| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.9984301412872841, | |
| "eval_steps": 100, | |
| "global_step": 477, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "grad_norm": 0.400390625, | |
| "learning_rate": 1.0416666666666667e-07, | |
| "logits/chosen": -2.2547454833984375, | |
| "logits/rejected": -2.401865005493164, | |
| "logps/chosen": -53.759212493896484, | |
| "logps/rejected": -48.83185958862305, | |
| "loss": 0.6931, | |
| "pred_label": 0.0, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 1, | |
| "use_label": 0.0 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 0.4609375, | |
| "learning_rate": 1.0416666666666667e-06, | |
| "logits/chosen": -2.2421462535858154, | |
| "logits/rejected": -2.2770614624023438, | |
| "logps/chosen": -51.98179626464844, | |
| "logps/rejected": -64.9604263305664, | |
| "loss": 0.6929, | |
| "pred_label": 0.0, | |
| "rewards/accuracies": 0.2222222238779068, | |
| "rewards/chosen": 0.001975727966055274, | |
| "rewards/margins": 0.00047667179023846984, | |
| "rewards/rejected": 0.001499056350439787, | |
| "step": 10, | |
| "use_label": 0.0 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 0.39453125, | |
| "learning_rate": 2.0833333333333334e-06, | |
| "logits/chosen": -2.2520272731781006, | |
| "logits/rejected": -2.255510091781616, | |
| "logps/chosen": -62.492515563964844, | |
| "logps/rejected": -72.63607788085938, | |
| "loss": 0.6919, | |
| "pred_label": 0.0, | |
| "rewards/accuracies": 0.2874999940395355, | |
| "rewards/chosen": 0.01601376011967659, | |
| "rewards/margins": 0.0011284304782748222, | |
| "rewards/rejected": 0.014885328710079193, | |
| "step": 20, | |
| "use_label": 0.0 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 0.5078125, | |
| "learning_rate": 3.125e-06, | |
| "logits/chosen": -2.3422012329101562, | |
| "logits/rejected": -2.3548905849456787, | |
| "logps/chosen": -79.14694213867188, | |
| "logps/rejected": -98.82722473144531, | |
| "loss": 0.6898, | |
| "pred_label": 0.0, | |
| "rewards/accuracies": 0.2874999940395355, | |
| "rewards/chosen": 0.030949687585234642, | |
| "rewards/margins": 0.0029636542312800884, | |
| "rewards/rejected": 0.027986034750938416, | |
| "step": 30, | |
| "use_label": 0.0 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 0.515625, | |
| "learning_rate": 4.166666666666667e-06, | |
| "logits/chosen": -2.322833776473999, | |
| "logits/rejected": -2.3010501861572266, | |
| "logps/chosen": -82.85880279541016, | |
| "logps/rejected": -82.40392303466797, | |
| "loss": 0.6866, | |
| "pred_label": 0.0, | |
| "rewards/accuracies": 0.2874999940395355, | |
| "rewards/chosen": 0.033333443105220795, | |
| "rewards/margins": 0.011918319389224052, | |
| "rewards/rejected": 0.021415119990706444, | |
| "step": 40, | |
| "use_label": 0.0 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 0.67578125, | |
| "learning_rate": 4.999731868769027e-06, | |
| "logits/chosen": -2.241189956665039, | |
| "logits/rejected": -2.263849973678589, | |
| "logps/chosen": -67.93062591552734, | |
| "logps/rejected": -81.85546875, | |
| "loss": 0.6805, | |
| "pred_label": 0.0, | |
| "rewards/accuracies": 0.32499998807907104, | |
| "rewards/chosen": 0.009002490900456905, | |
| "rewards/margins": 0.03016103245317936, | |
| "rewards/rejected": -0.02115854248404503, | |
| "step": 50, | |
| "use_label": 0.0 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "grad_norm": 1.09375, | |
| "learning_rate": 4.9903533134293035e-06, | |
| "logits/chosen": -2.218756914138794, | |
| "logits/rejected": -2.1594481468200684, | |
| "logps/chosen": -62.0407600402832, | |
| "logps/rejected": -71.9369888305664, | |
| "loss": 0.6748, | |
| "pred_label": 0.0, | |
| "rewards/accuracies": 0.3062500059604645, | |
| "rewards/chosen": -0.0231451578438282, | |
