| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 100, | |
| "global_step": 478, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "grad_norm": 27.516530702491266, | |
| "learning_rate": 1.0416666666666666e-08, | |
| "logits/chosen": -1.6311802864074707, | |
| "logits/rejected": -1.6300675868988037, | |
| "logps/chosen": -327.34149169921875, | |
| "logps/rejected": -429.85833740234375, | |
| "loss": 0.6931, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 25.454136109809017, | |
| "learning_rate": 1.0416666666666667e-07, | |
| "logits/chosen": -1.5537606477737427, | |
| "logits/rejected": -1.506614089012146, | |
| "logps/chosen": -312.7512512207031, | |
| "logps/rejected": -303.0074462890625, | |
| "loss": 0.6929, | |
| "rewards/accuracies": 0.4513888955116272, | |
| "rewards/chosen": 7.373202970484272e-05, | |
| "rewards/margins": -5.077762034488842e-05, | |
| "rewards/rejected": 0.000124509577290155, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 23.434204335413668, | |
| "learning_rate": 2.0833333333333333e-07, | |
| "logits/chosen": -1.6461513042449951, | |
| "logits/rejected": -1.5507893562316895, | |
| "logps/chosen": -329.1868896484375, | |
| "logps/rejected": -351.44891357421875, | |
| "loss": 0.6915, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.01134030893445015, | |
| "rewards/margins": 0.00160890631377697, | |
| "rewards/rejected": 0.009731403551995754, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 23.393322169409338, | |
| "learning_rate": 3.1249999999999997e-07, | |
| "logits/chosen": -1.6500400304794312, | |
| "logits/rejected": -1.5703976154327393, | |
| "logps/chosen": -346.0126953125, | |
| "logps/rejected": -312.0833435058594, | |
| "loss": 0.6837, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": 0.07147429138422012, | |
| "rewards/margins": 0.022869160398840904, | |
| "rewards/rejected": 0.04860512539744377, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 20.10353412052848, | |
| "learning_rate": 4.1666666666666667e-07, | |
| "logits/chosen": -1.5655517578125, | |
| "logits/rejected": -1.5073188543319702, | |
| "logps/chosen": -298.72418212890625, | |
| "logps/rejected": -287.02362060546875, | |
| "loss": 0.67, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": 0.16840842366218567, | |
| "rewards/margins": 0.05939362570643425, | |
| "rewards/rejected": 0.10901479423046112, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 18.756434278176222, | |
| "learning_rate": 4.999733114418725e-07, | |
| "logits/chosen": -1.57842218875885, | |
| "logits/rejected": -1.529872179031372, | |
| "logps/chosen": -299.90679931640625, | |
| "logps/rejected": -323.1398010253906, | |
| "loss": 0.6556, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": 0.2638908326625824, | |
| "rewards/margins": 0.05764131620526314, | |
| "rewards/rejected": 0.20624952018260956, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "grad_norm": 22.017328959957382, | |
| "learning_rate": 4.990398100856366e-07, | |
| "logits/chosen": -1.6750396490097046, | |
| "logits/rejected": -1.6164966821670532, | |
| "logps/chosen": -288.38311767578125, | |
| "logps/rejected": -338.00555419921875, | |
| "loss": 0.6408, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": 0.3388586640357971, | |
| "rewards/margins": 0.11625377833843231, | |
| "rewards/rejected": 0.222604900598526, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "grad_norm": 25.053484688153212, | |
| "learning_rate": 4.967775735898179e-07, | |
| "logits/chosen": -1.492963433265686, | |
| "logits/rejected": -1.4768620729446411, | |
| "logps/chosen": -287.0970764160156, | |
| "logps/rejected": -302.3998718261719, | |
| "loss": 0.6124, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": 0.2931686043739319, | |
