{ "best_metric": null, "best_model_checkpoint": null, "epoch": 1.0, "eval_steps": 500, "global_step": 478, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0020920502092050207, "grad_norm": 9.179426872640752, "learning_rate": 1.0416666666666666e-08, "logits/chosen": -3.15625, "logits/rejected": -3.21875, "logps/chosen": -194.0, "logps/rejected": -344.0, "loss": 0.6914, "rewards/accuracies": 0.0, "rewards/chosen": 0.0, "rewards/margins": 0.0, "rewards/rejected": 0.0, "step": 1 }, { "epoch": 0.02092050209205021, "grad_norm": 8.141488615674298, "learning_rate": 1.0416666666666667e-07, "logits/chosen": -3.25, "logits/rejected": -3.265625, "logps/chosen": -258.0, "logps/rejected": -284.0, "loss": 0.6922, "rewards/accuracies": 0.1944444477558136, "rewards/chosen": 0.00104522705078125, "rewards/margins": 0.0001544952392578125, "rewards/rejected": 0.000888824462890625, "step": 10 }, { "epoch": 0.04184100418410042, "grad_norm": 9.082683460317224, "learning_rate": 2.0833333333333333e-07, "logits/chosen": -3.15625, "logits/rejected": -3.203125, "logps/chosen": -284.0, "logps/rejected": -306.0, "loss": 0.6907, "rewards/accuracies": 0.4437499940395355, "rewards/chosen": 0.005645751953125, "rewards/margins": 0.00439453125, "rewards/rejected": 0.00127410888671875, "step": 20 }, { "epoch": 0.06276150627615062, "grad_norm": 8.16076946925618, "learning_rate": 3.1249999999999997e-07, "logits/chosen": -3.25, "logits/rejected": -3.25, "logps/chosen": -240.0, "logps/rejected": -280.0, "loss": 0.685, "rewards/accuracies": 0.625, "rewards/chosen": 0.023193359375, "rewards/margins": 0.018310546875, "rewards/rejected": 0.0048828125, "step": 30 }, { "epoch": 0.08368200836820083, "grad_norm": 8.287837872072064, "learning_rate": 4.1666666666666667e-07, "logits/chosen": -3.1875, "logits/rejected": -3.296875, "logps/chosen": -282.0, "logps/rejected": -288.0, "loss": 0.6643, "rewards/accuracies": 0.6937500238418579, "rewards/chosen": 0.0517578125, "rewards/margins": 0.05615234375, "rewards/rejected": -0.0045166015625, "step": 40 }, { "epoch": 0.10460251046025104, "grad_norm": 9.606561637669815, "learning_rate": 4.999733114418725e-07, "logits/chosen": -3.28125, "logits/rejected": -3.328125, "logps/chosen": -262.0, "logps/rejected": -304.0, "loss": 0.6325, "rewards/accuracies": 0.78125, "rewards/chosen": 0.02392578125, "rewards/margins": 0.1435546875, "rewards/rejected": -0.119140625, "step": 50 }, { "epoch": 0.12552301255230125, "grad_norm": 16.63131726373972, "learning_rate": 4.990398100856366e-07, "logits/chosen": -3.390625, "logits/rejected": -3.375, "logps/chosen": -312.0, "logps/rejected": -352.0, "loss": 0.5891, "rewards/accuracies": 0.7562500238418579, "rewards/chosen": -0.1298828125, "rewards/margins": 0.283203125, "rewards/rejected": -0.412109375, "step": 60 }, { "epoch": 0.14644351464435146, "grad_norm": 21.421381161941554, "learning_rate": 4.967775735898179e-07, "logits/chosen": -3.515625, "logits/rejected": -3.46875, "logps/chosen": -330.0, "logps/rejected": -422.0, "loss": 0.4871, "rewards/accuracies": 0.862500011920929, "rewards/chosen": -0.443359375, "rewards/margins": 0.6328125, "rewards/rejected": -1.078125, "step": 70 }, { "epoch": 0.16736401673640167, "grad_norm": 22.981089648710746, "learning_rate": 4.931986719649298e-07, "logits/chosen": -3.5625, "logits/rejected": -3.453125, "logps/chosen": -378.0, "logps/rejected": -524.0, "loss": 0.3838, "rewards/accuracies": 0.8812500238418579, "rewards/chosen": -0.9453125, "rewards/margins": 1.1953125, "rewards/rejected": -2.140625, "step": 80 }, { "epoch": 0.18828451882845187, "grad_norm": 49.65622310516201, "learning_rate": 4.883222001996351e-07, "logits/chosen": -3.5, "logits/rejected": -3.453125, "logps/chosen": -362.0, "logps/rejected": -596.0, "loss": 0.3306, "rewards/accuracies": 0.875, "rewards/chosen": -1.015625, "rewards/margins": 1.984375, "rewards/rejected": -3.0, "step": 90 }, { "epoch": 0.20920502092050208, "grad_norm": 40.70267796836403, "learning_rate": 4.821741763807186e-07, "logits/chosen": -3.5, "logits/rejected": -3.453125, "logps/chosen": -370.0, "logps/rejected": -632.0, "loss": 0.2865, "rewards/accuracies": 0.90625, "rewards/chosen": -1.1640625, "rewards/margins": 2.453125, "rewards/rejected": -3.625, "step": 100 }, { "epoch": 0.2301255230125523, "grad_norm": 21.60396640423782, "learning_rate": 4.747874028753375e-07, "logits/chosen": -3.421875, "logits/rejected": -3.40625, "logps/chosen": -406.0, "logps/rejected": -764.0, "loss": 0.2824, "rewards/accuracies": 0.9125000238418579, "rewards/chosen": -1.421875, "rewards/margins": 3.3125, "rewards/rejected": -4.75, "step": 110 }, { "epoch": 0.2510460251046025, "grad_norm": 30.693758944208717, "learning_rate": 4.662012913161997e-07, "logits/chosen": -3.453125, "logits/rejected": -3.453125, "logps/chosen": -382.0, "logps/rejected": -748.0, "loss": 0.2596, "rewards/accuracies": 0.925000011920929, "rewards/chosen": -1.359375, "rewards/margins": 3.328125, "rewards/rejected": -4.6875, "step": 120 }, { "epoch": 0.2719665271966527, "grad_norm": 38.17865118077698, "learning_rate": 4.5646165232345103e-07, "logits/chosen": -3.453125, "logits/rejected": -3.375, "logps/chosen": -406.0, "logps/rejected": -812.0, "loss": 0.2441, "rewards/accuracies": 0.8999999761581421, "rewards/chosen": -1.484375, "rewards/margins": 3.65625, "rewards/rejected": -5.125, "step": 130 }, { "epoch": 0.2928870292887029, "grad_norm": 37.560083058930886, "learning_rate": 4.456204510851956e-07, "logits/chosen": -3.390625, "logits/rejected": -3.28125, "logps/chosen": -446.0, "logps/rejected": -804.0, "loss": 0.2539, "rewards/accuracies": 0.8500000238418579, "rewards/chosen": -1.71875, "rewards/margins": 3.25, "rewards/rejected": -4.96875, "step": 140 }, { "epoch": 0.3138075313807531, "grad_norm": 19.482065261711334, "learning_rate": 4.337355301007335e-07, "logits/chosen": -3.421875, "logits/rejected": -3.328125, "logps/chosen": -416.0, "logps/rejected": -828.0, "loss": 0.2635, "rewards/accuracies": 0.875, "rewards/chosen": -1.546875, "rewards/margins": 3.828125, "rewards/rejected": -5.375, "step": 150 }, { "epoch": 0.33472803347280333, "grad_norm": 37.99112464719011, "learning_rate": 4.2087030056579986e-07, "logits/chosen": -3.359375, "logits/rejected": -3.28125, "logps/chosen": -390.0, "logps/rejected": -760.0, "loss": 0.2458, "rewards/accuracies": 0.9125000238418579, "rewards/chosen": -1.4765625, "rewards/margins": 3.484375, "rewards/rejected": -4.96875, "step": 160 }, { "epoch": 0.35564853556485354, "grad_norm": 38.3980683297455, "learning_rate": 4.070934040463998e-07, "logits/chosen": -3.25, "logits/rejected": -3.34375, "logps/chosen": -422.0, "logps/rejected": -964.0, "loss": 0.2064, "rewards/accuracies": 0.887499988079071, "rewards/chosen": -1.6640625, "rewards/margins": 4.875, "rewards/rejected": -6.5625, "step": 170 }, { "epoch": 0.37656903765690375, "grad_norm": 23.15841813857398, "learning_rate": 3.9247834624635404e-07, "logits/chosen": -3.3125, "logits/rejected": -3.3125, "logps/chosen": -494.0, "logps/rejected": -1080.0, "loss": 0.177, "rewards/accuracies": 0.925000011920929, "rewards/chosen": -2.078125, "rewards/margins": 5.71875, "rewards/rejected": -7.8125, "step": 180 }, { "epoch": 0.39748953974895396, "grad_norm": 33.22017416138887, "learning_rate": 