| "rewards/margins": 0.04653460532426834, | |
| "rewards/rejected": -0.06967976689338684, | |
| "step": 60, | |
| "use_label": 0.0 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "grad_norm": 0.8984375, | |
| "learning_rate": 4.967625656594782e-06, | |
| "logits/chosen": -2.08909273147583, | |
| "logits/rejected": -2.088801383972168, | |
| "logps/chosen": -68.09326171875, | |
| "logps/rejected": -81.9454116821289, | |
| "loss": 0.6684, | |
| "pred_label": 0.0, | |
| "rewards/accuracies": 0.25, | |
| "rewards/chosen": -0.12382155656814575, | |
| "rewards/margins": 0.03761869668960571, | |
| "rewards/rejected": -0.16144026815891266, | |
| "step": 70, | |
| "use_label": 0.0 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "grad_norm": 1.15625, | |
| "learning_rate": 4.93167072587771e-06, | |
| "logits/chosen": -2.20400071144104, | |
| "logits/rejected": -2.1452622413635254, | |
| "logps/chosen": -55.867881774902344, | |
| "logps/rejected": -70.91771697998047, | |
| "loss": 0.6588, | |
| "pred_label": 0.0, | |
| "rewards/accuracies": 0.26875001192092896, | |
| "rewards/chosen": -0.0733698159456253, | |
| "rewards/margins": 0.10403277724981308, | |
| "rewards/rejected": -0.17740261554718018, | |
| "step": 80, | |
| "use_label": 0.0 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "grad_norm": 1.0546875, | |
| "learning_rate": 4.882681251368549e-06, | |
| "logits/chosen": -1.991231918334961, | |
| "logits/rejected": -1.9964717626571655, | |
| "logps/chosen": -72.28443908691406, | |
| "logps/rejected": -90.79218292236328, | |
| "loss": 0.6587, | |
| "pred_label": 0.0, | |
| "rewards/accuracies": 0.30000001192092896, | |
| "rewards/chosen": -0.13902384042739868, | |
| "rewards/margins": 0.08125626295804977, | |
| "rewards/rejected": -0.22028008103370667, | |
| "step": 90, | |
| "use_label": 0.0 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "grad_norm": 2.359375, | |
| "learning_rate": 4.8209198325401815e-06, | |
| "logits/chosen": -1.9231764078140259, | |
| "logits/rejected": -1.9043807983398438, | |
| "logps/chosen": -103.5636978149414, | |
| "logps/rejected": -96.08602142333984, | |
| "loss": 0.6551, | |
| "pred_label": 0.0, | |
| "rewards/accuracies": 0.35624998807907104, | |
| "rewards/chosen": -0.2353379726409912, | |
| "rewards/margins": 0.08685441315174103, | |
| "rewards/rejected": -0.32219237089157104, | |
| "step": 100, | |
| "use_label": 0.0 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "eval_logits/chosen": -1.762041687965393, | |
| "eval_logits/rejected": -1.7460479736328125, | |
| "eval_logps/chosen": -87.55253601074219, | |
| "eval_logps/rejected": -114.47212219238281, | |
| "eval_loss": 0.652633547782898, | |
| "eval_pred_label": 0.0, | |
| "eval_rewards/accuracies": 0.3359375, | |
| "eval_rewards/chosen": -0.23640292882919312, | |
| "eval_rewards/margins": 0.136388897895813, | |
| "eval_rewards/rejected": -0.3727918267250061, | |
| "eval_runtime": 125.4491, | |
| "eval_samples_per_second": 15.943, | |
| "eval_steps_per_second": 0.255, | |
| "eval_use_label": 0.0, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "grad_norm": 1.59375, | |
| "learning_rate": 4.746717530629565e-06, | |
| "logits/chosen": -1.7847106456756592, | |
| "logits/rejected": -1.7590484619140625, | |
| "logps/chosen": -85.73925018310547, | |
| "logps/rejected": -106.20509338378906, | |
| "loss": 0.6557, | |
| "pred_label": 0.0, | |
| "rewards/accuracies": 0.3499999940395355, | |
| "rewards/chosen": -0.14638465642929077, | |
| "rewards/margins": 0.12975916266441345, | |
| "rewards/rejected": -0.2761438190937042, | |
| "step": 110, | |
| "use_label": 0.0 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "grad_norm": 1.828125, | |
| "learning_rate": 4.660472094042121e-06, | |
| "logits/chosen": -1.1902318000793457, | |
| "logits/rejected": -1.0542975664138794, | |
| "logps/chosen": -108.4779052734375, | |
| "logps/rejected": -127.95109558105469, | |
| "loss": 0.6493, | |
| "pred_label": 0.0, | |
| "rewards/accuracies": 0.36250001192092896, | |
| "rewards/chosen": -0.38532325625419617, | |
| "rewards/margins": 0.1649974286556244, | |
| "rewards/rejected": -0.5503206849098206, | |
| "step": 120, | |
| "use_label": 0.0 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "grad_norm": 1.9375, | |
| "learning_rate": 4.5626458262912745e-06, | |
| "logits/chosen": -0.818010687828064, | |
| "logits/rejected": -0.7847374081611633, | |
| "logps/chosen": -109.61775207519531, | |
| "logps/rejected": -133.42086791992188, | |
| "loss": 0.6524, | |
| "pred_label": 0.0, | |
| "rewards/accuracies": 0.33125001192092896, | |
| "rewards/chosen": -0.43839359283447266, | |
| "rewards/margins": 0.16735044121742249, | |
| "rewards/rejected": -0.6057440638542175, | |
| "step": 130, | |
| "use_label": 0.0 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "grad_norm": 1.71875, | |
| "learning_rate": 4.453763107901676e-06, | |
| "logits/chosen": -0.7395650148391724, | |
| "logits/rejected": -0.8444339036941528, | |
| "logps/chosen": -116.97528076171875, | |
| "logps/rejected": -130.2399139404297, | |
| "loss": 0.6381, | |
| "pred_label": 0.0, | |
| "rewards/accuracies": 0.33125001192092896, | |
| "rewards/chosen": -0.3622770607471466, | |
| "rewards/margins": 0.1490650475025177, | |
| "rewards/rejected": -0.5113420486450195, | |
| "step": 140, | |
| "use_label": 0.0 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "grad_norm": 2.125, | |
| "learning_rate": 4.33440758555951e-06, | |
| "logits/chosen": -0.6497868299484253, | |
| "logits/rejected": -0.6378159523010254, | |
| "logps/chosen": -89.60552978515625, | |
| "logps/rejected": -115.42192077636719, | |
| "loss": 0.6379, | |
| "pred_label": 0.0, | |
| "rewards/accuracies": 0.3187499940395355, | |
| "rewards/chosen": -0.2445882111787796, | |
| "rewards/margins": 0.23124215006828308, | |
| "rewards/rejected": -0.4758303761482239, | |
| "step": 150, | |
| "use_label": 0.0 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "grad_norm": 2.15625, | |
| "learning_rate": 4.205219043576955e-06, | |
| "logits/chosen": -0.3159053921699524, | |
| "logits/rejected": -0.33064812421798706, | |
| "logps/chosen": -99.68696594238281, | |
| "logps/rejected": -129.45729064941406, | |
| "loss": 0.6317, | |
| "pred_label": 0.0, | |
| "rewards/accuracies": 0.2874999940395355, | |
| "rewards/chosen": -0.35356926918029785, | |
| "rewards/margins": 0.16687795519828796, | |
| "rewards/rejected": -0.5204472541809082, | |
| "step": 160, | |
| "use_label": 0.0 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "grad_norm": 2.4375, | |
| "learning_rate": 4.066889974440757e-06, | |
| "logits/chosen": 0.14531800150871277, | |
| "logits/rejected": 0.18166163563728333, | |
| "logps/chosen": -95.45491027832031, | |
| "logps/rejected": -125.1463623046875, | |
| "loss": 0.6291, | |
| "pred_label": 0.0, | |
| "rewards/accuracies": 0.29374998807907104, | |
| "rewards/chosen": -0.39946848154067993, | |
| "rewards/margins": 0.20978550612926483, | |
| "rewards/rejected": -0.609254002571106, | |
| "step": 170, | |
| "use_label": 0.0 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "grad_norm": 2.453125, | |
| "learning_rate": 3.92016186682789e-06, | |
| "logits/chosen": -0.3282355070114136, | |
| "logits/rejected": -0.21966704726219177, | |
| "logps/chosen": -108.00712585449219, | |
| "logps/rejected": -128.67587280273438, | |
| "loss": 0.649, | |
| "pred_label": 0.0, | |
| "rewards/accuracies": 0.35624998807907104, | |
| "rewards/chosen": -0.4521949887275696, | |
| "rewards/margins": 0.27172034978866577, | |
| "rewards/rejected": -0.7239152789115906, | |
| "step": 180, | |
| "use_label": 0.0 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "grad_norm": 1.84375, | |
| "learning_rate": 3.7658212309857576e-06, | |
| "logits/chosen": -0.889633297920227, | |
| "logits/rejected": -0.6851574778556824, | |
| "logps/chosen": -91.25111389160156, | |
| "logps/rejected": -118.9649887084961, | |
| "loss": 0.6461, | |
| "pred_label": 0.0, | |
| "rewards/accuracies": 0.33125001192092896, | |
| "rewards/chosen": -0.32139474153518677, | |
| "rewards/margins": 0.22424864768981934, | |
| "rewards/rejected": -0.5456433892250061, | |
| "step": 190, | |
| "use_label": 0.0 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "grad_norm": 1.9453125, | |
| "learning_rate": 3.604695382782159e-06, | |
| "logits/chosen": -0.8204952478408813, | |
| "logits/rejected": -0.7186430096626282, | |
| "logps/chosen": -112.41142272949219, | |
| "logps/rejected": -120.7835693359375, | |
| "loss": 0.6376, | |
| "pred_label": 0.0, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": -0.30735117197036743, | |
| "rewards/margins": 0.169038325548172, | |
| "rewards/rejected": -0.47638946771621704, | |
| "step": 200, | |
| "use_label": 0.0 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "eval_logits/chosen": -0.023804781958460808, | |
| "eval_logits/rejected": 0.04317883029580116, | |
| "eval_logps/chosen": -97.96138000488281, | |
| "eval_logps/rejected": -137.9141845703125, | |
| "eval_loss": 0.6288520693778992, | |
| "eval_pred_label": 0.0, | |
| "eval_rewards/accuracies": 0.3671875, | |
| "eval_rewards/chosen": -0.34049129486083984, | |
| "eval_rewards/margins": 0.26672109961509705, | |
| "eval_rewards/rejected": -0.6072123646736145, | |
| "eval_runtime": 125.433, | |
| "eval_samples_per_second": 15.945, | |
| "eval_steps_per_second": 0.255, | |
| "eval_use_label": 0.0, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "grad_norm": 2.265625, | |
| "learning_rate": 3.437648009023905e-06, | |
| "logits/chosen": -0.05805685371160507, | |
| "logits/rejected": -0.06056814268231392, | |
| "logps/chosen": -88.78871154785156, | |
| "logps/rejected": -124.3318862915039, | |
| "loss": 0.6218, | |
| "pred_label": 0.0, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.3281395435333252, | |
| "rewards/margins": 0.28538644313812256, | |
| "rewards/rejected": -0.613525927066803, | |
| "step": 210, | |
| "use_label": 0.0 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "grad_norm": 2.21875, | |
| "learning_rate": 3.265574537815398e-06, | |
| "logits/chosen": -0.1400775909423828, | |
| "logits/rejected": -0.005620801355689764, | |
| "logps/chosen": -133.7158660888672, | |
| "logps/rejected": -136.84619140625, | |
| "loss": 0.627, | |
| "pred_label": 0.0, | |
| "rewards/accuracies": 0.30000001192092896, | |
| "rewards/chosen": -0.5408719778060913, | |
| "rewards/margins": 0.16390959918498993, | |
| "rewards/rejected": -0.7047815918922424, | |
| "step": 220, | |
| "use_label": 0.0 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "grad_norm": 1.8515625, | |
| "learning_rate": 3.089397338773569e-06, | |
| "logits/chosen": 0.16266627609729767, | |
| "logits/rejected": 0.2626825273036957, | |
| "logps/chosen": -93.3644027709961, | |
| "logps/rejected": -119.67996978759766, | |
| "loss": 0.6261, | |
| "pred_label": 0.0, | |
| "rewards/accuracies": 0.3187499940395355, | |
| "rewards/chosen": -0.28929832577705383, | |
| "rewards/margins": 0.27991363406181335, | |
| "rewards/rejected": -0.5692119598388672, | |
| "step": 230, | |
| "use_label": 0.0 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "grad_norm": 1.8984375, | |
| "learning_rate": 2.9100607788275547e-06, | |
| "logits/chosen": 0.854693591594696, | |
| "logits/rejected": 0.7261193990707397, | |
| "logps/chosen": -99.00528717041016, | |
| "logps/rejected": -135.73580932617188, | |
| "loss": 0.6295, | |
| "pred_label": 0.0, | |
| "rewards/accuracies": 0.3687500059604645, | |
| "rewards/chosen": -0.2997274696826935, | |
| "rewards/margins": 0.3153937757015228, | |
| "rewards/rejected": -0.6151211857795715, | |
| "step": 240, | |
| "use_label": 0.0 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "grad_norm": 2.03125, | |
| "learning_rate": 2.72852616010567e-06, | |
| "logits/chosen": 0.6816203594207764, | |
| "logits/rejected": 0.7033491134643555, | |
| "logps/chosen": -119.7255859375, | |
| "logps/rejected": -144.8857421875, | |
| "loss": 0.6376, | |
| "pred_label": 0.0, | |
| "rewards/accuracies": 0.3812499940395355, | |
| "rewards/chosen": -0.4632648825645447, | |
| "rewards/margins": 0.2932681143283844, | |
| "rewards/rejected": -0.7565330266952515, | |
| "step": 250, | |
| "use_label": 0.0 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "grad_norm": 1.8984375, | |
| "learning_rate": 2.5457665670441937e-06, | |
| "logits/chosen": 0.5938165187835693, | |
| "logits/rejected": 0.5592354536056519, | |
| "logps/chosen": -110.32804870605469, | |
| "logps/rejected": -146.76275634765625, | |
| "loss": 0.6162, | |
| "pred_label": 0.0, | |
| "rewards/accuracies": 0.34375, | |
| "rewards/chosen": -0.44222426414489746, | |
| "rewards/margins": 0.2809238135814667, | |
| "rewards/rejected": -0.7231480479240417, | |
| "step": 260, | |
| "use_label": 0.0 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "grad_norm": 2.90625, | |
| "learning_rate": 2.3627616503391813e-06, | |
| "logits/chosen": 0.6390979290008545, | |
| "logits/rejected": 0.5789315700531006, | |
| "logps/chosen": -123.83528137207031, | |
| "logps/rejected": -144.61489868164062, | |
| "loss": 0.6162, | |
| "pred_label": 0.0, | |
| "rewards/accuracies": 0.36250001192092896, | |
| "rewards/chosen": -0.5091949701309204, | |
| "rewards/margins": 0.24320097267627716, | |
| "rewards/rejected": -0.7523959279060364, | |
| "step": 270, | |
| "use_label": 0.0 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "grad_norm": 2.34375, | |
| "learning_rate": 2.1804923757009885e-06, | |
| "logits/chosen": 0.8771865963935852, | |
| "logits/rejected": 1.0158352851867676, | |
| "logps/chosen": -118.5296859741211, | |
| "logps/rejected": -138.31729125976562, | |
| "loss": 0.6357, | |
| "pred_label": 0.0, | |
| "rewards/accuracies": 0.30000001192092896, | |
| "rewards/chosen": -0.5302416086196899, | |
| "rewards/margins": 0.2237352430820465, | |
| "rewards/rejected": -0.7539768218994141, | |
| "step": 280, | |
| "use_label": 0.0 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "grad_norm": 2.59375, | |
| "learning_rate": 1.9999357655598894e-06, | |
| "logits/chosen": 0.44083184003829956, | |
| "logits/rejected": 0.41123947501182556, | |
| "logps/chosen": -112.27372741699219, | |
| "logps/rejected": -146.95498657226562, | |
| "loss": 0.6228, | |
| "pred_label": 0.0, | |
| "rewards/accuracies": 0.30000001192092896, | |
| "rewards/chosen": -0.4572528004646301, | |
| "rewards/margins": 0.24868395924568176, | |
| "rewards/rejected": -0.7059367299079895, | |
| "step": 290, | |
| "use_label": 0.0 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "grad_norm": 2.34375, | |
| "learning_rate": 1.8220596619089576e-06, | |
| "logits/chosen": 0.6273639798164368, | |
| "logits/rejected": 0.5140804052352905, | |
| "logps/chosen": -123.02046966552734, | |
| "logps/rejected": -168.80987548828125, | |
| "loss": 0.6196, | |
| "pred_label": 0.0, | |
| "rewards/accuracies": 0.40625, | |
| "rewards/chosen": -0.4542613625526428, | |
| "rewards/margins": 0.2926333546638489, | |
| "rewards/rejected": -0.7468947172164917, | |
| "step": 300, | |
| "use_label": 0.0 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "eval_logits/chosen": 1.0944873094558716, | |
| "eval_logits/rejected": 1.1831356287002563, | |
| "eval_logps/chosen": -102.62176513671875, | |
| "eval_logps/rejected": -150.12503051757812, | |
| "eval_loss": 0.618873655796051, | |
| "eval_pred_label": 0.0, | |
| "eval_rewards/accuracies": 0.375, | |
| "eval_rewards/chosen": -0.3870951533317566, | |
| "eval_rewards/margins": 0.34222573041915894, | |
| "eval_rewards/rejected": -0.7293209433555603, | |
| "eval_runtime": 125.4362, | |
| "eval_samples_per_second": 15.944, | |
| "eval_steps_per_second": 0.255, | |
| "eval_use_label": 0.0, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "grad_norm": 1.8515625, | |
| "learning_rate": 1.647817538357072e-06, | |
| "logits/chosen": 0.8131985664367676, | |
| "logits/rejected": 0.8752232789993286, | |
| "logps/chosen": -91.52378845214844, | |
| "logps/rejected": -139.95840454101562, | |
| "loss": 0.5999, | |
| "pred_label": 0.0, | |
| "rewards/accuracies": 0.3687500059604645, | |
| "rewards/chosen": -0.3592718541622162, | |
| "rewards/margins": 0.3578081727027893, | |
| "rewards/rejected": -0.7170799970626831, | |
| "step": 310, | |
| "use_label": 0.0 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "grad_norm": 2.40625, | |
| "learning_rate": 1.4781433892011132e-06, | |
| "logits/chosen": 0.9751952886581421, | |
| "logits/rejected": 1.1630818843841553, | |
| "logps/chosen": -135.82566833496094, | |
| "logps/rejected": -168.11805725097656, | |
| "loss": 0.6109, | |
| "pred_label": 0.0, | |
| "rewards/accuracies": 0.3687500059604645, | |
| "rewards/chosen": -0.6275521516799927, | |
| "rewards/margins": 0.3816707730293274, | |
| "rewards/rejected": -1.0092228651046753, | |
| "step": 320, | |
| "use_label": 0.0 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "grad_norm": 1.984375, | |
| "learning_rate": 1.3139467229135999e-06, | |
| "logits/chosen": 1.3293979167938232, | |
| "logits/rejected": 1.3260401487350464, | |
| "logps/chosen": -135.96664428710938, | |
| "logps/rejected": -166.52359008789062, | |
| "loss": 0.6295, | |
| "pred_label": 0.0, | |
| "rewards/accuracies": 0.33125001192092896, | |
| "rewards/chosen": -0.6585850715637207, | |
| "rewards/margins": 0.3205706775188446, | |
| "rewards/rejected": -0.9791557192802429, | |
| "step": 330, | |
| "use_label": 0.0 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "grad_norm": 2.09375, | |
| "learning_rate": 1.1561076868822756e-06, | |
| "logits/chosen": 0.7383319139480591, | |
| "logits/rejected": 0.6407849192619324, | |
| "logps/chosen": -150.60504150390625, | |
| "logps/rejected": -166.74940490722656, | |
| "loss": 0.6247, | |
| "pred_label": 0.0, | |
| "rewards/accuracies": 0.3499999940395355, | |
| "rewards/chosen": -0.658658504486084, | |
| "rewards/margins": 0.24373307824134827, | |
| "rewards/rejected": -0.9023915529251099, | |
| "step": 340, | |
| "use_label": 0.0 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "grad_norm": 2.21875, | |
| "learning_rate": 1.0054723495346484e-06, | |
| "logits/chosen": 0.6359546184539795, | |
| "logits/rejected": 0.7167641520500183, | |
| "logps/chosen": -163.8385772705078, | |
| "logps/rejected": -195.6297607421875, | |
| "loss": 0.6138, | |
| "pred_label": 0.0, | |
| "rewards/accuracies": 0.36250001192092896, | |
| "rewards/chosen": -0.7442194819450378, | |
| "rewards/margins": 0.3593973219394684, | |
| "rewards/rejected": -1.103616714477539, | |
| "step": 350, | |
| "use_label": 0.0 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "grad_norm": 1.859375, | |
| "learning_rate": 8.628481651367876e-07, | |
| "logits/chosen": 0.7298086881637573, | |
| "logits/rejected": 0.8517257571220398, | |
| "logps/chosen": -119.41548156738281, | |
| "logps/rejected": -165.3460235595703, | |
| "loss": 0.6137, | |
| "pred_label": 0.0, | |
| "rewards/accuracies": 0.3812499940395355, | |
| "rewards/chosen": -0.5577787160873413, | |
| "rewards/margins": 0.37339919805526733, | |
| "rewards/rejected": -0.9311779141426086, | |
| "step": 360, | |
| "use_label": 0.0 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "grad_norm": 2.421875, | |
| "learning_rate": 7.289996455765749e-07, | |
| "logits/chosen": 0.8383787274360657, | |
| "logits/rejected": 0.9305205345153809, | |
| "logps/chosen": -111.84449768066406, | |
| "logps/rejected": -153.93136596679688, | |
| "loss": 0.6125, | |
| "pred_label": 0.0, | |
| "rewards/accuracies": 0.3499999940395355, | |
| "rewards/chosen": -0.46409696340560913, | |
| "rewards/margins": 0.39606258273124695, | |
| "rewards/rejected": -0.8601595759391785, | |
| "step": 370, | |
| "use_label": 0.0 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "grad_norm": 1.8984375, | |
| "learning_rate": 6.046442623320145e-07, | |
| "logits/chosen": 0.5329448580741882, | |
| "logits/rejected": 0.513522744178772, | |
| "logps/chosen": -116.62841796875, | |
| "logps/rejected": -165.17893981933594, | |
| "loss": 0.6191, | |
| "pred_label": 0.0, | |
| "rewards/accuracies": 0.3187499940395355, | |
| "rewards/chosen": -0.5079302787780762, | |
| "rewards/margins": 0.2802185118198395, | |
| "rewards/rejected": -0.7881487607955933, | |
| "step": 380, | |
| "use_label": 0.0 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "grad_norm": 2.4375, | |
| "learning_rate": 4.904486005914027e-07, | |
| "logits/chosen": 0.8266662359237671, | |
| "logits/rejected": 0.5234752893447876, | |
| "logps/chosen": -159.83407592773438, | |
| "logps/rejected": -186.96768188476562, | |
| "loss": 0.6085, | |
| "pred_label": 0.0, | |
| "rewards/accuracies": 0.38749998807907104, | |
| "rewards/chosen": -0.6701575517654419, | |
| "rewards/margins": 0.36982032656669617, | |
| "rewards/rejected": -1.039977788925171, | |
| "step": 390, | |
| "use_label": 0.0 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "grad_norm": 2.46875, | |
| "learning_rate": 3.8702478614051353e-07, | |
| "logits/chosen": 0.511390745639801, | |
| "logits/rejected": 0.6720080971717834, | |
| "logps/chosen": -116.7987060546875, | |
| "logps/rejected": -141.3931884765625, | |
| "loss": 0.6139, | |
| "pred_label": 0.0, | |
| "rewards/accuracies": 0.3812499940395355, | |
| "rewards/chosen": -0.4430771768093109, | |
| "rewards/margins": 0.3362268805503845, | |
| "rewards/rejected": -0.779304027557373, | |
| "step": 400, | |
| "use_label": 0.0 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "eval_logits/chosen": 1.4532994031906128, | |
| "eval_logits/rejected": 1.5453113317489624, | |
| "eval_logps/chosen": -112.56050109863281, | |
| "eval_logps/rejected": -162.19764709472656, | |
| "eval_loss": 0.6157013177871704, | |
| "eval_pred_label": 0.0, | |
| "eval_rewards/accuracies": 0.37109375, | |
| "eval_rewards/chosen": -0.4864824414253235, | |
| "eval_rewards/margins": 0.36356455087661743, | |
| "eval_rewards/rejected": -0.8500469923019409, | |
| "eval_runtime": 125.4203, | |
| "eval_samples_per_second": 15.946, | |
| "eval_steps_per_second": 0.255, | |
| "eval_use_label": 0.0, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "grad_norm": 2.203125, | |
| "learning_rate": 2.9492720416985004e-07, | |
| "logits/chosen": 0.8359997868537903, | |
| "logits/rejected": 0.8144146800041199, | |
| "logps/chosen": -110.30177307128906, | |
| "logps/rejected": -143.6800079345703, | |
| "loss": 0.6222, | |
| "pred_label": 0.0, | |
| "rewards/accuracies": 0.39375001192092896, | |
| "rewards/chosen": -0.4887877404689789, | |
| "rewards/margins": 0.3508199453353882, | |
| "rewards/rejected": -0.8396075963973999, | |
| "step": 410, | |
| "use_label": 0.0 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "grad_norm": 1.984375, | |
| "learning_rate": 2.1464952759020857e-07, | |
| "logits/chosen": 1.027252435684204, | |
| "logits/rejected": 0.9827619791030884, | |
| "logps/chosen": -106.49784851074219, | |
| "logps/rejected": -116.97566223144531, | |
| "loss": 0.6216, | |
| "pred_label": 0.0, | |
| "rewards/accuracies": 0.2750000059604645, | |
| "rewards/chosen": -0.4555872976779938, | |
| "rewards/margins": 0.20033884048461914, | |
| "rewards/rejected": -0.6559261083602905, | |
| "step": 420, | |
| "use_label": 0.0 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "grad_norm": 1.96875, | |
| "learning_rate": 1.4662207078575685e-07, | |
| "logits/chosen": 0.9206047058105469, | |
| "logits/rejected": 0.8673297166824341, | |
| "logps/chosen": -151.376220703125, | |
| "logps/rejected": -178.04725646972656, | |
| "loss": 0.5986, | |
| "pred_label": 0.0, | |
| "rewards/accuracies": 0.45625001192092896, | |
| "rewards/chosen": -0.5210937261581421, | |
| "rewards/margins": 0.46580758690834045, | |
| "rewards/rejected": -0.9869012832641602, | |
| "step": 430, | |
| "use_label": 0.0 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "grad_norm": 2.125, | |
| "learning_rate": 9.120948298936422e-08, | |
| "logits/chosen": 0.9004503488540649, | |
| "logits/rejected": 1.0573413372039795, | |
| "logps/chosen": -119.21500396728516, | |
| "logps/rejected": -165.19241333007812, | |
| "loss": 0.6064, | |
| "pred_label": 0.0, | |
| "rewards/accuracies": 0.35624998807907104, | |
| "rewards/chosen": -0.5231102705001831, | |
| "rewards/margins": 0.37818416953086853, | |
| "rewards/rejected": -0.9012944102287292, | |
| "step": 440, | |
| "use_label": 0.0 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "grad_norm": 2.46875, | |
| "learning_rate": 4.870879364444109e-08, | |
| "logits/chosen": 1.300728440284729, | |
| "logits/rejected": 1.0580918788909912, | |
| "logps/chosen": -129.29281616210938, | |
| "logps/rejected": -178.3690948486328, | |
| "loss": 0.6111, | |
| "pred_label": 0.0, | |
| "rewards/accuracies": 0.3499999940395355, | |
| "rewards/chosen": -0.570349931716919, | |
| "rewards/margins": 0.3304445147514343, | |
| "rewards/rejected": -0.9007943868637085, | |
| "step": 450, | |
| "use_label": 0.0 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "grad_norm": 1.8359375, | |
| "learning_rate": 1.93478202307823e-08, | |
| "logits/chosen": 1.1906068325042725, | |
| "logits/rejected": 1.2149587869644165, | |
| "logps/chosen": -83.74864196777344, | |
| "logps/rejected": -130.91348266601562, | |
| "loss": 0.6154, | |
| "pred_label": 0.0, | |
| "rewards/accuracies": 0.3375000059604645, | |
| "rewards/chosen": -0.3762877583503723, | |
| "rewards/margins": 0.2993956208229065, | |
| "rewards/rejected": -0.6756833791732788, | |
| "step": 460, | |
| "use_label": 0.0 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "grad_norm": 2.375, | |
| "learning_rate": 3.283947088983663e-09, | |
| "logits/chosen": 1.1844379901885986, | |
| "logits/rejected": 0.9474547505378723, | |
| "logps/chosen": -113.1079330444336, | |
| "logps/rejected": -141.49147033691406, | |
| "loss": 0.6213, | |
| "pred_label": 0.0, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": -0.4577876627445221, | |
| "rewards/margins": 0.26655709743499756, | |
| "rewards/rejected": -0.7243447303771973, | |
| "step": 470, | |
| "use_label": 0.0 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 477, | |
| "total_flos": 0.0, | |
| "train_loss": 0.6357159084743924, | |
| "train_runtime": 9601.7268, | |
| "train_samples_per_second": 6.367, | |
| "train_steps_per_second": 0.05 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 477, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 50, | |
| "total_flos": 0.0, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |