| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 50, | |
| "global_step": 436, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.022935779816513763, | |
| "grad_norm": 1.642045632140662, | |
| "learning_rate": 1.1363636363636363e-07, | |
| "logits/chosen": -2.6192917823791504, | |
| "logits/rejected": -2.5524227619171143, | |
| "logps/chosen": -265.41119384765625, | |
| "logps/rejected": -236.11862182617188, | |
| "loss": 0.0154, | |
| "rewards/accuracies": 0.4000000059604645, | |
| "rewards/chosen": 0.00034645519917830825, | |
| "rewards/margins": 0.00023277592845261097, | |
| "rewards/rejected": 0.00011367930710548535, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.045871559633027525, | |
| "grad_norm": 1.5167637902207143, | |
| "learning_rate": 2.2727272727272726e-07, | |
| "logits/chosen": -2.657719135284424, | |
| "logits/rejected": -2.5759785175323486, | |
| "logps/chosen": -298.7945556640625, | |
| "logps/rejected": -274.304443359375, | |
| "loss": 0.0155, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -9.255408076569438e-05, | |
| "rewards/margins": 0.0018523468170315027, | |
| "rewards/rejected": -0.0019449004903435707, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.06880733944954129, | |
| "grad_norm": 1.3745597654790793, | |
| "learning_rate": 3.4090909090909085e-07, | |
| "logits/chosen": -2.6762423515319824, | |
| "logits/rejected": -2.6026246547698975, | |
| "logps/chosen": -290.37896728515625, | |
| "logps/rejected": -234.3507080078125, | |
| "loss": 0.0146, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": 0.009538007900118828, | |
| "rewards/margins": 0.013226142153143883, | |
| "rewards/rejected": -0.0036881337873637676, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.09174311926605505, | |
| "grad_norm": 1.272192316985563, | |
| "learning_rate": 4.545454545454545e-07, | |
| "logits/chosen": -2.660547971725464, | |
| "logits/rejected": -2.6108529567718506, | |
| "logps/chosen": -280.96484375, | |
| "logps/rejected": -267.6105041503906, | |
| "loss": 0.013, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": 0.04204345494508743, | |
| "rewards/margins": 0.04021826013922691, | |
| "rewards/rejected": 0.0018251972505822778, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.11467889908256881, | |
| "grad_norm": 1.4079236984792325, | |
| "learning_rate": 4.997110275491701e-07, | |
| "logits/chosen": -2.6261672973632812, | |
| "logits/rejected": -2.6208655834198, | |
| "logps/chosen": -289.76519775390625, | |
| "logps/rejected": -299.06353759765625, | |
| "loss": 0.012, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": 0.06486029922962189, | |
| "rewards/margins": 0.06365373730659485, | |
| "rewards/rejected": 0.0012065758928656578, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.11467889908256881, | |
| "eval_logits/chosen": -2.5786819458007812, | |
| "eval_logits/rejected": -2.502084732055664, | |
| "eval_logps/chosen": -277.3097839355469, | |
| "eval_logps/rejected": -247.28094482421875, | |
| "eval_loss": 0.011314952746033669, | |
| "eval_rewards/accuracies": 0.7025862336158752, | |
| "eval_rewards/chosen": 0.07780314981937408, | |
| "eval_rewards/margins": 0.08538833260536194, | |
| "eval_rewards/rejected": -0.007585177198052406, | |
| "eval_runtime": 94.9097, | |
| "eval_samples_per_second": 19.155, | |
| "eval_steps_per_second": 0.306, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.13761467889908258, | |
| "grad_norm": 1.418614516322105, | |
| "learning_rate": 4.979475034558115e-07, | |
| "logits/chosen": -2.5812747478485107, | |
| "logits/rejected": -2.5265355110168457, | |
| "logps/chosen": -284.41473388671875, | |
| "logps/rejected": -259.66094970703125, | |
| "loss": 0.0112, | |
| "rewards/accuracies": 0.6937500238418579, | |
| "rewards/chosen": 0.05069952458143234, | |
| "rewards/margins": 0.0717436671257019, | |
| "rewards/rejected": -0.021044140681624413, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.16055045871559634, | |
| "grad_norm": 1.4648752438020702, | |
| "learning_rate": 4.945923025551788e-07, | |
| "logits/chosen": -2.5097055435180664, | |
| "logits/rejected": -2.4729437828063965, | |
| "logps/chosen": -319.59161376953125, | |
| "logps/rejected": -265.26019287109375, | |
| "loss": 0.0115, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": 0.07979384809732437, | |
| "rewards/margins": 0.1183195561170578, | |
| "rewards/rejected": -0.03852573037147522, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.1834862385321101, | |
| "grad_norm": 1.379325651626498, | |
| "learning_rate": 4.896669632591651e-07, | |
| "logits/chosen": -2.4763333797454834, | |
| "logits/rejected": -2.385080099105835, | |
| "logps/chosen": -278.93988037109375, | |
| "logps/rejected": -255.78781127929688, | |
| "loss": 0.011, | |
| "rewards/accuracies": 0.706250011920929, | |
| "rewards/chosen": 0.06909944862127304, | |
| "rewards/margins": 0.10041693598031998, | |
| "rewards/rejected": -0.031317487359046936, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.20642201834862386, | |
| "grad_norm": 1.3303176283339873, | |
| "learning_rate": 4.832031033425662e-07, | |
| "logits/chosen": -2.423882246017456, | |
| "logits/rejected": -2.376399278640747, | |
| "logps/chosen": -280.73907470703125, | |
| "logps/rejected": -252.6663360595703, | |
| "loss": 0.0103, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": 0.06854326277971268, | |
| "rewards/margins": 0.12719407677650452, | |
| "rewards/rejected": -0.058650821447372437, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.22935779816513763, | |
| "grad_norm": 1.4785994754129348, | |
| "learning_rate": 4.752422169756047e-07, | |
| "logits/chosen": -2.3845717906951904, | |
| "logits/rejected": -2.3276596069335938, | |
| "logps/chosen": -268.87396240234375, | |
| "logps/rejected": -280.0634460449219, | |
| "loss": 0.011, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": 0.035507336258888245, | |
| "rewards/margins": 0.11685723066329956, | |
| "rewards/rejected": -0.08134988695383072, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.22935779816513763, | |
| "eval_logits/chosen": -2.4392149448394775, | |
| "eval_logits/rejected": -2.3370442390441895, | |
| "eval_logps/chosen": -280.90240478515625, | |
| "eval_logps/rejected": -254.33267211914062, | |
| "eval_loss": 0.009970244951546192, | |
| "eval_rewards/accuracies": 0.7068965435028076, | |
| "eval_rewards/chosen": 0.04187687486410141, | |
| "eval_rewards/margins": 0.1199791207909584, | |
| "eval_rewards/rejected": -0.078102245926857, | |
| "eval_runtime": 95.3397, | |
| "eval_samples_per_second": 19.069, | |
| "eval_steps_per_second": 0.304, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.25229357798165136, | |
| "grad_norm": 2.0224189698687702, | |
| "learning_rate": 4.658354083558188e-07, | |
| "logits/chosen": -2.4632654190063477, | |
| "logits/rejected": -2.354820966720581, | |
| "logps/chosen": -263.564453125, | |
| "logps/rejected": -241.26791381835938, | |
| "loss": 0.0113, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": 0.01623677834868431, | |
| "rewards/margins": 0.09158362448215485, | |
| "rewards/rejected": -0.07534684240818024, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.27522935779816515, | |
| "grad_norm": 1.4135396422800603, | |
| "learning_rate": 4.550430636492389e-07, | |
| "logits/chosen": -2.382753610610962, | |
| "logits/rejected": -2.350785970687866, | |
| "logps/chosen": -275.2984619140625, | |
| "logps/rejected": -260.0333557128906, | |
| "loss": 0.0102, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": 0.005208671558648348, | |
| "rewards/margins": 0.0909029170870781, | |
| "rewards/rejected": -0.08569424599409103, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.2981651376146789, | |
| "grad_norm": 2.470336886918338, | |
| "learning_rate": 4.429344633468004e-07, | |
| "logits/chosen": -2.404059648513794, | |
| "logits/rejected": -2.3569350242614746, | |
| "logps/chosen": -254.55068969726562, | |
| "logps/rejected": -252.3956756591797, | |
| "loss": 0.0105, | |
| "rewards/accuracies": 0.6937500238418579, | |
| "rewards/chosen": 0.016294140368700027, | |
| "rewards/margins": 0.11412493139505386, | |
| "rewards/rejected": -0.09783079475164413, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.3211009174311927, | |
| "grad_norm": 3.815944054063513, | |
| "learning_rate": 4.2958733752443187e-07, | |
| "logits/chosen": -2.4113519191741943, | |
| "logits/rejected": -2.340986728668213, | |
| "logps/chosen": -272.99420166015625, | |
| "logps/rejected": -231.37026977539062, | |
| "loss": 0.0123, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": 0.0319095216691494, | |
| "rewards/margins": 0.10264714062213898, | |
| "rewards/rejected": -0.07073761522769928, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.3440366972477064, | |
| "grad_norm": 1.3170543399714232, | |
| "learning_rate": 4.150873668617898e-07, | |
| "logits/chosen": -2.454432249069214, | |
| "logits/rejected": -2.370666265487671, | |
| "logps/chosen": -268.79541015625, | |
| "logps/rejected": -247.0717315673828, | |
| "loss": 0.0104, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -0.022518452256917953, | |
| "rewards/margins": 0.11632315069437027, | |
| "rewards/rejected": -0.13884159922599792, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.3440366972477064, | |
| "eval_logits/chosen": -2.494401454925537, | |
| "eval_logits/rejected": -2.39561128616333, | |
| "eval_logps/chosen": -285.85308837890625, | |
| "eval_logps/rejected": -260.5493469238281, | |
| "eval_loss": 0.009796149097383022, | |
| "eval_rewards/accuracies": 0.7198275923728943, | |
| "eval_rewards/chosen": -0.007629875559359789, | |
| "eval_rewards/margins": 0.13263897597789764, | |
| "eval_rewards/rejected": -0.14026884734630585, | |
| "eval_runtime": 95.8939, | |
| "eval_samples_per_second": 18.958, | |
| "eval_steps_per_second": 0.302, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.3669724770642202, | |
| "grad_norm": 2.6585100320508737, | |
| "learning_rate": 3.9952763262280397e-07, | |
| "logits/chosen": -2.4578630924224854, | |
| "logits/rejected": -2.3927180767059326, | |
| "logps/chosen": -293.46124267578125, | |
| "logps/rejected": -295.8601379394531, | |
| "loss": 0.01, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": 0.030898962169885635, | |
| "rewards/margins": 0.12981417775154114, | |
| "rewards/rejected": -0.0989152044057846, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.38990825688073394, | |
| "grad_norm": 1.3118055868117948, | |
| "learning_rate": 3.8300801912883414e-07, | |
| "logits/chosen": -2.504678249359131, | |
| "logits/rejected": -2.4544835090637207, | |
| "logps/chosen": -267.3302307128906, | |
| "logps/rejected": -279.33062744140625, | |
| "loss": 0.0091, | |
| "rewards/accuracies": 0.7562500238418579, | |
| "rewards/chosen": 0.04270657151937485, | |
| "rewards/margins": 0.13532397150993347, | |
| "rewards/rejected": -0.09261739999055862, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.41284403669724773, | |
| "grad_norm": 1.4248945431420277, | |
| "learning_rate": 3.6563457256020884e-07, | |
| "logits/chosen": -2.509859561920166, | |
| "logits/rejected": -2.4090194702148438, | |
| "logps/chosen": -306.2326965332031, | |
| "logps/rejected": -253.7351531982422, | |
| "loss": 0.0096, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.028747806325554848, | |
| "rewards/margins": 0.123654805123806, | |
| "rewards/rejected": -0.1524026244878769, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.43577981651376146, | |
| "grad_norm": 4.61575651504461, | |
| "learning_rate": 3.475188202022617e-07, | |
| "logits/chosen": -2.4155473709106445, | |
| "logits/rejected": -2.3843960762023926, | |
| "logps/chosen": -256.64117431640625, | |
| "logps/rejected": -276.67425537109375, | |
| "loss": 0.0105, | |
| "rewards/accuracies": 0.78125, | |
| "rewards/chosen": 0.014733311720192432, | |
| "rewards/margins": 0.13803140819072723, | |
| "rewards/rejected": -0.12329809367656708, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.45871559633027525, | |
| "grad_norm": 1.2837860780603194, | |
| "learning_rate": 3.287770545059052e-07, | |
| "logits/chosen": -2.529806613922119, | |
| "logits/rejected": -2.4307055473327637, | |
| "logps/chosen": -277.78741455078125, | |
| "logps/rejected": -256.05511474609375, | |
| "loss": 0.0096, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": 0.004942134954035282, | |
| "rewards/margins": 0.12693332135677338, | |
| "rewards/rejected": -0.12199117988348007, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.45871559633027525, | |
| "eval_logits/chosen": -2.514023542404175, | |
| "eval_logits/rejected": -2.4208788871765137, | |
| "eval_logps/chosen": -282.1954345703125, | |
| "eval_logps/rejected": -257.61944580078125, | |
| "eval_loss": 0.009310290217399597, | |
| "eval_rewards/accuracies": 0.7931034564971924, | |
| "eval_rewards/chosen": 0.028946416452527046, | |
| "eval_rewards/margins": 0.13991650938987732, | |
| "eval_rewards/rejected": -0.11097008734941483, | |
| "eval_runtime": 96.1764, | |
| "eval_samples_per_second": 18.903, | |
| "eval_steps_per_second": 0.302, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.481651376146789, | |
| "grad_norm": 1.090616369349917, | |
| "learning_rate": 3.0952958655864954e-07, | |
| "logits/chosen": -2.4676451683044434, | |
| "logits/rejected": -2.4505865573883057, | |
| "logps/chosen": -270.9748840332031, | |
| "logps/rejected": -265.1389465332031, | |
| "loss": 0.0094, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": 0.023295128718018532, | |
| "rewards/margins": 0.13126251101493835, | |
| "rewards/rejected": -0.10796739161014557, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.5045871559633027, | |
| "grad_norm": 1.493461344772632, | |
| "learning_rate": 2.898999737583448e-07, | |
| "logits/chosen": -2.504462957382202, | |
| "logits/rejected": -2.404648542404175, | |
| "logps/chosen": -322.62689208984375, | |
| "logps/rejected": -300.34686279296875, | |
| "loss": 0.0091, | |
| "rewards/accuracies": 0.78125, | |
| "rewards/chosen": 0.00053420226322487, | |
| "rewards/margins": 0.15430037677288055, | |
| "rewards/rejected": -0.1537661850452423, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.5275229357798165, | |
| "grad_norm": 1.2922195153842637, | |
| "learning_rate": 2.7001422664752333e-07, | |
| "logits/chosen": -2.3982276916503906, | |
| "logits/rejected": -2.3454272747039795, | |
| "logps/chosen": -269.4321594238281, | |
| "logps/rejected": -279.17926025390625, | |
| "loss": 0.0093, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -0.014886477962136269, | |
| "rewards/margins": 0.12070702016353607, | |
| "rewards/rejected": -0.1355935037136078, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.5504587155963303, | |
| "grad_norm": 1.3338425925379636, | |
| "learning_rate": 2.5e-07, | |
| "logits/chosen": -2.4747557640075684, | |
| "logits/rejected": -2.3822951316833496, | |
| "logps/chosen": -288.63037109375, | |
| "logps/rejected": -274.5633850097656, | |
| "loss": 0.0095, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -0.010984973981976509, | |
| "rewards/margins": 0.12114904075860977, | |
| "rewards/rejected": -0.13213400542736053, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.573394495412844, | |
| "grad_norm": 1.161806057366623, | |
| "learning_rate": 2.2998577335247667e-07, | |
| "logits/chosen": -2.513209819793701, | |
| "logits/rejected": -2.424811363220215, | |
| "logps/chosen": -303.5930480957031, | |
| "logps/rejected": -273.9664611816406, | |
| "loss": 0.0094, | |
| "rewards/accuracies": 0.7562500238418579, | |
| "rewards/chosen": 0.002007619244977832, | |
| "rewards/margins": 0.13723386824131012, | |
| "rewards/rejected": -0.1352262645959854, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.573394495412844, | |
| "eval_logits/chosen": -2.4590651988983154, | |
| "eval_logits/rejected": -2.362999200820923, | |
| "eval_logps/chosen": -286.21722412109375, | |
| "eval_logps/rejected": -263.2693786621094, | |
| "eval_loss": 0.008912510238587856, | |
| "eval_rewards/accuracies": 0.7801724076271057, | |
| "eval_rewards/chosen": -0.011271164752542973, | |
| "eval_rewards/margins": 0.15619821846485138, | |
| "eval_rewards/rejected": -0.16746938228607178, | |
| "eval_runtime": 94.78, | |
| "eval_samples_per_second": 19.181, | |
| "eval_steps_per_second": 0.306, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.5963302752293578, | |
| "grad_norm": 1.1995883831745127, | |
| "learning_rate": 2.1010002624165524e-07, | |
| "logits/chosen": -2.4454376697540283, | |
| "logits/rejected": -2.407090902328491, | |
| "logps/chosen": -273.8803405761719, | |
| "logps/rejected": -292.88421630859375, | |
| "loss": 0.0097, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": -0.03165815398097038, | |
| "rewards/margins": 0.15324734151363373, | |
| "rewards/rejected": -0.1849054992198944, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.6192660550458715, | |
| "grad_norm": 1.236047360327654, | |
| "learning_rate": 1.9047041344135043e-07, | |
| "logits/chosen": -2.4361116886138916, | |
| "logits/rejected": -2.4080004692077637, | |
| "logps/chosen": -270.04998779296875, | |
| "logps/rejected": -270.3052978515625, | |
| "loss": 0.0091, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -0.03987707942724228, | |
| "rewards/margins": 0.14470525085926056, | |
| "rewards/rejected": -0.18458232283592224, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.6422018348623854, | |
| "grad_norm": 1.3000373431490932, | |
| "learning_rate": 1.7122294549409482e-07, | |
| "logits/chosen": -2.512075424194336, | |
| "logits/rejected": -2.448713779449463, | |
| "logps/chosen": -273.0007019042969, | |
| "logps/rejected": -280.35565185546875, | |
| "loss": 0.0098, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -1.8634367734193802e-05, | |
| "rewards/margins": 0.16557954251766205, | |
| "rewards/rejected": -0.16559818387031555, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.6651376146788991, | |
| "grad_norm": 1.354856941159808, | |
| "learning_rate": 1.524811797977383e-07, | |
| "logits/chosen": -2.493873119354248, | |
| "logits/rejected": -2.416597843170166, | |
| "logps/chosen": -287.20404052734375, | |
| "logps/rejected": -267.6150817871094, | |
| "loss": 0.009, | |
| "rewards/accuracies": 0.768750011920929, | |
| "rewards/chosen": 0.005267986096441746, | |
| "rewards/margins": 0.1422511339187622, | |
| "rewards/rejected": -0.13698314130306244, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.6880733944954128, | |
| "grad_norm": 1.3081620144699164, | |
| "learning_rate": 1.3436542743979125e-07, | |
| "logits/chosen": -2.4964632987976074, | |
| "logits/rejected": -2.472830057144165, | |
| "logps/chosen": -307.45733642578125, | |
| "logps/rejected": -271.1383361816406, | |
| "loss": 0.0096, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.015080097131431103, | |
| "rewards/margins": 0.10770467668771744, | |
| "rewards/rejected": -0.12278477847576141, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.6880733944954128, | |
| "eval_logits/chosen": -2.477182388305664, | |
| "eval_logits/rejected": -2.382119655609131, | |
| "eval_logps/chosen": -286.4156188964844, | |
| "eval_logps/rejected": -262.90252685546875, | |
| "eval_loss": 0.008760624565184116, | |
| "eval_rewards/accuracies": 0.7844827771186829, | |
| "eval_rewards/chosen": -0.013255205936729908, | |
| "eval_rewards/margins": 0.15054550766944885, | |
| "eval_rewards/rejected": -0.16380071640014648, | |
| "eval_runtime": 95.1377, | |
| "eval_samples_per_second": 19.109, | |
| "eval_steps_per_second": 0.305, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.7110091743119266, | |
| "grad_norm": 1.3153487126066306, | |
| "learning_rate": 1.1699198087116588e-07, | |
| "logits/chosen": -2.5111746788024902, | |
| "logits/rejected": -2.422899007797241, | |
| "logps/chosen": -280.66143798828125, | |
| "logps/rejected": -280.439697265625, | |
| "loss": 0.0094, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.026962900534272194, | |
| "rewards/margins": 0.11293704807758331, | |
| "rewards/rejected": -0.13989993929862976, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.7339449541284404, | |
| "grad_norm": 1.337124035482138, | |
| "learning_rate": 1.00472367377196e-07, | |
| "logits/chosen": -2.4476630687713623, | |
| "logits/rejected": -2.397982120513916, | |
| "logps/chosen": -275.42584228515625, | |
| "logps/rejected": -251.29122924804688, | |
| "loss": 0.0093, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.011938780546188354, | |
| "rewards/margins": 0.15607169270515442, | |
| "rewards/rejected": -0.16801045835018158, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.7568807339449541, | |
| "grad_norm": 1.4113093599011106, | |
| "learning_rate": 8.49126331382102e-08, | |
| "logits/chosen": -2.435225009918213, | |
| "logits/rejected": -2.386702537536621, | |
| "logps/chosen": -275.8332214355469, | |
| "logps/rejected": -260.52587890625, | |
| "loss": 0.0095, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.013110649771988392, | |
| "rewards/margins": 0.12114731222391129, | |
| "rewards/rejected": -0.1342579573392868, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.7798165137614679, | |
| "grad_norm": 1.366187901878708, | |
| "learning_rate": 7.041266247556812e-08, | |
| "logits/chosen": -2.521238327026367, | |
| "logits/rejected": -2.4764904975891113, | |
| "logps/chosen": -289.8477478027344, | |
| "logps/rejected": -272.86456298828125, | |
| "loss": 0.009, | |
| "rewards/accuracies": 0.731249988079071, | |
| "rewards/chosen": -0.01855117455124855, | |
| "rewards/margins": 0.11025450378656387, | |
| "rewards/rejected": -0.12880566716194153, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.8027522935779816, | |
| "grad_norm": 1.3145992203531836, | |
| "learning_rate": 5.706553665319955e-08, | |
| "logits/chosen": -2.4982964992523193, | |
| "logits/rejected": -2.3964176177978516, | |
| "logps/chosen": -277.12042236328125, | |
| "logps/rejected": -253.37203979492188, | |
| "loss": 0.0096, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -0.026293223723769188, | |
| "rewards/margins": 0.13081298768520355, | |
| "rewards/rejected": -0.1571062207221985, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.8027522935779816, | |
| "eval_logits/chosen": -2.481409788131714, | |
| "eval_logits/rejected": -2.3850882053375244, | |
| "eval_logps/chosen": -285.64593505859375, | |
| "eval_logps/rejected": -262.6905517578125, | |
| "eval_loss": 0.008698553778231144, | |
| "eval_rewards/accuracies": 0.7801724076271057, | |
| "eval_rewards/chosen": -0.005558254197239876, | |
| "eval_rewards/margins": 0.15612287819385529, | |
| "eval_rewards/rejected": -0.1616811454296112, | |
| "eval_runtime": 94.1358, | |
| "eval_samples_per_second": 19.313, | |
| "eval_steps_per_second": 0.308, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.8256880733944955, | |
| "grad_norm": 1.1160006320842197, | |
| "learning_rate": 4.4956936350761005e-08, | |
| "logits/chosen": -2.4785516262054443, | |
| "logits/rejected": -2.427774429321289, | |
| "logps/chosen": -252.49581909179688, | |
| "logps/rejected": -271.5592041015625, | |
| "loss": 0.0088, | |
| "rewards/accuracies": 0.731249988079071, | |
| "rewards/chosen": -0.0017578303813934326, | |
| "rewards/margins": 0.12846335768699646, | |
| "rewards/rejected": -0.1302211880683899, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.8486238532110092, | |
| "grad_norm": 1.2680559242352918, | |
| "learning_rate": 3.416459164418123e-08, | |
| "logits/chosen": -2.514042854309082, | |
| "logits/rejected": -2.4509449005126953, | |
| "logps/chosen": -300.03057861328125, | |
| "logps/rejected": -276.8949279785156, | |
| "loss": 0.0087, | |
| "rewards/accuracies": 0.7562500238418579, | |
| "rewards/chosen": -0.004460807889699936, | |
| "rewards/margins": 0.13874222338199615, | |
| "rewards/rejected": -0.1432030349969864, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.8715596330275229, | |
| "grad_norm": 1.5408590452374413, | |
| "learning_rate": 2.475778302439524e-08, | |
| "logits/chosen": -2.4833552837371826, | |
| "logits/rejected": -2.4306118488311768, | |
| "logps/chosen": -296.1810607910156, | |
| "logps/rejected": -272.9661560058594, | |
| "loss": 0.009, | |
| "rewards/accuracies": 0.78125, | |
| "rewards/chosen": 0.001969636185094714, | |
| "rewards/margins": 0.15506890416145325, | |
| "rewards/rejected": -0.15309928357601166, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.8944954128440367, | |
| "grad_norm": 1.4629042335763474, | |
| "learning_rate": 1.6796896657433805e-08, | |
| "logits/chosen": -2.4795479774475098, | |
| "logits/rejected": -2.3772130012512207, | |
| "logps/chosen": -253.0847930908203, | |
| "logps/rejected": -242.50405883789062, | |
| "loss": 0.0095, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.011273050680756569, | |
| "rewards/margins": 0.12070544064044952, | |
| "rewards/rejected": -0.13197848200798035, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.9174311926605505, | |
| "grad_norm": 1.3631415527821928, | |
| "learning_rate": 1.0333036740834855e-08, | |
| "logits/chosen": -2.4071569442749023, | |
| "logits/rejected": -2.371851682662964, | |
| "logps/chosen": -224.0278778076172, | |
| "logps/rejected": -241.8699188232422, | |
| "loss": 0.0093, | |
| "rewards/accuracies": 0.731249988079071, | |
| "rewards/chosen": 0.0053174905478954315, | |
| "rewards/margins": 0.1318611204624176, | |
| "rewards/rejected": -0.12654362618923187, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.9174311926605505, | |
| "eval_logits/chosen": -2.4800400733947754, | |
| "eval_logits/rejected": -2.383610486984253, | |
| "eval_logps/chosen": -285.21075439453125, | |
| "eval_logps/rejected": -261.9634704589844, | |
| "eval_loss": 0.008683313615620136, | |
| "eval_rewards/accuracies": 0.767241358757019, | |
| "eval_rewards/chosen": -0.0012068306095898151, | |
| "eval_rewards/margins": 0.15320327877998352, | |
| "eval_rewards/rejected": -0.15441007912158966, | |
| "eval_runtime": 95.2651, | |
| "eval_samples_per_second": 19.084, | |
| "eval_steps_per_second": 0.304, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.9403669724770642, | |
| "grad_norm": 1.2959245163133988, | |
| "learning_rate": 5.4076974448211685e-09, | |
| "logits/chosen": -2.414137363433838, | |
| "logits/rejected": -2.364811420440674, | |
| "logps/chosen": -271.1864318847656, | |
| "logps/rejected": -253.48568725585938, | |
| "loss": 0.009, | |
| "rewards/accuracies": 0.768750011920929, | |
| "rewards/chosen": -0.0023452930618077517, | |
| "rewards/margins": 0.1520446538925171, | |
| "rewards/rejected": -0.15438991785049438, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.963302752293578, | |
| "grad_norm": 2.8287990040902504, | |
| "learning_rate": 2.052496544188487e-09, | |
| "logits/chosen": -2.4582016468048096, | |
| "logits/rejected": -2.3712384700775146, | |
| "logps/chosen": -260.9129638671875, | |
| "logps/rejected": -261.7887878417969, | |
| "loss": 0.0093, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.014739753678441048, | |
| "rewards/margins": 0.14437474310398102, | |
| "rewards/rejected": -0.1591145098209381, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.9862385321100917, | |
| "grad_norm": 1.3866473112375508, | |
| "learning_rate": 2.889724508297886e-10, | |
| "logits/chosen": -2.474024772644043, | |
| "logits/rejected": -2.357807159423828, | |
| "logps/chosen": -303.5400695800781, | |
| "logps/rejected": -256.8912658691406, | |
| "loss": 0.0094, | |
| "rewards/accuracies": 0.768750011920929, | |
| "rewards/chosen": -0.003646157681941986, | |
| "rewards/margins": 0.12490139901638031, | |
| "rewards/rejected": -0.1285475790500641, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 436, | |
| "total_flos": 0.0, | |
| "train_loss": 0.010265434613673512, | |
| "train_runtime": 11699.9785, | |
| "train_samples_per_second": 4.766, | |
| "train_steps_per_second": 0.037 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 436, | |
| "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 | |
| } | |