3.7710310482256523e-07, "logits/chosen": -3.328125, "logits/rejected": -3.296875, "logps/chosen": -494.0, "logps/rejected": -1216.0, "loss": 0.1968, "rewards/accuracies": 0.9125000238418579, "rewards/chosen": -2.46875, "rewards/margins": 6.71875, "rewards/rejected": -9.1875, "step": 190 }, { "epoch": 0.41841004184100417, "grad_norm": 36.59859322366375, "learning_rate": 3.610497133404795e-07, "logits/chosen": -3.203125, "logits/rejected": -3.265625, "logps/chosen": -540.0, "logps/rejected": -1088.0, "loss": 0.1855, "rewards/accuracies": 0.918749988079071, "rewards/chosen": -2.515625, "rewards/margins": 5.4375, "rewards/rejected": -7.96875, "step": 200 }, { "epoch": 0.4393305439330544, "grad_norm": 36.840734409464304, "learning_rate": 3.4440382358952115e-07, "logits/chosen": -3.234375, "logits/rejected": -3.171875, "logps/chosen": -446.0, "logps/rejected": -904.0, "loss": 0.1862, "rewards/accuracies": 0.8812500238418579, "rewards/chosen": -2.015625, "rewards/margins": 4.375, "rewards/rejected": -6.375, "step": 210 }, { "epoch": 0.4602510460251046, "grad_norm": 30.146625845778278, "learning_rate": 3.272542485937368e-07, "logits/chosen": -3.25, "logits/rejected": -3.234375, "logps/chosen": -442.0, "logps/rejected": -948.0, "loss": 0.1748, "rewards/accuracies": 0.9312499761581421, "rewards/chosen": -1.71875, "rewards/margins": 4.96875, "rewards/rejected": -6.6875, "step": 220 }, { "epoch": 0.4811715481171548, "grad_norm": 35.07669431518915, "learning_rate": 3.096924887558854e-07, "logits/chosen": -3.328125, "logits/rejected": -3.25, "logps/chosen": -548.0, "logps/rejected": -1296.0, "loss": 0.1756, "rewards/accuracies": 0.925000011920929, "rewards/chosen": -2.625, "rewards/margins": 6.875, "rewards/rejected": -9.5, "step": 230 }, { "epoch": 0.502092050209205, "grad_norm": 21.539495671811064, "learning_rate": 2.9181224366319943e-07, "logits/chosen": -3.1875, "logits/rejected": -3.0, "logps/chosen": -524.0, "logps/rejected": -1072.0, "loss": 0.1623, "rewards/accuracies": 0.925000011920929, "rewards/chosen": -2.65625, "rewards/margins": 5.5, "rewards/rejected": -8.1875, "step": 240 }, { "epoch": 0.5230125523012552, "grad_norm": 58.51867172243014, "learning_rate": 2.7370891215954565e-07, "logits/chosen": -3.171875, "logits/rejected": -3.078125, "logps/chosen": -476.0, "logps/rejected": -1024.0, "loss": 0.1623, "rewards/accuracies": 0.9312499761581421, "rewards/chosen": -2.203125, "rewards/margins": 5.1875, "rewards/rejected": -7.40625, "step": 250 }, { "epoch": 0.5439330543933054, "grad_norm": 25.48762152166077, "learning_rate": 2.55479083351317e-07, "logits/chosen": -3.1875, "logits/rejected": -3.09375, "logps/chosen": -486.0, "logps/rejected": -1056.0, "loss": 0.1609, "rewards/accuracies": 0.918749988079071, "rewards/chosen": -2.234375, "rewards/margins": 5.40625, "rewards/rejected": -7.625, "step": 260 }, { "epoch": 0.5648535564853556, "grad_norm": 21.73789476591961, "learning_rate": 2.3722002126275822e-07, "logits/chosen": -3.125, "logits/rejected": -3.09375, "logps/chosen": -474.0, "logps/rejected": -1032.0, "loss": 0.1593, "rewards/accuracies": 0.9375, "rewards/chosen": -2.296875, "rewards/margins": 5.40625, "rewards/rejected": -7.71875, "step": 270 }, { "epoch": 0.5857740585774058, "grad_norm": 30.111586407368343, "learning_rate": 2.19029145890313e-07, "logits/chosen": -3.125, "logits/rejected": -3.046875, "logps/chosen": -504.0, "logps/rejected": -1104.0, "loss": 0.1712, "rewards/accuracies": 0.925000011920929, "rewards/chosen": -2.25, "rewards/margins": 5.71875, "rewards/rejected": -7.96875, "step": 280 }, { "epoch": 0.606694560669456, "grad_norm": 