{ "best_metric": null, "best_model_checkpoint": null, "epoch": 0.9989071038251366, "eval_steps": 400, "global_step": 457, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.01092896174863388, "grad_norm": 353.8300178137516, "learning_rate": 1.0869565217391303e-07, "logits/chosen": -1.0153186321258545, "logits/rejected": -1.0100535154342651, "logps/chosen": -0.28067928552627563, "logps/rejected": -0.2859733998775482, "loss": 5.0876, "rewards/accuracies": 0.53125, "rewards/chosen": -2.806792974472046, "rewards/margins": 0.05294132977724075, "rewards/rejected": -2.859734296798706, "step": 5 }, { "epoch": 0.02185792349726776, "grad_norm": 225.98953916688063, "learning_rate": 2.1739130434782607e-07, "logits/chosen": -1.058803677558899, "logits/rejected": -1.0075746774673462, "logps/chosen": -0.25702008605003357, "logps/rejected": -0.27079910039901733, "loss": 5.0481, "rewards/accuracies": 0.543749988079071, "rewards/chosen": -2.5702006816864014, "rewards/margins": 0.13779030740261078, "rewards/rejected": -2.707991123199463, "step": 10 }, { "epoch": 0.03278688524590164, "grad_norm": 256.50567074528465, "learning_rate": 3.260869565217391e-07, "logits/chosen": -0.9965429306030273, "logits/rejected": -0.9519298672676086, "logps/chosen": -0.2673397660255432, "logps/rejected": -0.27320751547813416, "loss": 4.9705, "rewards/accuracies": 0.512499988079071, "rewards/chosen": -2.6733975410461426, "rewards/margins": 0.05867765098810196, "rewards/rejected": -2.7320752143859863, "step": 15 }, { "epoch": 0.04371584699453552, "grad_norm": 346.54496281760004, "learning_rate": 4.3478260869565214e-07, "logits/chosen": -0.9495851397514343, "logits/rejected": -0.8978370428085327, "logps/chosen": -0.2726512551307678, "logps/rejected": -0.28480401635169983, "loss": 5.0659, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -2.7265126705169678, "rewards/margins": 0.12152715772390366, "rewards/rejected": -2.8480401039123535, "step": 20 }, { "epoch": 0.0546448087431694, "grad_norm": 308.4996786317866, "learning_rate": 5.434782608695652e-07, "logits/chosen": -0.9501255750656128, "logits/rejected": -0.8753129839897156, "logps/chosen": -0.27574610710144043, "logps/rejected": -0.29435569047927856, "loss": 4.9704, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -2.7574610710144043, "rewards/margins": 0.18609575927257538, "rewards/rejected": -2.943556547164917, "step": 25 }, { "epoch": 0.06557377049180328, "grad_norm": 310.0691471680414, "learning_rate": 6.521739130434782e-07, "logits/chosen": -1.0519192218780518, "logits/rejected": -0.986653208732605, "logps/chosen": -0.26742228865623474, "logps/rejected": -0.28410372138023376, "loss": 5.0602, "rewards/accuracies": 0.5062500238418579, "rewards/chosen": -2.6742231845855713, "rewards/margins": 0.1668141633272171, "rewards/rejected": -2.8410372734069824, "step": 30 }, { "epoch": 0.07650273224043716, "grad_norm": 189.7575115104871, "learning_rate": 7.608695652173913e-07, "logits/chosen": -0.9969251751899719, "logits/rejected": -0.9315601587295532, "logps/chosen": -0.25781017541885376, "logps/rejected": -0.2797554135322571, "loss": 4.9378, "rewards/accuracies": 0.48124998807907104, "rewards/chosen": -2.578101634979248, "rewards/margins": 0.21945223212242126, "rewards/rejected": -2.797553777694702, "step": 35 }, { "epoch": 0.08743169398907104, "grad_norm": 225.69610863472298, "learning_rate": 8.695652173913043e-07, "logits/chosen": -0.9572874307632446, "logits/rejected": -0.8963949084281921, "logps/chosen": -0.28224068880081177, "logps/rejected": -0.2995293140411377, "loss": 5.1508, "rewards/accuracies": 0.5249999761581421, "rewards/chosen": -2.82240629196167, "rewards/margins": 0.17288652062416077, "rewards/rejected": -2.995293140411377, "step": 40 }, { "epoch": 0.09836065573770492, "grad_norm": 131.95682988894754, "learning_rate": 9.782608695652173e-07, "logits/chosen": -1.0119043588638306, "logits/rejected": -0.9314481019973755, "logps/chosen": -0.2863091230392456, "logps/rejected": -0.3087444305419922, "loss": 4.7676, "rewards/accuracies": 0.5, "rewards/chosen": -2.863091230392456, "rewards/margins": 0.2243528664112091, "rewards/rejected": -3.0874440670013428, "step": 45 }, { "epoch": 0.1092896174863388, "grad_norm": 345.07474799660855, "learning_rate": 9.997663088532014e-07, "logits/chosen": -0.9598929286003113, "logits/rejected": -0.8789777755737305, "logps/chosen": -0.2817566692829132, "logps/rejected": -0.2901131510734558, "loss": 4.9269, "rewards/accuracies": 0.48124998807907104, "rewards/chosen": -2.817566394805908, "rewards/margins": 0.08356456458568573, "rewards/rejected": -2.9011311531066895, "step": 50 }, { "epoch": 0.12021857923497267, "grad_norm": 162.76427513659027, "learning_rate": 9.98817312944725e-07, "logits/chosen": -0.9862833023071289, "logits/rejected": -0.8716124296188354, "logps/chosen": -0.28066322207450867, "logps/rejected": -0.31615525484085083, "loss": 4.707, "rewards/accuracies": 0.5562499761581421, "rewards/chosen": -2.8066320419311523, "rewards/margins": 0.35492032766342163, "rewards/rejected": -3.161552667617798, "step": 55 }, { "epoch": 0.13114754098360656, "grad_norm": 119.0877915995248, "learning_rate": 9.971397915250336e-07, "logits/chosen": -1.0102354288101196, "logits/rejected": -0.9650627970695496, "logps/chosen": -0.2704823911190033, "logps/rejected": -0.31392908096313477, "loss": 4.4957, "rewards/accuracies": 0.612500011920929, "rewards/chosen": -2.7048239707946777, "rewards/margins": 0.4344666600227356, "rewards/rejected": -3.1392903327941895, "step": 60 }, { "epoch": 0.14207650273224043, "grad_norm": 214.6265229333486, "learning_rate": 9.94736194623663e-07, "logits/chosen": -1.0099595785140991, "logits/rejected": -0.939832866191864, "logps/chosen": -0.3105173707008362, "logps/rejected": -0.34290796518325806, "loss": 4.8297, "rewards/accuracies": 0.53125, "rewards/chosen": -3.1051735877990723, "rewards/margins": 0.3239057958126068, "rewards/rejected": -3.42907977104187, "step": 65 }, { "epoch": 0.15300546448087432, "grad_norm": 473.8808962259616, "learning_rate": 9.916100327075037e-07, "logits/chosen": -0.9457240104675293, "logits/rejected": -0.9267830848693848, "logps/chosen": -0.29852330684661865, "logps/rejected": -0.32594314217567444, "loss": 4.6136, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -2.9852328300476074, "rewards/margins": 0.2741985023021698, "rewards/rejected": -3.2594313621520996, "step": 70 }, { "epoch": 0.16393442622950818, "grad_norm": 123.02284060076431, "learning_rate": 9.877658715537428e-07, "logits/chosen": -0.9411904215812683, "logits/rejected": -0.9241803288459778, "logps/chosen": -0.31740230321884155, "logps/rejected": -0.351471483707428, "loss": 4.6792, "rewards/accuracies": 0.625, "rewards/chosen": -3.174023389816284, "rewards/margins": 0.3406919836997986, "rewards/rejected": -3.5147147178649902, "step": 75 }, { "epoch": 0.17486338797814208, "grad_norm": 136.7631938838273, "learning_rate": 9.832093255815216e-07, "logits/chosen": -0.9370519518852234, "logits/rejected": -0.8748278617858887, "logps/chosen": -0.31292015314102173, "logps/rejected": -0.33769917488098145, "loss": 4.6083, "rewards/accuracies": 0.6187499761581421, "rewards/chosen": -3.1292014122009277, "rewards/margins": 0.2477904111146927, "rewards/rejected": -3.3769919872283936, "step": 80 }, { "epoch": 0.18579234972677597, "grad_norm": 198.11968829037178, "learning_rate": 9.779470496520441e-07, "logits/chosen": -0.9416698217391968, "logits/rejected": -0.891922116279602, "logps/chosen": -0.3160974383354187, "logps/rejected": -0.3771803379058838, "loss": 4.4814, "rewards/accuracies": 0.6312500238418579, "rewards/chosen": -3.1609745025634766, "rewards/margins": 0.6108287572860718, "rewards/rejected": -3.7718029022216797, "step": 85 }, { "epoch": 0.19672131147540983, "grad_norm": 145.2843942098038, "learning_rate": 9.719867293491144e-07, "logits/chosen": -1.0085046291351318, "logits/rejected": -0.9282514452934265, "logps/chosen": -0.3432404100894928, "logps/rejected": -0.381814569234848, "loss": 4.5692, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -3.4324042797088623, "rewards/margins": 0.38574135303497314, "rewards/rejected": -3.818145275115967, "step": 90 }, { "epoch": 0.20765027322404372, "grad_norm": 224.8728680422599, "learning_rate": 9.653370697542987e-07, "logits/chosen": -0.9396700859069824, "logits/rejected": -0.9407837986946106, "logps/chosen": -0.3434152603149414, "logps/rejected": -0.3718097507953644, "loss": 4.3339, "rewards/accuracies": 0.59375, "rewards/chosen": -3.434152603149414, "rewards/margins": 0.2839447855949402, "rewards/rejected": -3.718097686767578, "step": 95 }, { "epoch": 0.2185792349726776, "grad_norm": 200.93333735175037, "learning_rate": 9.580077827331037e-07, "logits/chosen": -0.9491629600524902, "logits/rejected": -0.9065796136856079, "logps/chosen": -0.38279426097869873, "logps/rejected": -0.4457218050956726, "loss": 4.2739, "rewards/accuracies": 0.6499999761581421, "rewards/chosen": -3.8279430866241455, "rewards/margins": 0.6292752027511597, "rewards/rejected": -4.457218170166016, "step": 100 }, { "epoch": 0.22950819672131148, "grad_norm": 166.91104379384572, "learning_rate": 9.500095727507419e-07, "logits/chosen": -1.0131243467330933, "logits/rejected": -0.9821707606315613, "logps/chosen": -0.36331382393836975, "logps/rejected": -0.416962206363678, "loss": 4.2844, "rewards/accuracies": 0.6187499761581421, "rewards/chosen": -3.6331381797790527, "rewards/margins": 0.5364840030670166, "rewards/rejected": -4.16962194442749, "step": 105 }, { "epoch": 0.24043715846994534, "grad_norm": 152.87786136664803, "learning_rate": 9.413541212382004e-07, "logits/chosen": -1.0002071857452393, "logits/rejected": -0.9825676083564758, "logps/chosen": -0.3780575394630432, "logps/rejected": -0.46578970551490784, "loss": 4.2075, "rewards/accuracies": 0.6625000238418579, "rewards/chosen": -3.7805755138397217, "rewards/margins": 0.8773216009140015, "rewards/rejected": -4.657896995544434, "step": 110 }, { "epoch": 0.25136612021857924, "grad_norm": 157.4597077495282, "learning_rate": 9.320540695314438e-07, "logits/chosen": -1.013012170791626, "logits/rejected": -0.9737430810928345, "logps/chosen": -0.38077500462532043, "logps/rejected": -0.4924758970737457, "loss": 4.0012, "rewards/accuracies": 0.6937500238418579, "rewards/chosen": -3.8077499866485596, "rewards/margins": 1.1170084476470947, "rewards/rejected": -4.924757957458496, "step": 115 }, { "epoch": 0.26229508196721313, "grad_norm": 125.21327412714959, "learning_rate": 9.221230004086721e-07, "logits/chosen": -1.029721975326538, "logits/rejected": -0.9594861268997192, "logps/chosen": -0.3884691596031189, "logps/rejected": -0.4513280987739563, "loss": 4.1127, "rewards/accuracies": 0.7124999761581421, "rewards/chosen": -3.8846917152404785, "rewards/margins": 0.6285887956619263, "rewards/rejected": -4.513280868530273, "step": 120 }, { "epoch": 0.273224043715847, "grad_norm": 285.25729007443874, "learning_rate": 9.11575418252596e-07, "logits/chosen": -0.9322704076766968, "logits/rejected": -0.8982660174369812, "logps/chosen": -0.4055355489253998, "logps/rejected": -0.4851369261741638, "loss": 3.8762, "rewards/accuracies": 0.7124999761581421, "rewards/chosen": -4.055355548858643, "rewards/margins": 0.7960137128829956, "rewards/rejected": -4.8513689041137695, "step": 125 }, { "epoch": 0.28415300546448086, "grad_norm": 140.757752850241, "learning_rate": 9.004267278667031e-07, "logits/chosen": -0.9658416509628296, "logits/rejected": -0.9603641629219055, "logps/chosen": -0.4266030788421631, "logps/rejected": -0.5589009523391724, "loss": 3.9144, "rewards/accuracies": 0.737500011920929, "rewards/chosen": -4.266031265258789, "rewards/margins": 1.3229780197143555, "rewards/rejected": -5.5890092849731445, "step": 130 }, { "epoch": 0.29508196721311475, "grad_norm": 112.11614815018599, "learning_rate": 8.886932119764565e-07, "logits/chosen": -1.0118809938430786, "logits/rejected": -0.9286627769470215, "logps/chosen": -0.41649705171585083, "logps/rejected": -0.5318835973739624, "loss": 3.7328, "rewards/accuracies": 0.675000011920929, "rewards/chosen": -4.164970397949219, "rewards/margins": 1.1538660526275635, "rewards/rejected": -5.318836212158203, "step": 135 }, { "epoch": 0.30601092896174864, "grad_norm": 151.56112609092273, "learning_rate": 8.763920074482809e-07, "logits/chosen": -1.00692880153656, "logits/rejected": -0.9528873562812805, "logps/chosen": -0.4440224766731262, "logps/rejected": -0.5973733067512512, "loss": 3.3329, "rewards/accuracies": 0.7250000238418579, "rewards/chosen": -4.440225124359131, "rewards/margins": 1.5335077047348022, "rewards/rejected": -5.9737324714660645, "step": 140 }, { "epoch": 0.31693989071038253, "grad_norm": 157.69245375817547, "learning_rate": 8.635410802610723e-07, "logits/chosen": -1.0103561878204346, "logits/rejected": -0.9901880025863647, "logps/chosen": -0.4360678195953369, "logps/rejected": -0.5111598372459412, "loss": 3.5759, "rewards/accuracies": 0.7250000238418579, "rewards/chosen": -4.360678672790527, "rewards/margins": 0.7509199380874634, "rewards/rejected": -5.111598014831543, "step": 145 }, { "epoch": 0.32786885245901637, "grad_norm": 131.42091257219136, "learning_rate": 8.501591992667849e-07, "logits/chosen": -1.0559594631195068, "logits/rejected": -1.0228536128997803, "logps/chosen": -0.47138285636901855, "logps/rejected": -0.6517553329467773, "loss": 3.4348, "rewards/accuracies": 0.762499988079071, "rewards/chosen": -4.7138285636901855, "rewards/margins": 1.8037246465682983, "rewards/rejected": -6.517552852630615, "step": 150 }, { "epoch": 0.33879781420765026, "grad_norm": 121.46660804843347, "learning_rate": 8.362659087784152e-07, "logits/chosen": -1.007810354232788, "logits/rejected": -0.952292799949646, "logps/chosen": -0.4735150933265686, "logps/rejected": -0.591022789478302, "loss": 3.4613, "rewards/accuracies": 0.6812499761581421, "rewards/chosen": -4.735150337219238, "rewards/margins": 1.175077199935913, "rewards/rejected": -5.9102277755737305, "step": 155 }, { "epoch": 0.34972677595628415, "grad_norm": 126.72976057892085, "learning_rate": 8.218815000254231e-07, "logits/chosen": -1.0282104015350342, "logits/rejected": -0.9731351733207703, "logps/chosen": -0.5363224744796753, "logps/rejected": -0.640962541103363, "loss": 3.4898, "rewards/accuracies": 0.6937500238418579, "rewards/chosen": -5.363224983215332, "rewards/margins": 1.046400785446167, "rewards/rejected": -6.409626007080078, "step": 160 }, { "epoch": 0.36065573770491804, "grad_norm": 153.56145857916775, "learning_rate": 8.07026981518276e-07, "logits/chosen": -1.0256257057189941, "logits/rejected": -0.9699680209159851, "logps/chosen": -0.5307550430297852, "logps/rejected": -0.6452345848083496, "loss": 3.3879, "rewards/accuracies": 0.7250000238418579, "rewards/chosen": -5.30755090713501, "rewards/margins": 1.1447944641113281, "rewards/rejected": -6.452345371246338, "step": 165 }, { "epoch": 0.37158469945355194, "grad_norm": 133.9853822185735, "learning_rate": 7.917240483654e-07, "logits/chosen": -1.0185896158218384, "logits/rejected": -0.9558917880058289, "logps/chosen": -0.5270066857337952, "logps/rejected": -0.6334024667739868, "loss": 3.5392, "rewards/accuracies": 0.7250000238418579, "rewards/chosen": -5.270066261291504, "rewards/margins": 1.0639582872390747, "rewards/rejected": -6.3340253829956055, "step": 170 }, { "epoch": 0.3825136612021858, "grad_norm": 170.7287592983698, "learning_rate": 7.759950505873521e-07, "logits/chosen": -1.0673246383666992, "logits/rejected": -1.0361274480819702, "logps/chosen": -0.558329701423645, "logps/rejected": -0.6510181427001953, "loss": 3.3022, "rewards/accuracies": 0.6499999761581421, "rewards/chosen": -5.583296775817871, "rewards/margins": 0.9268839955329895, "rewards/rejected": -6.510180473327637, "step": 175 }, { "epoch": 0.39344262295081966, "grad_norm": 134.7756954954796, "learning_rate": 7.598629604744872e-07, "logits/chosen": -1.0714390277862549, "logits/rejected": -1.0629351139068604, "logps/chosen": -0.5433841347694397, "logps/rejected": -0.7269644141197205, "loss": 3.1167, "rewards/accuracies": 0.7875000238418579, "rewards/chosen": -5.433840751647949, "rewards/margins": 1.8358032703399658, "rewards/rejected": -7.269643306732178, "step": 180 }, { "epoch": 0.40437158469945356, "grad_norm": 150.12430395788675, "learning_rate": 7.433513390357989e-07, "logits/chosen": -1.1003663539886475, "logits/rejected": -1.1102244853973389, "logps/chosen": -0.5805007219314575, "logps/rejected": -0.7595118880271912, "loss": 3.1145, "rewards/accuracies": 0.800000011920929, "rewards/chosen": -5.805006980895996, "rewards/margins": 1.7901118993759155, "rewards/rejected": -7.595119476318359, "step": 185 }, { "epoch": 0.41530054644808745, "grad_norm": 174.07312409793005, "learning_rate": 7.264843015879321e-07, "logits/chosen": -1.1195979118347168, "logits/rejected": -1.0757617950439453, "logps/chosen": -0.5829612612724304, "logps/rejected": -0.7835941314697266, "loss": 3.1898, "rewards/accuracies": 0.7875000238418579, "rewards/chosen": -5.829612731933594, "rewards/margins": 2.00632905960083, "rewards/rejected": -7.835942268371582, "step": 190 }, { "epoch": 0.4262295081967213, "grad_norm": 146.25540915098478, "learning_rate": 7.092864825346266e-07, "logits/chosen": -1.1447865962982178, "logits/rejected": -1.1201746463775635, "logps/chosen": -0.6956247091293335, "logps/rejected": -0.9184806942939758, "loss": 3.2154, "rewards/accuracies": 0.762499988079071, "rewards/chosen": -6.956246852874756, "rewards/margins": 2.2285592555999756, "rewards/rejected": -9.184806823730469, "step": 195 }, { "epoch": 0.4371584699453552, "grad_norm": 143.20432181294757, "learning_rate": 6.917829993880302e-07, "logits/chosen": -1.1187337636947632, "logits/rejected": -1.0346059799194336, "logps/chosen": -0.6610385179519653, "logps/rejected": -0.8449029922485352, "loss": 3.0607, "rewards/accuracies": 0.768750011920929, "rewards/chosen": -6.610384941101074, "rewards/margins": 1.83864426612854, "rewards/rejected": -8.449029922485352, "step": 200 }, { "epoch": 0.44808743169398907, "grad_norm": 114.99460631566296, "learning_rate": 6.739994160844309e-07, "logits/chosen": -1.0897611379623413, "logits/rejected": -1.1026214361190796, "logps/chosen": -0.6384384632110596, "logps/rejected": -0.8304228782653809, "loss": 2.9101, "rewards/accuracies": 0.8062499761581421, "rewards/chosen": -6.384384632110596, "rewards/margins": 1.9198448657989502, "rewards/rejected": -8.304228782653809, "step": 205 }, { "epoch": 0.45901639344262296, "grad_norm": 208.90885124367924, "learning_rate": 6.559617056479827e-07, "logits/chosen": -1.1222480535507202, "logits/rejected": -1.122051477432251, "logps/chosen": -0.6973533630371094, "logps/rejected": -0.942583441734314, "loss": 2.9151, "rewards/accuracies": 0.793749988079071, "rewards/chosen": -6.973533630371094, "rewards/margins": 2.452300548553467, "rewards/rejected": -9.425833702087402, "step": 210 }, { "epoch": 0.46994535519125685, "grad_norm": 158.10928377564565, "learning_rate": 6.376962122569567e-07, "logits/chosen": -1.1337159872055054, "logits/rejected": -1.0800633430480957, "logps/chosen": -0.7191341519355774, "logps/rejected": -0.9607221484184265, "loss": 3.1913, "rewards/accuracies": 0.8125, "rewards/chosen": -7.191342353820801, "rewards/margins": 2.4158806800842285, "rewards/rejected": -9.607221603393555, "step": 215 }, { "epoch": 0.4808743169398907, "grad_norm": 120.26950487116113, "learning_rate": 6.192296127679192e-07, "logits/chosen": -1.1914238929748535, "logits/rejected": -1.164651870727539, "logps/chosen": -0.7474638223648071, "logps/rejected": -0.9536181688308716, "loss": 2.9159, "rewards/accuracies": 0.7562500238418579, "rewards/chosen": -7.47463846206665, "rewards/margins": 2.0615434646606445, "rewards/rejected": -9.536182403564453, "step": 220 }, { "epoch": 0.4918032786885246, "grad_norm": 120.68908478001487, "learning_rate": 6.005888777540319e-07, "logits/chosen": -1.20607328414917, "logits/rejected": -1.1677665710449219, "logps/chosen": -0.7626639604568481, "logps/rejected": -1.0024259090423584, "loss": 2.9058, "rewards/accuracies": 0.824999988079071, "rewards/chosen": -7.626638889312744, "rewards/margins": 2.397620439529419, "rewards/rejected": -10.024259567260742, "step": 225 }, { "epoch": 0.5027322404371585, "grad_norm": 122.43410251633057, "learning_rate": 5.818012321143773e-07, "logits/chosen": -1.116031527519226, "logits/rejected": -1.117067575454712, "logps/chosen": -0.767257034778595, "logps/rejected": -1.0256441831588745, "loss": 2.9487, "rewards/accuracies": 0.8062499761581421, "rewards/chosen": -7.67257022857666, "rewards/margins": 2.5838723182678223, "rewards/rejected": -10.256442070007324, "step": 230 }, { "epoch": 0.5136612021857924, "grad_norm": 146.45120973143196, "learning_rate": 5.628941153118388e-07, "logits/chosen": -1.1330724954605103, "logits/rejected": -1.0960749387741089, "logps/chosen": -0.7881630659103394, "logps/rejected": -1.0228068828582764, "loss": 2.7419, "rewards/accuracies": 0.831250011920929, "rewards/chosen": -7.881631374359131, "rewards/margins": 2.346438407897949, "rewards/rejected": -10.228069305419922, "step": 235 }, { "epoch": 0.5245901639344263, "grad_norm": 131.87657600514555, "learning_rate": 5.438951412976098e-07, "logits/chosen": -1.1725775003433228, "logits/rejected": -1.182626485824585, "logps/chosen": -0.7777260541915894, "logps/rejected": -1.0662057399749756, "loss": 2.5085, "rewards/accuracies": 0.8187500238418579, "rewards/chosen": -7.777259826660156, "rewards/margins": 2.8847975730895996, "rewards/rejected": -10.66205883026123, "step": 240 }, { "epoch": 0.5355191256830601, "grad_norm": 135.24353045590152, "learning_rate": 5.248320581808619e-07, "logits/chosen": -1.1081842184066772, "logits/rejected": -1.0651283264160156, "logps/chosen": -0.7994285225868225, "logps/rejected": -1.083718180656433, "loss": 2.7299, "rewards/accuracies": 0.78125, "rewards/chosen": -7.994284152984619, "rewards/margins": 2.842897415161133, "rewards/rejected": -10.837182998657227, "step": 245 }, { "epoch": 0.546448087431694, "grad_norm": 124.21267267369478, "learning_rate": 5.057327077024744e-07, "logits/chosen": -1.17474365234375, "logits/rejected": -1.1420042514801025, "logps/chosen": -0.8092962503433228, "logps/rejected": -1.037235975265503, "loss": 2.8644, "rewards/accuracies": 0.737500011920929, "rewards/chosen": -8.092963218688965, "rewards/margins": 2.279397964477539, "rewards/rejected": -10.37236213684082, "step": 250 }, { "epoch": 0.5573770491803278, "grad_norm": 234.59899677846647, "learning_rate": 4.866249845720132e-07, "logits/chosen": -1.1719624996185303, "logits/rejected": -1.1423064470291138, "logps/chosen": -0.8820212483406067, "logps/rejected": -1.1843823194503784, "loss": 2.6121, "rewards/accuracies": 0.793749988079071, "rewards/chosen": -8.820211410522461, "rewards/margins": 3.0236105918884277, "rewards/rejected": -11.84382438659668, "step": 255 }, { "epoch": 0.5683060109289617, "grad_norm": 136.28015337191263, "learning_rate": 4.675367957273505e-07, "logits/chosen": -1.1379584074020386, "logits/rejected": -1.1297852993011475, "logps/chosen": -0.8502503633499146, "logps/rejected": -1.1393508911132812, "loss": 2.6642, "rewards/accuracies": 0.8125, "rewards/chosen": -8.502504348754883, "rewards/margins": 2.8910045623779297, "rewards/rejected": -11.393507957458496, "step": 260 }, { "epoch": 0.5792349726775956, "grad_norm": 140.269461590517, "learning_rate": 4.4849601957642285e-07, "logits/chosen": -1.1645795106887817, "logits/rejected": -1.1305253505706787, "logps/chosen": -0.8733395338058472, "logps/rejected": -1.1651700735092163, "loss": 2.6882, "rewards/accuracies": 0.800000011920929, "rewards/chosen": -8.73339557647705, "rewards/margins": 2.9183056354522705, "rewards/rejected": -11.651700973510742, "step": 265 }, { "epoch": 0.5901639344262295, "grad_norm": 141.71312924760568, "learning_rate": 4.295304652806592e-07, "logits/chosen": -1.159639835357666, "logits/rejected": -1.138240098953247, "logps/chosen": -0.8841695785522461, "logps/rejected": -1.2160850763320923, "loss": 2.5159, "rewards/accuracies": 0.8062499761581421, "rewards/chosen": -8.841692924499512, "rewards/margins": 3.3191559314727783, "rewards/rejected": -12.160849571228027, "step": 270 }, { "epoch": 0.6010928961748634, "grad_norm": 110.3486106272621, "learning_rate": 4.106678321395433e-07, "logits/chosen": -1.133528232574463, "logits/rejected": -1.0763746500015259, "logps/chosen": -0.9235790967941284, "logps/rejected": -1.1159861087799072, "loss": 2.7155, "rewards/accuracies": 0.6625000238418579, "rewards/chosen": -9.23579216003418, "rewards/margins": 1.9240700006484985, "rewards/rejected": -11.15986156463623, "step": 275 }, { "epoch": 0.6120218579234973, "grad_norm": 130.88653954832958, "learning_rate": 3.9193566913562915e-07, "logits/chosen": -1.0998766422271729, "logits/rejected": -1.1054426431655884, "logps/chosen": -0.9381259083747864, "logps/rejected": -1.2925664186477661, "loss": 2.6819, "rewards/accuracies": 0.737500011920929, "rewards/chosen": -9.38125991821289, "rewards/margins": 3.5444045066833496, "rewards/rejected": -12.925663948059082, "step": 280 }, { "epoch": 0.6229508196721312, "grad_norm": 120.47229604941695, "learning_rate": 3.7336133469909623e-07, "logits/chosen": -1.2325929403305054, "logits/rejected": -1.2060118913650513, "logps/chosen": -0.9257510900497437, "logps/rejected": -1.290015697479248, "loss": 2.3392, "rewards/accuracies": 0.8500000238418579, "rewards/chosen": -9.257511138916016, "rewards/margins": 3.6426455974578857, "rewards/rejected": -12.90015697479248, "step": 285 }, { "epoch": 0.6338797814207651, "grad_norm": 160.84314275747502, "learning_rate": 3.549719567506076e-07, "logits/chosen": -1.1784595251083374, "logits/rejected": -1.1460840702056885, "logps/chosen": -0.995724081993103, "logps/rejected": -1.3252066373825073, "loss": 2.6635, "rewards/accuracies": 0.793749988079071, "rewards/chosen": -9.95724105834961, "rewards/margins": 3.2948246002197266, "rewards/rejected": -13.252065658569336, "step": 290 }, { "epoch": 0.644808743169399, "grad_norm": 128.31998968470484, "learning_rate": 3.3679439308082774e-07, "logits/chosen": -1.162412405014038, "logits/rejected": -1.165961503982544, "logps/chosen": -0.995215117931366, "logps/rejected": -1.3827722072601318, "loss": 2.1144, "rewards/accuracies": 0.856249988079071, "rewards/chosen": -9.95215129852295, "rewards/margins": 3.875570774078369, "rewards/rejected": -13.827722549438477, "step": 295 }, { "epoch": 0.6557377049180327, "grad_norm": 113.0437640023239, "learning_rate": 3.1885519212446716e-07, "logits/chosen": -1.198628306388855, "logits/rejected": -1.1889218091964722, "logps/chosen": -1.035154104232788, "logps/rejected": -1.390194296836853, "loss": 2.3734, "rewards/accuracies": 0.800000011920929, "rewards/chosen": -10.351542472839355, "rewards/margins": 3.550400495529175, "rewards/rejected": -13.901944160461426, "step": 300 }, { "epoch": 0.6666666666666666, "grad_norm": 148.57393016872774, "learning_rate": 3.0118055418614295e-07, "logits/chosen": -1.1905871629714966, "logits/rejected": -1.147964596748352, "logps/chosen": -1.0231964588165283, "logps/rejected": -1.4057810306549072, "loss": 2.5431, "rewards/accuracies": 0.8062499761581421, "rewards/chosen": -10.231963157653809, "rewards/margins": 3.825847625732422, "rewards/rejected": -14.05781078338623, "step": 305 }, { "epoch": 0.6775956284153005, "grad_norm": 201.17993276494983, "learning_rate": 2.83796293174686e-07, "logits/chosen": -1.1385854482650757, "logits/rejected": -1.1492459774017334, "logps/chosen": -1.0367190837860107, "logps/rejected": -1.4528653621673584, "loss": 2.7689, "rewards/accuracies": 0.800000011920929, "rewards/chosen": -10.36719036102295, "rewards/margins": 4.161464214324951, "rewards/rejected": -14.528656005859375, "step": 310 }, { "epoch": 0.6885245901639344, "grad_norm": 150.32062982255817, "learning_rate": 2.6672779890178046e-07, "logits/chosen": -1.1942903995513916, "logits/rejected": -1.1974905729293823, "logps/chosen": -1.101068139076233, "logps/rejected": -1.3708066940307617, "loss": 2.4457, "rewards/accuracies": 0.7437499761581421, "rewards/chosen": -11.01068115234375, "rewards/margins": 2.6973862648010254, "rewards/rejected": -13.70806884765625, "step": 315 }, { "epoch": 0.6994535519125683, "grad_norm": 115.42514466297712, "learning_rate": 2.500000000000001e-07, "logits/chosen": -1.2662487030029297, "logits/rejected": -1.2223560810089111, "logps/chosen": -1.1014665365219116, "logps/rejected": -1.4792479276657104, "loss": 2.4845, "rewards/accuracies": 0.800000011920929, "rewards/chosen": -11.014665603637695, "rewards/margins": 3.77781343460083, "rewards/rejected": -14.79248046875, "step": 320 }, { "epoch": 0.7103825136612022, "grad_norm": 168.00405960921228, "learning_rate": 2.3363732751439923e-07, "logits/chosen": -1.225192666053772, "logits/rejected": -1.2128854990005493, "logps/chosen": -1.0635629892349243, "logps/rejected": -1.3986676931381226, "loss": 2.4994, "rewards/accuracies": 0.800000011920929, "rewards/chosen": -10.635629653930664, "rewards/margins": 3.3510470390319824, "rewards/rejected": -13.986676216125488, "step": 325 }, { "epoch": 0.7213114754098361, "grad_norm": 103.05949871144796, "learning_rate": 2.1766367922083283e-07, "logits/chosen": -1.160074234008789, "logits/rejected": -1.1407663822174072, "logps/chosen": -1.0139520168304443, "logps/rejected": -1.4850687980651855, "loss": 2.3858, "rewards/accuracies": 0.856249988079071, "rewards/chosen": -10.139519691467285, "rewards/margins": 4.7111687660217285, "rewards/rejected": -14.850689888000488, "step": 330 }, { "epoch": 0.73224043715847, "grad_norm": 140.1249582615815, "learning_rate": 2.021023847231202e-07, "logits/chosen": -1.1456273794174194, "logits/rejected": -1.115674614906311, "logps/chosen": -1.1265430450439453, "logps/rejected": -1.4902852773666382, "loss": 2.3602, "rewards/accuracies": 0.856249988079071, "rewards/chosen": -11.265429496765137, "rewards/margins": 3.637423276901245, "rewards/rejected": -14.902852058410645, "step": 335 }, { "epoch": 0.7431693989071039, "grad_norm": 144.16110329539526, "learning_rate": 1.869761713800254e-07, "logits/chosen": -1.1668914556503296, "logits/rejected": -1.1292145252227783, "logps/chosen": -1.0967395305633545, "logps/rejected": -1.4661519527435303, "loss": 2.5221, "rewards/accuracies": 0.768750011920929, "rewards/chosen": -10.967395782470703, "rewards/margins": 3.6941237449645996, "rewards/rejected": -14.661520004272461, "step": 340 }, { "epoch": 0.7540983606557377, "grad_norm": 149.08880264886358, "learning_rate": 1.7230713111182164e-07, "logits/chosen": -1.2098548412322998, "logits/rejected": -1.2148889303207397, "logps/chosen": -1.1389497518539429, "logps/rejected": -1.5369229316711426, "loss": 2.5085, "rewards/accuracies": 0.8125, "rewards/chosen": -11.389498710632324, "rewards/margins": 3.979731798171997, "rewards/rejected": -15.369229316711426, "step": 345 }, { "epoch": 0.7650273224043715, "grad_norm": 144.89905388084784, "learning_rate": 1.5811668813491696e-07, "logits/chosen": -1.2163605690002441, "logits/rejected": -1.2011533975601196, "logps/chosen": -1.0724060535430908, "logps/rejected": -1.4004809856414795, "loss": 2.4309, "rewards/accuracies": 0.7875000238418579, "rewards/chosen": -10.724058151245117, "rewards/margins": 3.280749559402466, "rewards/rejected": -14.004809379577637, "step": 350 }, { "epoch": 0.7759562841530054, "grad_norm": 146.85736518453848, "learning_rate": 1.4442556767166369e-07, "logits/chosen": -1.171179175376892, "logits/rejected": -1.144737958908081, "logps/chosen": -1.0694037675857544, "logps/rejected": -1.4522759914398193, "loss": 2.4166, "rewards/accuracies": 0.84375, "rewards/chosen": -10.694037437438965, "rewards/margins": 3.828723192214966, "rewards/rejected": -14.522760391235352, "step": 355 }, { "epoch": 0.7868852459016393, "grad_norm": 145.85471061365715, "learning_rate": 1.312537656810549e-07, "logits/chosen": -1.1415390968322754, "logits/rejected": -1.1467649936676025, "logps/chosen": -1.1098216772079468, "logps/rejected": -1.4691263437271118, "loss": 2.4771, "rewards/accuracies": 0.762499988079071, "rewards/chosen": -11.09821605682373, "rewards/margins": 3.5930473804473877, "rewards/rejected": -14.691264152526855, "step": 360 }, { "epoch": 0.7978142076502732, "grad_norm": 167.3238922002567, "learning_rate": 1.1862051965451214e-07, "logits/chosen": -1.2173941135406494, "logits/rejected": -1.2231066226959229, "logps/chosen": -1.143596887588501, "logps/rejected": -1.5389248132705688, "loss": 2.3622, "rewards/accuracies": 0.78125, "rewards/chosen": -11.435969352722168, "rewards/margins": 3.9532806873321533, "rewards/rejected": -15.389248847961426, "step": 365 }, { "epoch": 0.8087431693989071, "grad_norm": 153.79026677729559, "learning_rate": 1.0654428051942138e-07, "logits/chosen": -1.225081443786621, "logits/rejected": -1.1957951784133911, "logps/chosen": -1.163589596748352, "logps/rejected": -1.5947258472442627, "loss": 2.5258, "rewards/accuracies": 0.7875000238418579, "rewards/chosen": -11.635896682739258, "rewards/margins": 4.311361312866211, "rewards/rejected": -15.947257995605469, "step": 370 }, { "epoch": 0.819672131147541, "grad_norm": 112.49710778946844, "learning_rate": 9.504268569144763e-08, "logits/chosen": -1.2405080795288086, "logits/rejected": -1.1872971057891846, "logps/chosen": -1.1241363286972046, "logps/rejected": -1.525253415107727, "loss": 2.385, "rewards/accuracies": 0.8125, "rewards/chosen": -11.241364479064941, "rewards/margins": 4.011169910430908, "rewards/rejected": -15.252534866333008, "step": 375 }, { "epoch": 0.8306010928961749, "grad_norm": 133.3144411726237, "learning_rate": 8.413253331499049e-08, "logits/chosen": -1.1226742267608643, "logits/rejected": -1.1412006616592407, "logps/chosen": -1.1478770971298218, "logps/rejected": -1.5184364318847656, "loss": 2.3692, "rewards/accuracies": 0.8500000238418579, "rewards/chosen": -11.478772163391113, "rewards/margins": 3.705592393875122, "rewards/rejected": -15.184364318847656, "step": 380 }, { "epoch": 0.8415300546448088, "grad_norm": 131.51573012388099, "learning_rate": 7.382975772939865e-08, "logits/chosen": -1.233059287071228, "logits/rejected": -1.2242696285247803, "logps/chosen": -1.211987853050232, "logps/rejected": -1.6103414297103882, "loss": 2.6625, "rewards/accuracies": 0.8500000238418579, "rewards/chosen": -12.119878768920898, "rewards/margins": 3.983535051345825, "rewards/rejected": -16.103412628173828, "step": 385 }, { "epoch": 0.8524590163934426, "grad_norm": 179.13558974996818, "learning_rate": 6.414940619677734e-08, "logits/chosen": -1.2085000276565552, "logits/rejected": -1.1961381435394287, "logps/chosen": -1.1327707767486572, "logps/rejected": -1.6155046224594116, "loss": 2.3237, "rewards/accuracies": 0.831250011920929, "rewards/chosen": -11.327706336975098, "rewards/margins": 4.827339172363281, "rewards/rejected": -16.155044555664062, "step": 390 }, { "epoch": 0.8633879781420765, "grad_norm": 187.32213579680786, "learning_rate": 5.5105616925376296e-08, "logits/chosen": -1.1982018947601318, "logits/rejected": -1.1823650598526, "logps/chosen": -1.1895129680633545, "logps/rejected": -1.5091607570648193, "loss": 2.3095, "rewards/accuracies": 0.7749999761581421, "rewards/chosen": -11.895132064819336, "rewards/margins": 3.196476936340332, "rewards/rejected": -15.091608047485352, "step": 395 }, { "epoch": 0.8743169398907104, "grad_norm": 152.08655815719752, "learning_rate": 4.6711598420656976e-08, "logits/chosen": -1.1382460594177246, "logits/rejected": -1.1149357557296753, "logps/chosen": -1.1855409145355225, "logps/rejected": -1.6318585872650146, "loss": 2.1949, "rewards/accuracies": 0.875, "rewards/chosen": -11.855409622192383, "rewards/margins": 4.463176727294922, "rewards/rejected": -16.318586349487305, "step": 400 }, { "epoch": 0.8743169398907104, "eval_logits/chosen": -1.3829468488693237, "eval_logits/rejected": -1.3507002592086792, "eval_logps/chosen": -1.1304905414581299, "eval_logps/rejected": -1.5389478206634521, "eval_loss": 2.412938117980957, "eval_rewards/accuracies": 0.8132529854774475, "eval_rewards/chosen": -11.30490493774414, "eval_rewards/margins": 4.0845723152160645, "eval_rewards/rejected": -15.389474868774414, "eval_runtime": 37.7675, "eval_samples_per_second": 34.898, "eval_steps_per_second": 2.198, "step": 400 }, { "epoch": 0.8852459016393442, "grad_norm": 137.4747366766835, "learning_rate": 3.897961019419516e-08, "logits/chosen": -1.1692960262298584, "logits/rejected": -1.1131761074066162, "logps/chosen": -1.078470230102539, "logps/rejected": -1.4457483291625977, "loss": 2.3038, "rewards/accuracies": 0.793749988079071, "rewards/chosen": -10.78470230102539, "rewards/margins": 3.672783613204956, "rewards/rejected": -14.457486152648926, "step": 405 }, { "epoch": 0.8961748633879781, "grad_norm": 128.05923930291988, "learning_rate": 3.192094485859526e-08, "logits/chosen": -1.171673059463501, "logits/rejected": -1.2017241716384888, "logps/chosen": -1.1775602102279663, "logps/rejected": -1.6198632717132568, "loss": 2.3744, "rewards/accuracies": 0.78125, "rewards/chosen": -11.775602340698242, "rewards/margins": 4.423029899597168, "rewards/rejected": -16.198633193969727, "step": 410 }, { "epoch": 0.907103825136612, "grad_norm": 136.25040575701973, "learning_rate": 2.5545911634565265e-08, "logits/chosen": -1.2182672023773193, "logits/rejected": -1.222606897354126, "logps/chosen": -1.1761579513549805, "logps/rejected": -1.6315196752548218, "loss": 2.556, "rewards/accuracies": 0.856249988079071, "rewards/chosen": -11.761578559875488, "rewards/margins": 4.553616523742676, "rewards/rejected": -16.315195083618164, "step": 415 }, { "epoch": 0.9180327868852459, "grad_norm": 135.49217722627617, "learning_rate": 1.9863821294241522e-08, "logits/chosen": -1.2129249572753906, "logits/rejected": -1.1957124471664429, "logps/chosen": -1.1399694681167603, "logps/rejected": -1.5886274576187134, "loss": 2.2082, "rewards/accuracies": 0.8374999761581421, "rewards/chosen": -11.399694442749023, "rewards/margins": 4.486580848693848, "rewards/rejected": -15.886274337768555, "step": 420 }, { "epoch": 0.9289617486338798, "grad_norm": 135.8877044935921, "learning_rate": 1.4882972562753615e-08, "logits/chosen": -1.2110410928726196, "logits/rejected": -1.2060327529907227, "logps/chosen": -1.2429840564727783, "logps/rejected": -1.6804256439208984, "loss": 2.4895, "rewards/accuracies": 0.8125, "rewards/chosen": -12.429841041564941, "rewards/margins": 4.374415397644043, "rewards/rejected": -16.804256439208984, "step": 425 }, { "epoch": 0.9398907103825137, "grad_norm": 132.90623812452836, "learning_rate": 1.0610639997888915e-08, "logits/chosen": -1.1448732614517212, "logits/rejected": -1.1479747295379639, "logps/chosen": -1.1143007278442383, "logps/rejected": -1.5705087184906006, "loss": 2.0744, "rewards/accuracies": 0.893750011920929, "rewards/chosen": -11.143007278442383, "rewards/margins": 4.562080383300781, "rewards/rejected": -15.705087661743164, "step": 430 }, { "epoch": 0.9508196721311475, "grad_norm": 150.69294443867608, "learning_rate": 7.053063365559997e-09, "logits/chosen": -1.2052174806594849, "logits/rejected": -1.2350053787231445, "logps/chosen": -1.1626297235488892, "logps/rejected": -1.625112771987915, "loss": 2.2249, "rewards/accuracies": 0.875, "rewards/chosen": -11.626296997070312, "rewards/margins": 4.6248297691345215, "rewards/rejected": -16.251127243041992, "step": 435 }, { "epoch": 0.9617486338797814, "grad_norm": 126.29899935620595, "learning_rate": 4.215438526591064e-09, "logits/chosen": -1.1682944297790527, "logits/rejected": -1.1369967460632324, "logps/chosen": -1.205596685409546, "logps/rejected": -1.5388822555541992, "loss": 2.426, "rewards/accuracies": 0.768750011920929, "rewards/chosen": -12.055967330932617, "rewards/margins": 3.332855701446533, "rewards/rejected": -15.388822555541992, "step": 440 }, { "epoch": 0.9726775956284153, "grad_norm": 145.76337247257322, "learning_rate": 2.1019098481337426e-09, "logits/chosen": -1.2075254917144775, "logits/rejected": -1.1947346925735474, "logps/chosen": -1.124491810798645, "logps/rejected": -1.569777250289917, "loss": 2.2395, "rewards/accuracies": 0.8187500238418579, "rewards/chosen": -11.244916915893555, "rewards/margins": 4.452854156494141, "rewards/rejected": -15.697772026062012, "step": 445 }, { "epoch": 0.9836065573770492, "grad_norm": 163.83031864944653, "learning_rate": 7.155641507955445e-10, "logits/chosen": -1.1285139322280884, "logits/rejected": -1.1291377544403076, "logps/chosen": -1.2102129459381104, "logps/rejected": -1.590519666671753, "loss": 2.5308, "rewards/accuracies": 0.824999988079071, "rewards/chosen": -12.102127075195312, "rewards/margins": 3.803069591522217, "rewards/rejected": -15.90519905090332, "step": 450 }, { "epoch": 0.994535519125683, "grad_norm": 155.57860273708854, "learning_rate": 5.842620032053824e-11, "logits/chosen": -1.1562345027923584, "logits/rejected": -1.1542116403579712, "logps/chosen": -1.228244662284851, "logps/rejected": -1.5371659994125366, "loss": 2.7124, "rewards/accuracies": 0.762499988079071, "rewards/chosen": -12.282448768615723, "rewards/margins": 3.0892109870910645, "rewards/rejected": -15.371660232543945, "step": 455 }, { "epoch": 0.9989071038251366, "step": 457, "total_flos": 0.0, "train_loss": 3.252214796955267, "train_runtime": 5969.5229, "train_samples_per_second": 9.809, "train_steps_per_second": 0.077 } ], "logging_steps": 5, "max_steps": 457, "num_input_tokens_seen": 0, "num_train_epochs": 1, "save_steps": 1000000, "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": 2, "trial_name": null, "trial_params": null }