| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.986666666666667, | |
| "eval_steps": 500, | |
| "global_step": 315, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.047407407407407405, | |
| "grad_norm": 1072.0, | |
| "learning_rate": 7.8125e-06, | |
| "log_odds_chosen": -1.060240626335144, | |
| "log_odds_ratio": -11.007429122924805, | |
| "logps/chosen": -22.712669372558594, | |
| "logps/rejected": -21.652647018432617, | |
| "loss": 455.4558, | |
| "nll_loss": 9.838397026062012, | |
| "rewards/accuracies": 0.4781250059604645, | |
| "rewards/chosen": -11.356334686279297, | |
| "rewards/margins": -0.530011773109436, | |
| "rewards/rejected": -10.826323509216309, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.09481481481481481, | |
| "grad_norm": 592.0, | |
| "learning_rate": 1.5625e-05, | |
| "log_odds_chosen": -3.1204988956451416, | |
| "log_odds_ratio": -11.043037414550781, | |
| "logps/chosen": -21.994853973388672, | |
| "logps/rejected": -18.874696731567383, | |
| "loss": 424.5746, | |
| "nll_loss": 8.752840042114258, | |
| "rewards/accuracies": 0.4437499940395355, | |
| "rewards/chosen": -10.997426986694336, | |
| "rewards/margins": -1.5600786209106445, | |
| "rewards/rejected": -9.437348365783691, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.14222222222222222, | |
| "grad_norm": 454.0, | |
| "learning_rate": 2.34375e-05, | |
| "log_odds_chosen": -1.2841793298721313, | |
| "log_odds_ratio": -11.318960189819336, | |
| "logps/chosen": -21.992443084716797, | |
| "logps/rejected": -20.706798553466797, | |
| "loss": 399.2196, | |
| "nll_loss": 8.00178050994873, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": -10.996221542358398, | |
| "rewards/margins": -0.6428214907646179, | |
| "rewards/rejected": -10.353399276733398, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.18962962962962962, | |
| "grad_norm": 3008.0, | |
| "learning_rate": 3.125e-05, | |
| "log_odds_chosen": -4.260960578918457, | |
| "log_odds_ratio": -9.433350563049316, | |
| "logps/chosen": -17.13646125793457, | |
| "logps/rejected": -12.87755298614502, | |
| "loss": 320.5827, | |
| "nll_loss": 6.233663558959961, | |
| "rewards/accuracies": 0.44062501192092896, | |
| "rewards/chosen": -8.568230628967285, | |
| "rewards/margins": -2.129453659057617, | |
| "rewards/rejected": -6.43877649307251, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.23703703703703705, | |
| "grad_norm": 296.0, | |
| "learning_rate": 3.90625e-05, | |
| "log_odds_chosen": -0.10830163955688477, | |
| "log_odds_ratio": -1.7876460552215576, | |
| "logps/chosen": -3.861504316329956, | |
| "logps/rejected": -3.7353768348693848, | |
| "loss": 92.4915, | |
| "nll_loss": 2.6942696571350098, | |
| "rewards/accuracies": 0.5531250238418579, | |
| "rewards/chosen": -1.930752158164978, | |
| "rewards/margins": -0.06306387484073639, | |
| "rewards/rejected": -1.8676884174346924, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.28444444444444444, | |
| "grad_norm": 189.0, | |
| "learning_rate": 4.6875e-05, | |
| "log_odds_chosen": 0.08914291858673096, | |
| "log_odds_ratio": -0.9090879559516907, | |
| "logps/chosen": -1.9305442571640015, | |
| "logps/rejected": -2.0069050788879395, | |
| "loss": 65.6584, | |
| "nll_loss": 2.0820810794830322, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": -0.9652721285820007, | |
| "rewards/margins": 0.03818041831254959, | |
| "rewards/rejected": -1.0034525394439697, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.33185185185185184, | |
| "grad_norm": 247.0, | |
| "learning_rate": 4.998613757348784e-05, | |
| "log_odds_chosen": 0.3332863748073578, | |
| "log_odds_ratio": -0.8136388063430786, | |
| "logps/chosen": -1.8473803997039795, | |
| "logps/rejected": -2.1461009979248047, | |
| "loss": 58.187, | |
| "nll_loss": 1.9638898372650146, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": -0.9236901998519897, | |
| "rewards/margins": 0.14936023950576782, | |
| "rewards/rejected": -1.0730504989624023, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.37925925925925924, | |
| "grad_norm": 316.0, | |
| "learning_rate": 4.990147841143462e-05, | |
| "log_odds_chosen": 0.5912309885025024, | |
| "log_odds_ratio": -0.6898080110549927, | |
| "logps/chosen": -1.7279255390167236, | |
| "logps/rejected": -2.248690128326416, | |
| "loss": 53.4463, | |
| "nll_loss": 1.926064133644104, | |
| "rewards/accuracies": 0.65625, | |
| "rewards/chosen": -0.8639627695083618, | |
| "rewards/margins": 0.2603822350502014, | |
| "rewards/rejected": -1.124345064163208, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.4266666666666667, | |
| "grad_norm": 436.0, | |
| "learning_rate": 4.97401218720448e-05, | |
| "log_odds_chosen": 1.5738911628723145, | |
| "log_odds_ratio": -0.6931017637252808, | |
| "logps/chosen": -1.9965721368789673, | |
| "logps/rejected": -3.4913876056671143, | |
| "loss": 42.7119, | |
| "nll_loss": 1.9767124652862549, | |
| "rewards/accuracies": 0.684374988079071, | |
| "rewards/chosen": -0.9982860684394836, | |
| "rewards/margins": 0.7474077939987183, | |
| "rewards/rejected": -1.7456938028335571, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.4740740740740741, | |
| "grad_norm": 340.0, | |
| "learning_rate": 4.9502564938797946e-05, | |
| "log_odds_chosen": 0.7620365619659424, | |
| "log_odds_ratio": -0.7459171414375305, | |
| "logps/chosen": -1.7422816753387451, | |
| "logps/rejected": -2.4515380859375, | |
| "loss": 52.5226, | |
| "nll_loss": 1.9401613473892212, | |
| "rewards/accuracies": 0.6312500238418579, | |
| "rewards/chosen": -0.8711408376693726, | |
| "rewards/margins": 0.3546280860900879, | |
| "rewards/rejected": -1.22576904296875, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.5214814814814814, | |
| "grad_norm": 336.0, | |
| "learning_rate": 4.918953929490768e-05, | |
| "log_odds_chosen": 2.066755533218384, | |
| "log_odds_ratio": -0.703250527381897, | |
| "logps/chosen": -2.0667014122009277, | |
| "logps/rejected": -4.0496721267700195, | |
| "loss": 43.0197, | |
| "nll_loss": 2.0524208545684814, | |
| "rewards/accuracies": 0.6812499761581421, | |
| "rewards/chosen": -1.0333507061004639, | |
| "rewards/margins": 0.9914852380752563, | |
| "rewards/rejected": -2.0248360633850098, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.5688888888888889, | |
| "grad_norm": 310.0, | |
| "learning_rate": 4.88020090697132e-05, | |
| "log_odds_chosen": 1.32530677318573, | |
| "log_odds_ratio": -0.7277625799179077, | |
| "logps/chosen": -1.7378164529800415, | |
| "logps/rejected": -2.996918201446533, | |
| "loss": 46.4442, | |
| "nll_loss": 1.8602434396743774, | |
| "rewards/accuracies": 0.6656249761581421, | |
| "rewards/chosen": -0.8689082264900208, | |
| "rewards/margins": 0.6295510530471802, | |
| "rewards/rejected": -1.4984591007232666, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.6162962962962963, | |
| "grad_norm": 418.0, | |
| "learning_rate": 4.834116786912897e-05, | |
| "log_odds_chosen": 2.172917604446411, | |
| "log_odds_ratio": -0.7570880651473999, | |
| "logps/chosen": -1.923484444618225, | |
| "logps/rejected": -4.014767169952393, | |
| "loss": 43.8154, | |
| "nll_loss": 2.0151474475860596, | |
| "rewards/accuracies": 0.671875, | |
| "rewards/chosen": -0.9617422223091125, | |
| "rewards/margins": 1.0456410646438599, | |
| "rewards/rejected": -2.0073835849761963, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.6637037037037037, | |
| "grad_norm": 362.0, | |
| "learning_rate": 4.7808435099299045e-05, | |
| "log_odds_chosen": 3.7954392433166504, | |
| "log_odds_ratio": -0.6524917483329773, | |
| "logps/chosen": -2.0215845108032227, | |
| "logps/rejected": -5.7059245109558105, | |
| "loss": 31.4526, | |
| "nll_loss": 2.0003814697265625, | |
| "rewards/accuracies": 0.753125011920929, | |
| "rewards/chosen": -1.0107922554016113, | |
| "rewards/margins": 1.842170000076294, | |
| "rewards/rejected": -2.8529622554779053, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.7111111111111111, | |
| "grad_norm": 644.0, | |
| "learning_rate": 4.720545159477922e-05, | |
| "log_odds_chosen": 2.579057216644287, | |
| "log_odds_ratio": -0.9140008687973022, | |
| "logps/chosen": -2.144348382949829, | |
| "logps/rejected": -4.612305641174316, | |
| "loss": 40.0912, | |
| "nll_loss": 1.925865888595581, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -1.0721741914749146, | |
| "rewards/margins": 1.2339783906936646, | |
| "rewards/rejected": -2.306152820587158, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.7585185185185185, | |
| "grad_norm": 474.0, | |
| "learning_rate": 4.653407456471222e-05, | |
| "log_odds_chosen": 2.602337121963501, | |
| "log_odds_ratio": -0.8660507202148438, | |
| "logps/chosen": -2.392124891281128, | |
| "logps/rejected": -4.895805358886719, | |
| "loss": 45.2894, | |
| "nll_loss": 2.1908349990844727, | |
| "rewards/accuracies": 0.6781250238418579, | |
| "rewards/chosen": -1.196062445640564, | |
| "rewards/margins": 1.2518401145935059, | |
| "rewards/rejected": -2.4479026794433594, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.8059259259259259, | |
| "grad_norm": 268.0, | |
| "learning_rate": 4.579637187256222e-05, | |
| "log_odds_chosen": 1.2037420272827148, | |
| "log_odds_ratio": -0.6474961042404175, | |
| "logps/chosen": -1.5563756227493286, | |
| "logps/rejected": -2.667034864425659, | |
| "loss": 40.2868, | |
| "nll_loss": 1.7489856481552124, | |
| "rewards/accuracies": 0.65625, | |
| "rewards/chosen": -0.7781878113746643, | |
| "rewards/margins": 0.5553296208381653, | |
| "rewards/rejected": -1.3335174322128296, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.8533333333333334, | |
| "grad_norm": 372.0, | |
| "learning_rate": 4.499461566702685e-05, | |
| "log_odds_chosen": 5.79607629776001, | |
| "log_odds_ratio": -0.834893524646759, | |
| "logps/chosen": -2.230257987976074, | |
| "logps/rejected": -7.889917850494385, | |
| "loss": 33.4453, | |
| "nll_loss": 2.099855422973633, | |
| "rewards/accuracies": 0.7437499761581421, | |
| "rewards/chosen": -1.115128993988037, | |
| "rewards/margins": 2.8298299312591553, | |
| "rewards/rejected": -3.9449589252471924, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.9007407407407407, | |
| "grad_norm": 644.0, | |
| "learning_rate": 4.413127538374411e-05, | |
| "log_odds_chosen": 2.9538180828094482, | |
| "log_odds_ratio": -0.7920938730239868, | |
| "logps/chosen": -1.8438174724578857, | |
| "logps/rejected": -4.690665245056152, | |
| "loss": 34.7808, | |
| "nll_loss": 1.8597549200057983, | |
| "rewards/accuracies": 0.690625011920929, | |
| "rewards/chosen": -0.9219087362289429, | |
| "rewards/margins": 1.4234240055084229, | |
| "rewards/rejected": -2.345332622528076, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.9481481481481482, | |
| "grad_norm": 296.0, | |
| "learning_rate": 4.320901013934887e-05, | |
| "log_odds_chosen": 3.595071315765381, | |
| "log_odds_ratio": -0.8463969230651855, | |
| "logps/chosen": -2.0342302322387695, | |
| "logps/rejected": -5.500476837158203, | |
| "loss": 30.0336, | |
| "nll_loss": 2.020622491836548, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -1.0171151161193848, | |
| "rewards/margins": 1.7331234216690063, | |
| "rewards/rejected": -2.7502384185791016, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.9955555555555555, | |
| "grad_norm": 312.0, | |
| "learning_rate": 4.223066054130568e-05, | |
| "log_odds_chosen": 2.3394980430603027, | |
| "log_odds_ratio": -0.7494302988052368, | |
| "logps/chosen": -1.8197135925292969, | |
| "logps/rejected": -4.058799743652344, | |
| "loss": 35.016, | |
| "nll_loss": 1.9035847187042236, | |
| "rewards/accuracies": 0.668749988079071, | |
| "rewards/chosen": -0.9098567962646484, | |
| "rewards/margins": 1.1195429563522339, | |
| "rewards/rejected": -2.029399871826172, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 1.0429629629629629, | |
| "grad_norm": 1912.0, | |
| "learning_rate": 4.1199239938743797e-05, | |
| "log_odds_chosen": 4.236447334289551, | |
| "log_odds_ratio": -0.4881769120693207, | |
| "logps/chosen": -2.2219159603118896, | |
| "logps/rejected": -6.284074306488037, | |
| "loss": 22.0456, | |
| "nll_loss": 2.1491634845733643, | |
| "rewards/accuracies": 0.8031250238418579, | |
| "rewards/chosen": -1.1109579801559448, | |
| "rewards/margins": 2.031079053878784, | |
| "rewards/rejected": -3.1420371532440186, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 1.0903703703703704, | |
| "grad_norm": 382.0, | |
| "learning_rate": 4.0117925141242174e-05, | |
| "log_odds_chosen": 4.837921142578125, | |
| "log_odds_ratio": -0.5910585522651672, | |
| "logps/chosen": -1.997674584388733, | |
| "logps/rejected": -6.637750148773193, | |
| "loss": 16.3477, | |
| "nll_loss": 1.9437005519866943, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -0.9988372921943665, | |
| "rewards/margins": 2.320038318634033, | |
| "rewards/rejected": -3.3188750743865967, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 1.1377777777777778, | |
| "grad_norm": 516.0, | |
| "learning_rate": 3.899004663415084e-05, | |
| "log_odds_chosen": 4.582813739776611, | |
| "log_odds_ratio": -0.4979763627052307, | |
| "logps/chosen": -1.5545316934585571, | |
| "logps/rejected": -5.910370826721191, | |
| "loss": 11.8075, | |
| "nll_loss": 1.7471332550048828, | |
| "rewards/accuracies": 0.784375011920929, | |
| "rewards/chosen": -0.7772658467292786, | |
| "rewards/margins": 2.177919626235962, | |
| "rewards/rejected": -2.9551854133605957, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 1.1851851851851851, | |
| "grad_norm": 466.0, | |
| "learning_rate": 3.781907832058587e-05, | |
| "log_odds_chosen": 5.0084943771362305, | |
| "log_odds_ratio": -0.5646654367446899, | |
| "logps/chosen": -1.8939094543457031, | |
| "logps/rejected": -6.705332279205322, | |
| "loss": 12.4279, | |
| "nll_loss": 1.95928156375885, | |
| "rewards/accuracies": 0.778124988079071, | |
| "rewards/chosen": -0.9469547271728516, | |
| "rewards/margins": 2.4057114124298096, | |
| "rewards/rejected": -3.352666139602661, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 1.2325925925925927, | |
| "grad_norm": 1016.0, | |
| "learning_rate": 3.660862682169282e-05, | |
| "log_odds_chosen": 6.262756824493408, | |
| "log_odds_ratio": -0.5206202268600464, | |
| "logps/chosen": -1.6548172235488892, | |
| "logps/rejected": -7.668360710144043, | |
| "loss": 2.6131, | |
| "nll_loss": 1.758933663368225, | |
| "rewards/accuracies": 0.809374988079071, | |
| "rewards/chosen": -0.8274086117744446, | |
| "rewards/margins": 3.006772041320801, | |
| "rewards/rejected": -3.8341803550720215, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "grad_norm": 506.0, | |
| "learning_rate": 3.5362420368134356e-05, | |
| "log_odds_chosen": 4.817102909088135, | |
| "log_odds_ratio": -0.6338304281234741, | |
| "logps/chosen": -1.9214130640029907, | |
| "logps/rejected": -6.519143104553223, | |
| "loss": 18.6154, | |
| "nll_loss": 1.8606481552124023, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -0.9607065320014954, | |
| "rewards/margins": 2.2988650798797607, | |
| "rewards/rejected": -3.2595715522766113, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 1.3274074074074074, | |
| "grad_norm": 956.0, | |
| "learning_rate": 3.408429731701635e-05, | |
| "log_odds_chosen": 4.097002983093262, | |
| "log_odds_ratio": -0.4646216034889221, | |
| "logps/chosen": -1.740180253982544, | |
| "logps/rejected": -5.640007972717285, | |
| "loss": 11.3345, | |
| "nll_loss": 1.65927255153656, | |
| "rewards/accuracies": 0.8062499761581421, | |
| "rewards/chosen": -0.870090126991272, | |
| "rewards/margins": 1.9499139785766602, | |
| "rewards/rejected": -2.8200039863586426, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 1.374814814814815, | |
| "grad_norm": 1064.0, | |
| "learning_rate": 3.2778194329621104e-05, | |
| "log_odds_chosen": 7.051480770111084, | |
| "log_odds_ratio": -0.4304323196411133, | |
| "logps/chosen": -2.5247302055358887, | |
| "logps/rejected": -9.359436988830566, | |
| "loss": 5.1241, | |
| "nll_loss": 2.1758809089660645, | |
| "rewards/accuracies": 0.846875011920929, | |
| "rewards/chosen": -1.2623651027679443, | |
| "rewards/margins": 3.417353868484497, | |
| "rewards/rejected": -4.679718494415283, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 1.4222222222222223, | |
| "grad_norm": 604.0, | |
| "learning_rate": 3.144813424636031e-05, | |
| "log_odds_chosen": 6.2552809715271, | |
| "log_odds_ratio": -0.4223237931728363, | |
| "logps/chosen": -1.5957365036010742, | |
| "logps/rejected": -7.607705116271973, | |
| "loss": 2.8133, | |
| "nll_loss": 1.7030588388442993, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": -0.7978682518005371, | |
| "rewards/margins": 3.005984306335449, | |
| "rewards/rejected": -3.8038525581359863, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.4696296296296296, | |
| "grad_norm": 972.0, | |
| "learning_rate": 3.0098213696293542e-05, | |
| "log_odds_chosen": 5.868264198303223, | |
| "log_odds_ratio": -0.7328829169273376, | |
| "logps/chosen": -2.398165225982666, | |
| "logps/rejected": -8.077600479125977, | |
| "loss": 14.0782, | |
| "nll_loss": 1.9348533153533936, | |
| "rewards/accuracies": 0.78125, | |
| "rewards/chosen": -1.199082612991333, | |
| "rewards/margins": 2.839717388153076, | |
| "rewards/rejected": -4.038800239562988, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 1.5170370370370372, | |
| "grad_norm": 884.0, | |
| "learning_rate": 2.8732590479375165e-05, | |
| "log_odds_chosen": 7.269213676452637, | |
| "log_odds_ratio": -0.5041804909706116, | |
| "logps/chosen": -2.1709227561950684, | |
| "logps/rejected": -9.193258285522461, | |
| "loss": 1.1204, | |
| "nll_loss": 2.00057315826416, | |
| "rewards/accuracies": 0.824999988079071, | |
| "rewards/chosen": -1.0854613780975342, | |
| "rewards/margins": 3.511167526245117, | |
| "rewards/rejected": -4.5966291427612305, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 1.5644444444444443, | |
| "grad_norm": 580.0, | |
| "learning_rate": 2.7355470760292956e-05, | |
| "log_odds_chosen": 6.614538669586182, | |
| "log_odds_ratio": -0.462197482585907, | |
| "logps/chosen": -1.875078797340393, | |
| "logps/rejected": -8.235835075378418, | |
| "loss": -2.3102, | |
| "nll_loss": 1.8360283374786377, | |
| "rewards/accuracies": 0.824999988079071, | |
| "rewards/chosen": -0.9375393986701965, | |
| "rewards/margins": 3.1803784370422363, | |
| "rewards/rejected": -4.117917537689209, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 1.6118518518518519, | |
| "grad_norm": 1064.0, | |
| "learning_rate": 2.597109611334169e-05, | |
| "log_odds_chosen": 6.787781715393066, | |
| "log_odds_ratio": -0.5965229272842407, | |
| "logps/chosen": -2.1229805946350098, | |
| "logps/rejected": -8.671282768249512, | |
| "loss": -2.3052, | |
| "nll_loss": 1.9179661273956299, | |
| "rewards/accuracies": 0.828125, | |
| "rewards/chosen": -1.0614902973175049, | |
| "rewards/margins": 3.27415132522583, | |
| "rewards/rejected": -4.335641384124756, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 1.6592592592592592, | |
| "grad_norm": 872.0, | |
| "learning_rate": 2.458373045823404e-05, | |
| "log_odds_chosen": 6.521327972412109, | |
| "log_odds_ratio": -0.528164267539978, | |
| "logps/chosen": -2.1449532508850098, | |
| "logps/rejected": -8.415470123291016, | |
| "loss": 1.3161, | |
| "nll_loss": 2.0013487339019775, | |
| "rewards/accuracies": 0.809374988079071, | |
| "rewards/chosen": -1.0724766254425049, | |
| "rewards/margins": 3.135258197784424, | |
| "rewards/rejected": -4.207735061645508, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 1.7066666666666666, | |
| "grad_norm": 1256.0, | |
| "learning_rate": 2.3197646927086697e-05, | |
| "log_odds_chosen": 7.416574001312256, | |
| "log_odds_ratio": -0.5437217354774475, | |
| "logps/chosen": -2.012195587158203, | |
| "logps/rejected": -9.163311004638672, | |
| "loss": -6.5787, | |
| "nll_loss": 1.931572675704956, | |
| "rewards/accuracies": 0.815625011920929, | |
| "rewards/chosen": -1.0060977935791016, | |
| "rewards/margins": 3.575557231903076, | |
| "rewards/rejected": -4.581655502319336, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 1.7540740740740741, | |
| "grad_norm": 500.0, | |
| "learning_rate": 2.1817114703032176e-05, | |
| "log_odds_chosen": 7.6875762939453125, | |
| "log_odds_ratio": -0.6660786867141724, | |
| "logps/chosen": -2.3043220043182373, | |
| "logps/rejected": -9.788423538208008, | |
| "loss": -4.8462, | |
| "nll_loss": 1.9335501194000244, | |
| "rewards/accuracies": 0.784375011920929, | |
| "rewards/chosen": -1.1521610021591187, | |
| "rewards/margins": 3.742050886154175, | |
| "rewards/rejected": -4.894211769104004, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 1.8014814814814815, | |
| "grad_norm": 876.0, | |
| "learning_rate": 2.0446385870993467e-05, | |
| "log_odds_chosen": 6.940661430358887, | |
| "log_odds_ratio": -0.4403323233127594, | |
| "logps/chosen": -1.603877067565918, | |
| "logps/rejected": -8.270362854003906, | |
| "loss": -8.6702, | |
| "nll_loss": 1.6499793529510498, | |
| "rewards/accuracies": 0.8343750238418579, | |
| "rewards/chosen": -0.801938533782959, | |
| "rewards/margins": 3.333242893218994, | |
| "rewards/rejected": -4.135181427001953, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 1.8488888888888888, | |
| "grad_norm": 1320.0, | |
| "learning_rate": 1.9089682321121834e-05, | |
| "log_odds_chosen": 9.642197608947754, | |
| "log_odds_ratio": -0.6060672402381897, | |
| "logps/chosen": -2.7960872650146484, | |
| "logps/rejected": -12.217751502990723, | |
| "loss": -4.8162, | |
| "nll_loss": 2.395156145095825, | |
| "rewards/accuracies": 0.8374999761581421, | |
| "rewards/chosen": -1.3980436325073242, | |
| "rewards/margins": 4.710832595825195, | |
| "rewards/rejected": -6.108875751495361, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 1.8962962962962964, | |
| "grad_norm": 1240.0, | |
| "learning_rate": 1.775118274523545e-05, | |
| "log_odds_chosen": 6.639898777008057, | |
| "log_odds_ratio": -0.8482674360275269, | |
| "logps/chosen": -2.435774564743042, | |
| "logps/rejected": -8.842093467712402, | |
| "loss": 3.4566, | |
| "nll_loss": 1.9974777698516846, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": -1.217887282371521, | |
| "rewards/margins": 3.203160047531128, | |
| "rewards/rejected": -4.421046733856201, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.9437037037037037, | |
| "grad_norm": 900.0, | |
| "learning_rate": 1.643500976631037e-05, | |
| "log_odds_chosen": 6.316862106323242, | |
| "log_odds_ratio": -0.5562707781791687, | |
| "logps/chosen": -1.6930224895477295, | |
| "logps/rejected": -7.736858367919922, | |
| "loss": -2.5638, | |
| "nll_loss": 1.7886855602264404, | |
| "rewards/accuracies": 0.8374999761581421, | |
| "rewards/chosen": -0.8465112447738647, | |
| "rewards/margins": 3.0219178199768066, | |
| "rewards/rejected": -3.868429183959961, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 1.991111111111111, | |
| "grad_norm": 940.0, | |
| "learning_rate": 1.514521724066537e-05, | |
| "log_odds_chosen": 6.6161909103393555, | |
| "log_odds_ratio": -0.8282305598258972, | |
| "logps/chosen": -2.5031771659851074, | |
| "logps/rejected": -8.889016151428223, | |
| "loss": 3.2385, | |
| "nll_loss": 1.9650413990020752, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -1.2515885829925537, | |
| "rewards/margins": 3.1929194927215576, | |
| "rewards/rejected": -4.444508075714111, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 2.0385185185185186, | |
| "grad_norm": 652.0, | |
| "learning_rate": 1.3885777771950348e-05, | |
| "log_odds_chosen": 9.489248275756836, | |
| "log_odds_ratio": -0.48209771513938904, | |
| "logps/chosen": -1.8140331506729126, | |
| "logps/rejected": -10.948740005493164, | |
| "loss": -27.9736, | |
| "nll_loss": 1.7284278869628906, | |
| "rewards/accuracies": 0.8374999761581421, | |
| "rewards/chosen": -0.9070165753364563, | |
| "rewards/margins": 4.567353248596191, | |
| "rewards/rejected": -5.474370002746582, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 2.0859259259259257, | |
| "grad_norm": 1088.0, | |
| "learning_rate": 1.2660570475395683e-05, | |
| "log_odds_chosen": 9.268631935119629, | |
| "log_odds_ratio": -0.5644906759262085, | |
| "logps/chosen": -2.2069149017333984, | |
| "logps/rejected": -11.19309139251709, | |
| "loss": -25.8251, | |
| "nll_loss": 1.8661258220672607, | |
| "rewards/accuracies": 0.840624988079071, | |
| "rewards/chosen": -1.1034574508666992, | |
| "rewards/margins": 4.493088722229004, | |
| "rewards/rejected": -5.596545696258545, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 2.1333333333333333, | |
| "grad_norm": 1536.0, | |
| "learning_rate": 1.1473369030008974e-05, | |
| "log_odds_chosen": 13.42963695526123, | |
| "log_odds_ratio": -0.2936347424983978, | |
| "logps/chosen": -2.7818703651428223, | |
| "logps/rejected": -15.93250560760498, | |
| "loss": -38.3857, | |
| "nll_loss": 2.1401987075805664, | |
| "rewards/accuracies": 0.8999999761581421, | |
| "rewards/chosen": -1.3909351825714111, | |
| "rewards/margins": 6.5753173828125, | |
| "rewards/rejected": -7.96625280380249, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 2.180740740740741, | |
| "grad_norm": 1392.0, | |
| "learning_rate": 1.0327830055518842e-05, | |
| "log_odds_chosen": 10.711917877197266, | |
| "log_odds_ratio": -0.31785595417022705, | |
| "logps/chosen": -1.934106469154358, | |
| "logps/rejected": -12.327714920043945, | |
| "loss": -36.8357, | |
| "nll_loss": 1.7895313501358032, | |
| "rewards/accuracies": 0.8968750238418579, | |
| "rewards/chosen": -0.967053234577179, | |
| "rewards/margins": 5.196803092956543, | |
| "rewards/rejected": -6.163857460021973, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 2.228148148148148, | |
| "grad_norm": 668.0, | |
| "learning_rate": 9.227481849865235e-06, | |
| "log_odds_chosen": 10.63416862487793, | |
| "log_odds_ratio": -0.34915637969970703, | |
| "logps/chosen": -1.4962049722671509, | |
| "logps/rejected": -11.764998435974121, | |
| "loss": -36.5253, | |
| "nll_loss": 1.6011587381362915, | |
| "rewards/accuracies": 0.8687499761581421, | |
| "rewards/chosen": -0.7481024861335754, | |
| "rewards/margins": 5.134397029876709, | |
| "rewards/rejected": -5.8824992179870605, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 2.2755555555555556, | |
| "grad_norm": 644.0, | |
| "learning_rate": 8.175713521924978e-06, | |
| "log_odds_chosen": 12.23620891571045, | |
| "log_odds_ratio": -0.1865963637828827, | |
| "logps/chosen": -1.4724217653274536, | |
| "logps/rejected": -13.347505569458008, | |
| "loss": -51.2, | |
| "nll_loss": 1.510587215423584, | |
| "rewards/accuracies": 0.8968750238418579, | |
| "rewards/chosen": -0.7362108826637268, | |
| "rewards/margins": 5.93754243850708, | |
| "rewards/rejected": -6.673752784729004, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 2.322962962962963, | |
| "grad_norm": 1704.0, | |
| "learning_rate": 7.1757645529443665e-06, | |
| "log_odds_chosen": 15.528274536132812, | |
| "log_odds_ratio": -0.33081239461898804, | |
| "logps/chosen": -2.277860641479492, | |
| "logps/rejected": -17.496549606323242, | |
| "loss": -45.7833, | |
| "nll_loss": 2.0120162963867188, | |
| "rewards/accuracies": 0.9156249761581421, | |
| "rewards/chosen": -1.138930320739746, | |
| "rewards/margins": 7.609343528747559, | |
| "rewards/rejected": -8.748274803161621, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 2.3703703703703702, | |
| "grad_norm": 1384.0, | |
| "learning_rate": 6.230714818829733e-06, | |
| "log_odds_chosen": 14.397786140441895, | |
| "log_odds_ratio": -0.28585368394851685, | |
| "logps/chosen": -2.087308645248413, | |
| "logps/rejected": -16.145244598388672, | |
| "loss": -41.6852, | |
| "nll_loss": 1.8240935802459717, | |
| "rewards/accuracies": 0.90625, | |
| "rewards/chosen": -1.0436543226242065, | |
| "rewards/margins": 7.02896785736084, | |
| "rewards/rejected": -8.072622299194336, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 2.417777777777778, | |
| "grad_norm": 960.0, | |
| "learning_rate": 5.343475104027743e-06, | |
| "log_odds_chosen": 13.305925369262695, | |
| "log_odds_ratio": -0.2490675449371338, | |
| "logps/chosen": -1.9038841724395752, | |
| "logps/rejected": -14.872416496276855, | |
| "loss": -47.3843, | |
| "nll_loss": 1.7593896389007568, | |
| "rewards/accuracies": 0.90625, | |
| "rewards/chosen": -0.9519420862197876, | |
| "rewards/margins": 6.484266757965088, | |
| "rewards/rejected": -7.436208248138428, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 2.4651851851851854, | |
| "grad_norm": 1152.0, | |
| "learning_rate": 4.516778136213037e-06, | |
| "log_odds_chosen": 13.447871208190918, | |
| "log_odds_ratio": -0.271032452583313, | |
| "logps/chosen": -1.6334911584854126, | |
| "logps/rejected": -14.739973068237305, | |
| "loss": -45.342, | |
| "nll_loss": 1.6864122152328491, | |
| "rewards/accuracies": 0.893750011920929, | |
| "rewards/chosen": -0.8167455792427063, | |
| "rewards/margins": 6.553240776062012, | |
| "rewards/rejected": -7.369986534118652, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 2.5125925925925925, | |
| "grad_norm": 1080.0, | |
| "learning_rate": 3.7531701693965554e-06, | |
| "log_odds_chosen": 12.9279203414917, | |
| "log_odds_ratio": -0.3936213552951813, | |
| "logps/chosen": -1.8595199584960938, | |
| "logps/rejected": -14.45753288269043, | |
| "loss": -43.4362, | |
| "nll_loss": 1.7575229406356812, | |
| "rewards/accuracies": 0.8843749761581421, | |
| "rewards/chosen": -0.9297599792480469, | |
| "rewards/margins": 6.299006938934326, | |
| "rewards/rejected": -7.228766441345215, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "grad_norm": 1016.0, | |
| "learning_rate": 3.055003141378948e-06, | |
| "log_odds_chosen": 13.242762565612793, | |
| "log_odds_ratio": -0.2418208122253418, | |
| "logps/chosen": -1.5649698972702026, | |
| "logps/rejected": -14.447901725769043, | |
| "loss": -49.7394, | |
| "nll_loss": 1.635581612586975, | |
| "rewards/accuracies": 0.903124988079071, | |
| "rewards/chosen": -0.7824849486351013, | |
| "rewards/margins": 6.441465854644775, | |
| "rewards/rejected": -7.2239508628845215, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 2.6074074074074076, | |
| "grad_norm": 1080.0, | |
| "learning_rate": 2.424427429704365e-06, | |
| "log_odds_chosen": 13.131576538085938, | |
| "log_odds_ratio": -0.24517115950584412, | |
| "logps/chosen": -1.7483772039413452, | |
| "logps/rejected": -14.538126945495605, | |
| "loss": -50.0705, | |
| "nll_loss": 1.7109159231185913, | |
| "rewards/accuracies": 0.918749988079071, | |
| "rewards/chosen": -0.8741886019706726, | |
| "rewards/margins": 6.394874572753906, | |
| "rewards/rejected": -7.269063472747803, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 2.6548148148148147, | |
| "grad_norm": 1320.0, | |
| "learning_rate": 1.8633852284264508e-06, | |
| "log_odds_chosen": 12.429898262023926, | |
| "log_odds_ratio": -0.4006059169769287, | |
| "logps/chosen": -2.0325827598571777, | |
| "logps/rejected": -14.141799926757812, | |
| "loss": -34.9616, | |
| "nll_loss": 1.9175059795379639, | |
| "rewards/accuracies": 0.8656250238418579, | |
| "rewards/chosen": -1.0162913799285889, | |
| "rewards/margins": 6.054608345031738, | |
| "rewards/rejected": -7.070899963378906, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 2.7022222222222223, | |
| "grad_norm": 1280.0, | |
| "learning_rate": 1.3736045660864034e-06, | |
| "log_odds_chosen": 11.897767066955566, | |
| "log_odds_ratio": -0.36061421036720276, | |
| "logps/chosen": -1.7342910766601562, | |
| "logps/rejected": -13.319345474243164, | |
| "loss": -42.368, | |
| "nll_loss": 1.7154544591903687, | |
| "rewards/accuracies": 0.871874988079071, | |
| "rewards/chosen": -0.8671455383300781, | |
| "rewards/margins": 5.792527198791504, | |
| "rewards/rejected": -6.659672737121582, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 2.74962962962963, | |
| "grad_norm": 1472.0, | |
| "learning_rate": 9.565939833279192e-07, | |
| "log_odds_chosen": 12.41030216217041, | |
| "log_odds_ratio": -0.48926201462745667, | |
| "logps/chosen": -2.32023286819458, | |
| "logps/rejected": -14.431295394897461, | |
| "loss": -30.8903, | |
| "nll_loss": 2.0258736610412598, | |
| "rewards/accuracies": 0.8812500238418579, | |
| "rewards/chosen": -1.16011643409729, | |
| "rewards/margins": 6.0555315017700195, | |
| "rewards/rejected": -7.2156476974487305, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 2.797037037037037, | |
| "grad_norm": 920.0, | |
| "learning_rate": 6.136378865420872e-07, | |
| "log_odds_chosen": 12.211552619934082, | |
| "log_odds_ratio": -0.4335884153842926, | |
| "logps/chosen": -1.8138420581817627, | |
| "logps/rejected": -13.713777542114258, | |
| "loss": -37.5638, | |
| "nll_loss": 1.7503633499145508, | |
| "rewards/accuracies": 0.859375, | |
| "rewards/chosen": -0.9069210290908813, | |
| "rewards/margins": 5.949967384338379, | |
| "rewards/rejected": -6.856888771057129, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 2.8444444444444446, | |
| "grad_norm": 1648.0, | |
| "learning_rate": 3.45792591853214e-07, | |
| "log_odds_chosen": 13.25987434387207, | |
| "log_odds_ratio": -0.322501003742218, | |
| "logps/chosen": -1.6583086252212524, | |
| "logps/rejected": -14.573458671569824, | |
| "loss": -45.0139, | |
| "nll_loss": 1.6782853603363037, | |
| "rewards/accuracies": 0.8812500238418579, | |
| "rewards/chosen": -0.8291543126106262, | |
| "rewards/margins": 6.45757532119751, | |
| "rewards/rejected": -7.286729335784912, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 2.891851851851852, | |
| "grad_norm": 1232.0, | |
| "learning_rate": 1.538830716302092e-07, | |
| "log_odds_chosen": 12.726137161254883, | |
| "log_odds_ratio": -0.36502203345298767, | |
| "logps/chosen": -1.7828433513641357, | |
| "logps/rejected": -14.167094230651855, | |
| "loss": -41.4455, | |
| "nll_loss": 1.751901626586914, | |
| "rewards/accuracies": 0.8843749761581421, | |
| "rewards/chosen": -0.8914216756820679, | |
| "rewards/margins": 6.19212532043457, | |
| "rewards/rejected": -7.083547115325928, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 2.9392592592592592, | |
| "grad_norm": 1456.0, | |
| "learning_rate": 3.8500413544415025e-08, | |
| "log_odds_chosen": 13.806699752807617, | |
| "log_odds_ratio": -0.15578700602054596, | |
| "logps/chosen": -1.7442131042480469, | |
| "logps/rejected": -15.208877563476562, | |
| "loss": -47.4243, | |
| "nll_loss": 1.7841014862060547, | |
| "rewards/accuracies": 0.9312499761581421, | |
| "rewards/chosen": -0.8721065521240234, | |
| "rewards/margins": 6.732332706451416, | |
| "rewards/rejected": -7.604438781738281, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 2.986666666666667, | |
| "grad_norm": 1952.0, | |
| "learning_rate": 0.0, | |
| "log_odds_chosen": 12.691495895385742, | |
| "log_odds_ratio": -0.2689394950866699, | |
| "logps/chosen": -1.6713707447052002, | |
| "logps/rejected": -13.993799209594727, | |
| "loss": -43.2258, | |
| "nll_loss": 1.702843427658081, | |
| "rewards/accuracies": 0.8843749761581421, | |
| "rewards/chosen": -0.8356853723526001, | |
| "rewards/margins": 6.1612138748168945, | |
| "rewards/rejected": -6.996899604797363, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 2.986666666666667, | |
| "step": 315, | |
| "total_flos": 0.0, | |
| "train_loss": 25.709435809604706, | |
| "train_runtime": 7268.7111, | |
| "train_samples_per_second": 2.786, | |
| "train_steps_per_second": 0.043 | |
| } | |
| ], | |
| "logging_steps": 5, | |
| "max_steps": 315, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 100000, | |
| "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 | |
| } | |