14.995982162867495, "learning_rate": 2.0100351342479216e-07, "logits/chosen": -3.09375, "logits/rejected": -3.140625, "logps/chosen": -506.0, "logps/rejected": -1152.0, "loss": 0.1555, "rewards/accuracies": 0.9312499761581421, "rewards/chosen": -2.40625, "rewards/margins": 6.28125, "rewards/rejected": -8.6875, "step": 290 }, { "epoch": 0.6276150627615062, "grad_norm": 23.125573548897137, "learning_rate": 1.8323929841460178e-07, "logits/chosen": -3.15625, "logits/rejected": -3.109375, "logps/chosen": -442.0, "logps/rejected": -972.0, "loss": 0.1522, "rewards/accuracies": 0.96875, "rewards/chosen": -1.9375, "rewards/margins": 5.03125, "rewards/rejected": -6.96875, "step": 300 }, { "epoch": 0.6485355648535565, "grad_norm": 30.186334084740086, "learning_rate": 1.6583128063291573e-07, "logits/chosen": -3.15625, "logits/rejected": -3.125, "logps/chosen": -488.0, "logps/rejected": -1176.0, "loss": 0.1534, "rewards/accuracies": 0.9437500238418579, "rewards/chosen": -2.296875, "rewards/margins": 6.46875, "rewards/rejected": -8.75, "step": 310 }, { "epoch": 0.6694560669456067, "grad_norm": 26.599242178946497, "learning_rate": 1.488723393865766e-07, "logits/chosen": -3.125, "logits/rejected": -3.140625, "logps/chosen": -572.0, "logps/rejected": -1256.0, "loss": 0.1466, "rewards/accuracies": 0.925000011920929, "rewards/chosen": -2.71875, "rewards/margins": 6.8125, "rewards/rejected": -9.5625, "step": 320 }, { "epoch": 0.6903765690376569, "grad_norm": 37.34252084072252, "learning_rate": 1.3245295796480788e-07, "logits/chosen": -3.109375, "logits/rejected": -3.046875, "logps/chosen": -536.0, "logps/rejected": -1088.0, "loss": 0.1493, "rewards/accuracies": 0.949999988079071, "rewards/chosen": -2.5625, "rewards/margins": 5.4375, "rewards/rejected": -8.0, "step": 330 }, { "epoch": 0.7112970711297071, "grad_norm": 16.22820507236929, "learning_rate": 1.1666074087171627e-07, "logits/chosen": -3.125, "logits/rejected": -3.125, "logps/chosen": -496.0, "logps/rejected": -1080.0, "loss": 0.1674, "rewards/accuracies": 0.918749988079071, "rewards/chosen": -2.25, "rewards/margins": 5.5625, "rewards/rejected": -7.8125, "step": 340 }, { "epoch": 0.7322175732217573, "grad_norm": 23.58336941904134, "learning_rate": 1.0157994641835734e-07, "logits/chosen": -3.171875, "logits/rejected": -3.046875, "logps/chosen": -478.0, "logps/rejected": -1016.0, "loss": 0.1654, "rewards/accuracies": 0.918749988079071, "rewards/chosen": -2.390625, "rewards/margins": 5.125, "rewards/rejected": -7.5, "step": 350 }, { "epoch": 0.7531380753138075, "grad_norm": 26.434431463514024, "learning_rate": 8.729103716819111e-08, "logits/chosen": -3.078125, "logits/rejected": -3.015625, "logps/chosen": -508.0, "logps/rejected": -1120.0, "loss": 0.1582, "rewards/accuracies": 0.9375, "rewards/chosen": -2.34375, "rewards/margins": 6.0, "rewards/rejected": -8.3125, "step": 360 }, { "epoch": 0.7740585774058577, "grad_norm": 29.800428401303478, "learning_rate": 7.387025063449081e-08, "logits/chosen": -3.171875, "logits/rejected": -3.109375, "logps/chosen": -448.0, "logps/rejected": -1048.0, "loss": 0.1459, "rewards/accuracies": 0.9375, "rewards/chosen": -2.078125, "rewards/margins": 5.625, "rewards/rejected": -7.6875, "step": 370 }, { "epoch": 0.7949790794979079, "grad_norm": 31.34756549971759, "learning_rate": 6.138919252022435e-08, "logits/chosen": -3.140625, "logits/rejected": -3.09375, "logps/chosen": -508.0, "logps/rejected": -1200.0, "loss": 0.1439, "rewards/accuracies": 0.9437500238418579, "rewards/chosen": -2.515625, "rewards/margins": 6.4375, "rewards/rejected": -8.9375, "step": 380 }, { "epoch": 0.8158995815899581, "grad_norm": 22.257076474784643, "learning_rate": 4.991445467064689e-08, "logits/chosen": -3.140625, "logits/rejected": -3.0625, "logps/chosen": -528.0, "logps/rejected": -1200.0, "loss": 0.1479, "rewards/accuracies": 0.9437500238418579, "rewards/chosen": -2.75, "rewards/margins": 6.1875, "rewards/rejected": -8.9375, "step": 390 }, { "epoch": 0.8368200836820083, "grad_norm": 35.692448570951086, "learning_rate": 3.9507259776993954e-08, "logits/chosen": -3.0625, "logits/rejected": -3.03125, "logps/chosen": -520.0, "logps/rejected": -1128.0, "loss": 0.149, "rewards/accuracies": 0.925000011920929, "rewards/chosen": -2.640625, "rewards/margins": 5.78125, "rewards/rejected": -8.4375, "step": 400 }, { "epoch": 0.8577405857740585, "grad_norm": 32.863561080859476, "learning_rate": 3.022313472693447e-08, "logits/chosen": -3.1875, "logits/rejected": -3.125, "logps/chosen": -500.0, "logps/rejected": -1144.0, "loss": 0.1347, "rewards/accuracies": 0.925000011920929, "rewards/chosen": -2.46875, "rewards/margins": 5.9375, "rewards/rejected": -8.375, "step": 410 }, { "epoch": 0.8786610878661087, "grad_norm": 26.587923562401375, "learning_rate": 2.2111614344599684e-08, "logits/chosen": -3.078125, "logits/rejected": -3.03125, "logps/chosen": -506.0, "logps/rejected": -1184.0, "loss": 0.1351, "rewards/accuracies": 0.9375, "rewards/chosen": -2.4375, "rewards/margins": 6.5, "rewards/rejected": -8.9375, "step": 420 }, { "epoch": 0.899581589958159, "grad_norm": 27.928112295198204, "learning_rate": 1.521597710086439e-08, "logits/chosen": -3.1875, "logits/rejected": -3.109375, "logps/chosen": -494.0, "logps/rejected": -1136.0, "loss": 0.1451, "rewards/accuracies": 0.918749988079071, "rewards/chosen": -2.265625, "rewards/margins": 5.9375, "rewards/rejected": -8.1875, "step": 430 }, { "epoch": 0.9205020920502092, "grad_norm": 26.82028432012561, "learning_rate": 9.57301420397924e-09, "logits/chosen": -3.1875, "logits/rejected": -3.171875, "logps/chosen": -498.0, "logps/rejected": -1144.0, "loss": 0.1447, "rewards/accuracies": 0.9125000238418579, "rewards/chosen": -2.25, "rewards/margins": 6.1875, "rewards/rejected": -8.4375, "step": 440 }, { "epoch": 0.9414225941422594, "grad_norm": 29.41953589986162, "learning_rate": 5.212833302556258e-09, "logits/chosen": -3.09375, "logits/rejected": -3.09375, "logps/chosen": -492.0, "logps/rejected": -1128.0, "loss": 0.1319, "rewards/accuracies": 0.9312499761581421, "rewards/chosen": -2.375, "rewards/margins": 6.0, "rewards/rejected": -8.375, "step": 450 }, { "epoch": 0.9623430962343096, "grad_norm": 29.613565103499333, "learning_rate": 2.158697848236607e-09, "logits/chosen": -3.078125, "logits/rejected": -3.015625, "logps/chosen": -548.0, "logps/rejected": -1192.0, "loss": 0.1455, "rewards/accuracies": 0.9312499761581421, "rewards/chosen": -2.53125, "rewards/margins": 6.1875, "rewards/rejected": -8.75, "step": 460 }, { "epoch": 0.9832635983263598, "grad_norm": 37.33025704068877, "learning_rate": 4.269029751107489e-10, "logits/chosen": -3.1875, "logits/rejected": -3.046875, "logps/chosen": -464.0, "logps/rejected": -1104.0, "loss": 0.1534, "rewards/accuracies": 0.9312499761581421, "rewards/chosen": -2.28125, "rewards/margins": 6.125, "rewards/rejected": -8.375, "step": 470 }, { "epoch": 1.0, "step": 478, "total_flos": 0.0, "train_loss": 0.2521152513795318, "train_runtime": 7015.4228, "train_samples_per_second": 8.714, "train_steps_per_second": 0.068 } ], "logging_steps": 10, "max_steps": 478, "num_input_tokens_seen": 0, "num_train_epochs": 1, "save_steps": 100, "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": 8, "trial_name": null, "trial_params": null }