| "rewards/margins": 0.22670602798461914, | |
| "rewards/rejected": 0.06646257638931274, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "grad_norm": 35.26897936392246, | |
| "learning_rate": 4.931986719649298e-07, | |
| "logits/chosen": -1.5673052072525024, | |
| "logits/rejected": -1.4728682041168213, | |
| "logps/chosen": -367.7672119140625, | |
| "logps/rejected": -335.300048828125, | |
| "loss": 0.5894, | |
| "rewards/accuracies": 0.65625, | |
| "rewards/chosen": 0.1240416020154953, | |
| "rewards/margins": 0.28898105025291443, | |
| "rewards/rejected": -0.16493944823741913, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "grad_norm": 40.64595674335777, | |
| "learning_rate": 4.883222001996351e-07, | |
| "logits/chosen": -1.1538174152374268, | |
| "logits/rejected": -1.0762898921966553, | |
| "logps/chosen": -330.572265625, | |
| "logps/rejected": -377.3353576660156, | |
| "loss": 0.5545, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -0.18641509115695953, | |
| "rewards/margins": 0.5493143200874329, | |
| "rewards/rejected": -0.7357293963432312, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "grad_norm": 48.97381689745216, | |
| "learning_rate": 4.821741763807186e-07, | |
| "logits/chosen": -0.7731859683990479, | |
| "logits/rejected": -0.7151339054107666, | |
| "logps/chosen": -357.76898193359375, | |
| "logps/rejected": -418.14154052734375, | |
| "loss": 0.539, | |
| "rewards/accuracies": 0.706250011920929, | |
| "rewards/chosen": -0.4816914498806, | |
| "rewards/margins": 0.6462317705154419, | |
| "rewards/rejected": -1.1279233694076538, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "eval_logits/chosen": -0.8811883330345154, | |
| "eval_logits/rejected": -0.7240222096443176, | |
| "eval_logps/chosen": -371.65472412109375, | |
| "eval_logps/rejected": -449.4052734375, | |
| "eval_loss": 0.5451700091362, | |
| "eval_rewards/accuracies": 0.7421875, | |
| "eval_rewards/chosen": -0.5987821817398071, | |
| "eval_rewards/margins": 0.7200188040733337, | |
| "eval_rewards/rejected": -1.3188011646270752, | |
| "eval_runtime": 97.5928, | |
| "eval_samples_per_second": 20.493, | |
| "eval_steps_per_second": 0.328, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "grad_norm": 40.67779253000725, | |
| "learning_rate": 4.747874028753375e-07, | |
| "logits/chosen": -0.9343360066413879, | |
| "logits/rejected": -0.7007087469100952, | |
| "logps/chosen": -409.1253967285156, | |
| "logps/rejected": -429.0912170410156, | |
| "loss": 0.571, | |
| "rewards/accuracies": 0.731249988079071, | |
| "rewards/chosen": -0.4740072786808014, | |
| "rewards/margins": 0.6539417505264282, | |
| "rewards/rejected": -1.1279489994049072, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "grad_norm": 52.47159513458736, | |
| "learning_rate": 4.662012913161997e-07, | |
| "logits/chosen": -0.5735512971878052, | |
| "logits/rejected": -0.4024372696876526, | |
| "logps/chosen": -359.32373046875, | |
| "logps/rejected": -393.42974853515625, | |
| "loss": 0.5452, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.2522381842136383, | |
| "rewards/margins": 0.5496727228164673, | |
| "rewards/rejected": -0.8019108772277832, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "grad_norm": 41.93944182107726, | |
| "learning_rate": 4.5646165232345103e-07, | |
| "logits/chosen": -0.06039993837475777, | |
| "logits/rejected": 0.15637175738811493, | |
| "logps/chosen": -362.6398010253906, | |
| "logps/rejected": -421.77099609375, | |
| "loss": 0.5241, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.3607412576675415, | |
| "rewards/margins": 0.6238666772842407, | |
| "rewards/rejected": -0.9846078753471375, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "grad_norm": 58.39454249885024, | |
| "learning_rate": 4.456204510851956e-07, | |
| "logits/chosen": 0.14134523272514343, | |
| "logits/rejected": 0.4324416220188141, | |
| "logps/chosen": -426.36358642578125, | |
| "logps/rejected": -497.6366271972656, | |
| "loss": 0.5256, | |
| "rewards/accuracies": 0.768750011920929, | |
| "rewards/chosen": -0.5936006903648376, | |
| "rewards/margins": 0.8088639378547668, | |
| "rewards/rejected": -1.4024646282196045, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "grad_norm": 47.848586407040784, | |
| "learning_rate": 4.337355301007335e-07, | |
| "logits/chosen": 0.6952211260795593, | |
| "logits/rejected": 0.8886941075325012, | |
| "logps/chosen": -412.7098083496094, | |
| "logps/rejected": -467.67486572265625, | |
| "loss": 0.5424, | |
| "rewards/accuracies": 0.6937500238418579, | |
| "rewards/chosen": -0.8109712600708008, | |
| "rewards/margins": 0.5735460519790649, | |
| "rewards/rejected": -1.3845174312591553, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "grad_norm": 41.0369792560692, | |
| "learning_rate": 4.2087030056579986e-07, | |
| "logits/chosen": 0.18771931529045105, | |
| "logits/rejected": 0.7211147546768188, | |
| "logps/chosen": -363.5400695800781, | |
| "logps/rejected": -417.8056640625, | |
| "loss": 0.5424, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -0.3788861632347107, | |
| "rewards/margins": 0.7999863028526306, | |
| "rewards/rejected": -1.1788724660873413, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "grad_norm": 40.70764257859768, | |
| "learning_rate": 4.070934040463998e-07, | |
| "logits/chosen": 0.3842553198337555, | |
| "logits/rejected": 0.7558060884475708, | |
| "logps/chosen": -360.5339660644531, | |
| "logps/rejected": -410.0292053222656, | |
| "loss": 0.5243, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.6288506388664246, | |
| "rewards/margins": 0.626385509967804, | |
| "rewards/rejected": -1.2552361488342285, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "grad_norm": 42.009965736488304, | |
| "learning_rate": 3.9247834624635404e-07, | |
| "logits/chosen": 0.8420892953872681, | |
| "logits/rejected": 1.2455089092254639, | |
| "logps/chosen": -370.99090576171875, | |
| "logps/rejected": -420.73358154296875, | |
| "loss": 0.4907, | |
| "rewards/accuracies": 0.731249988079071, | |
| "rewards/chosen": -0.6862328052520752, | |
| "rewards/margins": 0.7635400891304016, | |
| "rewards/rejected": -1.449772834777832, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "grad_norm": 40.59427137175583, | |
| "learning_rate": 3.7710310482256523e-07, | |
| "logits/chosen": 0.35120826959609985, | |
| "logits/rejected": 0.5841476321220398, | |
| "logps/chosen": -369.50762939453125, | |
| "logps/rejected": -431.60430908203125, | |
| "loss": 0.5135, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.45533379912376404, | |
| "rewards/margins": 0.7292474508285522, | |
| "rewards/rejected": -1.1845813989639282, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "grad_norm": 74.46574358755885, | |
| "learning_rate": 3.610497133404795e-07, | |
| "logits/chosen": 0.5723182559013367, | |
| "logits/rejected": 0.8850580453872681, | |
| "logps/chosen": -384.32122802734375, | |
| "logps/rejected": -460.3968200683594, | |
| "loss": 0.5294, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -0.7563287019729614, | |
| "rewards/margins": 0.8355164527893066, | |
| "rewards/rejected": -1.591845154762268, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "eval_logits/chosen": 0.45017001032829285, | |
| "eval_logits/rejected": 0.8080278038978577, | |
| "eval_logps/chosen": -378.31610107421875, | |
| "eval_logps/rejected": -471.284912109375, | |
| "eval_loss": 0.49995097517967224, | |
| "eval_rewards/accuracies": 0.76171875, | |
| "eval_rewards/chosen": -0.6653960347175598, | |
| "eval_rewards/margins": 0.8722014427185059, | |
| "eval_rewards/rejected": -1.537597417831421, | |
| "eval_runtime": 97.5595, | |
| "eval_samples_per_second": 20.5, | |
| "eval_steps_per_second": 0.328, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "grad_norm": 46.031984990497406, | |
| "learning_rate": 3.4440382358952115e-07, | |
| "logits/chosen": 0.48276287317276, | |
| "logits/rejected": 0.8357312083244324, | |
| "logps/chosen": -404.5564880371094, | |
| "logps/rejected": -446.3399963378906, | |
| "loss": 0.5221, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.5977016687393188, | |
| "rewards/margins": 0.8030574917793274, | |
| "rewards/rejected": -1.400759220123291, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "grad_norm": 48.160697637910886, | |
| "learning_rate": 3.272542485937368e-07, | |
| "logits/chosen": 0.5655020475387573, | |
| "logits/rejected": 0.9708032608032227, | |
| "logps/chosen": -374.70733642578125, | |
| "logps/rejected": -441.2498474121094, | |
| "loss": 0.5028, | |
| "rewards/accuracies": 0.7437499761581421, | |
| "rewards/chosen": -0.6240397691726685, | |
| "rewards/margins": 0.8947107195854187, | |
| "rewards/rejected": -1.5187504291534424, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "grad_norm": 48.6209922138632, | |
| "learning_rate": 3.096924887558854e-07, | |
| "logits/chosen": 0.6631933450698853, | |
| "logits/rejected": 1.059995412826538, | |
| "logps/chosen": -387.43206787109375, | |
| "logps/rejected": -495.94976806640625, | |
| "loss": 0.5111, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -0.9049599766731262, | |
| "rewards/margins": 1.0336183309555054, | |
| "rewards/rejected": -1.9385783672332764, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "grad_norm": 50.906502494422604, | |
| "learning_rate": 2.9181224366319943e-07, | |
| "logits/chosen": 0.07225779443979263, | |
| "logits/rejected": 0.43359050154685974, | |
| "logps/chosen": -382.9250183105469, | |
| "logps/rejected": -458.04705810546875, | |
| "loss": 0.4847, | |
| "rewards/accuracies": 0.78125, | |
| "rewards/chosen": -0.6258467435836792, | |
| "rewards/margins": 0.9445594549179077, | |
| "rewards/rejected": -1.570406198501587, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "grad_norm": 51.78937629264524, | |
| "learning_rate": 2.7370891215954565e-07, | |
| "logits/chosen": 0.30628976225852966, | |
| "logits/rejected": 0.6650404930114746, | |
| "logps/chosen": -413.821044921875, | |
| "logps/rejected": -486.84588623046875, | |
| "loss": 0.5071, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": -0.6004015207290649, | |
| "rewards/margins": 1.1020289659500122, | |
| "rewards/rejected": -1.7024303674697876, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "grad_norm": 45.99084165109373, | |
| "learning_rate": 2.55479083351317e-07, | |
| "logits/chosen": 0.2199174165725708, | |
| "logits/rejected": 0.5914181470870972, | |
| "logps/chosen": -416.1441955566406, | |
| "logps/rejected": -481.27020263671875, | |
| "loss": 0.4928, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -0.6798352003097534, | |
| "rewards/margins": 1.0327410697937012, | |
| "rewards/rejected": -1.7125762701034546, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "grad_norm": 44.6265768332988, | |
| "learning_rate": 2.3722002126275822e-07, | |
| "logits/chosen": 0.2704251706600189, | |
| "logits/rejected": 0.6909293532371521, | |
| "logps/chosen": -405.48089599609375, | |
| "logps/rejected": -471.8568420410156, | |
| "loss": 0.5062, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -0.7410668134689331, | |
| "rewards/margins": 0.8480156064033508, | |
| "rewards/rejected": -1.5890824794769287, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "grad_norm": 45.745689932530695, | |
| "learning_rate": 2.19029145890313e-07, | |
| "logits/chosen": 0.5122145414352417, | |
| "logits/rejected": 0.9128357172012329, | |
| "logps/chosen": -386.9629821777344, | |
| "logps/rejected": -474.33697509765625, | |
| "loss": 0.5119, | |
| "rewards/accuracies": 0.78125, | |
| "rewards/chosen": -0.7393491864204407, | |
| "rewards/margins": 1.0442432165145874, | |
| "rewards/rejected": -1.7835925817489624, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "grad_norm": 39.82940708039207, | |
| "learning_rate": 2.0100351342479216e-07, | |
| "logits/chosen": 0.5138941407203674, | |
| "logits/rejected": 0.7629462480545044, | |
| "logps/chosen": -382.9958801269531, | |
| "logps/rejected": -461.4434509277344, | |
| "loss": 0.4913, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -0.8513978719711304, | |
| "rewards/margins": 0.8561092615127563, | |
| "rewards/rejected": -1.7075071334838867, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "grad_norm": 46.49918219263532, | |
| "learning_rate": 1.8323929841460178e-07, | |
| "logits/chosen": 0.3830031752586365, | |
| "logits/rejected": 0.9853779077529907, | |
| "logps/chosen": -448.693359375, | |
| "logps/rejected": -515.8469848632812, | |
| "loss": 0.4704, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -0.9706672430038452, | |
| "rewards/margins": 1.0561556816101074, | |
| "rewards/rejected": -2.026822805404663, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "eval_logits/chosen": 0.3976738452911377, | |
| "eval_logits/rejected": 0.8997932076454163, | |
| "eval_logps/chosen": -396.1895446777344, | |
| "eval_logps/rejected": -516.5240478515625, | |
| "eval_loss": 0.48780035972595215, | |
| "eval_rewards/accuracies": 0.75390625, | |
| "eval_rewards/chosen": -0.8441304564476013, | |
| "eval_rewards/margins": 1.1458582878112793, | |
| "eval_rewards/rejected": -1.9899888038635254, | |
| "eval_runtime": 97.6501, | |
| "eval_samples_per_second": 20.481, | |
| "eval_steps_per_second": 0.328, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "grad_norm": 52.076181120314466, | |
| "learning_rate": 1.6583128063291573e-07, | |
| "logits/chosen": 0.43108898401260376, | |
| "logits/rejected": 0.7893391251564026, | |
| "logps/chosen": -438.8291931152344, | |
| "logps/rejected": -502.6856994628906, | |
| "loss": 0.485, | |
| "rewards/accuracies": 0.7562500238418579, | |
| "rewards/chosen": -0.9125974774360657, | |
| "rewards/margins": 0.9991793632507324, | |
| "rewards/rejected": -1.9117769002914429, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "grad_norm": 53.82916961985562, | |
| "learning_rate": 1.488723393865766e-07, | |
| "logits/chosen": 0.5213574171066284, | |
| "logits/rejected": 1.0497839450836182, | |
| "logps/chosen": -437.28936767578125, | |
| "logps/rejected": -490.869873046875, | |
| "loss": 0.4654, | |
| "rewards/accuracies": 0.7437499761581421, | |
| "rewards/chosen": -0.8886297941207886, | |
| "rewards/margins": 1.0424896478652954, | |
| "rewards/rejected": -1.9311193227767944, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "grad_norm": 52.0979391510305, | |
| "learning_rate": 1.3245295796480788e-07, | |
| "logits/chosen": 0.27822378277778625, | |
| "logits/rejected": 0.6966886520385742, | |
| "logps/chosen": -394.94219970703125, | |
| "logps/rejected": -488.16571044921875, | |
| "loss": 0.4721, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.8261836767196655, | |
| "rewards/margins": 0.8932548761367798, | |
| "rewards/rejected": -1.7194385528564453, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "grad_norm": 53.006317641552194, | |
| "learning_rate": 1.1666074087171627e-07, | |
| "logits/chosen": 0.24821150302886963, | |
| "logits/rejected": 0.725691556930542, | |
| "logps/chosen": -424.24761962890625, | |
| "logps/rejected": -520.8505249023438, | |
| "loss": 0.485, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.772143542766571, | |
| "rewards/margins": 1.1264902353286743, | |
| "rewards/rejected": -1.8986337184906006, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "grad_norm": 50.07894063945848, | |
| "learning_rate": 1.0157994641835734e-07, | |
| "logits/chosen": 0.5090032815933228, | |
| "logits/rejected": 1.0217373371124268, | |
| "logps/chosen": -399.2255859375, | |
| "logps/rejected": -497.14910888671875, | |
| "loss": 0.4596, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": -0.9144848585128784, | |
| "rewards/margins": 1.1523492336273193, | |
| "rewards/rejected": -2.0668342113494873, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "grad_norm": 46.12071728187378, | |
| "learning_rate": 8.729103716819111e-08, | |
| "logits/chosen": 0.29714930057525635, | |
| "logits/rejected": 0.8836923837661743, | |
| "logps/chosen": -449.24822998046875, | |
| "logps/rejected": -515.9562377929688, | |
| "loss": 0.507, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -0.9280809164047241, | |
| "rewards/margins": 1.0685356855392456, | |
| "rewards/rejected": -1.9966167211532593, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "grad_norm": 44.466114694620565, | |
| "learning_rate": 7.387025063449081e-08, | |
| "logits/chosen": 0.3406144082546234, | |
| "logits/rejected": 0.8408617973327637, | |
| "logps/chosen": -423.6982421875, | |
| "logps/rejected": -472.16302490234375, | |
| "loss": 0.4937, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -0.973863959312439, | |
| "rewards/margins": 0.8518773317337036, | |
| "rewards/rejected": -1.825741171836853, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "grad_norm": 40.2546545345207, | |
| "learning_rate": 6.138919252022435e-08, | |
| "logits/chosen": 0.5687306523323059, | |
| "logits/rejected": 0.8843731880187988, | |
| "logps/chosen": -388.6229248046875, | |
| "logps/rejected": -532.6536865234375, | |
| "loss": 0.4777, | |
| "rewards/accuracies": 0.793749988079071, | |
| "rewards/chosen": -1.018700361251831, | |
| "rewards/margins": 1.1921350955963135, | |
| "rewards/rejected": -2.2108356952667236, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "grad_norm": 50.26373841675365, | |
| "learning_rate": 4.991445467064689e-08, | |
| "logits/chosen": 0.3668118119239807, | |
| "logits/rejected": 0.6758331060409546, | |
| "logps/chosen": -445.70294189453125, | |
| "logps/rejected": -524.3595581054688, | |
| "loss": 0.4756, | |
| "rewards/accuracies": 0.6937500238418579, | |
| "rewards/chosen": -0.906339168548584, | |
| "rewards/margins": 0.9413349032402039, | |
| "rewards/rejected": -1.8476741313934326, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "grad_norm": 45.03150381268163, | |
| "learning_rate": 3.9507259776993954e-08, | |
| "logits/chosen": 0.47453969717025757, | |
| "logits/rejected": 0.8979718089103699, | |
| "logps/chosen": -417.6993713378906, | |
| "logps/rejected": -514.3675537109375, | |
| "loss": 0.4856, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -0.9480649828910828, | |
| "rewards/margins": 1.0335150957107544, | |
| "rewards/rejected": -1.981580138206482, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "eval_logits/chosen": 0.22729139029979706, | |
| "eval_logits/rejected": 0.7285040616989136, | |
| "eval_logps/chosen": -388.4799499511719, | |
| "eval_logps/rejected": -508.7244873046875, | |
| "eval_loss": 0.4800879657268524, | |
| "eval_rewards/accuracies": 0.76171875, | |
| "eval_rewards/chosen": -0.7670345306396484, | |
| "eval_rewards/margins": 1.1449588537216187, | |
| "eval_rewards/rejected": -1.911993384361267, | |
| "eval_runtime": 97.4275, | |
| "eval_samples_per_second": 20.528, | |
| "eval_steps_per_second": 0.328, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "grad_norm": 43.3283841726019, | |
| "learning_rate": 3.022313472693447e-08, | |
| "logits/chosen": 0.2033652514219284, | |
| "logits/rejected": 0.8105279803276062, | |
| "logps/chosen": -432.13153076171875, | |
| "logps/rejected": -513.7069702148438, | |
| "loss": 0.4936, | |
| "rewards/accuracies": 0.731249988079071, | |
| "rewards/chosen": -0.8029797673225403, | |
| "rewards/margins": 1.0931544303894043, | |
| "rewards/rejected": -1.8961341381072998, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "grad_norm": 43.3329326100997, | |
| "learning_rate": 2.2111614344599684e-08, | |
| "logits/chosen": 0.23255594074726105, | |
| "logits/rejected": 0.6442626118659973, | |
| "logps/chosen": -432.7608337402344, | |
| "logps/rejected": -520.2620239257812, | |
| "loss": 0.4641, | |
| "rewards/accuracies": 0.7437499761581421, | |
| "rewards/chosen": -0.8288372159004211, | |
| "rewards/margins": 1.0977327823638916, | |
| "rewards/rejected": -1.926569938659668, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "grad_norm": 49.02509873372083, | |
| "learning_rate": 1.521597710086439e-08, | |
| "logits/chosen": 0.39820823073387146, | |
| "logits/rejected": 0.7836225628852844, | |
| "logps/chosen": -431.13739013671875, | |
| "logps/rejected": -512.7986450195312, | |
| "loss": 0.471, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": -0.9906813502311707, | |
| "rewards/margins": 1.0531766414642334, | |
| "rewards/rejected": -2.043858051300049, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "grad_norm": 46.801343844231866, | |
| "learning_rate": 9.57301420397924e-09, | |
| "logits/chosen": 0.2471725195646286, | |
| "logits/rejected": 0.7467008829116821, | |
| "logps/chosen": -417.58258056640625, | |
| "logps/rejected": -499.89215087890625, | |
| "loss": 0.4942, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.8245165944099426, | |
| "rewards/margins": 0.9129160642623901, | |
| "rewards/rejected": -1.737432837486267, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "grad_norm": 46.37037120468359, | |
| "learning_rate": 5.212833302556258e-09, | |
| "logits/chosen": 0.4000117778778076, | |
| "logits/rejected": 0.6783515810966492, | |
| "logps/chosen": -436.942138671875, | |
| "logps/rejected": -551.7266845703125, | |
| "loss": 0.4786, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.9379842877388, | |
| "rewards/margins": 0.9772690534591675, | |
| "rewards/rejected": -1.9152534008026123, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "grad_norm": 46.5176855624682, | |
| "learning_rate": 2.158697848236607e-09, | |
| "logits/chosen": 0.3254472017288208, | |
| "logits/rejected": 0.7334408164024353, | |
| "logps/chosen": -410.4520568847656, | |
| "logps/rejected": -479.54351806640625, | |
| "loss": 0.4804, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -0.8168198466300964, | |
| "rewards/margins": 0.9972711801528931, | |
| "rewards/rejected": -1.8140909671783447, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "grad_norm": 42.92566705814915, | |
| "learning_rate": 4.269029751107489e-10, | |
| "logits/chosen": 0.2615105211734772, | |
| "logits/rejected": 0.6600261926651001, | |
| "logps/chosen": -408.7879638671875, | |
| "logps/rejected": -514.4189453125, | |
| "loss": 0.4762, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.8471711874008179, | |
| "rewards/margins": 1.000536322593689, | |
| "rewards/rejected": -1.8477073907852173, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 478, | |
| "total_flos": 0.0, | |
| "train_loss": 0.5266154430900158, | |
| "train_runtime": 7640.9818, | |
| "train_samples_per_second": 8.001, | |
| "train_steps_per_second": 0.063 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 478, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 100, | |
| "total_flos": 0.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |