": -0.3821715712547302, "rewards/margins": 0.9115235805511475, "rewards/rejected": -1.2936952114105225, "step": 1238 }, { "epoch": 1.414011830945763, "grad_norm": 52.37085158764497, "learning_rate": 4.3772789463583627e-08, "logits/chosen": -1.3224272727966309, "logits/rejected": -1.3503855466842651, "logps/chosen": -172.18899536132812, "logps/rejected": -193.6123504638672, "loss": 0.4386, "rewards/accuracies": 0.78125, "rewards/chosen": -0.44380372762680054, "rewards/margins": 0.857738733291626, "rewards/rejected": -1.3015424013137817, "step": 1240 }, { "epoch": 1.4162924951892237, "grad_norm": 64.55355325183278, "learning_rate": 4.345867544942353e-08, "logits/chosen": -1.2631657123565674, "logits/rejected": -1.3294970989227295, "logps/chosen": -181.64718627929688, "logps/rejected": -230.25701904296875, "loss": 0.4121, "rewards/accuracies": 0.71875, "rewards/chosen": -0.5138639211654663, "rewards/margins": 0.8278074264526367, "rewards/rejected": -1.341671347618103, "step": 1242 }, { "epoch": 1.4185731594326847, "grad_norm": 65.60045840124785, "learning_rate": 4.314537934448628e-08, "logits/chosen": -1.2260847091674805, "logits/rejected": -1.2379428148269653, "logps/chosen": -175.8683624267578, "logps/rejected": -202.67051696777344, "loss": 0.4525, "rewards/accuracies": 0.875, "rewards/chosen": -0.4962596893310547, "rewards/margins": 1.1486889123916626, "rewards/rejected": -1.6449487209320068, "step": 1244 }, { "epoch": 1.4208538236761457, "grad_norm": 77.13313632492196, "learning_rate": 4.283290568081591e-08, "logits/chosen": -1.1557011604309082, "logits/rejected": -1.2198400497436523, "logps/chosen": -171.91201782226562, "logps/rejected": -200.71144104003906, "loss": 0.4431, "rewards/accuracies": 0.84375, "rewards/chosen": -0.4828857481479645, "rewards/margins": 0.8654166460037231, "rewards/rejected": -1.3483023643493652, "step": 1246 }, { "epoch": 1.4231344879196066, "grad_norm": 55.00661044224625, "learning_rate": 4.2521258978559314e-08, "logits/chosen": -1.258105993270874, "logits/rejected": -1.282645583152771, "logps/chosen": -180.73272705078125, "logps/rejected": -241.80764770507812, "loss": 0.4077, "rewards/accuracies": 0.875, "rewards/chosen": -0.5227749347686768, "rewards/margins": 1.5435858964920044, "rewards/rejected": -2.0663607120513916, "step": 1248 }, { "epoch": 1.4254151521630676, "grad_norm": 53.935655347865456, "learning_rate": 4.2210443745900804e-08, "logits/chosen": -1.1817071437835693, "logits/rejected": -1.247178077697754, "logps/chosen": -151.1819610595703, "logps/rejected": -170.9989776611328, "loss": 0.4273, "rewards/accuracies": 0.8125, "rewards/chosen": -0.3710220456123352, "rewards/margins": 0.9180817604064941, "rewards/rejected": -1.2891038656234741, "step": 1250 }, { "epoch": 1.4276958164065285, "grad_norm": 57.828569007499375, "learning_rate": 4.190046447899689e-08, "logits/chosen": -1.2078405618667603, "logits/rejected": -1.2860413789749146, "logps/chosen": -141.80389404296875, "logps/rejected": -168.6719512939453, "loss": 0.4219, "rewards/accuracies": 0.75, "rewards/chosen": -0.4949069023132324, "rewards/margins": 0.8605579137802124, "rewards/rejected": -1.3554648160934448, "step": 1252 }, { "epoch": 1.4299764806499893, "grad_norm": 49.820693081313166, "learning_rate": 4.159132566191129e-08, "logits/chosen": -1.3393913507461548, "logits/rejected": -1.393333911895752, "logps/chosen": -139.9639434814453, "logps/rejected": -185.2383575439453, "loss": 0.401, "rewards/accuracies": 0.78125, "rewards/chosen": -0.3605578541755676, "rewards/margins": 1.1692156791687012, "rewards/rejected": -1.5297735929489136, "step": 1254 }, { "epoch": 1.4322571448934502, "grad_norm": 56.66007237622535, "learning_rate": 4.1283031766550014e-08, "logits/chosen": -1.159570336341858, "logits/rejected": -1.2179524898529053, "logps/chosen": -167.4163818359375, "logps/rejected": -245.31614685058594, "loss": 0.4344, "rewards/accuracies": 0.75, "rewards/chosen": -0.5021023750305176, "rewards/margins": 1.504926085472107, "rewards/rejected": -2.007028341293335, "step": 1256 }, { "epoch": 1.4345378091369112, "grad_norm": 70.62489205877687, "learning_rate": 4.097558725259672e-08, "logits/chosen": -1.33053457736969, "logits/rejected": -1.388944149017334, "logps/chosen": -178.24188232421875, "logps/rejected": -234.7947998046875, "loss": 0.3998, "rewards/accuracies": 0.78125, "rewards/chosen": -0.40949881076812744, "rewards/margins": 1.050101399421692, "rewards/rejected": -1.4596002101898193, "step": 1258 }, { "epoch": 1.436818473380372, "grad_norm": 66.98990507182315, "learning_rate": 4.0668996567448154e-08, "logits/chosen": -1.414581060409546, "logits/rejected": -1.4014796018600464, "logps/chosen": -169.44393920898438, "logps/rejected": -187.62033081054688, "loss": 0.4344, "rewards/accuracies": 0.65625, "rewards/chosen": -0.3710756003856659, "rewards/margins": 0.5481195449829102, "rewards/rejected": -0.9191950559616089, "step": 1260 }, { "epoch": 1.439099137623833, "grad_norm": 57.85916556533107, "learning_rate": 4.0363264146149844e-08, "logits/chosen": -1.191425085067749, "logits/rejected": -1.2502682209014893, "logps/chosen": -190.24957275390625, "logps/rejected": -221.98013305664062, "loss": 0.4816, "rewards/accuracies": 0.875, "rewards/chosen": -0.5489203333854675, "rewards/margins": 1.0626386404037476, "rewards/rejected": -1.6115591526031494, "step": 1262 }, { "epoch": 1.4413798018672939, "grad_norm": 54.90589553967945, "learning_rate": 4.005839441133198e-08, "logits/chosen": -1.243033766746521, "logits/rejected": -1.3693134784698486, "logps/chosen": -153.51661682128906, "logps/rejected": -217.17115783691406, "loss": 0.4357, "rewards/accuracies": 0.75, "rewards/chosen": -0.41171663999557495, "rewards/margins": 1.114880084991455, "rewards/rejected": -1.5265967845916748, "step": 1264 }, { "epoch": 1.4436604661107548, "grad_norm": 62.949403049310924, "learning_rate": 3.9754391773145326e-08, "logits/chosen": -1.3396437168121338, "logits/rejected": -1.3314851522445679, "logps/chosen": -195.9013671875, "logps/rejected": -226.99224853515625, "loss": 0.4549, "rewards/accuracies": 0.78125, "rewards/chosen": -0.6362702250480652, "rewards/margins": 0.9720792174339294, "rewards/rejected": -1.6083494424819946, "step": 1266 }, { "epoch": 1.4459411303542158, "grad_norm": 109.10943345175065, "learning_rate": 3.945126062919756e-08, "logits/chosen": -1.4142718315124512, "logits/rejected": -1.3863128423690796, "logps/chosen": -232.01536560058594, "logps/rejected": -258.9195861816406, "loss": 0.4683, "rewards/accuracies": 0.84375, "rewards/chosen": -0.5452659130096436, "rewards/margins": 1.1544592380523682, "rewards/rejected": -1.6997252702713013, "step": 1268 }, { "epoch": 1.4482217945976765, "grad_norm": 66.47161188134956, "learning_rate": 3.914900536448959e-08, "logits/chosen": -1.277639389038086, "logits/rejected": -1.2443594932556152, "logps/chosen": -167.44473266601562, "logps/rejected": -201.05442810058594, "loss": 0.468, "rewards/accuracies": 0.75, "rewards/chosen": -0.40504151582717896, "rewards/margins": 1.4868779182434082, "rewards/rejected": -1.8919193744659424, "step": 1270 }, { "epoch": 1.4505024588411375, "grad_norm": 65.21617644771925, "learning_rate": 3.8847630351352045e-08, "logits/chosen": -1.2669049501419067, "logits/rejected": -1.2787154912948608, "logps/chosen": -244.93118286132812, "logps/rejected": -307.34344482421875, "loss": 0.4429, "rewards/accuracies": 0.78125, "rewards/chosen": -0.9488785862922668, "rewards/margins": 1.5935635566711426, "rewards/rejected": -2.5424418449401855, "step": 1272 }, { "epoch": 1.4527831230845984, "grad_norm": 67.80268214039134, "learning_rate": 3.854713994938221e-08, "logits/chosen": -1.3307723999023438, "logits/rejected": -1.400517463684082, "logps/chosen": -156.8104248046875, "logps/rejected": -182.17807006835938, "loss": 0.4631, "rewards/accuracies": 0.78125, "rewards/chosen": -0.2960550785064697, "rewards/margins": 0.9029641151428223, "rewards/rejected": -1.1990193128585815, "step": 1274 }, { "epoch": 1.4550637873280592, "grad_norm": 66.08251651441701, "learning_rate": 3.8247538505380816e-08, "logits/chosen": -1.3477903604507446, "logits/rejected": -1.4483450651168823, "logps/chosen": -194.99876403808594, "logps/rejected": -235.32986450195312, "loss": 0.4232, "rewards/accuracies": 0.78125, "rewards/chosen": -0.44531339406967163, "rewards/margins": 0.8939595222473145, "rewards/rejected": -1.3392727375030518, "step": 1276 }, { "epoch": 1.4573444515715201, "grad_norm": 64.19995573754183, "learning_rate": 3.794883035328921e-08, "logits/chosen": -1.2755396366119385, "logits/rejected": -1.3425655364990234, "logps/chosen": -186.4855499267578, "logps/rejected": -233.87648010253906, "loss": 0.4074, "rewards/accuracies": 0.84375, "rewards/chosen": -0.3420637249946594, "rewards/margins": 1.2022613286972046, "rewards/rejected": -1.5443251132965088, "step": 1278 }, { "epoch": 1.4596251158149811, "grad_norm": 51.505850800471386, "learning_rate": 3.765101981412665e-08, "logits/chosen": -1.0226508378982544, "logits/rejected": -1.1150188446044922, "logps/chosen": -165.54244995117188, "logps/rejected": -209.08641052246094, "loss": 0.4498, "rewards/accuracies": 0.78125, "rewards/chosen": -0.6313174962997437, "rewards/margins": 0.6761065721511841, "rewards/rejected": -1.3074240684509277, "step": 1280 }, { "epoch": 1.461905780058442, "grad_norm": 65.45552204323278, "learning_rate": 3.735411119592782e-08, "logits/chosen": -1.1841048002243042, "logits/rejected": -1.1225578784942627, "logps/chosen": -210.95909118652344, "logps/rejected": -228.45896911621094, "loss": 0.4092, "rewards/accuracies": 0.90625, "rewards/chosen": -0.6781859397888184, "rewards/margins": 1.1642651557922363, "rewards/rejected": -1.8424510955810547, "step": 1282 }, { "epoch": 1.464186444301903, "grad_norm": 54.509138851706474, "learning_rate": 3.705810879368047e-08, "logits/chosen": -1.260365605354309, "logits/rejected": -1.2909530401229858, "logps/chosen": -191.3780517578125, "logps/rejected": -205.57510375976562, "loss": 0.4331, "rewards/accuracies": 0.84375, "rewards/chosen": -0.3257947564125061, "rewards/margins": 1.1021668910980225, "rewards/rejected": -1.4279615879058838, "step": 1284 }, { "epoch": 1.4664671085453638, "grad_norm": 62.120618048817526, "learning_rate": 3.6763016889263345e-08, "logits/chosen": -1.233807921409607, "logits/rejected": -1.1972962617874146, "logps/chosen": -130.00839233398438, "logps/rejected": -157.5312957763672, "loss": 0.4483, "rewards/accuracies": 0.78125, "rewards/chosen": -0.3486970067024231, "rewards/margins": 0.8312156796455383, "rewards/rejected": -1.1799125671386719, "step": 1286 }, { "epoch": 1.4687477727888247, "grad_norm": 62.47302706565703, "learning_rate": 3.6468839751384206e-08, "logits/chosen": -1.2912284135818481, "logits/rejected": -1.286245584487915, "logps/chosen": -225.38636779785156, "logps/rejected": -253.08624267578125, "loss": 0.4193, "rewards/accuracies": 0.8125, "rewards/chosen": -0.7340028285980225, "rewards/margins": 1.2588945627212524, "rewards/rejected": -1.9928972721099854, "step": 1288 }, { "epoch": 1.4710284370322857, "grad_norm": 56.815010583661, "learning_rate": 3.6175581635518015e-08, "logits/chosen": -1.3371250629425049, "logits/rejected": -1.3072669506072998, "logps/chosen": -188.99107360839844, "logps/rejected": -214.9119873046875, "loss": 0.4364, "rewards/accuracies": 0.875, "rewards/chosen": -0.46143385767936707, "rewards/margins": 1.1111385822296143, "rewards/rejected": -1.5725722312927246, "step": 1290 }, { "epoch": 1.4733091012757464, "grad_norm": 74.02679784754338, "learning_rate": 3.5883246783845543e-08, "logits/chosen": -1.2495771646499634, "logits/rejected": -1.2960941791534424, "logps/chosen": -157.36351013183594, "logps/rejected": -196.3470916748047, "loss": 0.4401, "rewards/accuracies": 0.84375, "rewards/chosen": -0.23140710592269897, "rewards/margins": 1.0461037158966064, "rewards/rejected": -1.2775108814239502, "step": 1292 }, { "epoch": 1.4755897655192074, "grad_norm": 61.44266871054178, "learning_rate": 3.559183942519188e-08, "logits/chosen": -1.3195384740829468, "logits/rejected": -1.3464099168777466, "logps/chosen": -170.23886108398438, "logps/rejected": -182.47579956054688, "loss": 0.4044, "rewards/accuracies": 0.75, "rewards/chosen": -0.5699801445007324, "rewards/margins": 0.663261890411377, "rewards/rejected": -1.233242154121399, "step": 1294 }, { "epoch": 1.4778704297626684, "grad_norm": 65.18540943608312, "learning_rate": 3.530136377496525e-08, "logits/chosen": -1.278255581855774, "logits/rejected": -1.2963995933532715, "logps/chosen": -214.0897216796875, "logps/rejected": -238.54718017578125, "loss": 0.4458, "rewards/accuracies": 0.90625, "rewards/chosen": -0.5689273476600647, "rewards/margins": 1.0575151443481445, "rewards/rejected": -1.6264426708221436, "step": 1296 }, { "epoch": 1.4801510940061293, "grad_norm": 56.51869240720061, "learning_rate": 3.50118240350961e-08, "logits/chosen": -1.2410857677459717, "logits/rejected": -1.3088514804840088, "logps/chosen": -158.88926696777344, "logps/rejected": -195.86138916015625, "loss": 0.449, "rewards/accuracies": 0.625, "rewards/chosen": -0.5141651630401611, "rewards/margins": 0.7266778945922852, "rewards/rejected": -1.2408430576324463, "step": 1298 }, { "epoch": 1.4824317582495903, "grad_norm": 69.4034723256464, "learning_rate": 3.472322439397635e-08, "logits/chosen": -1.3286279439926147, "logits/rejected": -1.384574294090271, "logps/chosen": -219.34544372558594, "logps/rejected": -237.2283172607422, "loss": 0.4893, "rewards/accuracies": 0.53125, "rewards/chosen": -0.6901402473449707, "rewards/margins": 0.6859029531478882, "rewards/rejected": -1.3760432004928589, "step": 1300 }, { "epoch": 1.4824317582495903, "eval_logits/chosen": -1.3521403074264526, "eval_logits/rejected": -1.3340317010879517, "eval_logps/chosen": -131.97569274902344, "eval_logps/rejected": -138.84446716308594, "eval_loss": 0.5475608110427856, "eval_rewards/accuracies": 0.7200000286102295, "eval_rewards/chosen": -0.20777291059494019, "eval_rewards/margins": 0.47040116786956787, "eval_rewards/rejected": -0.6781739592552185, "eval_runtime": 21.6329, "eval_samples_per_second": 4.623, "eval_steps_per_second": 1.156, "step": 1300 }, { "epoch": 1.484712422493051, "grad_norm": 66.5561433557891, "learning_rate": 3.4435569026398645e-08, "logits/chosen": -1.1897640228271484, "logits/rejected": -1.3370938301086426, "logps/chosen": -141.83460998535156, "logps/rejected": -182.74176025390625, "loss": 0.4463, "rewards/accuracies": 0.8125, "rewards/chosen": -0.18475398421287537, "rewards/margins": 0.82102370262146, "rewards/rejected": -1.0057775974273682, "step": 1302 }, { "epoch": 1.486993086736512, "grad_norm": 48.88961410554484, "learning_rate": 3.4148862093496145e-08, "logits/chosen": -1.2648987770080566, "logits/rejected": -1.274294376373291, "logps/chosen": -161.31605529785156, "logps/rejected": -165.04278564453125, "loss": 0.3889, "rewards/accuracies": 0.75, "rewards/chosen": -0.4667995870113373, "rewards/margins": 0.7909737229347229, "rewards/rejected": -1.2577731609344482, "step": 1304 }, { "epoch": 1.489273750979973, "grad_norm": 67.73951681369277, "learning_rate": 3.386310774268214e-08, "logits/chosen": -1.2678455114364624, "logits/rejected": -1.320731520652771, "logps/chosen": -204.93576049804688, "logps/rejected": -220.75344848632812, "loss": 0.433, "rewards/accuracies": 0.84375, "rewards/chosen": -0.5178290605545044, "rewards/margins": 0.9241759777069092, "rewards/rejected": -1.4420050382614136, "step": 1306 }, { "epoch": 1.491554415223434, "grad_norm": 53.60885013196278, "learning_rate": 3.3578310107590255e-08, "logits/chosen": -1.2518330812454224, "logits/rejected": -1.3134666681289673, "logps/chosen": -128.73098754882812, "logps/rejected": -141.5028839111328, "loss": 0.4175, "rewards/accuracies": 0.78125, "rewards/chosen": -0.4798870086669922, "rewards/margins": 0.7654281854629517, "rewards/rejected": -1.2453151941299438, "step": 1308 }, { "epoch": 1.4938350794668946, "grad_norm": 64.42291386610013, "learning_rate": 3.329447330801455e-08, "logits/chosen": -1.2452740669250488, "logits/rejected": -1.2846417427062988, "logps/chosen": -127.42951202392578, "logps/rejected": -177.83033752441406, "loss": 0.4905, "rewards/accuracies": 0.78125, "rewards/chosen": -0.4574730694293976, "rewards/margins": 1.024438500404358, "rewards/rejected": -1.481911540031433, "step": 1310 }, { "epoch": 1.4961157437103556, "grad_norm": 60.99913378119905, "learning_rate": 3.3011601449849914e-08, "logits/chosen": -1.2262144088745117, "logits/rejected": -1.200211763381958, "logps/chosen": -175.07473754882812, "logps/rejected": -194.3573455810547, "loss": 0.419, "rewards/accuracies": 0.78125, "rewards/chosen": -0.5272756814956665, "rewards/margins": 1.1627264022827148, "rewards/rejected": -1.6900020837783813, "step": 1312 }, { "epoch": 1.4983964079538166, "grad_norm": 61.30357952779087, "learning_rate": 3.272969862503271e-08, "logits/chosen": -1.258878231048584, "logits/rejected": -1.3279513120651245, "logps/chosen": -169.57151794433594, "logps/rejected": -205.15086364746094, "loss": 0.4254, "rewards/accuracies": 0.9375, "rewards/chosen": -0.45883575081825256, "rewards/margins": 1.1377463340759277, "rewards/rejected": -1.5965821743011475, "step": 1314 }, { "epoch": 1.5006770721972775, "grad_norm": 60.7970290747735, "learning_rate": 3.2448768911481574e-08, "logits/chosen": -1.3344089984893799, "logits/rejected": -1.3924615383148193, "logps/chosen": -242.09625244140625, "logps/rejected": -284.8103942871094, "loss": 0.3898, "rewards/accuracies": 0.90625, "rewards/chosen": -0.42801302671432495, "rewards/margins": 1.6990736722946167, "rewards/rejected": -2.127086639404297, "step": 1316 }, { "epoch": 1.5029577364407385, "grad_norm": 58.23756711117375, "learning_rate": 3.216881637303839e-08, "logits/chosen": -1.2963494062423706, "logits/rejected": -1.2964308261871338, "logps/chosen": -194.5859375, "logps/rejected": -207.51177978515625, "loss": 0.4461, "rewards/accuracies": 0.84375, "rewards/chosen": -0.37401753664016724, "rewards/margins": 0.9556913375854492, "rewards/rejected": -1.3297089338302612, "step": 1318 }, { "epoch": 1.5052384006841992, "grad_norm": 52.85976726825495, "learning_rate": 3.188984505940955e-08, "logits/chosen": -1.310511589050293, "logits/rejected": -1.3698493242263794, "logps/chosen": -155.9166717529297, "logps/rejected": -185.9949951171875, "loss": 0.4608, "rewards/accuracies": 0.78125, "rewards/chosen": -0.28941965103149414, "rewards/margins": 0.9209386110305786, "rewards/rejected": -1.2103582620620728, "step": 1320 }, { "epoch": 1.5075190649276602, "grad_norm": 65.51821056925988, "learning_rate": 3.161185900610737e-08, "logits/chosen": -1.2264206409454346, "logits/rejected": -1.2563592195510864, "logps/chosen": -169.9464111328125, "logps/rejected": -215.42799377441406, "loss": 0.4746, "rewards/accuracies": 0.84375, "rewards/chosen": -0.45204079151153564, "rewards/margins": 0.9063868522644043, "rewards/rejected": -1.35842764377594, "step": 1322 }, { "epoch": 1.509799729171121, "grad_norm": 65.41148602924677, "learning_rate": 3.1334862234391624e-08, "logits/chosen": -1.1124111413955688, "logits/rejected": -1.2182986736297607, "logps/chosen": -168.18359375, "logps/rejected": -212.33782958984375, "loss": 0.3879, "rewards/accuracies": 0.9375, "rewards/chosen": -0.4647515118122101, "rewards/margins": 1.2126293182373047, "rewards/rejected": -1.677380919456482, "step": 1324 }, { "epoch": 1.512080393414582, "grad_norm": 56.95611387521237, "learning_rate": 3.105885875121151e-08, "logits/chosen": -1.2453256845474243, "logits/rejected": -1.3663585186004639, "logps/chosen": -183.23095703125, "logps/rejected": -234.28765869140625, "loss": 0.4144, "rewards/accuracies": 0.90625, "rewards/chosen": -0.22449856996536255, "rewards/margins": 1.4339529275894165, "rewards/rejected": -1.6584514379501343, "step": 1326 }, { "epoch": 1.5143610576580429, "grad_norm": 75.28120333419699, "learning_rate": 3.078385254914764e-08, "logits/chosen": -1.246031403541565, "logits/rejected": -1.3311541080474854, "logps/chosen": -181.19082641601562, "logps/rejected": -224.01809692382812, "loss": 0.4438, "rewards/accuracies": 0.71875, "rewards/chosen": -0.5463556051254272, "rewards/margins": 1.012133240699768, "rewards/rejected": -1.5584888458251953, "step": 1328 }, { "epoch": 1.5166417219015038, "grad_norm": 59.21659370389, "learning_rate": 3.0509847606354214e-08, "logits/chosen": -1.1559014320373535, "logits/rejected": -1.175613522529602, "logps/chosen": -175.91024780273438, "logps/rejected": -201.29910278320312, "loss": 0.4106, "rewards/accuracies": 0.9375, "rewards/chosen": -0.5821335315704346, "rewards/margins": 1.0530939102172852, "rewards/rejected": -1.6352273225784302, "step": 1330 }, { "epoch": 1.5189223861449648, "grad_norm": 65.02090257303733, "learning_rate": 3.0236847886501535e-08, "logits/chosen": -1.3365192413330078, "logits/rejected": -1.3719249963760376, "logps/chosen": -195.5978546142578, "logps/rejected": -219.9347686767578, "loss": 0.3962, "rewards/accuracies": 0.90625, "rewards/chosen": -0.3096795678138733, "rewards/margins": 1.1052062511444092, "rewards/rejected": -1.4148855209350586, "step": 1332 }, { "epoch": 1.5212030503884257, "grad_norm": 59.28308867818603, "learning_rate": 2.9964857338718716e-08, "logits/chosen": -1.357577919960022, "logits/rejected": -1.2968313694000244, "logps/chosen": -199.23211669921875, "logps/rejected": -185.07896423339844, "loss": 0.4879, "rewards/accuracies": 0.71875, "rewards/chosen": -0.36101239919662476, "rewards/margins": 0.6830317974090576, "rewards/rejected": -1.0440441370010376, "step": 1334 }, { "epoch": 1.5234837146318867, "grad_norm": 55.43183849143864, "learning_rate": 2.9693879897536432e-08, "logits/chosen": -1.1980834007263184, "logits/rejected": -1.229064702987671, "logps/chosen": -199.63746643066406, "logps/rejected": -221.99234008789062, "loss": 0.4335, "rewards/accuracies": 0.84375, "rewards/chosen": -0.5939749479293823, "rewards/margins": 1.0861446857452393, "rewards/rejected": -1.6801198720932007, "step": 1336 }, { "epoch": 1.5257643788753474, "grad_norm": 46.58400293343511, "learning_rate": 2.9423919482830118e-08, "logits/chosen": -1.1741948127746582, "logits/rejected": -1.2695064544677734, "logps/chosen": -147.90426635742188, "logps/rejected": -210.15591430664062, "loss": 0.4373, "rewards/accuracies": 0.90625, "rewards/chosen": -0.16375023126602173, "rewards/margins": 1.4425245523452759, "rewards/rejected": -1.6062746047973633, "step": 1338 }, { "epoch": 1.5280450431188084, "grad_norm": 60.486372599547316, "learning_rate": 2.9154979999763197e-08, "logits/chosen": -1.2853294610977173, "logits/rejected": -1.3705867528915405, "logps/chosen": -159.66326904296875, "logps/rejected": -226.9650421142578, "loss": 0.4299, "rewards/accuracies": 0.90625, "rewards/chosen": -0.5052796006202698, "rewards/margins": 1.3695988655090332, "rewards/rejected": -1.8748785257339478, "step": 1340 }, { "epoch": 1.5303257073622691, "grad_norm": 66.91951190930519, "learning_rate": 2.8887065338730633e-08, "logits/chosen": -1.1936756372451782, "logits/rejected": -1.2785755395889282, "logps/chosen": -156.64866638183594, "logps/rejected": -195.68453979492188, "loss": 0.4813, "rewards/accuracies": 0.8125, "rewards/chosen": -0.30220431089401245, "rewards/margins": 0.8970733880996704, "rewards/rejected": -1.1992775201797485, "step": 1342 }, { "epoch": 1.53260637160573, "grad_norm": 60.327377050918805, "learning_rate": 2.86201793753026e-08, "logits/chosen": -1.2364442348480225, "logits/rejected": -1.2909033298492432, "logps/chosen": -191.64169311523438, "logps/rejected": -228.86416625976562, "loss": 0.4307, "rewards/accuracies": 0.8125, "rewards/chosen": -0.5236424207687378, "rewards/margins": 0.9988542795181274, "rewards/rejected": -1.5224968194961548, "step": 1344 }, { "epoch": 1.534887035849191, "grad_norm": 71.82683810198446, "learning_rate": 2.835432597016848e-08, "logits/chosen": -1.19918692111969, "logits/rejected": -1.2287514209747314, "logps/chosen": -203.8031768798828, "logps/rejected": -217.00051879882812, "loss": 0.4555, "rewards/accuracies": 0.9375, "rewards/chosen": -0.5840819478034973, "rewards/margins": 1.0689098834991455, "rewards/rejected": -1.6529918909072876, "step": 1346 }, { "epoch": 1.537167700092652, "grad_norm": 51.67351577452576, "learning_rate": 2.8089508969081e-08, "logits/chosen": -1.1072896718978882, "logits/rejected": -1.1964483261108398, "logps/chosen": -167.41482543945312, "logps/rejected": -227.43304443359375, "loss": 0.3874, "rewards/accuracies": 0.84375, "rewards/chosen": -0.4981079697608948, "rewards/margins": 1.5203238725662231, "rewards/rejected": -2.0184319019317627, "step": 1348 }, { "epoch": 1.539448364336113, "grad_norm": 54.16559868864558, "learning_rate": 2.7825732202800544e-08, "logits/chosen": -1.1472342014312744, "logits/rejected": -1.2177824974060059, "logps/chosen": -162.28453063964844, "logps/rejected": -183.14820861816406, "loss": 0.4325, "rewards/accuracies": 0.75, "rewards/chosen": -0.6399708986282349, "rewards/margins": 0.8657874464988708, "rewards/rejected": -1.5057581663131714, "step": 1350 }, { "epoch": 1.541729028579574, "grad_norm": 59.09514477278818, "learning_rate": 2.756299948703982e-08, "logits/chosen": -1.1995211839675903, "logits/rejected": -1.2396866083145142, "logps/chosen": -132.65550231933594, "logps/rejected": -141.451416015625, "loss": 0.4078, "rewards/accuracies": 0.75, "rewards/chosen": -0.38874107599258423, "rewards/margins": 0.7118159532546997, "rewards/rejected": -1.1005568504333496, "step": 1352 }, { "epoch": 1.5440096928230347, "grad_norm": 61.53811528707539, "learning_rate": 2.7301314622408612e-08, "logits/chosen": -1.2403637170791626, "logits/rejected": -1.3120653629302979, "logps/chosen": -156.661865234375, "logps/rejected": -218.12490844726562, "loss": 0.4467, "rewards/accuracies": 0.8125, "rewards/chosen": -0.468991219997406, "rewards/margins": 1.6298167705535889, "rewards/rejected": -2.0988078117370605, "step": 1354 }, { "epoch": 1.5462903570664956, "grad_norm": 68.28115300743727, "learning_rate": 2.704068139435881e-08, "logits/chosen": -1.2020457983016968, "logits/rejected": -1.2723631858825684, "logps/chosen": -167.51483154296875, "logps/rejected": -186.7481231689453, "loss": 0.4608, "rewards/accuracies": 0.71875, "rewards/chosen": -0.5097277760505676, "rewards/margins": 0.8848594427108765, "rewards/rejected": -1.3945870399475098, "step": 1356 }, { "epoch": 1.5485710213099564, "grad_norm": 65.06182702724148, "learning_rate": 2.6781103573129703e-08, "logits/chosen": -1.270340085029602, "logits/rejected": -1.2233667373657227, "logps/chosen": -141.70269775390625, "logps/rejected": -159.1035919189453, "loss": 0.4297, "rewards/accuracies": 0.875, "rewards/chosen": -0.3412969708442688, "rewards/margins": 0.8257958292961121, "rewards/rejected": -1.1670928001403809, "step": 1358 }, { "epoch": 1.5508516855534173, "grad_norm": 54.785688494455215, "learning_rate": 2.652258491369329e-08, "logits/chosen": -1.2447706460952759, "logits/rejected": -1.2511292695999146, "logps/chosen": -171.25672912597656, "logps/rejected": -208.33084106445312, "loss": 0.4368, "rewards/accuracies": 0.875, "rewards/chosen": -0.3487054705619812, "rewards/margins": 1.1124207973480225, "rewards/rejected": -1.4611263275146484, "step": 1360 }, { "epoch": 1.5531323497968783, "grad_norm": 61.41628077442576, "learning_rate": 2.626512915570015e-08, "logits/chosen": -1.328946590423584, "logits/rejected": -1.3554866313934326, "logps/chosen": -125.71770477294922, "logps/rejected": -141.460693359375, "loss": 0.4368, "rewards/accuracies": 0.8125, "rewards/chosen": -0.13910508155822754, "rewards/margins": 0.7971990704536438, "rewards/rejected": -0.9363042116165161, "step": 1362 }, { "epoch": 1.5554130140403393, "grad_norm": 75.66249491591283, "learning_rate": 2.6008740023425247e-08, "logits/chosen": -1.188770055770874, "logits/rejected": -1.2130908966064453, "logps/chosen": -183.416748046875, "logps/rejected": -207.94090270996094, "loss": 0.4306, "rewards/accuracies": 0.71875, "rewards/chosen": -0.651962399482727, "rewards/margins": 1.215145230293274, "rewards/rejected": -1.867107629776001, "step": 1364 }, { "epoch": 1.5576936782838002, "grad_norm": 55.08493508678333, "learning_rate": 2.5753421225714055e-08, "logits/chosen": -1.2770978212356567, "logits/rejected": -1.3901137113571167, "logps/chosen": -182.26524353027344, "logps/rejected": -213.17454528808594, "loss": 0.4494, "rewards/accuracies": 0.78125, "rewards/chosen": -0.602812647819519, "rewards/margins": 1.0171971321105957, "rewards/rejected": -1.6200097799301147, "step": 1366 }, { "epoch": 1.5599743425272612, "grad_norm": 61.38622790534087, "learning_rate": 2.549917645592893e-08, "logits/chosen": -1.0256890058517456, "logits/rejected": -1.0421488285064697, "logps/chosen": -125.034423828125, "logps/rejected": -140.7981414794922, "loss": 0.4129, "rewards/accuracies": 0.78125, "rewards/chosen": -0.5211226344108582, "rewards/margins": 0.7981542348861694, "rewards/rejected": -1.3192768096923828, "step": 1368 }, { "epoch": 1.562255006770722, "grad_norm": 56.63177895037392, "learning_rate": 2.524600939189566e-08, "logits/chosen": -1.155221939086914, "logits/rejected": -1.1944361925125122, "logps/chosen": -161.79409790039062, "logps/rejected": -194.83041381835938, "loss": 0.4127, "rewards/accuracies": 0.84375, "rewards/chosen": -0.42054063081741333, "rewards/margins": 1.119522213935852, "rewards/rejected": -1.5400630235671997, "step": 1370 }, { "epoch": 1.564535671014183, "grad_norm": 83.86033988088741, "learning_rate": 2.4993923695850305e-08, "logits/chosen": -1.293369174003601, "logits/rejected": -1.3639140129089355, "logps/chosen": -198.75289916992188, "logps/rejected": -229.84117126464844, "loss": 0.4988, "rewards/accuracies": 0.8125, "rewards/chosen": -0.41869914531707764, "rewards/margins": 0.7414001226425171, "rewards/rejected": -1.1600991487503052, "step": 1372 }, { "epoch": 1.5668163352576436, "grad_norm": 62.24749551918009, "learning_rate": 2.4742923014386154e-08, "logits/chosen": -1.3067548274993896, "logits/rejected": -1.3476440906524658, "logps/chosen": -224.86627197265625, "logps/rejected": -235.1451416015625, "loss": 0.4154, "rewards/accuracies": 0.875, "rewards/chosen": -0.6012443900108337, "rewards/margins": 0.9985144138336182, "rewards/rejected": -1.5997586250305176, "step": 1374 }, { "epoch": 1.5690969995011046, "grad_norm": 64.51543232805054, "learning_rate": 2.4493010978401063e-08, "logits/chosen": -1.0690737962722778, "logits/rejected": -1.084768295288086, "logps/chosen": -163.8816375732422, "logps/rejected": -171.3943634033203, "loss": 0.4249, "rewards/accuracies": 0.875, "rewards/chosen": -0.611855685710907, "rewards/margins": 0.8946461081504822, "rewards/rejected": -1.5065017938613892, "step": 1376 }, { "epoch": 1.5713776637445656, "grad_norm": 54.357068938946284, "learning_rate": 2.4244191203044805e-08, "logits/chosen": -1.4049066305160522, "logits/rejected": -1.4406367540359497, "logps/chosen": -193.67234802246094, "logps/rejected": -223.92030334472656, "loss": 0.3912, "rewards/accuracies": 0.8125, "rewards/chosen": -0.3962157368659973, "rewards/margins": 1.082966685295105, "rewards/rejected": -1.479182481765747, "step": 1378 }, { "epoch": 1.5736583279880265, "grad_norm": 56.319793805013724, "learning_rate": 2.399646728766691e-08, "logits/chosen": -1.3638195991516113, "logits/rejected": -1.3412432670593262, "logps/chosen": -164.48809814453125, "logps/rejected": -196.62823486328125, "loss": 0.4493, "rewards/accuracies": 0.78125, "rewards/chosen": -0.2889966666698456, "rewards/margins": 1.261388897895813, "rewards/rejected": -1.5503859519958496, "step": 1380 }, { "epoch": 1.5759389922314875, "grad_norm": 60.4303807612328, "learning_rate": 2.3749842815764498e-08, "logits/chosen": -1.1395071744918823, "logits/rejected": -1.1399792432785034, "logps/chosen": -139.91824340820312, "logps/rejected": -184.93896484375, "loss": 0.4426, "rewards/accuracies": 0.75, "rewards/chosen": -0.37107953429222107, "rewards/margins": 0.8352210521697998, "rewards/rejected": -1.2063006162643433, "step": 1382 }, { "epoch": 1.5782196564749484, "grad_norm": 55.142567366980046, "learning_rate": 2.3504321354930568e-08, "logits/chosen": -1.1650046110153198, "logits/rejected": -1.2775689363479614, "logps/chosen": -144.93995666503906, "logps/rejected": -177.63087463378906, "loss": 0.4272, "rewards/accuracies": 0.78125, "rewards/chosen": -0.5145533084869385, "rewards/margins": 0.9500089287757874, "rewards/rejected": -1.4645624160766602, "step": 1384 }, { "epoch": 1.5805003207184094, "grad_norm": 63.43007453665504, "learning_rate": 2.3259906456802213e-08, "logits/chosen": -1.216495394706726, "logits/rejected": -1.2768280506134033, "logps/chosen": -169.9758758544922, "logps/rejected": -194.25665283203125, "loss": 0.4427, "rewards/accuracies": 0.78125, "rewards/chosen": -0.43023961782455444, "rewards/margins": 0.8201103210449219, "rewards/rejected": -1.250349998474121, "step": 1386 }, { "epoch": 1.5827809849618701, "grad_norm": 58.56739556116542, "learning_rate": 2.301660165700936e-08, "logits/chosen": -1.291746735572815, "logits/rejected": -1.3648316860198975, "logps/chosen": -166.92015075683594, "logps/rejected": -194.9810028076172, "loss": 0.4506, "rewards/accuracies": 0.8125, "rewards/chosen": -0.5265369415283203, "rewards/margins": 0.8070346713066101, "rewards/rejected": -1.3335715532302856, "step": 1388 }, { "epoch": 1.585061649205331, "grad_norm": 54.69042761154998, "learning_rate": 2.2774410475123608e-08, "logits/chosen": -1.3127899169921875, "logits/rejected": -1.2719758749008179, "logps/chosen": -178.777587890625, "logps/rejected": -247.97511291503906, "loss": 0.3964, "rewards/accuracies": 0.875, "rewards/chosen": -0.19407829642295837, "rewards/margins": 1.4336225986480713, "rewards/rejected": -1.6277010440826416, "step": 1390 }, { "epoch": 1.5873423134487918, "grad_norm": 62.874447838260565, "learning_rate": 2.2533336414607317e-08, "logits/chosen": -1.2624969482421875, "logits/rejected": -1.319579839706421, "logps/chosen": -169.37161254882812, "logps/rejected": -186.8485870361328, "loss": 0.4139, "rewards/accuracies": 0.90625, "rewards/chosen": -0.39721712470054626, "rewards/margins": 0.8743160367012024, "rewards/rejected": -1.2715333700180054, "step": 1392 }, { "epoch": 1.5896229776922528, "grad_norm": 56.69647177405222, "learning_rate": 2.2293382962762853e-08, "logits/chosen": -1.1312240362167358, "logits/rejected": -1.1108318567276, "logps/chosen": -149.8402557373047, "logps/rejected": -147.10037231445312, "loss": 0.4255, "rewards/accuracies": 0.71875, "rewards/chosen": -0.5710242986679077, "rewards/margins": 0.5461159348487854, "rewards/rejected": -1.117140293121338, "step": 1394 }, { "epoch": 1.5919036419357138, "grad_norm": 71.94947354276687, "learning_rate": 2.2054553590682268e-08, "logits/chosen": -1.1476179361343384, "logits/rejected": -1.2181971073150635, "logps/chosen": -152.3661346435547, "logps/rejected": -192.77369689941406, "loss": 0.4157, "rewards/accuracies": 0.875, "rewards/chosen": -0.5518044233322144, "rewards/margins": 1.2432973384857178, "rewards/rejected": -1.7951017618179321, "step": 1396 }, { "epoch": 1.5941843061791747, "grad_norm": 65.43519539314455, "learning_rate": 2.1816851753197018e-08, "logits/chosen": -1.2672624588012695, "logits/rejected": -1.3876826763153076, "logps/chosen": -198.90475463867188, "logps/rejected": -248.56439208984375, "loss": 0.479, "rewards/accuracies": 0.875, "rewards/chosen": -0.7569680213928223, "rewards/margins": 1.3200610876083374, "rewards/rejected": -2.077029228210449, "step": 1398 }, { "epoch": 1.5964649704226357, "grad_norm": 55.75209984482193, "learning_rate": 2.1580280888827997e-08, "logits/chosen": -1.3570483922958374, "logits/rejected": -1.39362633228302, "logps/chosen": -168.9151153564453, "logps/rejected": -199.656982421875, "loss": 0.4361, "rewards/accuracies": 0.8125, "rewards/chosen": -0.38282498717308044, "rewards/margins": 1.1389210224151611, "rewards/rejected": -1.521746039390564, "step": 1400 }, { "epoch": 1.5964649704226357, "eval_logits/chosen": -1.349047064781189, "eval_logits/rejected": -1.3316236734390259, "eval_logps/chosen": -131.90457153320312, "eval_logps/rejected": -138.97027587890625, "eval_loss": 0.5412697792053223, "eval_rewards/accuracies": 0.7200000286102295, "eval_rewards/chosen": -0.2006600797176361, "eval_rewards/margins": 0.49009186029434204, "eval_rewards/rejected": -0.6907519698143005, "eval_runtime": 20.6376, "eval_samples_per_second": 4.846, "eval_steps_per_second": 1.211, "step": 1400 }, { "epoch": 1.5987456346660966, "grad_norm": 58.426952642835886, "learning_rate": 2.1344844419735752e-08, "logits/chosen": -1.3734134435653687, "logits/rejected": -1.4403023719787598, "logps/chosen": -138.4149627685547, "logps/rejected": -162.69033813476562, "loss": 0.4393, "rewards/accuracies": 0.875, "rewards/chosen": -0.2428184598684311, "rewards/margins": 0.8137100338935852, "rewards/rejected": -1.0565285682678223, "step": 1402 }, { "epoch": 1.6010262989095574, "grad_norm": 57.12690162473469, "learning_rate": 2.1110545751671073e-08, "logits/chosen": -1.2013407945632935, "logits/rejected": -1.3104689121246338, "logps/chosen": -137.76437377929688, "logps/rejected": -174.31373596191406, "loss": 0.4367, "rewards/accuracies": 0.875, "rewards/chosen": -0.14584028720855713, "rewards/margins": 0.999373197555542, "rewards/rejected": -1.1452134847640991, "step": 1404 }, { "epoch": 1.6033069631530183, "grad_norm": 60.35122974906749, "learning_rate": 2.0877388273925644e-08, "logits/chosen": -1.227086067199707, "logits/rejected": -1.1707607507705688, "logps/chosen": -209.45407104492188, "logps/rejected": -257.14776611328125, "loss": 0.3969, "rewards/accuracies": 0.8125, "rewards/chosen": -0.581037163734436, "rewards/margins": 1.283540964126587, "rewards/rejected": -1.864578127861023, "step": 1406 }, { "epoch": 1.605587627396479, "grad_norm": 52.46325406296021, "learning_rate": 2.0645375359283045e-08, "logits/chosen": -1.3980076313018799, "logits/rejected": -1.4242044687271118, "logps/chosen": -116.77565002441406, "logps/rejected": -137.52145385742188, "loss": 0.4733, "rewards/accuracies": 0.8125, "rewards/chosen": -0.23290672898292542, "rewards/margins": 0.7732049226760864, "rewards/rejected": -1.0061116218566895, "step": 1408 }, { "epoch": 1.60786829163994, "grad_norm": 55.90140225780788, "learning_rate": 2.0414510363970018e-08, "logits/chosen": -1.1187829971313477, "logits/rejected": -1.1768170595169067, "logps/chosen": -150.73451232910156, "logps/rejected": -189.43748474121094, "loss": 0.3741, "rewards/accuracies": 0.84375, "rewards/chosen": -0.3249818682670593, "rewards/margins": 0.9408416748046875, "rewards/rejected": -1.2658236026763916, "step": 1410 }, { "epoch": 1.610148955883401, "grad_norm": 58.50925426124599, "learning_rate": 2.0184796627607725e-08, "logits/chosen": -1.1422569751739502, "logits/rejected": -1.1473525762557983, "logps/chosen": -188.11888122558594, "logps/rejected": -191.87713623046875, "loss": 0.4537, "rewards/accuracies": 0.78125, "rewards/chosen": -0.35315239429473877, "rewards/margins": 0.663144588470459, "rewards/rejected": -1.0162967443466187, "step": 1412 }, { "epoch": 1.612429620126862, "grad_norm": 64.2495432668556, "learning_rate": 1.9956237473163718e-08, "logits/chosen": -1.2085424661636353, "logits/rejected": -1.306652545928955, "logps/chosen": -97.00778198242188, "logps/rejected": -127.79579162597656, "loss": 0.4977, "rewards/accuracies": 0.75, "rewards/chosen": -0.26015713810920715, "rewards/margins": 0.6060620546340942, "rewards/rejected": -0.866219162940979, "step": 1414 }, { "epoch": 1.614710284370323, "grad_norm": 50.97231528427863, "learning_rate": 1.9728836206903654e-08, "logits/chosen": -1.4538676738739014, "logits/rejected": -1.437317967414856, "logps/chosen": -138.0614776611328, "logps/rejected": -137.07269287109375, "loss": 0.4251, "rewards/accuracies": 0.8125, "rewards/chosen": -0.09943617880344391, "rewards/margins": 0.7697718143463135, "rewards/rejected": -0.8692080974578857, "step": 1416 }, { "epoch": 1.616990948613784, "grad_norm": 51.24642546995866, "learning_rate": 1.9502596118343552e-08, "logits/chosen": -1.0932207107543945, "logits/rejected": -1.1693406105041504, "logps/chosen": -126.78893280029297, "logps/rejected": -140.98338317871094, "loss": 0.4273, "rewards/accuracies": 0.6875, "rewards/chosen": -0.1916692703962326, "rewards/margins": 0.7681133151054382, "rewards/rejected": -0.959782600402832, "step": 1418 }, { "epoch": 1.6192716128572446, "grad_norm": 61.40277169360943, "learning_rate": 1.9277520480202203e-08, "logits/chosen": -1.3312891721725464, "logits/rejected": -1.2821072340011597, "logps/chosen": -151.71487426757812, "logps/rejected": -167.5812225341797, "loss": 0.4401, "rewards/accuracies": 0.9375, "rewards/chosen": -0.4683433473110199, "rewards/margins": 1.1215391159057617, "rewards/rejected": -1.589882493019104, "step": 1420 }, { "epoch": 1.6215522771007056, "grad_norm": 54.3969390887107, "learning_rate": 1.9053612548353803e-08, "logits/chosen": -1.3286281824111938, "logits/rejected": -1.467494249343872, "logps/chosen": -185.03489685058594, "logps/rejected": -245.1704559326172, "loss": 0.3985, "rewards/accuracies": 0.84375, "rewards/chosen": -0.6573659181594849, "rewards/margins": 1.2262755632400513, "rewards/rejected": -1.8836416006088257, "step": 1422 }, { "epoch": 1.6238329413441663, "grad_norm": 54.259477393424476, "learning_rate": 1.8830875561780902e-08, "logits/chosen": -1.2293699979782104, "logits/rejected": -1.3463877439498901, "logps/chosen": -140.27874755859375, "logps/rejected": -198.82611083984375, "loss": 0.3951, "rewards/accuracies": 0.8125, "rewards/chosen": -0.30968916416168213, "rewards/margins": 0.9538030624389648, "rewards/rejected": -1.2634921073913574, "step": 1424 }, { "epoch": 1.6261136055876273, "grad_norm": 61.69027813672776, "learning_rate": 1.8609312742527493e-08, "logits/chosen": -1.2697855234146118, "logits/rejected": -1.2990537881851196, "logps/chosen": -155.84657287597656, "logps/rejected": -212.16445922851562, "loss": 0.4447, "rewards/accuracies": 0.78125, "rewards/chosen": -0.38760051131248474, "rewards/margins": 0.9349652528762817, "rewards/rejected": -1.3225656747817993, "step": 1426 }, { "epoch": 1.6283942698310883, "grad_norm": 54.55311985521747, "learning_rate": 1.8388927295652446e-08, "logits/chosen": -1.285940170288086, "logits/rejected": -1.2539726495742798, "logps/chosen": -165.970947265625, "logps/rejected": -213.34823608398438, "loss": 0.399, "rewards/accuracies": 0.90625, "rewards/chosen": -0.4254089295864105, "rewards/margins": 1.3802815675735474, "rewards/rejected": -1.8056904077529907, "step": 1428 }, { "epoch": 1.6306749340745492, "grad_norm": 76.43067475489244, "learning_rate": 1.81697224091831e-08, "logits/chosen": -1.3153434991836548, "logits/rejected": -1.3224899768829346, "logps/chosen": -164.8282470703125, "logps/rejected": -173.127685546875, "loss": 0.4394, "rewards/accuracies": 0.875, "rewards/chosen": -0.13270916044712067, "rewards/margins": 0.787804901599884, "rewards/rejected": -0.9205139875411987, "step": 1430 }, { "epoch": 1.6329555983180102, "grad_norm": 63.28082306297566, "learning_rate": 1.7951701254069208e-08, "logits/chosen": -1.2579729557037354, "logits/rejected": -1.3529393672943115, "logps/chosen": -172.70582580566406, "logps/rejected": -199.5102081298828, "loss": 0.4179, "rewards/accuracies": 0.875, "rewards/chosen": -0.26907747983932495, "rewards/margins": 0.9893758296966553, "rewards/rejected": -1.258453369140625, "step": 1432 }, { "epoch": 1.6352362625614711, "grad_norm": 66.45448871459682, "learning_rate": 1.773486698413701e-08, "logits/chosen": -1.321483850479126, "logits/rejected": -1.2988489866256714, "logps/chosen": -278.5541076660156, "logps/rejected": -317.18048095703125, "loss": 0.4615, "rewards/accuracies": 0.6875, "rewards/chosen": -0.9866227507591248, "rewards/margins": 1.5252459049224854, "rewards/rejected": -2.511868715286255, "step": 1434 }, { "epoch": 1.6375169268049319, "grad_norm": 61.23114152941888, "learning_rate": 1.751922273604366e-08, "logits/chosen": -1.2703336477279663, "logits/rejected": -1.3587698936462402, "logps/chosen": -119.15101623535156, "logps/rejected": -158.48219299316406, "loss": 0.4167, "rewards/accuracies": 0.875, "rewards/chosen": -0.3894941508769989, "rewards/margins": 0.8275178074836731, "rewards/rejected": -1.2170119285583496, "step": 1436 }, { "epoch": 1.6397975910483928, "grad_norm": 55.95939622635403, "learning_rate": 1.7304771629231797e-08, "logits/chosen": -1.3791868686676025, "logits/rejected": -1.32236909866333, "logps/chosen": -139.13572692871094, "logps/rejected": -141.41555786132812, "loss": 0.4648, "rewards/accuracies": 0.84375, "rewards/chosen": -0.30967453122138977, "rewards/margins": 0.7233924269676208, "rewards/rejected": -1.033066987991333, "step": 1438 }, { "epoch": 1.6420782552918538, "grad_norm": 65.12174124488925, "learning_rate": 1.709151676588446e-08, "logits/chosen": -1.3115848302841187, "logits/rejected": -1.3686813116073608, "logps/chosen": -187.90061950683594, "logps/rejected": -219.3175506591797, "loss": 0.4063, "rewards/accuracies": 0.8125, "rewards/chosen": -0.2774468958377838, "rewards/margins": 1.2934999465942383, "rewards/rejected": -1.5709468126296997, "step": 1440 }, { "epoch": 1.6443589195353145, "grad_norm": 63.019089490249065, "learning_rate": 1.687946123088021e-08, "logits/chosen": -1.148698091506958, "logits/rejected": -1.2162479162216187, "logps/chosen": -128.07449340820312, "logps/rejected": -160.94265747070312, "loss": 0.3965, "rewards/accuracies": 0.78125, "rewards/chosen": -0.44137704372406006, "rewards/margins": 0.892095685005188, "rewards/rejected": -1.3334728479385376, "step": 1442 }, { "epoch": 1.6466395837787755, "grad_norm": 53.81256325707842, "learning_rate": 1.6668608091748494e-08, "logits/chosen": -1.3144904375076294, "logits/rejected": -1.3196699619293213, "logps/chosen": -147.97418212890625, "logps/rejected": -184.45281982421875, "loss": 0.4082, "rewards/accuracies": 0.78125, "rewards/chosen": -0.3717328608036041, "rewards/margins": 1.0338945388793945, "rewards/rejected": -1.4056274890899658, "step": 1444 }, { "epoch": 1.6489202480222365, "grad_norm": 58.43605410338442, "learning_rate": 1.6458960398625288e-08, "logits/chosen": -1.3495270013809204, "logits/rejected": -1.3430432081222534, "logps/chosen": -221.313232421875, "logps/rejected": -240.23922729492188, "loss": 0.4216, "rewards/accuracies": 0.875, "rewards/chosen": -0.4131534695625305, "rewards/margins": 1.1750929355621338, "rewards/rejected": -1.588246464729309, "step": 1446 }, { "epoch": 1.6512009122656974, "grad_norm": 62.53542903290901, "learning_rate": 1.6250521184208888e-08, "logits/chosen": -1.2720297574996948, "logits/rejected": -1.2884955406188965, "logps/chosen": -170.361328125, "logps/rejected": -195.9408721923828, "loss": 0.4433, "rewards/accuracies": 0.84375, "rewards/chosen": -0.25558391213417053, "rewards/margins": 0.8461555242538452, "rewards/rejected": -1.1017394065856934, "step": 1448 }, { "epoch": 1.6534815765091584, "grad_norm": 59.021319623999084, "learning_rate": 1.60432934637162e-08, "logits/chosen": -1.2962281703948975, "logits/rejected": -1.3130199909210205, "logps/chosen": -220.19346618652344, "logps/rejected": -234.11521911621094, "loss": 0.3885, "rewards/accuracies": 0.9375, "rewards/chosen": -0.5436195731163025, "rewards/margins": 0.8701571226119995, "rewards/rejected": -1.4137766361236572, "step": 1450 }, { "epoch": 1.6557622407526194, "grad_norm": 65.21677604603016, "learning_rate": 1.5837280234839012e-08, "logits/chosen": -1.258408546447754, "logits/rejected": -1.2448734045028687, "logps/chosen": -189.3031005859375, "logps/rejected": -186.64773559570312, "loss": 0.4388, "rewards/accuracies": 0.75, "rewards/chosen": -0.5159119367599487, "rewards/margins": 0.5587047338485718, "rewards/rejected": -1.0746166706085205, "step": 1452 }, { "epoch": 1.65804290499608, "grad_norm": 56.877107879066756, "learning_rate": 1.5632484477700635e-08, "logits/chosen": -1.3145238161087036, "logits/rejected": -1.3254127502441406, "logps/chosen": -225.0052490234375, "logps/rejected": -260.56292724609375, "loss": 0.4098, "rewards/accuracies": 0.84375, "rewards/chosen": -0.6009418964385986, "rewards/margins": 1.1089880466461182, "rewards/rejected": -1.7099300622940063, "step": 1454 }, { "epoch": 1.660323569239541, "grad_norm": 68.97770948668654, "learning_rate": 1.542890915481282e-08, "logits/chosen": -1.2907415628433228, "logits/rejected": -1.2642382383346558, "logps/chosen": -122.03399658203125, "logps/rejected": -136.55223083496094, "loss": 0.4316, "rewards/accuracies": 0.90625, "rewards/chosen": -0.2291896939277649, "rewards/margins": 0.7178550362586975, "rewards/rejected": -0.9470447897911072, "step": 1456 }, { "epoch": 1.6626042334830018, "grad_norm": 68.21135169118118, "learning_rate": 1.5226557211032908e-08, "logits/chosen": -1.2149584293365479, "logits/rejected": -1.2694858312606812, "logps/chosen": -163.82122802734375, "logps/rejected": -217.37213134765625, "loss": 0.4581, "rewards/accuracies": 0.78125, "rewards/chosen": -0.4557928442955017, "rewards/margins": 0.9242541193962097, "rewards/rejected": -1.3800469636917114, "step": 1458 }, { "epoch": 1.6648848977264628, "grad_norm": 58.468740016047576, "learning_rate": 1.5025431573521207e-08, "logits/chosen": -1.3966403007507324, "logits/rejected": -1.395541787147522, "logps/chosen": -136.16111755371094, "logps/rejected": -158.5441436767578, "loss": 0.4215, "rewards/accuracies": 0.875, "rewards/chosen": -0.3195006549358368, "rewards/margins": 0.7229774594306946, "rewards/rejected": -1.042478084564209, "step": 1460 }, { "epoch": 1.6671655619699237, "grad_norm": 56.594315997333034, "learning_rate": 1.4825535151698653e-08, "logits/chosen": -1.3145086765289307, "logits/rejected": -1.3318628072738647, "logps/chosen": -218.93898010253906, "logps/rejected": -239.53721618652344, "loss": 0.4041, "rewards/accuracies": 0.8125, "rewards/chosen": -0.4259910583496094, "rewards/margins": 0.9239952564239502, "rewards/rejected": -1.34998619556427, "step": 1462 }, { "epoch": 1.6694462262133847, "grad_norm": 55.89445679724181, "learning_rate": 1.4626870837204775e-08, "logits/chosen": -1.379180908203125, "logits/rejected": -1.4072362184524536, "logps/chosen": -153.11817932128906, "logps/rejected": -182.13790893554688, "loss": 0.4185, "rewards/accuracies": 0.75, "rewards/chosen": -0.38719913363456726, "rewards/margins": 0.9092380404472351, "rewards/rejected": -1.2964370250701904, "step": 1464 }, { "epoch": 1.6717268904568456, "grad_norm": 64.47758334862267, "learning_rate": 1.4429441503855722e-08, "logits/chosen": -1.3387432098388672, "logits/rejected": -1.436204433441162, "logps/chosen": -214.61618041992188, "logps/rejected": -248.8955078125, "loss": 0.4463, "rewards/accuracies": 0.84375, "rewards/chosen": -0.5446640849113464, "rewards/margins": 1.1796414852142334, "rewards/rejected": -1.7243056297302246, "step": 1466 }, { "epoch": 1.6740075547003066, "grad_norm": 59.93784188124436, "learning_rate": 1.4233250007602871e-08, "logits/chosen": -1.1467586755752563, "logits/rejected": -1.1875630617141724, "logps/chosen": -198.951416015625, "logps/rejected": -230.94252014160156, "loss": 0.4599, "rewards/accuracies": 0.8125, "rewards/chosen": -0.7742232084274292, "rewards/margins": 1.2332063913345337, "rewards/rejected": -2.007429599761963, "step": 1468 }, { "epoch": 1.6762882189437673, "grad_norm": 62.25982751081324, "learning_rate": 1.4038299186491442e-08, "logits/chosen": -1.1409118175506592, "logits/rejected": -1.2762134075164795, "logps/chosen": -144.74981689453125, "logps/rejected": -227.7673797607422, "loss": 0.4217, "rewards/accuracies": 0.6875, "rewards/chosen": -0.43610525131225586, "rewards/margins": 1.684708833694458, "rewards/rejected": -2.120814085006714, "step": 1470 }, { "epoch": 1.6785688831872283, "grad_norm": 49.66084963571848, "learning_rate": 1.3844591860619382e-08, "logits/chosen": -1.3702046871185303, "logits/rejected": -1.4003832340240479, "logps/chosen": -164.3496856689453, "logps/rejected": -176.85682678222656, "loss": 0.4076, "rewards/accuracies": 0.71875, "rewards/chosen": -0.23472319543361664, "rewards/margins": 0.8279229402542114, "rewards/rejected": -1.0626461505889893, "step": 1472 }, { "epoch": 1.680849547430689, "grad_norm": 52.98373196788465, "learning_rate": 1.3652130832096653e-08, "logits/chosen": -1.1784981489181519, "logits/rejected": -1.2945374250411987, "logps/chosen": -185.53562927246094, "logps/rejected": -221.29660034179688, "loss": 0.4167, "rewards/accuracies": 0.9375, "rewards/chosen": -0.5693493485450745, "rewards/margins": 1.0049147605895996, "rewards/rejected": -1.5742641687393188, "step": 1474 }, { "epoch": 1.68313021167415, "grad_norm": 76.34629112032515, "learning_rate": 1.3460918885004658e-08, "logits/chosen": -1.2638037204742432, "logits/rejected": -1.3342554569244385, "logps/chosen": -177.4965362548828, "logps/rejected": -229.96853637695312, "loss": 0.4293, "rewards/accuracies": 0.8125, "rewards/chosen": -0.4774022698402405, "rewards/margins": 1.2457822561264038, "rewards/rejected": -1.723184585571289, "step": 1476 }, { "epoch": 1.685410875917611, "grad_norm": 66.57532589289495, "learning_rate": 1.3270958785355979e-08, "logits/chosen": -1.2002267837524414, "logits/rejected": -1.2736998796463013, "logps/chosen": -172.23741149902344, "logps/rejected": -227.90280151367188, "loss": 0.4668, "rewards/accuracies": 0.84375, "rewards/chosen": -0.5302340984344482, "rewards/margins": 1.142404556274414, "rewards/rejected": -1.6726385354995728, "step": 1478 }, { "epoch": 1.687691540161072, "grad_norm": 69.95077709646252, "learning_rate": 1.308225328105439e-08, "logits/chosen": -1.376049518585205, "logits/rejected": -1.4532899856567383, "logps/chosen": -178.05099487304688, "logps/rejected": -243.27200317382812, "loss": 0.4221, "rewards/accuracies": 0.875, "rewards/chosen": -0.3648741543292999, "rewards/margins": 1.095149040222168, "rewards/rejected": -1.460023045539856, "step": 1480 }, { "epoch": 1.6899722044045329, "grad_norm": 64.49691609932415, "learning_rate": 1.2894805101854989e-08, "logits/chosen": -1.3072996139526367, "logits/rejected": -1.2989863157272339, "logps/chosen": -166.4363555908203, "logps/rejected": -176.7548828125, "loss": 0.4753, "rewards/accuracies": 0.78125, "rewards/chosen": -0.39158201217651367, "rewards/margins": 0.8482180833816528, "rewards/rejected": -1.239800214767456, "step": 1482 }, { "epoch": 1.6922528686479938, "grad_norm": 59.42081736231207, "learning_rate": 1.270861695932489e-08, "logits/chosen": -1.3978495597839355, "logits/rejected": -1.4046682119369507, "logps/chosen": -214.11187744140625, "logps/rejected": -258.4549865722656, "loss": 0.4494, "rewards/accuracies": 0.625, "rewards/chosen": -0.5993155241012573, "rewards/margins": 0.9518192410469055, "rewards/rejected": -1.5511348247528076, "step": 1484 }, { "epoch": 1.6945335328914546, "grad_norm": 56.004460149491955, "learning_rate": 1.2523691546803872e-08, "logits/chosen": -1.1605501174926758, "logits/rejected": -1.2564440965652466, "logps/chosen": -139.624267578125, "logps/rejected": -178.62393188476562, "loss": 0.4557, "rewards/accuracies": 0.71875, "rewards/chosen": -0.27934566140174866, "rewards/margins": 0.7967109680175781, "rewards/rejected": -1.0760565996170044, "step": 1486 }, { "epoch": 1.6968141971349155, "grad_norm": 53.58208037132539, "learning_rate": 1.234003153936548e-08, "logits/chosen": -1.1957886219024658, "logits/rejected": -1.3672311305999756, "logps/chosen": -163.6760711669922, "logps/rejected": -232.0875244140625, "loss": 0.4126, "rewards/accuracies": 0.84375, "rewards/chosen": -0.378174364566803, "rewards/margins": 1.1063618659973145, "rewards/rejected": -1.4845364093780518, "step": 1488 }, { "epoch": 1.6990948613783763, "grad_norm": 75.62123712030618, "learning_rate": 1.2157639593778268e-08, "logits/chosen": -1.249180555343628, "logits/rejected": -1.257821798324585, "logps/chosen": -143.6417236328125, "logps/rejected": -205.5145721435547, "loss": 0.4895, "rewards/accuracies": 0.78125, "rewards/chosen": -0.3994945287704468, "rewards/margins": 1.2723394632339478, "rewards/rejected": -1.671833872795105, "step": 1490 }, { "epoch": 1.7013755256218372, "grad_norm": 64.99974592800548, "learning_rate": 1.1976518348467424e-08, "logits/chosen": -1.2328625917434692, "logits/rejected": -1.2472069263458252, "logps/chosen": -199.8415069580078, "logps/rejected": -222.4342803955078, "loss": 0.4444, "rewards/accuracies": 0.84375, "rewards/chosen": -0.544438898563385, "rewards/margins": 1.3196830749511719, "rewards/rejected": -1.8641220331192017, "step": 1492 }, { "epoch": 1.7036561898652982, "grad_norm": 66.84615124590252, "learning_rate": 1.1796670423476574e-08, "logits/chosen": -1.364052414894104, "logits/rejected": -1.3560256958007812, "logps/chosen": -133.53671264648438, "logps/rejected": -152.27175903320312, "loss": 0.4364, "rewards/accuracies": 0.75, "rewards/chosen": -0.3934406042098999, "rewards/margins": 0.749383270740509, "rewards/rejected": -1.1428238153457642, "step": 1494 }, { "epoch": 1.7059368541087592, "grad_norm": 63.13360469846886, "learning_rate": 1.1618098420429879e-08, "logits/chosen": -1.2544901371002197, "logits/rejected": -1.303771734237671, "logps/chosen": -166.77374267578125, "logps/rejected": -190.30552673339844, "loss": 0.4538, "rewards/accuracies": 0.84375, "rewards/chosen": -0.4510793685913086, "rewards/margins": 0.7863295674324036, "rewards/rejected": -1.2374088764190674, "step": 1496 }, { "epoch": 1.7082175183522201, "grad_norm": 50.3590196948944, "learning_rate": 1.1440804922494441e-08, "logits/chosen": -1.352832317352295, "logits/rejected": -1.357743263244629, "logps/chosen": -193.49298095703125, "logps/rejected": -207.95184326171875, "loss": 0.3787, "rewards/accuracies": 0.875, "rewards/chosen": -0.36873170733451843, "rewards/margins": 1.0819979906082153, "rewards/rejected": -1.4507297277450562, "step": 1498 }, { "epoch": 1.710498182595681, "grad_norm": 59.01395721554923, "learning_rate": 1.1264792494342856e-08, "logits/chosen": -1.2877607345581055, "logits/rejected": -1.3114897012710571, "logps/chosen": -158.8022003173828, "logps/rejected": -197.77500915527344, "loss": 0.4406, "rewards/accuracies": 0.875, "rewards/chosen": -0.48249971866607666, "rewards/margins": 0.8965498208999634, "rewards/rejected": -1.37904953956604, "step": 1500 }, { "epoch": 1.710498182595681, "eval_logits/chosen": -1.3420703411102295, "eval_logits/rejected": -1.3242188692092896, "eval_logps/chosen": -132.36380004882812, "eval_logps/rejected": -138.97621154785156, "eval_loss": 0.5477466583251953, "eval_rewards/accuracies": 0.7200000286102295, "eval_rewards/chosen": -0.24658337235450745, "eval_rewards/margins": 0.44476309418678284, "eval_rewards/rejected": -0.6913464069366455, "eval_runtime": 21.5135, "eval_samples_per_second": 4.648, "eval_steps_per_second": 1.162, "step": 1500 }, { "epoch": 1.712778846839142, "grad_norm": 59.337175090307106, "learning_rate": 1.1090063682116202e-08, "logits/chosen": -1.2650439739227295, "logits/rejected": -1.3502650260925293, "logps/chosen": -183.25546264648438, "logps/rejected": -224.42698669433594, "loss": 0.4722, "rewards/accuracies": 0.6875, "rewards/chosen": -0.49561411142349243, "rewards/margins": 1.1255251169204712, "rewards/rejected": -1.6211390495300293, "step": 1502 }, { "epoch": 1.7150595110826028, "grad_norm": 65.4693723853712, "learning_rate": 1.091662101338714e-08, "logits/chosen": -1.198454737663269, "logits/rejected": -1.1747905015945435, "logps/chosen": -176.22344970703125, "logps/rejected": -209.0, "loss": 0.428, "rewards/accuracies": 0.875, "rewards/chosen": -0.4550539553165436, "rewards/margins": 1.0402750968933105, "rewards/rejected": -1.4953290224075317, "step": 1504 }, { "epoch": 1.7173401753260638, "grad_norm": 56.55232069175886, "learning_rate": 1.0744466997123425e-08, "logits/chosen": -1.131535291671753, "logits/rejected": -1.2650810480117798, "logps/chosen": -149.54214477539062, "logps/rejected": -207.89288330078125, "loss": 0.4043, "rewards/accuracies": 0.78125, "rewards/chosen": -0.31441810727119446, "rewards/margins": 1.342848777770996, "rewards/rejected": -1.6572668552398682, "step": 1506 }, { "epoch": 1.7196208395695245, "grad_norm": 51.8349753003821, "learning_rate": 1.0573604123651503e-08, "logits/chosen": -1.3305891752243042, "logits/rejected": -1.357151985168457, "logps/chosen": -193.94952392578125, "logps/rejected": -216.302490234375, "loss": 0.3987, "rewards/accuracies": 0.875, "rewards/chosen": -0.3934152126312256, "rewards/margins": 1.1823620796203613, "rewards/rejected": -1.575777292251587, "step": 1508 }, { "epoch": 1.7219015038129855, "grad_norm": 63.60872727771506, "learning_rate": 1.0404034864620603e-08, "logits/chosen": -1.174712061882019, "logits/rejected": -1.2022875547409058, "logps/chosen": -148.9698486328125, "logps/rejected": -179.8749542236328, "loss": 0.4382, "rewards/accuracies": 0.9375, "rewards/chosen": -0.2437065690755844, "rewards/margins": 1.0777262449264526, "rewards/rejected": -1.3214329481124878, "step": 1510 }, { "epoch": 1.7241821680564464, "grad_norm": 56.263603035763026, "learning_rate": 1.0235761672966902e-08, "logits/chosen": -1.2788983583450317, "logits/rejected": -1.2880148887634277, "logps/chosen": -158.33840942382812, "logps/rejected": -187.15737915039062, "loss": 0.397, "rewards/accuracies": 0.9375, "rewards/chosen": -0.34705379605293274, "rewards/margins": 1.0756170749664307, "rewards/rejected": -1.422670841217041, "step": 1512 }, { "epoch": 1.7264628322999074, "grad_norm": 54.16757407624544, "learning_rate": 1.0068786982878087e-08, "logits/chosen": -1.3718278408050537, "logits/rejected": -1.337512731552124, "logps/chosen": -133.3784942626953, "logps/rejected": -142.33865356445312, "loss": 0.3983, "rewards/accuracies": 0.84375, "rewards/chosen": -0.2702075242996216, "rewards/margins": 0.9396376013755798, "rewards/rejected": -1.2098451852798462, "step": 1514 }, { "epoch": 1.7287434965433683, "grad_norm": 65.43992152706379, "learning_rate": 9.903113209758097e-09, "logits/chosen": -1.174783706665039, "logits/rejected": -1.1753756999969482, "logps/chosen": -141.90370178222656, "logps/rejected": -165.7357940673828, "loss": 0.3958, "rewards/accuracies": 0.84375, "rewards/chosen": -0.39702969789505005, "rewards/margins": 1.0974403619766235, "rewards/rejected": -1.494470238685608, "step": 1516 }, { "epoch": 1.7310241607868293, "grad_norm": 53.07155795474272, "learning_rate": 9.73874275019223e-09, "logits/chosen": -1.2876778841018677, "logits/rejected": -1.2978581190109253, "logps/chosen": -172.82862854003906, "logps/rejected": -199.86740112304688, "loss": 0.4196, "rewards/accuracies": 0.84375, "rewards/chosen": -0.360828161239624, "rewards/margins": 0.7934844493865967, "rewards/rejected": -1.1543126106262207, "step": 1518 }, { "epoch": 1.73330482503029, "grad_norm": 56.60655829896917, "learning_rate": 9.575677981912456e-09, "logits/chosen": -1.3419064283370972, "logits/rejected": -1.4197977781295776, "logps/chosen": -215.0706787109375, "logps/rejected": -234.6818084716797, "loss": 0.4216, "rewards/accuracies": 0.875, "rewards/chosen": -0.36836767196655273, "rewards/margins": 0.9383161067962646, "rewards/rejected": -1.3066837787628174, "step": 1520 }, { "epoch": 1.735585489273751, "grad_norm": 60.4365358347784, "learning_rate": 9.413921263763024e-09, "logits/chosen": -1.1901061534881592, "logits/rejected": -1.1732714176177979, "logps/chosen": -148.09774780273438, "logps/rejected": -138.72329711914062, "loss": 0.4147, "rewards/accuracies": 0.84375, "rewards/chosen": -0.24259831011295319, "rewards/margins": 0.738941490650177, "rewards/rejected": -0.9815397262573242, "step": 1522 }, { "epoch": 1.7378661535172117, "grad_norm": 61.21131572090355, "learning_rate": 9.253474935666338e-09, "logits/chosen": -1.319941520690918, "logits/rejected": -1.32881498336792, "logps/chosen": -186.58230590820312, "logps/rejected": -211.57525634765625, "loss": 0.5162, "rewards/accuracies": 0.84375, "rewards/chosen": -0.4148866832256317, "rewards/margins": 1.0180983543395996, "rewards/rejected": -1.4329850673675537, "step": 1524 }, { "epoch": 1.7401468177606727, "grad_norm": 55.57275725271543, "learning_rate": 9.094341318589071e-09, "logits/chosen": -1.1223199367523193, "logits/rejected": -1.2205249071121216, "logps/chosen": -123.37957763671875, "logps/rejected": -155.87530517578125, "loss": 0.4639, "rewards/accuracies": 0.875, "rewards/chosen": -0.4156235158443451, "rewards/margins": 0.8936042785644531, "rewards/rejected": -1.309227705001831, "step": 1526 }, { "epoch": 1.7424274820041337, "grad_norm": 75.32187086371678, "learning_rate": 8.936522714508676e-09, "logits/chosen": -1.1820390224456787, "logits/rejected": -1.227403163909912, "logps/chosen": -240.14578247070312, "logps/rejected": -284.03839111328125, "loss": 0.4268, "rewards/accuracies": 0.71875, "rewards/chosen": -0.9300968647003174, "rewards/margins": 1.1027811765670776, "rewards/rejected": -2.0328779220581055, "step": 1528 }, { "epoch": 1.7447081462475946, "grad_norm": 56.43871330540775, "learning_rate": 8.780021406380011e-09, "logits/chosen": -1.1268565654754639, "logits/rejected": -1.2307937145233154, "logps/chosen": -176.69204711914062, "logps/rejected": -203.42315673828125, "loss": 0.433, "rewards/accuracies": 0.84375, "rewards/chosen": -0.7230830788612366, "rewards/margins": 1.0807398557662964, "rewards/rejected": -1.8038227558135986, "step": 1530 }, { "epoch": 1.7469888104910556, "grad_norm": 53.44401452377932, "learning_rate": 8.624839658102345e-09, "logits/chosen": -1.172848105430603, "logits/rejected": -1.2623870372772217, "logps/chosen": -172.2826385498047, "logps/rejected": -218.0788116455078, "loss": 0.4167, "rewards/accuracies": 0.9375, "rewards/chosen": -0.4317859411239624, "rewards/margins": 1.20097017288208, "rewards/rejected": -1.632756233215332, "step": 1532 }, { "epoch": 1.7492694747345165, "grad_norm": 44.672996526234876, "learning_rate": 8.470979714486626e-09, "logits/chosen": -1.2581678628921509, "logits/rejected": -1.3436583280563354, "logps/chosen": -164.71560668945312, "logps/rejected": -197.1131134033203, "loss": 0.3891, "rewards/accuracies": 0.84375, "rewards/chosen": -0.48163390159606934, "rewards/margins": 1.20646333694458, "rewards/rejected": -1.688097357749939, "step": 1534 }, { "epoch": 1.7515501389779773, "grad_norm": 59.7183121930943, "learning_rate": 8.318443801222918e-09, "logits/chosen": -1.1442248821258545, "logits/rejected": -1.2501918077468872, "logps/chosen": -155.9255828857422, "logps/rejected": -185.97073364257812, "loss": 0.4332, "rewards/accuracies": 0.875, "rewards/chosen": -0.4522356390953064, "rewards/margins": 0.7766329050064087, "rewards/rejected": -1.2288686037063599, "step": 1536 }, { "epoch": 1.7538308032214383, "grad_norm": 52.4727599794761, "learning_rate": 8.167234124848343e-09, "logits/chosen": -1.2721006870269775, "logits/rejected": -1.3032159805297852, "logps/chosen": -213.73348999023438, "logps/rejected": -231.73304748535156, "loss": 0.4188, "rewards/accuracies": 0.875, "rewards/chosen": -0.38457486033439636, "rewards/margins": 1.214958667755127, "rewards/rejected": -1.5995336771011353, "step": 1538 }, { "epoch": 1.756111467464899, "grad_norm": 56.28899819514317, "learning_rate": 8.017352872715077e-09, "logits/chosen": -1.246867299079895, "logits/rejected": -1.321312665939331, "logps/chosen": -166.28347778320312, "logps/rejected": -189.37855529785156, "loss": 0.3864, "rewards/accuracies": 0.9375, "rewards/chosen": -0.6479480266571045, "rewards/margins": 0.916532039642334, "rewards/rejected": -1.5644800662994385, "step": 1540 }, { "epoch": 1.75839213170836, "grad_norm": 65.96019718362385, "learning_rate": 7.868802212958703e-09, "logits/chosen": -1.2408630847930908, "logits/rejected": -1.3144943714141846, "logps/chosen": -148.76805114746094, "logps/rejected": -169.2152099609375, "loss": 0.4679, "rewards/accuracies": 0.8125, "rewards/chosen": -0.313798189163208, "rewards/margins": 0.8195222020149231, "rewards/rejected": -1.1333203315734863, "step": 1542 }, { "epoch": 1.760672795951821, "grad_norm": 67.11700647965242, "learning_rate": 7.721584294466899e-09, "logits/chosen": -1.3456957340240479, "logits/rejected": -1.3798415660858154, "logps/chosen": -180.2974090576172, "logps/rejected": -203.2277374267578, "loss": 0.4066, "rewards/accuracies": 0.78125, "rewards/chosen": -0.5308500528335571, "rewards/margins": 0.968846321105957, "rewards/rejected": -1.4996963739395142, "step": 1544 }, { "epoch": 1.7629534601952819, "grad_norm": 59.467637180664155, "learning_rate": 7.575701246848299e-09, "logits/chosen": -1.2550122737884521, "logits/rejected": -1.4017306566238403, "logps/chosen": -176.01060485839844, "logps/rejected": -221.2160186767578, "loss": 0.4151, "rewards/accuracies": 0.84375, "rewards/chosen": -0.6037588715553284, "rewards/margins": 1.1545339822769165, "rewards/rejected": -1.7582929134368896, "step": 1546 }, { "epoch": 1.7652341244387428, "grad_norm": 59.34463361306739, "learning_rate": 7.431155180401704e-09, "logits/chosen": -1.1449761390686035, "logits/rejected": -1.2112215757369995, "logps/chosen": -175.2537384033203, "logps/rejected": -213.36187744140625, "loss": 0.3787, "rewards/accuracies": 0.90625, "rewards/chosen": -0.547699511051178, "rewards/margins": 1.1472948789596558, "rewards/rejected": -1.694994568824768, "step": 1548 }, { "epoch": 1.7675147886822038, "grad_norm": 107.3697379891581, "learning_rate": 7.287948186085613e-09, "logits/chosen": -1.1891409158706665, "logits/rejected": -1.2180662155151367, "logps/chosen": -101.96951293945312, "logps/rejected": -131.66488647460938, "loss": 0.5419, "rewards/accuracies": 0.90625, "rewards/chosen": -0.35958331823349, "rewards/margins": 0.8678293228149414, "rewards/rejected": -1.2274125814437866, "step": 1550 }, { "epoch": 1.7697954529256648, "grad_norm": 58.46793417479867, "learning_rate": 7.146082335487824e-09, "logits/chosen": -1.2643113136291504, "logits/rejected": -1.2983956336975098, "logps/chosen": -184.83563232421875, "logps/rejected": -210.6478729248047, "loss": 0.417, "rewards/accuracies": 0.75, "rewards/chosen": -0.5568149089813232, "rewards/margins": 0.8875846862792969, "rewards/rejected": -1.4443995952606201, "step": 1552 }, { "epoch": 1.7720761171691255, "grad_norm": 68.29885233859635, "learning_rate": 7.005559680795658e-09, "logits/chosen": -1.255906105041504, "logits/rejected": -1.2899165153503418, "logps/chosen": -123.41840362548828, "logps/rejected": -138.2176055908203, "loss": 0.4641, "rewards/accuracies": 0.84375, "rewards/chosen": -0.31046992540359497, "rewards/margins": 0.6126350164413452, "rewards/rejected": -0.9231049418449402, "step": 1554 }, { "epoch": 1.7743567814125865, "grad_norm": 54.33302324585405, "learning_rate": 6.866382254766156e-09, "logits/chosen": -1.330174207687378, "logits/rejected": -1.363855242729187, "logps/chosen": -196.70680236816406, "logps/rejected": -213.6434326171875, "loss": 0.4372, "rewards/accuracies": 0.78125, "rewards/chosen": -0.5205326676368713, "rewards/margins": 1.0018987655639648, "rewards/rejected": -1.5224316120147705, "step": 1556 }, { "epoch": 1.7766374456560472, "grad_norm": 66.94592762976625, "learning_rate": 6.7285520706966914e-09, "logits/chosen": -1.3225196599960327, "logits/rejected": -1.3494971990585327, "logps/chosen": -185.50836181640625, "logps/rejected": -207.73159790039062, "loss": 0.4265, "rewards/accuracies": 0.8125, "rewards/chosen": -0.5292472839355469, "rewards/margins": 0.988534152507782, "rewards/rejected": -1.5177814960479736, "step": 1558 }, { "epoch": 1.7789181098995082, "grad_norm": 66.14902165695068, "learning_rate": 6.592071122395848e-09, "logits/chosen": -1.2869586944580078, "logits/rejected": -1.2955509424209595, "logps/chosen": -203.05213928222656, "logps/rejected": -221.87176513671875, "loss": 0.4555, "rewards/accuracies": 0.90625, "rewards/chosen": -0.4785195589065552, "rewards/margins": 0.8982763886451721, "rewards/rejected": -1.3767958879470825, "step": 1560 }, { "epoch": 1.7811987741429691, "grad_norm": 56.82906531266912, "learning_rate": 6.4569413841546124e-09, "logits/chosen": -1.3752797842025757, "logits/rejected": -1.342197060585022, "logps/chosen": -213.2560577392578, "logps/rejected": -226.26138305664062, "loss": 0.3949, "rewards/accuracies": 0.78125, "rewards/chosen": -0.6319646835327148, "rewards/margins": 0.7403106093406677, "rewards/rejected": -1.3722753524780273, "step": 1562 }, { "epoch": 1.78347943838643, "grad_norm": 57.01107575636888, "learning_rate": 6.323164810717751e-09, "logits/chosen": -1.4058406352996826, "logits/rejected": -1.4284158945083618, "logps/chosen": -164.4686279296875, "logps/rejected": -175.3655548095703, "loss": 0.4426, "rewards/accuracies": 0.84375, "rewards/chosen": -0.2919442057609558, "rewards/margins": 0.8540188074111938, "rewards/rejected": -1.1459629535675049, "step": 1564 }, { "epoch": 1.785760102629891, "grad_norm": 53.78725622915069, "learning_rate": 6.1907433372555885e-09, "logits/chosen": -1.227329969406128, "logits/rejected": -1.2803211212158203, "logps/chosen": -215.01806640625, "logps/rejected": -240.11415100097656, "loss": 0.4856, "rewards/accuracies": 0.78125, "rewards/chosen": -0.37019410729408264, "rewards/margins": 1.2054839134216309, "rewards/rejected": -1.5756779909133911, "step": 1566 }, { "epoch": 1.788040766873352, "grad_norm": 50.06277787185665, "learning_rate": 6.0596788793360055e-09, "logits/chosen": -1.2019214630126953, "logits/rejected": -1.167099952697754, "logps/chosen": -128.18609619140625, "logps/rejected": -156.6576690673828, "loss": 0.3964, "rewards/accuracies": 0.84375, "rewards/chosen": -0.24505016207695007, "rewards/margins": 0.7954214811325073, "rewards/rejected": -1.0404715538024902, "step": 1568 }, { "epoch": 1.7903214311168127, "grad_norm": 58.820202526748965, "learning_rate": 5.929973332896676e-09, "logits/chosen": -1.295718789100647, "logits/rejected": -1.3617008924484253, "logps/chosen": -177.0591583251953, "logps/rejected": -229.13153076171875, "loss": 0.3765, "rewards/accuracies": 0.84375, "rewards/chosen": -0.3693293035030365, "rewards/margins": 1.0683492422103882, "rewards/rejected": -1.437678575515747, "step": 1570 }, { "epoch": 1.7926020953602737, "grad_norm": 69.22569134675261, "learning_rate": 5.801628574217732e-09, "logits/chosen": -1.231567621231079, "logits/rejected": -1.2869716882705688, "logps/chosen": -115.6909408569336, "logps/rejected": -169.88333129882812, "loss": 0.4175, "rewards/accuracies": 0.8125, "rewards/chosen": -0.2287057340145111, "rewards/margins": 0.9424973130226135, "rewards/rejected": -1.1712028980255127, "step": 1572 }, { "epoch": 1.7948827596037344, "grad_norm": 57.45741658297443, "learning_rate": 5.674646459894539e-09, "logits/chosen": -1.3439488410949707, "logits/rejected": -1.3449054956436157, "logps/chosen": -169.16116333007812, "logps/rejected": -214.2422332763672, "loss": 0.4044, "rewards/accuracies": 0.84375, "rewards/chosen": -0.29708918929100037, "rewards/margins": 1.2709224224090576, "rewards/rejected": -1.5680116415023804, "step": 1574 }, { "epoch": 1.7971634238471954, "grad_norm": 53.78313839896124, "learning_rate": 5.549028826810886e-09, "logits/chosen": -1.3002756834030151, "logits/rejected": -1.305463433265686, "logps/chosen": -191.0225067138672, "logps/rejected": -210.54293823242188, "loss": 0.4336, "rewards/accuracies": 0.90625, "rewards/chosen": -0.6976014971733093, "rewards/margins": 1.1452404260635376, "rewards/rejected": -1.8428419828414917, "step": 1576 }, { "epoch": 1.7994440880906564, "grad_norm": 54.73230632912025, "learning_rate": 5.42477749211242e-09, "logits/chosen": -1.1779245138168335, "logits/rejected": -1.1519317626953125, "logps/chosen": -162.1126708984375, "logps/rejected": -177.6374053955078, "loss": 0.4314, "rewards/accuracies": 0.875, "rewards/chosen": -0.23180986940860748, "rewards/margins": 1.1439672708511353, "rewards/rejected": -1.375777244567871, "step": 1578 }, { "epoch": 1.8017247523341173, "grad_norm": 72.74484543771715, "learning_rate": 5.301894253180295e-09, "logits/chosen": -1.179969310760498, "logits/rejected": -1.2561529874801636, "logps/chosen": -142.0485382080078, "logps/rejected": -183.1311492919922, "loss": 0.4323, "rewards/accuracies": 0.75, "rewards/chosen": -0.3770468533039093, "rewards/margins": 0.8587576150894165, "rewards/rejected": -1.2358046770095825, "step": 1580 }, { "epoch": 1.8040054165775783, "grad_norm": 52.220335591750064, "learning_rate": 5.180380887605252e-09, "logits/chosen": -1.292273998260498, "logits/rejected": -1.3601034879684448, "logps/chosen": -198.21466064453125, "logps/rejected": -246.2793731689453, "loss": 0.3947, "rewards/accuracies": 0.90625, "rewards/chosen": -0.4539499282836914, "rewards/margins": 1.7145434617996216, "rewards/rejected": -2.1684935092926025, "step": 1582 }, { "epoch": 1.8062860808210393, "grad_norm": 53.485899864483656, "learning_rate": 5.060239153161872e-09, "logits/chosen": -1.1770296096801758, "logits/rejected": -1.2744455337524414, "logps/chosen": -188.80868530273438, "logps/rejected": -251.5846405029297, "loss": 0.4114, "rewards/accuracies": 0.84375, "rewards/chosen": -0.5241818428039551, "rewards/margins": 1.110929250717163, "rewards/rejected": -1.6351109743118286, "step": 1584 }, { "epoch": 1.8085667450645, "grad_norm": 71.54349228655005, "learning_rate": 4.941470787783131e-09, "logits/chosen": -1.2420802116394043, "logits/rejected": -1.2955926656723022, "logps/chosen": -153.46408081054688, "logps/rejected": -185.44371032714844, "loss": 0.4323, "rewards/accuracies": 0.8125, "rewards/chosen": -0.4371834695339203, "rewards/margins": 1.0911630392074585, "rewards/rejected": -1.5283464193344116, "step": 1586 }, { "epoch": 1.810847409307961, "grad_norm": 43.02051086457159, "learning_rate": 4.8240775095352515e-09, "logits/chosen": -1.2829195261001587, "logits/rejected": -1.3175885677337646, "logps/chosen": -147.54412841796875, "logps/rejected": -172.52725219726562, "loss": 0.3904, "rewards/accuracies": 0.84375, "rewards/chosen": -0.4066880941390991, "rewards/margins": 0.901046633720398, "rewards/rejected": -1.307734727859497, "step": 1588 }, { "epoch": 1.8131280735514217, "grad_norm": 56.502783078071204, "learning_rate": 4.708061016592923e-09, "logits/chosen": -1.327852725982666, "logits/rejected": -1.3697575330734253, "logps/chosen": -178.9647216796875, "logps/rejected": -206.71951293945312, "loss": 0.4688, "rewards/accuracies": 0.78125, "rewards/chosen": -0.5031104683876038, "rewards/margins": 1.0050235986709595, "rewards/rejected": -1.5081340074539185, "step": 1590 }, { "epoch": 1.8154087377948827, "grad_norm": 52.124874452064724, "learning_rate": 4.593422987214668e-09, "logits/chosen": -1.2299569845199585, "logits/rejected": -1.276241421699524, "logps/chosen": -146.35475158691406, "logps/rejected": -174.0493621826172, "loss": 0.4263, "rewards/accuracies": 0.96875, "rewards/chosen": 0.01635855622589588, "rewards/margins": 1.0491864681243896, "rewards/rejected": -1.0328278541564941, "step": 1592 }, { "epoch": 1.8176894020383436, "grad_norm": 52.42542929704426, "learning_rate": 4.480165079718568e-09, "logits/chosen": -1.2522296905517578, "logits/rejected": -1.2631944417953491, "logps/chosen": -195.42518615722656, "logps/rejected": -221.43161010742188, "loss": 0.4216, "rewards/accuracies": 0.90625, "rewards/chosen": -0.45845019817352295, "rewards/margins": 1.135830044746399, "rewards/rejected": -1.5942802429199219, "step": 1594 }, { "epoch": 1.8199700662818046, "grad_norm": 64.83240480126588, "learning_rate": 4.368288932458308e-09, "logits/chosen": -1.3056426048278809, "logits/rejected": -1.2827690839767456, "logps/chosen": -172.3594207763672, "logps/rejected": -187.4986114501953, "loss": 0.4437, "rewards/accuracies": 0.8125, "rewards/chosen": -0.4361146092414856, "rewards/margins": 0.8268535733222961, "rewards/rejected": -1.2629680633544922, "step": 1596 }, { "epoch": 1.8222507305252655, "grad_norm": 52.10599638998205, "learning_rate": 4.257796163799454e-09, "logits/chosen": -1.1653319597244263, "logits/rejected": -1.2118003368377686, "logps/chosen": -109.62616729736328, "logps/rejected": -133.0435028076172, "loss": 0.4041, "rewards/accuracies": 0.75, "rewards/chosen": -0.14802826941013336, "rewards/margins": 0.935161828994751, "rewards/rejected": -1.0831902027130127, "step": 1598 }, { "epoch": 1.8245313947687265, "grad_norm": 55.52964788045627, "learning_rate": 4.1486883720960435e-09, "logits/chosen": -1.1933330297470093, "logits/rejected": -1.228639006614685, "logps/chosen": -149.2159423828125, "logps/rejected": -200.82742309570312, "loss": 0.3988, "rewards/accuracies": 0.875, "rewards/chosen": -0.32056179642677307, "rewards/margins": 1.0430424213409424, "rewards/rejected": -1.3636044263839722, "step": 1600 }, { "epoch": 1.8245313947687265, "eval_logits/chosen": -1.3431406021118164, "eval_logits/rejected": -1.3253653049468994, "eval_logps/chosen": -132.28553771972656, "eval_logps/rejected": -139.2881317138672, "eval_loss": 0.5448750853538513, "eval_rewards/accuracies": 0.7200000286102295, "eval_rewards/chosen": -0.23875679075717926, "eval_rewards/margins": 0.48378121852874756, "eval_rewards/rejected": -0.7225379943847656, "eval_runtime": 20.9101, "eval_samples_per_second": 4.782, "eval_steps_per_second": 1.196, "step": 1600 }, { "epoch": 1.8268120590121875, "grad_norm": 52.449005735628376, "learning_rate": 4.040967135667472e-09, "logits/chosen": -1.3316993713378906, "logits/rejected": -1.4035625457763672, "logps/chosen": -134.9228057861328, "logps/rejected": -165.58567810058594, "loss": 0.408, "rewards/accuracies": 0.75, "rewards/chosen": -0.16615627706050873, "rewards/margins": 0.7779840230941772, "rewards/rejected": -0.9441402554512024, "step": 1602 }, { "epoch": 1.8290927232556482, "grad_norm": 58.59433738136844, "learning_rate": 3.9346340127756616e-09, "logits/chosen": -1.3039021492004395, "logits/rejected": -1.3089298009872437, "logps/chosen": -196.00987243652344, "logps/rejected": -218.62310791015625, "loss": 0.4161, "rewards/accuracies": 0.875, "rewards/chosen": -0.46055513620376587, "rewards/margins": 0.8772752285003662, "rewards/rejected": -1.3378304243087769, "step": 1604 }, { "epoch": 1.8313733874991092, "grad_norm": 63.586529609086426, "learning_rate": 3.829690541602504e-09, "logits/chosen": -1.173471450805664, "logits/rejected": -1.27016282081604, "logps/chosen": -162.02359008789062, "logps/rejected": -195.39901733398438, "loss": 0.3966, "rewards/accuracies": 0.84375, "rewards/chosen": -0.4807528853416443, "rewards/margins": 1.1842964887619019, "rewards/rejected": -1.665049433708191, "step": 1606 }, { "epoch": 1.83365405174257, "grad_norm": 59.32975592376087, "learning_rate": 3.726138240227628e-09, "logits/chosen": -1.255246639251709, "logits/rejected": -1.3199628591537476, "logps/chosen": -132.19796752929688, "logps/rejected": -187.007568359375, "loss": 0.433, "rewards/accuracies": 0.875, "rewards/chosen": -0.2542805075645447, "rewards/margins": 1.1002681255340576, "rewards/rejected": -1.3545485734939575, "step": 1608 }, { "epoch": 1.8359347159860309, "grad_norm": 53.62231825347675, "learning_rate": 3.623978606606426e-09, "logits/chosen": -1.165490984916687, "logits/rejected": -1.1475247144699097, "logps/chosen": -173.6094970703125, "logps/rejected": -202.46153259277344, "loss": 0.3824, "rewards/accuracies": 0.75, "rewards/chosen": -0.47014373540878296, "rewards/margins": 0.9737652540206909, "rewards/rejected": -1.4439090490341187, "step": 1610 }, { "epoch": 1.8382153802294918, "grad_norm": 54.19282245857111, "learning_rate": 3.523213118548407e-09, "logits/chosen": -1.2416139841079712, "logits/rejected": -1.278630256652832, "logps/chosen": -152.09185791015625, "logps/rejected": -186.2200164794922, "loss": 0.4421, "rewards/accuracies": 0.78125, "rewards/chosen": -0.47047847509384155, "rewards/margins": 0.9149271845817566, "rewards/rejected": -1.3854056596755981, "step": 1612 }, { "epoch": 1.8404960444729528, "grad_norm": 66.35928406172948, "learning_rate": 3.423843233695789e-09, "logits/chosen": -1.2137271165847778, "logits/rejected": -1.209067940711975, "logps/chosen": -162.6099853515625, "logps/rejected": -168.80848693847656, "loss": 0.5097, "rewards/accuracies": 0.8125, "rewards/chosen": -0.4765735864639282, "rewards/margins": 0.7680255174636841, "rewards/rejected": -1.2445989847183228, "step": 1614 }, { "epoch": 1.8427767087164137, "grad_norm": 76.5553780804754, "learning_rate": 3.3258703895024386e-09, "logits/chosen": -1.2277370691299438, "logits/rejected": -1.2694729566574097, "logps/chosen": -167.89581298828125, "logps/rejected": -190.85592651367188, "loss": 0.448, "rewards/accuracies": 0.6875, "rewards/chosen": -0.586859941482544, "rewards/margins": 0.8319023847579956, "rewards/rejected": -1.4187625646591187, "step": 1616 }, { "epoch": 1.8450573729598747, "grad_norm": 62.481075662440986, "learning_rate": 3.2292960032130557e-09, "logits/chosen": -1.3022971153259277, "logits/rejected": -1.3909885883331299, "logps/chosen": -138.24668884277344, "logps/rejected": -144.76327514648438, "loss": 0.5064, "rewards/accuracies": 0.75, "rewards/chosen": -0.29410520195961, "rewards/margins": 0.5049476623535156, "rewards/rejected": -0.7990528345108032, "step": 1618 }, { "epoch": 1.8473380372033354, "grad_norm": 49.40037975448036, "learning_rate": 3.134121471842688e-09, "logits/chosen": -1.3413515090942383, "logits/rejected": -1.3459126949310303, "logps/chosen": -168.7082977294922, "logps/rejected": -210.79649353027344, "loss": 0.3703, "rewards/accuracies": 0.8125, "rewards/chosen": -0.37254247069358826, "rewards/margins": 1.4622108936309814, "rewards/rejected": -1.8347532749176025, "step": 1620 }, { "epoch": 1.8496187014467964, "grad_norm": 64.64467682395964, "learning_rate": 3.0403481721565373e-09, "logits/chosen": -1.165281891822815, "logits/rejected": -1.2723007202148438, "logps/chosen": -136.62118530273438, "logps/rejected": -162.6814422607422, "loss": 0.4646, "rewards/accuracies": 0.78125, "rewards/chosen": -0.21550993621349335, "rewards/margins": 0.7950283288955688, "rewards/rejected": -1.0105382204055786, "step": 1622 }, { "epoch": 1.8518993656902571, "grad_norm": 52.728229173624364, "learning_rate": 2.947977460649975e-09, "logits/chosen": -1.2271634340286255, "logits/rejected": -1.2183120250701904, "logps/chosen": -120.3056640625, "logps/rejected": -152.6262664794922, "loss": 0.4505, "rewards/accuracies": 0.8125, "rewards/chosen": -0.4195891320705414, "rewards/margins": 0.9882142543792725, "rewards/rejected": -1.4078034162521362, "step": 1624 }, { "epoch": 1.854180029933718, "grad_norm": 69.89183140204983, "learning_rate": 2.8570106735290144e-09, "logits/chosen": -1.2754062414169312, "logits/rejected": -1.285915732383728, "logps/chosen": -148.9326171875, "logps/rejected": -176.26077270507812, "loss": 0.4184, "rewards/accuracies": 0.75, "rewards/chosen": -0.42058467864990234, "rewards/margins": 1.0723397731781006, "rewards/rejected": -1.492924451828003, "step": 1626 }, { "epoch": 1.856460694177179, "grad_norm": 78.54847866654863, "learning_rate": 2.7674491266909016e-09, "logits/chosen": -1.2117140293121338, "logits/rejected": -1.2298004627227783, "logps/chosen": -187.1058807373047, "logps/rejected": -215.0411376953125, "loss": 0.4243, "rewards/accuracies": 0.875, "rewards/chosen": -0.9483327865600586, "rewards/margins": 1.1213573217391968, "rewards/rejected": -2.069690227508545, "step": 1628 }, { "epoch": 1.85874135842064, "grad_norm": 55.582996135186875, "learning_rate": 2.679294115705144e-09, "logits/chosen": -1.2989763021469116, "logits/rejected": -1.425642967224121, "logps/chosen": -191.18502807617188, "logps/rejected": -235.31173706054688, "loss": 0.3977, "rewards/accuracies": 0.84375, "rewards/chosen": -0.30440258979797363, "rewards/margins": 1.0316964387893677, "rewards/rejected": -1.3360989093780518, "step": 1630 }, { "epoch": 1.861022022664101, "grad_norm": 51.611066982113556, "learning_rate": 2.5925469157947135e-09, "logits/chosen": -1.2190608978271484, "logits/rejected": -1.2017196416854858, "logps/chosen": -209.8916778564453, "logps/rejected": -251.41043090820312, "loss": 0.3837, "rewards/accuracies": 0.875, "rewards/chosen": -0.5788516402244568, "rewards/margins": 1.5164175033569336, "rewards/rejected": -2.095268964767456, "step": 1632 }, { "epoch": 1.863302686907562, "grad_norm": 61.654874361770986, "learning_rate": 2.507208781817638e-09, "logits/chosen": -1.2814161777496338, "logits/rejected": -1.4046311378479004, "logps/chosen": -190.91163635253906, "logps/rejected": -240.64810180664062, "loss": 0.4435, "rewards/accuracies": 0.84375, "rewards/chosen": -0.6178247928619385, "rewards/margins": 1.0759724378585815, "rewards/rejected": -1.6937971115112305, "step": 1634 }, { "epoch": 1.8655833511510227, "grad_norm": 65.59685553528632, "learning_rate": 2.4232809482488403e-09, "logits/chosen": -1.204872965812683, "logits/rejected": -1.2012598514556885, "logps/chosen": -180.18150329589844, "logps/rejected": -206.04872131347656, "loss": 0.4425, "rewards/accuracies": 0.75, "rewards/chosen": -0.6783896684646606, "rewards/margins": 1.0068707466125488, "rewards/rejected": -1.68526029586792, "step": 1636 }, { "epoch": 1.8678640153944837, "grad_norm": 60.54718272304763, "learning_rate": 2.340764629162284e-09, "logits/chosen": -1.1621663570404053, "logits/rejected": -1.2637563943862915, "logps/chosen": -178.81527709960938, "logps/rejected": -228.64007568359375, "loss": 0.3834, "rewards/accuracies": 0.8125, "rewards/chosen": -0.3456554114818573, "rewards/margins": 0.9540256261825562, "rewards/rejected": -1.2996809482574463, "step": 1638 }, { "epoch": 1.8701446796379444, "grad_norm": 55.533431370620015, "learning_rate": 2.2596610182133325e-09, "logits/chosen": -1.2759058475494385, "logits/rejected": -1.3463444709777832, "logps/chosen": -156.00009155273438, "logps/rejected": -192.7170867919922, "loss": 0.4157, "rewards/accuracies": 0.90625, "rewards/chosen": -0.003083046991378069, "rewards/margins": 1.3031508922576904, "rewards/rejected": -1.3062340021133423, "step": 1640 }, { "epoch": 1.8724253438814054, "grad_norm": 62.10012065390903, "learning_rate": 2.1799712886216628e-09, "logits/chosen": -1.1718287467956543, "logits/rejected": -1.1970500946044922, "logps/chosen": -146.75628662109375, "logps/rejected": -208.56558227539062, "loss": 0.3858, "rewards/accuracies": 0.875, "rewards/chosen": -0.1122959554195404, "rewards/margins": 1.5379095077514648, "rewards/rejected": -1.650205373764038, "step": 1642 }, { "epoch": 1.8747060081248663, "grad_norm": 56.04768532231553, "learning_rate": 2.1016965931541007e-09, "logits/chosen": -1.253339171409607, "logits/rejected": -1.3260321617126465, "logps/chosen": -207.69583129882812, "logps/rejected": -243.12586975097656, "loss": 0.4223, "rewards/accuracies": 0.84375, "rewards/chosen": -0.5657081604003906, "rewards/margins": 1.0796537399291992, "rewards/rejected": -1.6453620195388794, "step": 1644 }, { "epoch": 1.8769866723683273, "grad_norm": 60.14746373631622, "learning_rate": 2.02483806410807e-09, "logits/chosen": -1.114426851272583, "logits/rejected": -1.2585283517837524, "logps/chosen": -144.21237182617188, "logps/rejected": -216.4067840576172, "loss": 0.4165, "rewards/accuracies": 0.875, "rewards/chosen": -0.6168836951255798, "rewards/margins": 1.211737036705017, "rewards/rejected": -1.8286206722259521, "step": 1646 }, { "epoch": 1.8792673366117882, "grad_norm": 58.35041588601004, "learning_rate": 1.9493968132951455e-09, "logits/chosen": -1.2065017223358154, "logits/rejected": -1.2185966968536377, "logps/chosen": -144.52651977539062, "logps/rejected": -182.16015625, "loss": 0.4119, "rewards/accuracies": 0.875, "rewards/chosen": -0.5021862387657166, "rewards/margins": 1.0205962657928467, "rewards/rejected": -1.522782564163208, "step": 1648 }, { "epoch": 1.8815480008552492, "grad_norm": 61.66500286457481, "learning_rate": 1.875373932025015e-09, "logits/chosen": -1.113441824913025, "logits/rejected": -1.2277448177337646, "logps/chosen": -88.60232543945312, "logps/rejected": -123.64103698730469, "loss": 0.5025, "rewards/accuracies": 0.6875, "rewards/chosen": -0.3603760600090027, "rewards/margins": 0.6354212760925293, "rewards/rejected": -0.9957974553108215, "step": 1650 }, { "epoch": 1.8838286650987102, "grad_norm": 66.53236233989348, "learning_rate": 1.8027704910896668e-09, "logits/chosen": -1.2696802616119385, "logits/rejected": -1.293856143951416, "logps/chosen": -178.1632843017578, "logps/rejected": -208.1569061279297, "loss": 0.4141, "rewards/accuracies": 0.84375, "rewards/chosen": -0.4311864674091339, "rewards/margins": 1.18040931224823, "rewards/rejected": -1.611595869064331, "step": 1652 }, { "epoch": 1.886109329342171, "grad_norm": 73.576105680566, "learning_rate": 1.731587540747903e-09, "logits/chosen": -1.3243728876113892, "logits/rejected": -1.3624733686447144, "logps/chosen": -178.9629669189453, "logps/rejected": -203.73611450195312, "loss": 0.4112, "rewards/accuracies": 0.84375, "rewards/chosen": -0.35923391580581665, "rewards/margins": 1.0950078964233398, "rewards/rejected": -1.4542417526245117, "step": 1654 }, { "epoch": 1.8883899935856319, "grad_norm": 60.917071878292, "learning_rate": 1.6618261107101628e-09, "logits/chosen": -1.2426798343658447, "logits/rejected": -1.299846887588501, "logps/chosen": -166.9674530029297, "logps/rejected": -200.53256225585938, "loss": 0.4225, "rewards/accuracies": 0.875, "rewards/chosen": -0.36187708377838135, "rewards/margins": 1.0805463790893555, "rewards/rejected": -1.4424233436584473, "step": 1656 }, { "epoch": 1.8906706578290926, "grad_norm": 58.315768026639354, "learning_rate": 1.5934872101235785e-09, "logits/chosen": -1.1988379955291748, "logits/rejected": -1.276864767074585, "logps/chosen": -117.3199691772461, "logps/rejected": -166.17037963867188, "loss": 0.4669, "rewards/accuracies": 0.6875, "rewards/chosen": -0.37138426303863525, "rewards/margins": 0.7784909009933472, "rewards/rejected": -1.1498751640319824, "step": 1658 }, { "epoch": 1.8929513220725536, "grad_norm": 73.00783189642374, "learning_rate": 1.5265718275574656e-09, "logits/chosen": -1.199881672859192, "logits/rejected": -1.294306755065918, "logps/chosen": -163.18499755859375, "logps/rejected": -240.10389709472656, "loss": 0.4091, "rewards/accuracies": 0.90625, "rewards/chosen": -0.5089311599731445, "rewards/margins": 1.5384039878845215, "rewards/rejected": -2.047335147857666, "step": 1660 }, { "epoch": 1.8952319863160145, "grad_norm": 59.87965880073438, "learning_rate": 1.4610809309889338e-09, "logits/chosen": -1.1384081840515137, "logits/rejected": -1.2086718082427979, "logps/chosen": -228.3030548095703, "logps/rejected": -291.9572448730469, "loss": 0.3708, "rewards/accuracies": 0.875, "rewards/chosen": -0.6747040748596191, "rewards/margins": 1.962480068206787, "rewards/rejected": -2.6371843814849854, "step": 1662 }, { "epoch": 1.8975126505594755, "grad_norm": 65.04049810626964, "learning_rate": 1.3970154677889312e-09, "logits/chosen": -1.2462139129638672, "logits/rejected": -1.3531625270843506, "logps/chosen": -166.90621948242188, "logps/rejected": -219.62258911132812, "loss": 0.4477, "rewards/accuracies": 0.78125, "rewards/chosen": -0.6883362531661987, "rewards/margins": 1.3010790348052979, "rewards/rejected": -1.989415168762207, "step": 1664 }, { "epoch": 1.8997933148029365, "grad_norm": 67.17708620355747, "learning_rate": 1.3343763647085337e-09, "logits/chosen": -1.149139404296875, "logits/rejected": -1.2173185348510742, "logps/chosen": -155.94863891601562, "logps/rejected": -230.28384399414062, "loss": 0.4712, "rewards/accuracies": 0.84375, "rewards/chosen": -0.5960929989814758, "rewards/margins": 1.4794337749481201, "rewards/rejected": -2.075526714324951, "step": 1666 }, { "epoch": 1.9020739790463974, "grad_norm": 56.97928801975319, "learning_rate": 1.2731645278655445e-09, "logits/chosen": -1.2667725086212158, "logits/rejected": -1.3436360359191895, "logps/chosen": -160.5927734375, "logps/rejected": -186.63018798828125, "loss": 0.4321, "rewards/accuracies": 0.8125, "rewards/chosen": -0.48528626561164856, "rewards/margins": 0.931164562702179, "rewards/rejected": -1.4164507389068604, "step": 1668 }, { "epoch": 1.9043546432898582, "grad_norm": 63.8162468661416, "learning_rate": 1.2133808427313485e-09, "logits/chosen": -1.2804149389266968, "logits/rejected": -1.3368322849273682, "logps/chosen": -131.45181274414062, "logps/rejected": -180.07037353515625, "loss": 0.4551, "rewards/accuracies": 0.65625, "rewards/chosen": -0.5746269226074219, "rewards/margins": 0.8660170435905457, "rewards/rejected": -1.4406440258026123, "step": 1670 }, { "epoch": 1.9066353075333191, "grad_norm": 52.46121205171607, "learning_rate": 1.1550261741181565e-09, "logits/chosen": -1.2648401260375977, "logits/rejected": -1.239923119544983, "logps/chosen": -163.7249298095703, "logps/rejected": -190.42611694335938, "loss": 0.3885, "rewards/accuracies": 0.84375, "rewards/chosen": -0.3933155834674835, "rewards/margins": 1.0268510580062866, "rewards/rejected": -1.4201666116714478, "step": 1672 }, { "epoch": 1.9089159717767799, "grad_norm": 64.22449427271324, "learning_rate": 1.0981013661664706e-09, "logits/chosen": -1.293068289756775, "logits/rejected": -1.4030743837356567, "logps/chosen": -174.76348876953125, "logps/rejected": -219.83175659179688, "loss": 0.4548, "rewards/accuracies": 0.75, "rewards/chosen": -0.3287160098552704, "rewards/margins": 0.9871301651000977, "rewards/rejected": -1.3158462047576904, "step": 1674 }, { "epoch": 1.9111966360202408, "grad_norm": 55.77569297347896, "learning_rate": 1.042607242332838e-09, "logits/chosen": -1.2760488986968994, "logits/rejected": -1.31367027759552, "logps/chosen": -148.86257934570312, "logps/rejected": -173.52359008789062, "loss": 0.4349, "rewards/accuracies": 0.78125, "rewards/chosen": -0.24598023295402527, "rewards/margins": 0.7863931655883789, "rewards/rejected": -1.0323734283447266, "step": 1676 }, { "epoch": 1.9134773002637018, "grad_norm": 57.697067279665426, "learning_rate": 9.885446053780277e-10, "logits/chosen": -1.3217543363571167, "logits/rejected": -1.3588594198226929, "logps/chosen": -234.18218994140625, "logps/rejected": -244.6886444091797, "loss": 0.4099, "rewards/accuracies": 0.78125, "rewards/chosen": -0.6677453517913818, "rewards/margins": 0.864335834980011, "rewards/rejected": -1.5320810079574585, "step": 1678 }, { "epoch": 1.9157579645071627, "grad_norm": 52.93198869808111, "learning_rate": 9.359142373553286e-10, "logits/chosen": -1.2812589406967163, "logits/rejected": -1.374354362487793, "logps/chosen": -181.33377075195312, "logps/rejected": -226.33584594726562, "loss": 0.3969, "rewards/accuracies": 0.84375, "rewards/chosen": -0.40316393971443176, "rewards/margins": 1.0113019943237305, "rewards/rejected": -1.4144660234451294, "step": 1680 }, { "epoch": 1.9180386287506237, "grad_norm": 56.81221018582624, "learning_rate": 8.847168995992915e-10, "logits/chosen": -1.3222296237945557, "logits/rejected": -1.3470890522003174, "logps/chosen": -144.92401123046875, "logps/rejected": -160.66360473632812, "loss": 0.4557, "rewards/accuracies": 0.8125, "rewards/chosen": -0.2944653630256653, "rewards/margins": 0.691783607006073, "rewards/rejected": -0.9862489104270935, "step": 1682 }, { "epoch": 1.9203192929940847, "grad_norm": 59.96217669380799, "learning_rate": 8.349533327146719e-10, "logits/chosen": -1.1826375722885132, "logits/rejected": -1.2797571420669556, "logps/chosen": -173.0347442626953, "logps/rejected": -217.1268768310547, "loss": 0.421, "rewards/accuracies": 0.875, "rewards/chosen": -0.6681329011917114, "rewards/margins": 1.3352155685424805, "rewards/rejected": -2.0033483505249023, "step": 1684 }, { "epoch": 1.9225999572375454, "grad_norm": 53.71192157200429, "learning_rate": 7.866242565657599e-10, "logits/chosen": -1.133514642715454, "logits/rejected": -1.2035727500915527, "logps/chosen": -123.30015563964844, "logps/rejected": -165.7233123779297, "loss": 0.4164, "rewards/accuracies": 0.90625, "rewards/chosen": -0.38910388946533203, "rewards/margins": 1.0756738185882568, "rewards/rejected": -1.4647778272628784, "step": 1686 }, { "epoch": 1.9248806214810064, "grad_norm": 84.94214636470907, "learning_rate": 7.397303702659674e-10, "logits/chosen": -1.2931269407272339, "logits/rejected": -1.3728893995285034, "logps/chosen": -153.67669677734375, "logps/rejected": -198.1190185546875, "loss": 0.4512, "rewards/accuracies": 0.78125, "rewards/chosen": -0.3215530216693878, "rewards/margins": 1.3663368225097656, "rewards/rejected": -1.6878899335861206, "step": 1688 }, { "epoch": 1.927161285724467, "grad_norm": 60.620511783810954, "learning_rate": 6.942723521676464e-10, "logits/chosen": -1.2509236335754395, "logits/rejected": -1.3288919925689697, "logps/chosen": -171.9304962158203, "logps/rejected": -200.7103729248047, "loss": 0.406, "rewards/accuracies": 0.90625, "rewards/chosen": -0.2758704423904419, "rewards/margins": 0.8382859230041504, "rewards/rejected": -1.1141563653945923, "step": 1690 }, { "epoch": 1.929441949967928, "grad_norm": 61.96268516778484, "learning_rate": 6.502508598523748e-10, "logits/chosen": -1.1139042377471924, "logits/rejected": -1.113561987876892, "logps/chosen": -145.94715881347656, "logps/rejected": -177.67091369628906, "loss": 0.4153, "rewards/accuracies": 0.84375, "rewards/chosen": -0.3293991684913635, "rewards/margins": 0.9399027824401855, "rewards/rejected": -1.2693020105361938, "step": 1692 }, { "epoch": 1.931722614211389, "grad_norm": 64.38956995843415, "learning_rate": 6.076665301213646e-10, "logits/chosen": -1.2710590362548828, "logits/rejected": -1.3398025035858154, "logps/chosen": -258.7908935546875, "logps/rejected": -270.1191711425781, "loss": 0.4663, "rewards/accuracies": 0.75, "rewards/chosen": -1.183610200881958, "rewards/margins": 1.2962732315063477, "rewards/rejected": -2.4798836708068848, "step": 1694 }, { "epoch": 1.93400327845485, "grad_norm": 70.46099460182324, "learning_rate": 5.665199789862907e-10, "logits/chosen": -1.3359112739562988, "logits/rejected": -1.3275290727615356, "logps/chosen": -183.69839477539062, "logps/rejected": -183.45762634277344, "loss": 0.5587, "rewards/accuracies": 0.6875, "rewards/chosen": -0.5863617658615112, "rewards/margins": 0.39982593059539795, "rewards/rejected": -0.9861876368522644, "step": 1696 }, { "epoch": 1.936283942698311, "grad_norm": 59.14480772543764, "learning_rate": 5.268118016603651e-10, "logits/chosen": -1.2939709424972534, "logits/rejected": -1.2879596948623657, "logps/chosen": -242.49627685546875, "logps/rejected": -255.47325134277344, "loss": 0.5263, "rewards/accuracies": 0.875, "rewards/chosen": -0.6082537174224854, "rewards/margins": 0.8644734621047974, "rewards/rejected": -1.4727270603179932, "step": 1698 }, { "epoch": 1.938564606941772, "grad_norm": 65.14033380217037, "learning_rate": 4.88542572549755e-10, "logits/chosen": -1.258131742477417, "logits/rejected": -1.341509461402893, "logps/chosen": -280.3218688964844, "logps/rejected": -349.66766357421875, "loss": 0.4044, "rewards/accuracies": 0.875, "rewards/chosen": -0.9879501461982727, "rewards/margins": 1.8015985488891602, "rewards/rejected": -2.789548873901367, "step": 1700 }, { "epoch": 1.938564606941772, "eval_logits/chosen": -1.3368662595748901, "eval_logits/rejected": -1.3193824291229248, "eval_logps/chosen": -132.35743713378906, "eval_logps/rejected": -139.43797302246094, "eval_loss": 0.5376756191253662, "eval_rewards/accuracies": 0.7200000286102295, "eval_rewards/chosen": -0.2459474503993988, "eval_rewards/margins": 0.49157509207725525, "eval_rewards/rejected": -0.7375224828720093, "eval_runtime": 20.9786, "eval_samples_per_second": 4.767, "eval_steps_per_second": 1.192, "step": 1700 }, { "epoch": 1.9408452711852326, "grad_norm": 67.5831820584933, "learning_rate": 4.5171284524521127e-10, "logits/chosen": -1.1273996829986572, "logits/rejected": -1.200268030166626, "logps/chosen": -141.1074981689453, "logps/rejected": -161.90756225585938, "loss": 0.3986, "rewards/accuracies": 0.875, "rewards/chosen": -0.13401609659194946, "rewards/margins": 0.9844987392425537, "rewards/rejected": -1.118514895439148, "step": 1702 }, { "epoch": 1.9431259354286936, "grad_norm": 64.49222478839803, "learning_rate": 4.163231525141309e-10, "logits/chosen": -1.4064816236495972, "logits/rejected": -1.4361658096313477, "logps/chosen": -240.72976684570312, "logps/rejected": -262.8630676269531, "loss": 0.4409, "rewards/accuracies": 0.9375, "rewards/chosen": -0.6487561464309692, "rewards/margins": 1.2385480403900146, "rewards/rejected": -1.8873043060302734, "step": 1704 }, { "epoch": 1.9454065996721546, "grad_norm": 50.83672179405874, "learning_rate": 3.8237400629280714e-10, "logits/chosen": -1.0193315744400024, "logits/rejected": -0.9887692928314209, "logps/chosen": -120.17151641845703, "logps/rejected": -136.94674682617188, "loss": 0.4543, "rewards/accuracies": 0.78125, "rewards/chosen": -0.27739959955215454, "rewards/margins": 0.6725433468818665, "rewards/rejected": -0.949942946434021, "step": 1706 }, { "epoch": 1.9476872639156153, "grad_norm": 53.5967901694209, "learning_rate": 3.4986589767902476e-10, "logits/chosen": -1.2411226034164429, "logits/rejected": -1.3345215320587158, "logps/chosen": -90.0027084350586, "logps/rejected": -116.62909698486328, "loss": 0.4495, "rewards/accuracies": 0.78125, "rewards/chosen": -0.078713558614254, "rewards/margins": 0.6798267364501953, "rewards/rejected": -0.7585403323173523, "step": 1708 }, { "epoch": 1.9499679281590763, "grad_norm": 59.266256355757655, "learning_rate": 3.187992969249875e-10, "logits/chosen": -1.2389315366744995, "logits/rejected": -1.2326477766036987, "logps/chosen": -127.89494323730469, "logps/rejected": -154.4901580810547, "loss": 0.4366, "rewards/accuracies": 0.78125, "rewards/chosen": -0.4514698088169098, "rewards/margins": 0.6477700471878052, "rewards/rejected": -1.0992399454116821, "step": 1710 }, { "epoch": 1.9522485924025372, "grad_norm": 62.77597760270175, "learning_rate": 2.8917465343047954e-10, "logits/chosen": -1.3058414459228516, "logits/rejected": -1.374163269996643, "logps/chosen": -171.95448303222656, "logps/rejected": -207.97914123535156, "loss": 0.3709, "rewards/accuracies": 0.84375, "rewards/chosen": -0.32011568546295166, "rewards/margins": 1.2897684574127197, "rewards/rejected": -1.6098840236663818, "step": 1712 }, { "epoch": 1.9545292566459982, "grad_norm": 56.41697287299428, "learning_rate": 2.609923957363702e-10, "logits/chosen": -1.2662739753723145, "logits/rejected": -1.3142364025115967, "logps/chosen": -149.61746215820312, "logps/rejected": -184.44931030273438, "loss": 0.4197, "rewards/accuracies": 0.84375, "rewards/chosen": -0.28848132491111755, "rewards/margins": 1.1091067790985107, "rewards/rejected": -1.3975881338119507, "step": 1714 }, { "epoch": 1.9568099208894592, "grad_norm": 58.90785637715505, "learning_rate": 2.3425293151845273e-10, "logits/chosen": -1.2464115619659424, "logits/rejected": -1.2890043258666992, "logps/chosen": -146.8938446044922, "logps/rejected": -162.07830810546875, "loss": 0.4128, "rewards/accuracies": 0.8125, "rewards/chosen": -0.2892056405544281, "rewards/margins": 0.7288376092910767, "rewards/rejected": -1.0180431604385376, "step": 1716 }, { "epoch": 1.9590905851329201, "grad_norm": 66.05988715134325, "learning_rate": 2.0895664758151521e-10, "logits/chosen": -1.1935899257659912, "logits/rejected": -1.254826545715332, "logps/chosen": -194.64492797851562, "logps/rejected": -225.4056854248047, "loss": 0.4007, "rewards/accuracies": 0.8125, "rewards/chosen": -0.525715172290802, "rewards/margins": 1.1838706731796265, "rewards/rejected": -1.7095859050750732, "step": 1718 }, { "epoch": 1.9613712493763809, "grad_norm": 64.18176403648182, "learning_rate": 1.8510390985371216e-10, "logits/chosen": -1.1969249248504639, "logits/rejected": -1.2043269872665405, "logps/chosen": -201.9025115966797, "logps/rejected": -233.6451873779297, "loss": 0.4386, "rewards/accuracies": 0.84375, "rewards/chosen": -0.5428147315979004, "rewards/margins": 0.8293758630752563, "rewards/rejected": -1.3721905946731567, "step": 1720 }, { "epoch": 1.9636519136198418, "grad_norm": 58.86675845104934, "learning_rate": 1.626950633813351e-10, "logits/chosen": -1.2230490446090698, "logits/rejected": -1.2853928804397583, "logps/chosen": -160.97621154785156, "logps/rejected": -223.6080780029297, "loss": 0.4377, "rewards/accuracies": 0.8125, "rewards/chosen": -0.3992640972137451, "rewards/margins": 1.2427630424499512, "rewards/rejected": -1.6420272588729858, "step": 1722 }, { "epoch": 1.9659325778633026, "grad_norm": 63.84847943868757, "learning_rate": 1.4173043232380554e-10, "logits/chosen": -1.1713950634002686, "logits/rejected": -1.2190814018249512, "logps/chosen": -160.047119140625, "logps/rejected": -184.2767791748047, "loss": 0.4372, "rewards/accuracies": 0.84375, "rewards/chosen": -0.4601638913154602, "rewards/margins": 0.9584896564483643, "rewards/rejected": -1.4186536073684692, "step": 1724 }, { "epoch": 1.9682132421067635, "grad_norm": 64.3589769132475, "learning_rate": 1.222103199489455e-10, "logits/chosen": -1.3236342668533325, "logits/rejected": -1.3482894897460938, "logps/chosen": -182.65269470214844, "logps/rejected": -217.88861083984375, "loss": 0.4358, "rewards/accuracies": 0.875, "rewards/chosen": -0.43886712193489075, "rewards/margins": 1.474854588508606, "rewards/rejected": -1.9137215614318848, "step": 1726 }, { "epoch": 1.9704939063502245, "grad_norm": 71.97812711207425, "learning_rate": 1.0413500862864743e-10, "logits/chosen": -1.377000093460083, "logits/rejected": -1.4007121324539185, "logps/chosen": -237.48316955566406, "logps/rejected": -259.9788818359375, "loss": 0.4143, "rewards/accuracies": 0.90625, "rewards/chosen": -0.6071898937225342, "rewards/margins": 1.190570592880249, "rewards/rejected": -1.7977604866027832, "step": 1728 }, { "epoch": 1.9727745705936854, "grad_norm": 61.7108383512953, "learning_rate": 8.750475983472227e-11, "logits/chosen": -1.2676740884780884, "logits/rejected": -1.2757924795150757, "logps/chosen": -190.92140197753906, "logps/rejected": -197.59115600585938, "loss": 0.4226, "rewards/accuracies": 0.90625, "rewards/chosen": -0.5921288132667542, "rewards/margins": 1.0576057434082031, "rewards/rejected": -1.6497344970703125, "step": 1730 }, { "epoch": 1.9750552348371464, "grad_norm": 188.64844039216806, "learning_rate": 7.231981413520217e-11, "logits/chosen": -1.263267993927002, "logits/rejected": -1.2889572381973267, "logps/chosen": -157.240478515625, "logps/rejected": -197.0875701904297, "loss": 0.4582, "rewards/accuracies": 0.875, "rewards/chosen": -0.24919547140598297, "rewards/margins": 0.7916581630706787, "rewards/rejected": -1.0408536195755005, "step": 1732 }, { "epoch": 1.9773358990806074, "grad_norm": 65.4504778217422, "learning_rate": 5.858039119077673e-11, "logits/chosen": -1.2087372541427612, "logits/rejected": -1.2781308889389038, "logps/chosen": -176.43057250976562, "logps/rejected": -216.8165740966797, "loss": 0.4563, "rewards/accuracies": 0.84375, "rewards/chosen": -0.41199278831481934, "rewards/margins": 0.922531008720398, "rewards/rejected": -1.3345237970352173, "step": 1734 }, { "epoch": 1.979616563324068, "grad_norm": 61.72345732533114, "learning_rate": 4.628668975166228e-11, "logits/chosen": -1.4069479703903198, "logits/rejected": -1.425885558128357, "logps/chosen": -155.35098266601562, "logps/rejected": -172.41546630859375, "loss": 0.3891, "rewards/accuracies": 0.84375, "rewards/chosen": -0.3574288785457611, "rewards/margins": 1.1354269981384277, "rewards/rejected": -1.4928559064865112, "step": 1736 }, { "epoch": 1.981897227567529, "grad_norm": 50.29362647543594, "learning_rate": 3.5438887654737346e-11, "logits/chosen": -1.1932240724563599, "logits/rejected": -1.2252675294876099, "logps/chosen": -104.49500274658203, "logps/rejected": -128.2619171142578, "loss": 0.3998, "rewards/accuracies": 0.78125, "rewards/chosen": -0.0990162342786789, "rewards/margins": 0.8109432458877563, "rewards/rejected": -0.9099595546722412, "step": 1738 }, { "epoch": 1.9841778918109898, "grad_norm": 63.67796104250234, "learning_rate": 2.603714182093375e-11, "logits/chosen": -1.2760851383209229, "logits/rejected": -1.3577792644500732, "logps/chosen": -172.22714233398438, "logps/rejected": -215.27749633789062, "loss": 0.4565, "rewards/accuracies": 0.84375, "rewards/chosen": -0.560702919960022, "rewards/margins": 1.0820647478103638, "rewards/rejected": -1.6427676677703857, "step": 1740 }, { "epoch": 1.9864585560544508, "grad_norm": 58.508228872135426, "learning_rate": 1.808158825297168e-11, "logits/chosen": -1.3208973407745361, "logits/rejected": -1.3905422687530518, "logps/chosen": -147.84127807617188, "logps/rejected": -178.2074737548828, "loss": 0.4758, "rewards/accuracies": 0.6875, "rewards/chosen": -0.48498719930648804, "rewards/margins": 0.888887882232666, "rewards/rejected": -1.3738751411437988, "step": 1742 }, { "epoch": 1.9887392202979117, "grad_norm": 58.05411220405763, "learning_rate": 1.1572342033416838e-11, "logits/chosen": -1.3570505380630493, "logits/rejected": -1.3865540027618408, "logps/chosen": -171.88287353515625, "logps/rejected": -210.6277618408203, "loss": 0.4492, "rewards/accuracies": 0.90625, "rewards/chosen": -0.28449732065200806, "rewards/margins": 1.1635254621505737, "rewards/rejected": -1.448022723197937, "step": 1744 }, { "epoch": 1.9910198845413727, "grad_norm": 60.292954042523704, "learning_rate": 6.50949732301509e-12, "logits/chosen": -1.2242615222930908, "logits/rejected": -1.3328139781951904, "logps/chosen": -164.70863342285156, "logps/rejected": -213.20814514160156, "loss": 0.3912, "rewards/accuracies": 0.78125, "rewards/chosen": -0.497256338596344, "rewards/margins": 1.1804652214050293, "rewards/rejected": -1.677721619606018, "step": 1746 }, { "epoch": 1.9933005487848336, "grad_norm": 68.33201893652561, "learning_rate": 2.893127359282488e-12, "logits/chosen": -1.2822688817977905, "logits/rejected": -1.3241004943847656, "logps/chosen": -207.72021484375, "logps/rejected": -262.7379455566406, "loss": 0.3881, "rewards/accuracies": 0.875, "rewards/chosen": -0.34897035360336304, "rewards/margins": 1.5757369995117188, "rewards/rejected": -1.924707293510437, "step": 1748 }, { "epoch": 1.9955812130282946, "grad_norm": 62.482127853658426, "learning_rate": 7.232844555282725e-13, "logits/chosen": -1.3659021854400635, "logits/rejected": -1.3771145343780518, "logps/chosen": -201.26531982421875, "logps/rejected": -219.556640625, "loss": 0.3859, "rewards/accuracies": 0.875, "rewards/chosen": -0.7029599547386169, "rewards/margins": 1.120865821838379, "rewards/rejected": -1.823825716972351, "step": 1750 }, { "epoch": 1.9978618772717553, "grad_norm": 67.65957613657208, "learning_rate": 0.0, "logits/chosen": -1.171149492263794, "logits/rejected": -1.2374571561813354, "logps/chosen": -129.96214294433594, "logps/rejected": -161.25244140625, "loss": 0.4188, "rewards/accuracies": 0.90625, "rewards/chosen": -0.34549519419670105, "rewards/margins": 1.0673820972442627, "rewards/rejected": -1.4128773212432861, "step": 1752 } ], "logging_steps": 2, "max_steps": 1752, "num_input_tokens_seen": 0, "num_train_epochs": 2, "save_steps": 300, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 0.0, "train_batch_size": 1, "trial_name": null, "trial_params": null }