Invalid JSON: Unexpected token 'N', ..."ad_norm": NaN,
"... is not valid JSON
| { | |
| "best_metric": 21.601177215576172, | |
| "best_model_checkpoint": "./output/checkpoints/2024-05-27_09-02-19/checkpoint-600", | |
| "epoch": 0.47206923682140045, | |
| "eval_steps": 100, | |
| "global_step": 600, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.003933910306845004, | |
| "grad_norm": 26.446468353271484, | |
| "learning_rate": 9.375000000000001e-07, | |
| "logits/chosen": -0.2329835593700409, | |
| "logits/rejected": -0.7131723165512085, | |
| "logps/chosen": -1.0090148448944092, | |
| "logps/rejected": -1.6766555309295654, | |
| "loss": 25.0031, | |
| "rewards/accuracies": 0.1875, | |
| "rewards/chosen": 8.527375939593185e-06, | |
| "rewards/margins": -3.058705624425784e-05, | |
| "rewards/rejected": 3.911443127435632e-05, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.007867820613690008, | |
| "grad_norm": 11.936881065368652, | |
| "learning_rate": 2.5e-06, | |
| "logits/chosen": -0.396948903799057, | |
| "logits/rejected": -0.7360211610794067, | |
| "logps/chosen": -0.8984262347221375, | |
| "logps/rejected": -1.1693015098571777, | |
| "loss": 24.9925, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": -4.258692206349224e-05, | |
| "rewards/margins": 7.496408943552524e-05, | |
| "rewards/rejected": -0.00011755101149901748, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.011801730920535013, | |
| "grad_norm": 13.576423645019531, | |
| "learning_rate": 4.0625000000000005e-06, | |
| "logits/chosen": -0.3573324680328369, | |
| "logits/rejected": -0.6578253507614136, | |
| "logps/chosen": -0.8142125010490417, | |
| "logps/rejected": -1.0063048601150513, | |
| "loss": 24.98, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": -0.00027080357540398836, | |
| "rewards/margins": 0.00020028329163324088, | |
| "rewards/rejected": -0.00047108688158914447, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.015735641227380016, | |
| "grad_norm": 34.40192794799805, | |
| "learning_rate": 5.3125e-06, | |
| "logits/chosen": -0.3880882263183594, | |
| "logits/rejected": -0.7228592038154602, | |
| "logps/chosen": -1.1428436040878296, | |
| "logps/rejected": -1.567692756652832, | |
| "loss": 24.8648, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": -0.001833672053180635, | |
| "rewards/margins": 0.0014055297942832112, | |
| "rewards/rejected": -0.003239201847463846, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.01966955153422502, | |
| "grad_norm": 16.62054443359375, | |
| "learning_rate": 6.875e-06, | |
| "logits/chosen": -0.25890520215034485, | |
| "logits/rejected": -0.7020931839942932, | |
| "logps/chosen": -1.212777853012085, | |
| "logps/rejected": -1.3589212894439697, | |
| "loss": 24.939, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.0046936506405472755, | |
| "rewards/margins": 0.0006289022276178002, | |
| "rewards/rejected": -0.005322552751749754, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.023603461841070025, | |
| "grad_norm": 22.588180541992188, | |
| "learning_rate": 8.4375e-06, | |
| "logits/chosen": -0.3284154236316681, | |
| "logits/rejected": -0.6061900854110718, | |
| "logps/chosen": -0.9161252975463867, | |
| "logps/rejected": -1.1756784915924072, | |
| "loss": 24.7008, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -0.004226677119731903, | |
| "rewards/margins": 0.003145116614177823, | |
| "rewards/rejected": -0.0073717935010790825, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.02753737214791503, | |
| "grad_norm": 35.44740295410156, | |
| "learning_rate": 1e-05, | |
| "logits/chosen": -0.4888971447944641, | |
| "logits/rejected": -0.7553955912590027, | |
| "logps/chosen": -1.252327561378479, | |
| "logps/rejected": -1.473224401473999, | |
| "loss": 24.5665, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": -0.013578332960605621, | |
| "rewards/margins": 0.004552872385829687, | |
| "rewards/rejected": -0.018131205812096596, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.03147128245476003, | |
| "grad_norm": 32.20027160644531, | |
| "learning_rate": 1.1562500000000002e-05, | |
| "logits/chosen": -0.4067641794681549, | |
| "logits/rejected": -0.7352877855300903, | |
| "logps/chosen": -1.055959939956665, | |
| "logps/rejected": -1.4485868215560913, | |
| "loss": 24.0967, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": -0.016065727919340134, | |
| "rewards/margins": 0.011377329006791115, | |
| "rewards/rejected": -0.02744305692613125, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.03540519276160504, | |
| "grad_norm": NaN, | |
| "learning_rate": 1.2812500000000001e-05, | |
| "logits/chosen": -0.7447024583816528, | |
| "logits/rejected": -1.0448763370513916, | |
| "logps/chosen": -1.723064661026001, | |
| "logps/rejected": -2.249486207962036, | |
| "loss": 24.0293, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": -0.041548244655132294, | |
| "rewards/margins": 0.013970533385872841, | |
| "rewards/rejected": -0.05551878362894058, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.03933910306845004, | |
| "grad_norm": 26.65926742553711, | |
| "learning_rate": 1.4375e-05, | |
| "logits/chosen": -0.37696924805641174, | |
| "logits/rejected": -0.46783286333084106, | |
| "logps/chosen": -1.015779733657837, | |
| "logps/rejected": -1.599536418914795, | |
| "loss": 24.0592, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": -0.020270783454179764, | |
| "rewards/margins": 0.019904401153326035, | |
| "rewards/rejected": -0.0401751883327961, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.043273013375295044, | |
| "grad_norm": 30.033103942871094, | |
| "learning_rate": 1.59375e-05, | |
| "logits/chosen": -0.7907823324203491, | |
| "logits/rejected": -0.990174412727356, | |
| "logps/chosen": -1.7764304876327515, | |
| "logps/rejected": -1.9972736835479736, | |
| "loss": 24.6893, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": -0.05663704872131348, | |
| "rewards/margins": 0.012498116120696068, | |
| "rewards/rejected": -0.0691351667046547, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.04720692368214005, | |
| "grad_norm": 155.12327575683594, | |
| "learning_rate": 1.7500000000000002e-05, | |
| "logits/chosen": -0.5883419513702393, | |
| "logits/rejected": -0.9729728698730469, | |
| "logps/chosen": -1.749121904373169, | |
| "logps/rejected": -2.5301637649536133, | |
| "loss": 23.3501, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": -0.06644631177186966, | |
| "rewards/margins": 0.04214775934815407, | |
| "rewards/rejected": -0.10859407484531403, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.05114083398898505, | |
| "grad_norm": 106.24934387207031, | |
| "learning_rate": 1.8750000000000002e-05, | |
| "logits/chosen": -0.8320428133010864, | |
| "logits/rejected": -1.0106093883514404, | |
| "logps/chosen": -1.1940712928771973, | |
| "logps/rejected": -2.7599964141845703, | |
| "loss": 23.1753, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": -0.03951374441385269, | |
| "rewards/margins": 0.08484560251235962, | |
| "rewards/rejected": -0.12435934692621231, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.05507474429583006, | |
| "grad_norm": 82.20561981201172, | |
| "learning_rate": 2.0312500000000002e-05, | |
| "logits/chosen": -0.9974054098129272, | |
| "logits/rejected": -1.2483726739883423, | |
| "logps/chosen": -1.6322723627090454, | |
| "logps/rejected": -2.4456088542938232, | |
| "loss": 21.933, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -0.07070576399564743, | |
| "rewards/margins": 0.055024802684783936, | |
| "rewards/rejected": -0.12573055922985077, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.059008654602675056, | |
| "grad_norm": 1317.8922119140625, | |
| "learning_rate": 2.1562500000000002e-05, | |
| "logits/chosen": -1.0032284259796143, | |
| "logits/rejected": -1.2543575763702393, | |
| "logps/chosen": -1.648602843284607, | |
| "logps/rejected": -3.2052788734436035, | |
| "loss": 30.9638, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.08028480410575867, | |
| "rewards/margins": 0.09190882742404938, | |
| "rewards/rejected": -0.17219363152980804, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.06294256490952006, | |
| "grad_norm": 135.98031616210938, | |
| "learning_rate": 2.3125000000000003e-05, | |
| "logits/chosen": -1.1862694025039673, | |
| "logits/rejected": -1.268090844154358, | |
| "logps/chosen": -1.7252533435821533, | |
| "logps/rejected": -2.730776786804199, | |
| "loss": 23.1291, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -0.08682449907064438, | |
| "rewards/margins": 0.07040676474571228, | |
| "rewards/rejected": -0.15723127126693726, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.06687647521636507, | |
| "grad_norm": 83.77543640136719, | |
| "learning_rate": 2.46875e-05, | |
| "logits/chosen": -1.4288493394851685, | |
| "logits/rejected": -1.6324199438095093, | |
| "logps/chosen": -1.858473539352417, | |
| "logps/rejected": -2.4925625324249268, | |
| "loss": 22.47, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -0.10322336852550507, | |
| "rewards/margins": 0.040188394486904144, | |
| "rewards/rejected": -0.143411785364151, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.07081038552321008, | |
| "grad_norm": 112.26095581054688, | |
| "learning_rate": 2.625e-05, | |
| "logits/chosen": -1.5094571113586426, | |
| "logits/rejected": -1.6503517627716064, | |
| "logps/chosen": -2.1784815788269043, | |
| "logps/rejected": -2.8247978687286377, | |
| "loss": 26.8652, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": -0.12195611000061035, | |
| "rewards/margins": 0.04399397224187851, | |
| "rewards/rejected": -0.16595008969306946, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.07474429583005507, | |
| "grad_norm": 116.14981079101562, | |
| "learning_rate": 2.7812500000000002e-05, | |
| "logits/chosen": -1.705255150794983, | |
| "logits/rejected": -1.8774993419647217, | |
| "logps/chosen": -2.0536391735076904, | |
| "logps/rejected": -2.913367748260498, | |
| "loss": 23.1756, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": -0.11939144134521484, | |
| "rewards/margins": 0.03494938462972641, | |
| "rewards/rejected": -0.15434083342552185, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.07867820613690008, | |
| "grad_norm": 171.99806213378906, | |
| "learning_rate": 2.9375000000000003e-05, | |
| "logits/chosen": -1.7922956943511963, | |
| "logits/rejected": -1.805193305015564, | |
| "logps/chosen": -2.2087619304656982, | |
| "logps/rejected": -2.953051805496216, | |
| "loss": 23.3214, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": -0.1307235062122345, | |
| "rewards/margins": 0.03435593843460083, | |
| "rewards/rejected": -0.16507944464683533, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.07867820613690008, | |
| "eval_logits/chosen": -1.966138243675232, | |
| "eval_logits/rejected": -2.09938645362854, | |
| "eval_logps/chosen": -2.3467371463775635, | |
| "eval_logps/rejected": -2.9913861751556396, | |
| "eval_loss": 22.537601470947266, | |
| "eval_rewards/accuracies": 0.643750011920929, | |
| "eval_rewards/chosen": -0.13215361535549164, | |
| "eval_rewards/margins": 0.04243787005543709, | |
| "eval_rewards/rejected": -0.17459148168563843, | |
| "eval_runtime": 254.2532, | |
| "eval_samples_per_second": 2.517, | |
| "eval_steps_per_second": 0.157, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.08261211644374508, | |
| "grad_norm": 90.37792205810547, | |
| "learning_rate": 3.09375e-05, | |
| "logits/chosen": -1.7015612125396729, | |
| "logits/rejected": -1.8363323211669922, | |
| "logps/chosen": -2.096281051635742, | |
| "logps/rejected": -3.1252198219299316, | |
| "loss": 27.6536, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": -0.11915542930364609, | |
| "rewards/margins": 0.06005290150642395, | |
| "rewards/rejected": -0.17920835316181183, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.08654602675059009, | |
| "grad_norm": 88.4887466430664, | |
| "learning_rate": 3.2500000000000004e-05, | |
| "logits/chosen": -1.657274842262268, | |
| "logits/rejected": -1.8521808385849, | |
| "logps/chosen": -1.8666527271270752, | |
| "logps/rejected": -2.9845376014709473, | |
| "loss": 21.5034, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": -0.10541417449712753, | |
| "rewards/margins": 0.05897489935159683, | |
| "rewards/rejected": -0.16438907384872437, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.0904799370574351, | |
| "grad_norm": 134.99591064453125, | |
| "learning_rate": 3.40625e-05, | |
| "logits/chosen": -1.6925245523452759, | |
| "logits/rejected": -1.7081615924835205, | |
| "logps/chosen": -2.6438632011413574, | |
| "logps/rejected": -3.7139625549316406, | |
| "loss": 22.8601, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.14907808601856232, | |
| "rewards/margins": 0.05926816537976265, | |
| "rewards/rejected": -0.20834624767303467, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.0944138473642801, | |
| "grad_norm": 105.24943542480469, | |
| "learning_rate": 3.5625000000000005e-05, | |
| "logits/chosen": -1.591507911682129, | |
| "logits/rejected": -1.58656907081604, | |
| "logps/chosen": -1.9171836376190186, | |
| "logps/rejected": -2.5378520488739014, | |
| "loss": 23.0976, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": -0.10242807865142822, | |
| "rewards/margins": 0.03649063780903816, | |
| "rewards/rejected": -0.1389187127351761, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.0983477576711251, | |
| "grad_norm": 87.82530212402344, | |
| "learning_rate": 3.71875e-05, | |
| "logits/chosen": -1.3108810186386108, | |
| "logits/rejected": -1.4434765577316284, | |
| "logps/chosen": -2.124854564666748, | |
| "logps/rejected": -3.0788867473602295, | |
| "loss": 24.5017, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": -0.1214764267206192, | |
| "rewards/margins": 0.04934501647949219, | |
| "rewards/rejected": -0.1708214432001114, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.1022816679779701, | |
| "grad_norm": 69.61011505126953, | |
| "learning_rate": 3.875e-05, | |
| "logits/chosen": -1.286228895187378, | |
| "logits/rejected": -1.5050832033157349, | |
| "logps/chosen": -2.5508532524108887, | |
| "logps/rejected": -3.24528169631958, | |
| "loss": 21.4116, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": -0.15334758162498474, | |
| "rewards/margins": 0.04556337743997574, | |
| "rewards/rejected": -0.19891095161437988, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.10621557828481511, | |
| "grad_norm": 101.8541259765625, | |
| "learning_rate": 3.999992445477636e-05, | |
| "logits/chosen": -1.3636066913604736, | |
| "logits/rejected": -1.5931237936019897, | |
| "logps/chosen": -3.0847220420837402, | |
| "logps/rejected": -3.839167356491089, | |
| "loss": 21.3367, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -0.2012835294008255, | |
| "rewards/margins": 0.05702406167984009, | |
| "rewards/rejected": -0.2583075761795044, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.11014948859166011, | |
| "grad_norm": 701.3626098632812, | |
| "learning_rate": 3.999728043187288e-05, | |
| "logits/chosen": -1.4217129945755005, | |
| "logits/rejected": -1.4933011531829834, | |
| "logps/chosen": -3.9832420349121094, | |
| "logps/rejected": -5.435095310211182, | |
| "loss": 23.8821, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -0.2787173390388489, | |
| "rewards/margins": 0.09217057377099991, | |
| "rewards/rejected": -0.3708879351615906, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.11408339889850512, | |
| "grad_norm": 163.25831604003906, | |
| "learning_rate": 3.9990859718476166e-05, | |
| "logits/chosen": -1.4570497274398804, | |
| "logits/rejected": -1.473787784576416, | |
| "logps/chosen": -3.3077120780944824, | |
| "logps/rejected": -4.8310956954956055, | |
| "loss": 20.1222, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": -0.2514913082122803, | |
| "rewards/margins": 0.11996223777532578, | |
| "rewards/rejected": -0.37145355343818665, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.11801730920535011, | |
| "grad_norm": 274.31146240234375, | |
| "learning_rate": 3.998066352720348e-05, | |
| "logits/chosen": -1.4901472330093384, | |
| "logits/rejected": -1.5641014575958252, | |
| "logps/chosen": -4.366189956665039, | |
| "logps/rejected": -5.792475700378418, | |
| "loss": 24.8947, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": -0.3093239367008209, | |
| "rewards/margins": 0.1069142073392868, | |
| "rewards/rejected": -0.4162382185459137, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.12195121951219512, | |
| "grad_norm": 204.0838623046875, | |
| "learning_rate": 3.9966693783709596e-05, | |
| "logits/chosen": -1.775489091873169, | |
| "logits/rejected": -1.6681087017059326, | |
| "logps/chosen": -3.3862712383270264, | |
| "logps/rejected": -3.850006580352783, | |
| "loss": 25.0897, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.22455720603466034, | |
| "rewards/margins": 0.03894208371639252, | |
| "rewards/rejected": -0.26349928975105286, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.12588512981904013, | |
| "grad_norm": 117.58898162841797, | |
| "learning_rate": 3.9948953126323144e-05, | |
| "logits/chosen": -1.7140939235687256, | |
| "logits/rejected": -1.8748031854629517, | |
| "logps/chosen": -2.808436632156372, | |
| "logps/rejected": -3.4621639251708984, | |
| "loss": 22.909, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": -0.17321309447288513, | |
| "rewards/margins": 0.03799115866422653, | |
| "rewards/rejected": -0.21120426058769226, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.12981904012588513, | |
| "grad_norm": 75.52522277832031, | |
| "learning_rate": 3.992744490554832e-05, | |
| "logits/chosen": -1.5584402084350586, | |
| "logits/rejected": -1.6980777978897095, | |
| "logps/chosen": -2.5916876792907715, | |
| "logps/rejected": -3.250844955444336, | |
| "loss": 22.8805, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": -0.16778233647346497, | |
| "rewards/margins": 0.046813301742076874, | |
| "rewards/rejected": -0.21459563076496124, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.13375295043273014, | |
| "grad_norm": 210.5413818359375, | |
| "learning_rate": 3.990217318343214e-05, | |
| "logits/chosen": -1.6046726703643799, | |
| "logits/rejected": -1.785196304321289, | |
| "logps/chosen": -3.144779920578003, | |
| "logps/rejected": -4.314496040344238, | |
| "loss": 21.1638, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.2154751569032669, | |
| "rewards/margins": 0.07778888940811157, | |
| "rewards/rejected": -0.2932640314102173, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.13768686073957515, | |
| "grad_norm": 137.43014526367188, | |
| "learning_rate": 3.987314273279721e-05, | |
| "logits/chosen": -1.538189172744751, | |
| "logits/rejected": -1.7611982822418213, | |
| "logps/chosen": -3.314256191253662, | |
| "logps/rejected": -4.361076354980469, | |
| "loss": 22.1568, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -0.24516558647155762, | |
| "rewards/margins": 0.08248710632324219, | |
| "rewards/rejected": -0.3276526927947998, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.14162077104642015, | |
| "grad_norm": 162.08778381347656, | |
| "learning_rate": 3.9840359036340424e-05, | |
| "logits/chosen": -1.5785366296768188, | |
| "logits/rejected": -1.6759214401245117, | |
| "logps/chosen": -3.9628891944885254, | |
| "logps/rejected": -4.663954257965088, | |
| "loss": 23.3996, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.2786514163017273, | |
| "rewards/margins": 0.06107773259282112, | |
| "rewards/rejected": -0.3397291302680969, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.14555468135326516, | |
| "grad_norm": 183.8037567138672, | |
| "learning_rate": 3.980382828559743e-05, | |
| "logits/chosen": -1.8036388158798218, | |
| "logits/rejected": -1.9109561443328857, | |
| "logps/chosen": -4.924368381500244, | |
| "logps/rejected": -5.794642448425293, | |
| "loss": 22.9192, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": -0.357871949672699, | |
| "rewards/margins": 0.06735062599182129, | |
| "rewards/rejected": -0.42522257566452026, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.14948859166011014, | |
| "grad_norm": 137.1122283935547, | |
| "learning_rate": 3.9763557379773316e-05, | |
| "logits/chosen": -1.7101930379867554, | |
| "logits/rejected": -1.8198864459991455, | |
| "logps/chosen": -3.584138870239258, | |
| "logps/rejected": -4.54425048828125, | |
| "loss": 20.9665, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.27856582403182983, | |
| "rewards/margins": 0.0747852548956871, | |
| "rewards/rejected": -0.3533511161804199, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.15342250196695514, | |
| "grad_norm": 164.39016723632812, | |
| "learning_rate": 3.971955392443965e-05, | |
| "logits/chosen": -1.697361707687378, | |
| "logits/rejected": -1.7193193435668945, | |
| "logps/chosen": -3.8646721839904785, | |
| "logps/rejected": -5.100863456726074, | |
| "loss": 21.2867, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": -0.29381299018859863, | |
| "rewards/margins": 0.07520242035388947, | |
| "rewards/rejected": -0.3690153658390045, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.15735641227380015, | |
| "grad_norm": 125.37354278564453, | |
| "learning_rate": 3.9671826230098045e-05, | |
| "logits/chosen": -1.6001428365707397, | |
| "logits/rejected": -1.736572504043579, | |
| "logps/chosen": -3.7506911754608154, | |
| "logps/rejected": -4.7256364822387695, | |
| "loss": 21.3918, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": -0.2859867215156555, | |
| "rewards/margins": 0.07923749834299088, | |
| "rewards/rejected": -0.3652242124080658, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.15735641227380015, | |
| "eval_logits/chosen": -1.5953483581542969, | |
| "eval_logits/rejected": -1.725934624671936, | |
| "eval_logps/chosen": -4.086066246032715, | |
| "eval_logps/rejected": -5.129216194152832, | |
| "eval_loss": 23.703336715698242, | |
| "eval_rewards/accuracies": 0.6546875238418579, | |
| "eval_rewards/chosen": -0.3060864806175232, | |
| "eval_rewards/margins": 0.0822879821062088, | |
| "eval_rewards/rejected": -0.3883745074272156, | |
| "eval_runtime": 254.3055, | |
| "eval_samples_per_second": 2.517, | |
| "eval_steps_per_second": 0.157, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.16129032258064516, | |
| "grad_norm": 167.259033203125, | |
| "learning_rate": 3.962038331061065e-05, | |
| "logits/chosen": -1.4170001745224, | |
| "logits/rejected": -1.6189939975738525, | |
| "logps/chosen": -3.634209156036377, | |
| "logps/rejected": -5.206550121307373, | |
| "loss": 26.5886, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": -0.2781711220741272, | |
| "rewards/margins": 0.10549378395080566, | |
| "rewards/rejected": -0.38366490602493286, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.16522423288749016, | |
| "grad_norm": 109.77017211914062, | |
| "learning_rate": 3.9565234881497835e-05, | |
| "logits/chosen": -1.5879325866699219, | |
| "logits/rejected": -1.6509323120117188, | |
| "logps/chosen": -2.8328652381896973, | |
| "logps/rejected": -3.345362901687622, | |
| "loss": 22.6272, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.20151302218437195, | |
| "rewards/margins": 0.04314180836081505, | |
| "rewards/rejected": -0.2446548491716385, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.16915814319433517, | |
| "grad_norm": 162.56524658203125, | |
| "learning_rate": 3.950639135810326e-05, | |
| "logits/chosen": -1.6067664623260498, | |
| "logits/rejected": -1.7900478839874268, | |
| "logps/chosen": -3.400160312652588, | |
| "logps/rejected": -4.594171047210693, | |
| "loss": 20.8583, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.2197611778974533, | |
| "rewards/margins": 0.0824299305677414, | |
| "rewards/rejected": -0.3021911084651947, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.17309205350118018, | |
| "grad_norm": 134.93789672851562, | |
| "learning_rate": 3.944386385362683e-05, | |
| "logits/chosen": -1.7304567098617554, | |
| "logits/rejected": -1.7651231288909912, | |
| "logps/chosen": -4.1914567947387695, | |
| "logps/rejected": -5.027632713317871, | |
| "loss": 21.3514, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.2970493733882904, | |
| "rewards/margins": 0.06264514476060867, | |
| "rewards/rejected": -0.3596945106983185, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.17702596380802518, | |
| "grad_norm": 106.4392318725586, | |
| "learning_rate": 3.937766417702591e-05, | |
| "logits/chosen": -1.645422339439392, | |
| "logits/rejected": -1.7480659484863281, | |
| "logps/chosen": -4.8556694984436035, | |
| "logps/rejected": -5.551773548126221, | |
| "loss": 25.2991, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": -0.39037808775901794, | |
| "rewards/margins": 0.040518540889024734, | |
| "rewards/rejected": -0.4308966100215912, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.1809598741148702, | |
| "grad_norm": 75.40190124511719, | |
| "learning_rate": 3.9307804830785033e-05, | |
| "logits/chosen": -1.710780382156372, | |
| "logits/rejected": -1.759790062904358, | |
| "logps/chosen": -4.1053643226623535, | |
| "logps/rejected": -5.580018520355225, | |
| "loss": 20.5995, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": -0.32871341705322266, | |
| "rewards/margins": 0.09709561616182327, | |
| "rewards/rejected": -0.42580899596214294, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.1848937844217152, | |
| "grad_norm": 129.5159912109375, | |
| "learning_rate": 3.923429900855468e-05, | |
| "logits/chosen": -1.5250613689422607, | |
| "logits/rejected": -1.7375835180282593, | |
| "logps/chosen": -3.8208870887756348, | |
| "logps/rejected": -5.471742630004883, | |
| "loss": 19.4905, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -0.2826174199581146, | |
| "rewards/margins": 0.10505588352680206, | |
| "rewards/rejected": -0.3876733183860779, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.1888276947285602, | |
| "grad_norm": 358.99334716796875, | |
| "learning_rate": 3.915716059265956e-05, | |
| "logits/chosen": -1.2919423580169678, | |
| "logits/rejected": -1.5058711767196655, | |
| "logps/chosen": -4.358604907989502, | |
| "logps/rejected": -5.214530944824219, | |
| "loss": 21.5458, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -0.30291062593460083, | |
| "rewards/margins": 0.06571656465530396, | |
| "rewards/rejected": -0.3686271905899048, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.19276160503540518, | |
| "grad_norm": 83.99627685546875, | |
| "learning_rate": 3.907640415147675e-05, | |
| "logits/chosen": -1.1905521154403687, | |
| "logits/rejected": -1.4401142597198486, | |
| "logps/chosen": -3.447228193283081, | |
| "logps/rejected": -4.264595985412598, | |
| "loss": 21.469, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.2503766417503357, | |
| "rewards/margins": 0.06672003120183945, | |
| "rewards/rejected": -0.31709665060043335, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.1966955153422502, | |
| "grad_norm": 97.8551025390625, | |
| "learning_rate": 3.8992044936684326e-05, | |
| "logits/chosen": -1.167415976524353, | |
| "logits/rejected": -1.3312307596206665, | |
| "logps/chosen": -3.2072510719299316, | |
| "logps/rejected": -3.7459397315979004, | |
| "loss": 24.394, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.23177361488342285, | |
| "rewards/margins": 0.04146546125411987, | |
| "rewards/rejected": -0.2732390761375427, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.2006294256490952, | |
| "grad_norm": 81.79573822021484, | |
| "learning_rate": 3.8904098880380946e-05, | |
| "logits/chosen": -1.0507287979125977, | |
| "logits/rejected": -1.1515988111495972, | |
| "logps/chosen": -2.6618685722351074, | |
| "logps/rejected": -3.6504600048065186, | |
| "loss": 21.5489, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": -0.17464396357536316, | |
| "rewards/margins": 0.08120186626911163, | |
| "rewards/rejected": -0.2558458149433136, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.2045633359559402, | |
| "grad_norm": 74.57398986816406, | |
| "learning_rate": 3.881258259207688e-05, | |
| "logits/chosen": -1.026132583618164, | |
| "logits/rejected": -1.1835418939590454, | |
| "logps/chosen": -3.0442395210266113, | |
| "logps/rejected": -3.4265129566192627, | |
| "loss": 24.2589, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": -0.18103419244289398, | |
| "rewards/margins": 0.02801087498664856, | |
| "rewards/rejected": -0.20904505252838135, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.2084972462627852, | |
| "grad_norm": 136.9661102294922, | |
| "learning_rate": 3.8717513355557156e-05, | |
| "logits/chosen": -1.0285115242004395, | |
| "logits/rejected": -1.2167742252349854, | |
| "logps/chosen": -2.512547016143799, | |
| "logps/rejected": -3.510410785675049, | |
| "loss": 22.8105, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -0.1588296890258789, | |
| "rewards/margins": 0.07337381690740585, | |
| "rewards/rejected": -0.23220351338386536, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.21243115656963021, | |
| "grad_norm": 69.99120330810547, | |
| "learning_rate": 3.861890912561731e-05, | |
| "logits/chosen": -0.8553465604782104, | |
| "logits/rejected": -1.1523014307022095, | |
| "logps/chosen": -2.420487880706787, | |
| "logps/rejected": -3.4689393043518066, | |
| "loss": 20.5014, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": -0.15432177484035492, | |
| "rewards/margins": 0.08016739785671234, | |
| "rewards/rejected": -0.23448920249938965, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.21636506687647522, | |
| "grad_norm": 75.06553649902344, | |
| "learning_rate": 3.85167885246725e-05, | |
| "logits/chosen": -1.0096137523651123, | |
| "logits/rejected": -0.9508267641067505, | |
| "logps/chosen": -3.4687724113464355, | |
| "logps/rejected": -4.444920539855957, | |
| "loss": 22.2787, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": -0.2557123303413391, | |
| "rewards/margins": 0.07582716643810272, | |
| "rewards/rejected": -0.3315395414829254, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.22029897718332023, | |
| "grad_norm": 80.17613983154297, | |
| "learning_rate": 3.8411170839240394e-05, | |
| "logits/chosen": -0.9037753939628601, | |
| "logits/rejected": -0.9584333300590515, | |
| "logps/chosen": -3.32795786857605, | |
| "logps/rejected": -4.637690544128418, | |
| "loss": 23.2314, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": -0.2455664873123169, | |
| "rewards/margins": 0.09604751318693161, | |
| "rewards/rejected": -0.3416139781475067, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.22423288749016523, | |
| "grad_norm": 51.01013946533203, | |
| "learning_rate": 3.8302076016298786e-05, | |
| "logits/chosen": -0.7821402549743652, | |
| "logits/rejected": -0.8784409761428833, | |
| "logps/chosen": -3.464825391769409, | |
| "logps/rejected": -4.3120927810668945, | |
| "loss": 26.3572, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": -0.22129957377910614, | |
| "rewards/margins": 0.05804131552577019, | |
| "rewards/rejected": -0.27934086322784424, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.22816679779701024, | |
| "grad_norm": 50.36003494262695, | |
| "learning_rate": 3.818952465951836e-05, | |
| "logits/chosen": -0.8527859449386597, | |
| "logits/rejected": -1.0032708644866943, | |
| "logps/chosen": -2.805267810821533, | |
| "logps/rejected": -3.6299731731414795, | |
| "loss": 22.6246, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.19116182625293732, | |
| "rewards/margins": 0.04059046879410744, | |
| "rewards/rejected": -0.23175227642059326, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.23210070810385522, | |
| "grad_norm": 73.20655059814453, | |
| "learning_rate": 3.80735380253715e-05, | |
| "logits/chosen": -1.153649926185608, | |
| "logits/rejected": -1.3198211193084717, | |
| "logps/chosen": -3.344242811203003, | |
| "logps/rejected": -3.948678970336914, | |
| "loss": 23.0667, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": -0.24411065876483917, | |
| "rewards/margins": 0.03483257442712784, | |
| "rewards/rejected": -0.2789432406425476, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.23603461841070023, | |
| "grad_norm": 74.69874572753906, | |
| "learning_rate": 3.7954138019117764e-05, | |
| "logits/chosen": -1.2777029275894165, | |
| "logits/rejected": -1.4231250286102295, | |
| "logps/chosen": -3.8555595874786377, | |
| "logps/rejected": -4.460744380950928, | |
| "loss": 23.7264, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.2938821017742157, | |
| "rewards/margins": 0.04352904483675957, | |
| "rewards/rejected": -0.33741116523742676, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.23603461841070023, | |
| "eval_logits/chosen": -1.299351453781128, | |
| "eval_logits/rejected": -1.5808088779449463, | |
| "eval_logps/chosen": -3.9815101623535156, | |
| "eval_logps/rejected": -4.813979148864746, | |
| "eval_loss": 22.46109962463379, | |
| "eval_rewards/accuracies": 0.6578124761581421, | |
| "eval_rewards/chosen": -0.29563087224960327, | |
| "eval_rewards/margins": 0.06121987849473953, | |
| "eval_rewards/rejected": -0.3568507432937622, | |
| "eval_runtime": 256.5735, | |
| "eval_samples_per_second": 2.494, | |
| "eval_steps_per_second": 0.156, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.23996852871754523, | |
| "grad_norm": 100.70768737792969, | |
| "learning_rate": 3.7831347190666886e-05, | |
| "logits/chosen": -1.4278929233551025, | |
| "logits/rejected": -1.5912885665893555, | |
| "logps/chosen": -4.3169355392456055, | |
| "logps/rejected": -5.223280906677246, | |
| "loss": 21.9094, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": -0.32976609468460083, | |
| "rewards/margins": 0.05834154412150383, | |
| "rewards/rejected": -0.38810762763023376, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.24390243902439024, | |
| "grad_norm": 220.49977111816406, | |
| "learning_rate": 3.770518873031997e-05, | |
| "logits/chosen": -1.3963868618011475, | |
| "logits/rejected": -1.5353944301605225, | |
| "logps/chosen": -4.6622209548950195, | |
| "logps/rejected": -5.298556327819824, | |
| "loss": 25.9132, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": -0.34076324105262756, | |
| "rewards/margins": 0.03211987391114235, | |
| "rewards/rejected": -0.3728831112384796, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.24783634933123525, | |
| "grad_norm": 45.653079986572266, | |
| "learning_rate": 3.757568646438977e-05, | |
| "logits/chosen": -1.3604671955108643, | |
| "logits/rejected": -1.4712865352630615, | |
| "logps/chosen": -4.468562126159668, | |
| "logps/rejected": -5.09032678604126, | |
| "loss": 23.4367, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": -0.32228899002075195, | |
| "rewards/margins": 0.037079013884067535, | |
| "rewards/rejected": -0.3593679964542389, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.25177025963808025, | |
| "grad_norm": 66.92400360107422, | |
| "learning_rate": 3.744286485070085e-05, | |
| "logits/chosen": -1.082240343093872, | |
| "logits/rejected": -1.4719394445419312, | |
| "logps/chosen": -4.237619876861572, | |
| "logps/rejected": -5.15458345413208, | |
| "loss": 23.5263, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.33307066559791565, | |
| "rewards/margins": 0.0633757933974266, | |
| "rewards/rejected": -0.39644646644592285, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.25570416994492523, | |
| "grad_norm": 177.88168334960938, | |
| "learning_rate": 3.730674897397048e-05, | |
| "logits/chosen": -1.114916443824768, | |
| "logits/rejected": -1.6648147106170654, | |
| "logps/chosen": -4.053045749664307, | |
| "logps/rejected": -5.189083099365234, | |
| "loss": 22.1226, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.3164909780025482, | |
| "rewards/margins": 0.07537268847227097, | |
| "rewards/rejected": -0.3918636739253998, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.25963808025177026, | |
| "grad_norm": 100.30528259277344, | |
| "learning_rate": 3.7167364541071115e-05, | |
| "logits/chosen": -0.988497257232666, | |
| "logits/rejected": -1.1937472820281982, | |
| "logps/chosen": -4.265296459197998, | |
| "logps/rejected": -4.853640556335449, | |
| "loss": 20.9283, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": -0.3129270374774933, | |
| "rewards/margins": 0.0673552006483078, | |
| "rewards/rejected": -0.3802822232246399, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.26357199055861524, | |
| "grad_norm": 83.86753845214844, | |
| "learning_rate": 3.7024737876175406e-05, | |
| "logits/chosen": -0.8350197076797485, | |
| "logits/rejected": -1.1670969724655151, | |
| "logps/chosen": -5.4864630699157715, | |
| "logps/rejected": -6.544106960296631, | |
| "loss": 20.1224, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": -0.4627310335636139, | |
| "rewards/margins": 0.0860290378332138, | |
| "rewards/rejected": -0.5487600564956665, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.2675059008654603, | |
| "grad_norm": 73.52009582519531, | |
| "learning_rate": 3.6878895915784616e-05, | |
| "logits/chosen": -0.6294984221458435, | |
| "logits/rejected": -0.6951633095741272, | |
| "logps/chosen": -6.682524681091309, | |
| "logps/rejected": -7.499720573425293, | |
| "loss": 23.3106, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": -0.576111912727356, | |
| "rewards/margins": 0.06401752680540085, | |
| "rewards/rejected": -0.6401294469833374, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.27143981117230526, | |
| "grad_norm": 47.29294204711914, | |
| "learning_rate": 3.6729866203641346e-05, | |
| "logits/chosen": -0.30728015303611755, | |
| "logits/rejected": -0.7238900065422058, | |
| "logps/chosen": -5.2912421226501465, | |
| "logps/rejected": -6.754895210266113, | |
| "loss": 20.956, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": -0.44258102774620056, | |
| "rewards/margins": 0.09092569351196289, | |
| "rewards/rejected": -0.5335067510604858, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.2753737214791503, | |
| "grad_norm": 92.66437530517578, | |
| "learning_rate": 3.6577676885527676e-05, | |
| "logits/chosen": -0.4043883681297302, | |
| "logits/rejected": -0.6512165069580078, | |
| "logps/chosen": -4.932716369628906, | |
| "logps/rejected": -6.014449596405029, | |
| "loss": 19.8567, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -0.39097142219543457, | |
| "rewards/margins": 0.09061526507139206, | |
| "rewards/rejected": -0.48158663511276245, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.27930763178599527, | |
| "grad_norm": 147.38983154296875, | |
| "learning_rate": 3.6422356703949525e-05, | |
| "logits/chosen": -0.1327817142009735, | |
| "logits/rejected": -0.62315833568573, | |
| "logps/chosen": -5.30325984954834, | |
| "logps/rejected": -6.832304954528809, | |
| "loss": 21.7519, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -0.4314250946044922, | |
| "rewards/margins": 0.1263391077518463, | |
| "rewards/rejected": -0.5577641725540161, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.2832415420928403, | |
| "grad_norm": 31.5225772857666, | |
| "learning_rate": 3.62639349927083e-05, | |
| "logits/chosen": -0.2402796745300293, | |
| "logits/rejected": -0.8563323020935059, | |
| "logps/chosen": -4.622067928314209, | |
| "logps/rejected": -6.146195411682129, | |
| "loss": 17.5716, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -0.37094250321388245, | |
| "rewards/margins": 0.14315392076969147, | |
| "rewards/rejected": -0.5140964388847351, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.2871754523996853, | |
| "grad_norm": 62.2461051940918, | |
| "learning_rate": 3.610244167136095e-05, | |
| "logits/chosen": -0.09466689825057983, | |
| "logits/rejected": -0.4544064402580261, | |
| "logps/chosen": -5.546882629394531, | |
| "logps/rejected": -6.525860786437988, | |
| "loss": 24.1383, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": -0.47001561522483826, | |
| "rewards/margins": 0.06077291816473007, | |
| "rewards/rejected": -0.5307885408401489, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.2911093627065303, | |
| "grad_norm": 48.28466796875, | |
| "learning_rate": 3.593790723956935e-05, | |
| "logits/chosen": -0.2374683916568756, | |
| "logits/rejected": -0.373137503862381, | |
| "logps/chosen": -8.164457321166992, | |
| "logps/rejected": -8.156596183776855, | |
| "loss": 27.2267, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.7038506269454956, | |
| "rewards/margins": 0.0024669456761330366, | |
| "rewards/rejected": -0.7063175439834595, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.2950432730133753, | |
| "grad_norm": 51.48979568481445, | |
| "learning_rate": 3.577036277134012e-05, | |
| "logits/chosen": 0.5883103609085083, | |
| "logits/rejected": 0.345896452665329, | |
| "logps/chosen": -7.786595344543457, | |
| "logps/rejected": -8.453828811645508, | |
| "loss": 23.7109, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": -0.6850046515464783, | |
| "rewards/margins": 0.03222974017262459, | |
| "rewards/rejected": -0.7172344326972961, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.2989771833202203, | |
| "grad_norm": 47.203521728515625, | |
| "learning_rate": 3.5599839909155954e-05, | |
| "logits/chosen": 0.8187308311462402, | |
| "logits/rejected": 0.5813020467758179, | |
| "logps/chosen": -7.474339962005615, | |
| "logps/rejected": -8.487818717956543, | |
| "loss": 21.9183, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": -0.649897038936615, | |
| "rewards/margins": 0.06481163203716278, | |
| "rewards/rejected": -0.714708685874939, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.3029110936270653, | |
| "grad_norm": 66.43785095214844, | |
| "learning_rate": 3.542637085799967e-05, | |
| "logits/chosen": 0.7243896722793579, | |
| "logits/rejected": 0.633999228477478, | |
| "logps/chosen": -6.237640380859375, | |
| "logps/rejected": -7.760465145111084, | |
| "loss": 21.8449, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.5326268672943115, | |
| "rewards/margins": 0.1084168553352356, | |
| "rewards/rejected": -0.6410436630249023, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.3068450039339103, | |
| "grad_norm": 37.43156814575195, | |
| "learning_rate": 3.524998837927192e-05, | |
| "logits/chosen": 0.06674204766750336, | |
| "logits/rejected": -0.2028542459011078, | |
| "logps/chosen": -3.726958751678467, | |
| "logps/rejected": -4.262044906616211, | |
| "loss": 24.4969, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": -0.28034740686416626, | |
| "rewards/margins": 0.04259229078888893, | |
| "rewards/rejected": -0.3229396939277649, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.3107789142407553, | |
| "grad_norm": 43.61391067504883, | |
| "learning_rate": 3.5070725784603906e-05, | |
| "logits/chosen": -0.40263357758522034, | |
| "logits/rejected": -0.7386992573738098, | |
| "logps/chosen": -2.9942004680633545, | |
| "logps/rejected": -3.7471251487731934, | |
| "loss": 22.408, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": -0.19999414682388306, | |
| "rewards/margins": 0.05711622163653374, | |
| "rewards/rejected": -0.2571103572845459, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.3147128245476003, | |
| "grad_norm": 39.55924606323242, | |
| "learning_rate": 3.488861692956612e-05, | |
| "logits/chosen": -0.3814232647418976, | |
| "logits/rejected": -0.7326269149780273, | |
| "logps/chosen": -3.743687391281128, | |
| "logps/rejected": -4.247666835784912, | |
| "loss": 22.5808, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": -0.2401810586452484, | |
| "rewards/margins": 0.03669751435518265, | |
| "rewards/rejected": -0.27687856554985046, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.3147128245476003, | |
| "eval_logits/chosen": 1.05367910861969, | |
| "eval_logits/rejected": 0.8413508534431458, | |
| "eval_logps/chosen": -3.476516008377075, | |
| "eval_logps/rejected": -4.1527628898620605, | |
| "eval_loss": 22.322988510131836, | |
| "eval_rewards/accuracies": 0.643750011920929, | |
| "eval_rewards/chosen": -0.2451314926147461, | |
| "eval_rewards/margins": 0.045597635209560394, | |
| "eval_rewards/rejected": -0.2907291054725647, | |
| "eval_runtime": 262.6324, | |
| "eval_samples_per_second": 2.437, | |
| "eval_steps_per_second": 0.152, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.31864673485444533, | |
| "grad_norm": 106.93277740478516, | |
| "learning_rate": 3.470369620727433e-05, | |
| "logits/chosen": -0.2646043300628662, | |
| "logits/rejected": -0.5908231735229492, | |
| "logps/chosen": -4.6682844161987305, | |
| "logps/rejected": -4.9077653884887695, | |
| "loss": 24.9016, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": -0.31124863028526306, | |
| "rewards/margins": 0.024502381682395935, | |
| "rewards/rejected": -0.3357509970664978, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.3225806451612903, | |
| "grad_norm": 74.05709838867188, | |
| "learning_rate": 3.451599854189419e-05, | |
| "logits/chosen": 0.022719597443938255, | |
| "logits/rejected": -0.2066432684659958, | |
| "logps/chosen": -4.9405951499938965, | |
| "logps/rejected": -5.324969291687012, | |
| "loss": 24.7214, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": -0.37897494435310364, | |
| "rewards/margins": 0.023352503776550293, | |
| "rewards/rejected": -0.40232744812965393, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.32651455546813535, | |
| "grad_norm": 39.29796600341797, | |
| "learning_rate": 3.4325559382045344e-05, | |
| "logits/chosen": 0.5940214395523071, | |
| "logits/rejected": 0.3333088755607605, | |
| "logps/chosen": -4.753328323364258, | |
| "logps/rejected": -5.081197738647461, | |
| "loss": 24.0772, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": -0.36598071455955505, | |
| "rewards/margins": 0.0194702185690403, | |
| "rewards/rejected": -0.38545092940330505, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.3304484657749803, | |
| "grad_norm": 77.69596862792969, | |
| "learning_rate": 3.413241469410669e-05, | |
| "logits/chosen": 0.5746585726737976, | |
| "logits/rejected": 0.37664318084716797, | |
| "logps/chosen": -4.366871356964111, | |
| "logps/rejected": -4.923527717590332, | |
| "loss": 23.106, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.33017784357070923, | |
| "rewards/margins": 0.03850778192281723, | |
| "rewards/rejected": -0.36868563294410706, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.33438237608182536, | |
| "grad_norm": 37.37318801879883, | |
| "learning_rate": 3.3936600955423684e-05, | |
| "logits/chosen": 0.5197581052780151, | |
| "logits/rejected": 0.22815366089344025, | |
| "logps/chosen": -4.27942419052124, | |
| "logps/rejected": -5.002130031585693, | |
| "loss": 21.5593, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": -0.3183726668357849, | |
| "rewards/margins": 0.0549759566783905, | |
| "rewards/rejected": -0.3733486235141754, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.33831628638867034, | |
| "grad_norm": 51.467933654785156, | |
| "learning_rate": 3.373815514741928e-05, | |
| "logits/chosen": 0.5920094847679138, | |
| "logits/rejected": 0.27607864141464233, | |
| "logps/chosen": -4.812392234802246, | |
| "logps/rejected": -6.322897434234619, | |
| "loss": 20.3197, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": -0.3940238356590271, | |
| "rewards/margins": 0.08363697677850723, | |
| "rewards/rejected": -0.47766080498695374, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.3422501966955153, | |
| "grad_norm": 92.83573150634766, | |
| "learning_rate": 3.353711474860957e-05, | |
| "logits/chosen": 0.2608449459075928, | |
| "logits/rejected": 0.022218376398086548, | |
| "logps/chosen": -6.094309329986572, | |
| "logps/rejected": -6.749911308288574, | |
| "loss": 22.7358, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": -0.48861637711524963, | |
| "rewards/margins": 0.06393507868051529, | |
| "rewards/rejected": -0.5525515079498291, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.34618410700236035, | |
| "grad_norm": 117.08031463623047, | |
| "learning_rate": 3.333351772752559e-05, | |
| "logits/chosen": 0.36764952540397644, | |
| "logits/rejected": 0.11572384834289551, | |
| "logps/chosen": -5.9948601722717285, | |
| "logps/rejected": -7.810961723327637, | |
| "loss": 22.5633, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -0.4657825827598572, | |
| "rewards/margins": 0.0877792239189148, | |
| "rewards/rejected": -0.553561806678772, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.35011801730920533, | |
| "grad_norm": 42.3343620300293, | |
| "learning_rate": 3.31274025355426e-05, | |
| "logits/chosen": -0.1417434960603714, | |
| "logits/rejected": -0.3869401216506958, | |
| "logps/chosen": -4.742400169372559, | |
| "logps/rejected": -5.590722560882568, | |
| "loss": 23.695, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.37819725275039673, | |
| "rewards/margins": 0.05572297424077988, | |
| "rewards/rejected": -0.4339202046394348, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.35405192761605037, | |
| "grad_norm": 48.5923957824707, | |
| "learning_rate": 3.2918808099618145e-05, | |
| "logits/chosen": -0.13320264220237732, | |
| "logits/rejected": -0.4407239854335785, | |
| "logps/chosen": -4.052382946014404, | |
| "logps/rejected": -5.202944755554199, | |
| "loss": 23.9333, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": -0.2898481488227844, | |
| "rewards/margins": 0.06827215105295181, | |
| "rewards/rejected": -0.35812026262283325, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.35798583792289534, | |
| "grad_norm": 40.526222229003906, | |
| "learning_rate": 3.270777381494025e-05, | |
| "logits/chosen": 0.15568742156028748, | |
| "logits/rejected": -0.1973900943994522, | |
| "logps/chosen": -3.3848178386688232, | |
| "logps/rejected": -4.244564056396484, | |
| "loss": 22.4256, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": -0.24747803807258606, | |
| "rewards/margins": 0.0624859556555748, | |
| "rewards/rejected": -0.30996400117874146, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.3619197482297404, | |
| "grad_norm": 39.41511917114258, | |
| "learning_rate": 3.2494339537487316e-05, | |
| "logits/chosen": 0.2709997296333313, | |
| "logits/rejected": -0.02324852906167507, | |
| "logps/chosen": -4.2176713943481445, | |
| "logps/rejected": -4.712052345275879, | |
| "loss": 22.1339, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -0.30599862337112427, | |
| "rewards/margins": 0.05780113860964775, | |
| "rewards/rejected": -0.3637998104095459, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.36585365853658536, | |
| "grad_norm": 64.34850311279297, | |
| "learning_rate": 3.227854557650086e-05, | |
| "logits/chosen": 0.43827542662620544, | |
| "logits/rejected": 0.3022671937942505, | |
| "logps/chosen": -4.146628379821777, | |
| "logps/rejected": -4.721283912658691, | |
| "loss": 24.9713, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.32393088936805725, | |
| "rewards/margins": 0.03894919902086258, | |
| "rewards/rejected": -0.36288008093833923, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.3697875688434304, | |
| "grad_norm": 48.07391357421875, | |
| "learning_rate": 3.206043268687271e-05, | |
| "logits/chosen": 0.8742543458938599, | |
| "logits/rejected": 0.6306554079055786, | |
| "logps/chosen": -4.235721111297607, | |
| "logps/rejected": -4.655104160308838, | |
| "loss": 24.504, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": -0.3277904689311981, | |
| "rewards/margins": 0.03070756234228611, | |
| "rewards/rejected": -0.3584980368614197, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.37372147915027537, | |
| "grad_norm": 48.7944450378418, | |
| "learning_rate": 3.1840042061448034e-05, | |
| "logits/chosen": 0.8953019380569458, | |
| "logits/rejected": 0.6509414315223694, | |
| "logps/chosen": -3.809953212738037, | |
| "logps/rejected": -4.727410793304443, | |
| "loss": 21.5063, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": -0.2697634696960449, | |
| "rewards/margins": 0.04955270141363144, | |
| "rewards/rejected": -0.31931617856025696, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.3776553894571204, | |
| "grad_norm": 50.050758361816406, | |
| "learning_rate": 3.161741532324567e-05, | |
| "logits/chosen": 0.6901669502258301, | |
| "logits/rejected": 0.4217056632041931, | |
| "logps/chosen": -3.9408392906188965, | |
| "logps/rejected": -4.815566062927246, | |
| "loss": 20.9824, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.3004041314125061, | |
| "rewards/margins": 0.06339363753795624, | |
| "rewards/rejected": -0.36379775404930115, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.3815892997639654, | |
| "grad_norm": 31.20033836364746, | |
| "learning_rate": 3.139259451759715e-05, | |
| "logits/chosen": 0.13000288605690002, | |
| "logits/rejected": -0.044496648013591766, | |
| "logps/chosen": -3.5587058067321777, | |
| "logps/rejected": -4.0875444412231445, | |
| "loss": 23.3392, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": -0.25211071968078613, | |
| "rewards/margins": 0.0408257320523262, | |
| "rewards/rejected": -0.2929364740848541, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.38552321007081036, | |
| "grad_norm": 42.778175354003906, | |
| "learning_rate": 3.116562210420604e-05, | |
| "logits/chosen": 0.001088732504285872, | |
| "logits/rejected": -0.32067522406578064, | |
| "logps/chosen": -3.792126178741455, | |
| "logps/rejected": -5.263632297515869, | |
| "loss": 19.5225, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": -0.2830085754394531, | |
| "rewards/margins": 0.11108819395303726, | |
| "rewards/rejected": -0.3940967619419098, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.3894571203776554, | |
| "grad_norm": 51.158023834228516, | |
| "learning_rate": 3.093654094912901e-05, | |
| "logits/chosen": 0.14770345389842987, | |
| "logits/rejected": -0.27998632192611694, | |
| "logps/chosen": -3.2016499042510986, | |
| "logps/rejected": -4.281766414642334, | |
| "loss": 20.9501, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -0.23279622197151184, | |
| "rewards/margins": 0.07809507101774216, | |
| "rewards/rejected": -0.3108913004398346, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.3933910306845004, | |
| "grad_norm": 56.20425033569336, | |
| "learning_rate": 3.070539431668008e-05, | |
| "logits/chosen": 0.262935608625412, | |
| "logits/rejected": 0.04751387611031532, | |
| "logps/chosen": -3.5432257652282715, | |
| "logps/rejected": -4.892638683319092, | |
| "loss": 19.8621, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": -0.2724161148071289, | |
| "rewards/margins": 0.10621275752782822, | |
| "rewards/rejected": -0.3786288797855377, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.3933910306845004, | |
| "eval_logits/chosen": 0.6740007400512695, | |
| "eval_logits/rejected": 0.4591088891029358, | |
| "eval_logps/chosen": -4.5571136474609375, | |
| "eval_logps/rejected": -5.613867282867432, | |
| "eval_loss": 21.918312072753906, | |
| "eval_rewards/accuracies": 0.659375011920929, | |
| "eval_rewards/chosen": -0.35319122672080994, | |
| "eval_rewards/margins": 0.08364833891391754, | |
| "eval_rewards/rejected": -0.43683958053588867, | |
| "eval_runtime": 263.6435, | |
| "eval_samples_per_second": 2.428, | |
| "eval_steps_per_second": 0.152, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.3973249409913454, | |
| "grad_norm": 58.8707275390625, | |
| "learning_rate": 3.0472225861259792e-05, | |
| "logits/chosen": 0.5369864702224731, | |
| "logits/rejected": 0.22990770637989044, | |
| "logps/chosen": -4.736105442047119, | |
| "logps/rejected": -6.215359687805176, | |
| "loss": 19.8037, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -0.3894490599632263, | |
| "rewards/margins": 0.10317282378673553, | |
| "rewards/rejected": -0.49262189865112305, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 0.4012588512981904, | |
| "grad_norm": 70.14684295654297, | |
| "learning_rate": 3.023707961911056e-05, | |
| "logits/chosen": 0.8286817669868469, | |
| "logits/rejected": 0.553676187992096, | |
| "logps/chosen": -5.788529872894287, | |
| "logps/rejected": -7.536166191101074, | |
| "loss": 17.8624, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -0.49081355333328247, | |
| "rewards/margins": 0.14760908484458923, | |
| "rewards/rejected": -0.6384226083755493, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.4051927616050354, | |
| "grad_norm": 64.01305389404297, | |
| "learning_rate": 3.0000000000000004e-05, | |
| "logits/chosen": 0.9336326718330383, | |
| "logits/rejected": 0.7778112292289734, | |
| "logps/chosen": -6.726840972900391, | |
| "logps/rejected": -7.550817966461182, | |
| "loss": 22.4924, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": -0.57944256067276, | |
| "rewards/margins": 0.06806603819131851, | |
| "rewards/rejected": -0.6475085616111755, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 0.4091266719118804, | |
| "grad_norm": 78.60610961914062, | |
| "learning_rate": 2.976103177883374e-05, | |
| "logits/chosen": 1.424285650253296, | |
| "logits/rejected": 1.2528765201568604, | |
| "logps/chosen": -6.3146467208862305, | |
| "logps/rejected": -7.42047643661499, | |
| "loss": 22.4632, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": -0.5491231083869934, | |
| "rewards/margins": 0.08809302002191544, | |
| "rewards/rejected": -0.6372160911560059, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.41306058221872544, | |
| "grad_norm": 53.94759750366211, | |
| "learning_rate": 2.9520220087199142e-05, | |
| "logits/chosen": 2.070854663848877, | |
| "logits/rejected": 1.9312273263931274, | |
| "logps/chosen": -7.275876522064209, | |
| "logps/rejected": -7.790124416351318, | |
| "loss": 22.9161, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.6180551052093506, | |
| "rewards/margins": 0.04511018842458725, | |
| "rewards/rejected": -0.663165271282196, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.4169944925255704, | |
| "grad_norm": 56.0859489440918, | |
| "learning_rate": 2.9277610404841792e-05, | |
| "logits/chosen": 2.1373679637908936, | |
| "logits/rejected": 1.8990551233291626, | |
| "logps/chosen": -6.323026180267334, | |
| "logps/rejected": -7.068973541259766, | |
| "loss": 21.5577, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": -0.5458236932754517, | |
| "rewards/margins": 0.058930903673172, | |
| "rewards/rejected": -0.6047546863555908, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.4209284028324154, | |
| "grad_norm": 38.24552536010742, | |
| "learning_rate": 2.903324855107617e-05, | |
| "logits/chosen": 1.698553442955017, | |
| "logits/rejected": 1.4752168655395508, | |
| "logps/chosen": -6.036180019378662, | |
| "logps/rejected": -7.159039497375488, | |
| "loss": 20.6918, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": -0.4921696186065674, | |
| "rewards/margins": 0.09151138365268707, | |
| "rewards/rejected": -0.5836809873580933, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 0.42486231313926043, | |
| "grad_norm": 78.7889633178711, | |
| "learning_rate": 2.8787180676132222e-05, | |
| "logits/chosen": 1.3787410259246826, | |
| "logits/rejected": 1.1702592372894287, | |
| "logps/chosen": -5.483891010284424, | |
| "logps/rejected": -6.863039493560791, | |
| "loss": 21.8569, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -0.4555833339691162, | |
| "rewards/margins": 0.10881421715021133, | |
| "rewards/rejected": -0.5643975138664246, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.4287962234461054, | |
| "grad_norm": 58.13717269897461, | |
| "learning_rate": 2.8539453252439388e-05, | |
| "logits/chosen": 1.238527536392212, | |
| "logits/rejected": 1.0659363269805908, | |
| "logps/chosen": -4.524688243865967, | |
| "logps/rejected": -5.795032978057861, | |
| "loss": 19.8882, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": -0.365116685628891, | |
| "rewards/margins": 0.0977514460682869, | |
| "rewards/rejected": -0.4628681540489197, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 0.43273013375295044, | |
| "grad_norm": 57.31916809082031, | |
| "learning_rate": 2.829011306584983e-05, | |
| "logits/chosen": 1.0495655536651611, | |
| "logits/rejected": 0.9189395904541016, | |
| "logps/chosen": -4.698520183563232, | |
| "logps/rejected": -5.4618024826049805, | |
| "loss": 22.771, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.3808877170085907, | |
| "rewards/margins": 0.05942262336611748, | |
| "rewards/rejected": -0.4403103291988373, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.4366640440597954, | |
| "grad_norm": 57.70218276977539, | |
| "learning_rate": 2.8039207206802444e-05, | |
| "logits/chosen": 1.0372337102890015, | |
| "logits/rejected": 0.7637672424316406, | |
| "logps/chosen": -5.192538261413574, | |
| "logps/rejected": -6.2920708656311035, | |
| "loss": 21.0183, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.4280625283718109, | |
| "rewards/margins": 0.07608579099178314, | |
| "rewards/rejected": -0.5041483640670776, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 0.44059795436664045, | |
| "grad_norm": 38.57474899291992, | |
| "learning_rate": 2.778678306142936e-05, | |
| "logits/chosen": 1.3346863985061646, | |
| "logits/rejected": 1.2482701539993286, | |
| "logps/chosen": -4.537243843078613, | |
| "logps/rejected": -5.363685607910156, | |
| "loss": 21.725, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": -0.367723673582077, | |
| "rewards/margins": 0.07422361522912979, | |
| "rewards/rejected": -0.4419472813606262, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.44453186467348543, | |
| "grad_norm": 47.30175018310547, | |
| "learning_rate": 2.753288830260655e-05, | |
| "logits/chosen": 1.081312894821167, | |
| "logits/rejected": 1.0209182500839233, | |
| "logps/chosen": -4.504608154296875, | |
| "logps/rejected": -5.094980716705322, | |
| "loss": 23.979, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.34262794256210327, | |
| "rewards/margins": 0.04645577073097229, | |
| "rewards/rejected": -0.38908374309539795, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 0.44846577498033047, | |
| "grad_norm": 41.14936065673828, | |
| "learning_rate": 2.727757088095037e-05, | |
| "logits/chosen": 1.2744576930999756, | |
| "logits/rejected": 0.9736446142196655, | |
| "logps/chosen": -4.100034236907959, | |
| "logps/rejected": -5.128809452056885, | |
| "loss": 20.4892, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -0.3064883053302765, | |
| "rewards/margins": 0.08413257449865341, | |
| "rewards/rejected": -0.3906208872795105, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.45239968528717545, | |
| "grad_norm": 29.25827407836914, | |
| "learning_rate": 2.7020879015761555e-05, | |
| "logits/chosen": 1.315836787223816, | |
| "logits/rejected": 1.091429352760315, | |
| "logps/chosen": -4.02718448638916, | |
| "logps/rejected": -4.86886739730835, | |
| "loss": 21.384, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": -0.2975132465362549, | |
| "rewards/margins": 0.07079926878213882, | |
| "rewards/rejected": -0.3683125078678131, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.4563335955940205, | |
| "grad_norm": 156.1727752685547, | |
| "learning_rate": 2.6762861185918532e-05, | |
| "logits/chosen": 1.271761178970337, | |
| "logits/rejected": 1.0751326084136963, | |
| "logps/chosen": -4.122300624847412, | |
| "logps/rejected": -5.07947301864624, | |
| "loss": 20.2656, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.3042159378528595, | |
| "rewards/margins": 0.0826338604092598, | |
| "rewards/rejected": -0.3868497610092163, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.46026750590086546, | |
| "grad_norm": 33.745567321777344, | |
| "learning_rate": 2.6503566120721685e-05, | |
| "logits/chosen": 1.1284042596817017, | |
| "logits/rejected": 0.8111549615859985, | |
| "logps/chosen": -4.277104377746582, | |
| "logps/rejected": -5.083390235900879, | |
| "loss": 20.6651, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": -0.3438721299171448, | |
| "rewards/margins": 0.06679800897836685, | |
| "rewards/rejected": -0.41067013144493103, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 0.46420141620771044, | |
| "grad_norm": 51.66657638549805, | |
| "learning_rate": 2.6243042790690332e-05, | |
| "logits/chosen": 0.8191879987716675, | |
| "logits/rejected": 0.7653782367706299, | |
| "logps/chosen": -5.171725273132324, | |
| "logps/rejected": -5.932669162750244, | |
| "loss": 22.7086, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": -0.4288380742073059, | |
| "rewards/margins": 0.0465051606297493, | |
| "rewards/rejected": -0.4753432869911194, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.46813532651455547, | |
| "grad_norm": 44.73387908935547, | |
| "learning_rate": 2.5981340398314148e-05, | |
| "logits/chosen": 0.5237664580345154, | |
| "logits/rejected": 0.2646290957927704, | |
| "logps/chosen": -4.674435615539551, | |
| "logps/rejected": -6.298100471496582, | |
| "loss": 19.2422, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -0.3852913975715637, | |
| "rewards/margins": 0.1304633468389511, | |
| "rewards/rejected": -0.5157546997070312, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 0.47206923682140045, | |
| "grad_norm": 45.94012451171875, | |
| "learning_rate": 2.571850836876074e-05, | |
| "logits/chosen": 0.5761805176734924, | |
| "logits/rejected": 0.38747546076774597, | |
| "logps/chosen": -5.026305198669434, | |
| "logps/rejected": -7.055234432220459, | |
| "loss": 19.6134, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.4065484404563904, | |
| "rewards/margins": 0.10743912309408188, | |
| "rewards/rejected": -0.5139876008033752, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.47206923682140045, | |
| "eval_logits/chosen": 0.718625545501709, | |
| "eval_logits/rejected": 0.5276994705200195, | |
| "eval_logps/chosen": -5.122862815856934, | |
| "eval_logps/rejected": -6.157461643218994, | |
| "eval_loss": 21.601177215576172, | |
| "eval_rewards/accuracies": 0.6812499761581421, | |
| "eval_rewards/chosen": -0.40976619720458984, | |
| "eval_rewards/margins": 0.08143284171819687, | |
| "eval_rewards/rejected": -0.4911990761756897, | |
| "eval_runtime": 265.5044, | |
| "eval_samples_per_second": 2.411, | |
| "eval_steps_per_second": 0.151, | |
| "step": 600 | |
| } | |
| ], | |
| "logging_steps": 5, | |
| "max_steps": 1271, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 10, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 0.0, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |