| { |
| "best_global_step": null, |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 2.9871794871794872, |
| "eval_steps": 500, |
| "global_step": 699, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.021367521367521368, |
| "grad_norm": 16.125, |
| "learning_rate": 1.7556090538745385e-06, |
| "logits/chosen": -3.5723183155059814, |
| "logits/rejected": -3.5035064220428467, |
| "logps/chosen": -41.11209487915039, |
| "logps/rejected": -79.84024047851562, |
| "loss": 0.693, |
| "rewards/accuracies": 0.3125, |
| "rewards/chosen": -0.00025137339252978563, |
| "rewards/margins": 0.0003745006979443133, |
| "rewards/rejected": -0.0006258742650970817, |
| "step": 5 |
| }, |
| { |
| "epoch": 0.042735042735042736, |
| "grad_norm": 16.625, |
| "learning_rate": 3.950120371217711e-06, |
| "logits/chosen": -3.5866427421569824, |
| "logits/rejected": -3.506498336791992, |
| "logps/chosen": -40.02031707763672, |
| "logps/rejected": -78.2303237915039, |
| "loss": 0.6804, |
| "rewards/accuracies": 0.8083333969116211, |
| "rewards/chosen": -0.0030814227648079395, |
| "rewards/margins": 0.025840703397989273, |
| "rewards/rejected": -0.028922125697135925, |
| "step": 10 |
| }, |
| { |
| "epoch": 0.0641025641025641, |
| "grad_norm": 14.75, |
| "learning_rate": 6.144631688560886e-06, |
| "logits/chosen": -3.5990729331970215, |
| "logits/rejected": -3.507888078689575, |
| "logps/chosen": -39.841556549072266, |
| "logps/rejected": -81.43814086914062, |
| "loss": 0.6304, |
| "rewards/accuracies": 0.987500011920929, |
| "rewards/chosen": -0.007047838531434536, |
| "rewards/margins": 0.131792351603508, |
| "rewards/rejected": -0.1388401836156845, |
| "step": 15 |
| }, |
| { |
| "epoch": 0.08547008547008547, |
| "grad_norm": 13.6875, |
| "learning_rate": 8.339143005904057e-06, |
| "logits/chosen": -3.5527968406677246, |
| "logits/rejected": -3.4919159412384033, |
| "logps/chosen": -39.75408935546875, |
| "logps/rejected": -80.72636413574219, |
| "loss": 0.5433, |
| "rewards/accuracies": 0.9750000238418579, |
| "rewards/chosen": -0.014718407765030861, |
| "rewards/margins": 0.33949846029281616, |
| "rewards/rejected": -0.35421687364578247, |
| "step": 20 |
| }, |
| { |
| "epoch": 0.10683760683760683, |
| "grad_norm": 11.1875, |
| "learning_rate": 1.0533654323247232e-05, |
| "logits/chosen": -3.5133519172668457, |
| "logits/rejected": -3.472460985183716, |
| "logps/chosen": -40.00214385986328, |
| "logps/rejected": -85.55389404296875, |
| "loss": 0.396, |
| "rewards/accuracies": 0.98333340883255, |
| "rewards/chosen": -0.041852615773677826, |
| "rewards/margins": 0.7774692177772522, |
| "rewards/rejected": -0.8193218111991882, |
| "step": 25 |
| }, |
| { |
| "epoch": 0.1282051282051282, |
| "grad_norm": 7.59375, |
| "learning_rate": 1.2728165640590407e-05, |
| "logits/chosen": -3.4411749839782715, |
| "logits/rejected": -3.445669651031494, |
| "logps/chosen": -42.452301025390625, |
| "logps/rejected": -93.04366302490234, |
| "loss": 0.259, |
| "rewards/accuracies": 0.9958332777023315, |
| "rewards/chosen": -0.13290514051914215, |
| "rewards/margins": 1.4268956184387207, |
| "rewards/rejected": -1.5598008632659912, |
| "step": 30 |
| }, |
| { |
| "epoch": 0.14957264957264957, |
| "grad_norm": 4.25, |
| "learning_rate": 1.4922676957933578e-05, |
| "logits/chosen": -3.2215240001678467, |
| "logits/rejected": -3.2669529914855957, |
| "logps/chosen": -44.15301513671875, |
| "logps/rejected": -108.67581939697266, |
| "loss": 0.1196, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -0.4359213709831238, |
| "rewards/margins": 2.7526631355285645, |
| "rewards/rejected": -3.1885838508605957, |
| "step": 35 |
| }, |
| { |
| "epoch": 0.17094017094017094, |
| "grad_norm": 2.046875, |
| "learning_rate": 1.5360556888469565e-05, |
| "logits/chosen": -2.778581142425537, |
| "logits/rejected": -2.812112331390381, |
| "logps/chosen": -52.46855926513672, |
| "logps/rejected": -143.13113403320312, |
| "loss": 0.0425, |
| "rewards/accuracies": 0.9916666746139526, |
| "rewards/chosen": -1.2584294080734253, |
| "rewards/margins": 5.229200839996338, |
| "rewards/rejected": -6.4876298904418945, |
| "step": 40 |
| }, |
| { |
| "epoch": 0.19230769230769232, |
| "grad_norm": 1.6875, |
| "learning_rate": 1.535640428282884e-05, |
| "logits/chosen": -2.3370110988616943, |
| "logits/rejected": -2.2969162464141846, |
| "logps/chosen": -51.32811737060547, |
| "logps/rejected": -161.91485595703125, |
| "loss": 0.0166, |
| "rewards/accuracies": 0.9958333969116211, |
| "rewards/chosen": -1.2713468074798584, |
| "rewards/margins": 7.2617645263671875, |
| "rewards/rejected": -8.533111572265625, |
| "step": 45 |
| }, |
| { |
| "epoch": 0.21367521367521367, |
| "grad_norm": 9.8125, |
| "learning_rate": 1.5349059809872097e-05, |
| "logits/chosen": -1.9922136068344116, |
| "logits/rejected": -1.9316717386245728, |
| "logps/chosen": -57.087364196777344, |
| "logps/rejected": -180.17538452148438, |
| "loss": 0.0198, |
| "rewards/accuracies": 0.9958333969116211, |
| "rewards/chosen": -1.6679397821426392, |
| "rewards/margins": 8.57298755645752, |
| "rewards/rejected": -10.240926742553711, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.23504273504273504, |
| "grad_norm": 0.71484375, |
| "learning_rate": 1.5338527542732884e-05, |
| "logits/chosen": -1.988929033279419, |
| "logits/rejected": -1.921478271484375, |
| "logps/chosen": -58.5947380065918, |
| "logps/rejected": -196.42152404785156, |
| "loss": 0.0097, |
| "rewards/accuracies": 0.9958332777023315, |
| "rewards/chosen": -1.8699321746826172, |
| "rewards/margins": 9.876224517822266, |
| "rewards/rejected": -11.746156692504883, |
| "step": 55 |
| }, |
| { |
| "epoch": 0.2564102564102564, |
| "grad_norm": 3.0625, |
| "learning_rate": 1.532481332244717e-05, |
| "logits/chosen": -1.9385712146759033, |
| "logits/rejected": -1.8249061107635498, |
| "logps/chosen": -57.23170852661133, |
| "logps/rejected": -198.87722778320312, |
| "loss": 0.0046, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -1.809206247329712, |
| "rewards/margins": 10.275887489318848, |
| "rewards/rejected": -12.08509349822998, |
| "step": 60 |
| }, |
| { |
| "epoch": 0.2777777777777778, |
| "grad_norm": 8.625, |
| "learning_rate": 1.5307924754713968e-05, |
| "logits/chosen": -1.939802885055542, |
| "logits/rejected": -1.8658767938613892, |
| "logps/chosen": -59.283775329589844, |
| "logps/rejected": -196.01666259765625, |
| "loss": 0.0104, |
| "rewards/accuracies": 0.9958333969116211, |
| "rewards/chosen": -1.8633003234863281, |
| "rewards/margins": 9.977038383483887, |
| "rewards/rejected": -11.840339660644531, |
| "step": 65 |
| }, |
| { |
| "epoch": 0.29914529914529914, |
| "grad_norm": 0.32421875, |
| "learning_rate": 1.528787120567736e-05, |
| "logits/chosen": -2.0353214740753174, |
| "logits/rejected": -1.9403060674667358, |
| "logps/chosen": -53.66817092895508, |
| "logps/rejected": -192.67019653320312, |
| "loss": 0.0032, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -1.436279058456421, |
| "rewards/margins": 10.137629508972168, |
| "rewards/rejected": -11.573908805847168, |
| "step": 70 |
| }, |
| { |
| "epoch": 0.32051282051282054, |
| "grad_norm": 0.33203125, |
| "learning_rate": 1.526466379673215e-05, |
| "logits/chosen": -2.051652669906616, |
| "logits/rejected": -1.9418226480484009, |
| "logps/chosen": -55.503456115722656, |
| "logps/rejected": -201.16795349121094, |
| "loss": 0.006, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -1.6412267684936523, |
| "rewards/margins": 10.790639877319336, |
| "rewards/rejected": -12.431867599487305, |
| "step": 75 |
| }, |
| { |
| "epoch": 0.3418803418803419, |
| "grad_norm": 6.375, |
| "learning_rate": 1.5238315398356126e-05, |
| "logits/chosen": -2.026411533355713, |
| "logits/rejected": -1.8895785808563232, |
| "logps/chosen": -58.959983825683594, |
| "logps/rejected": -205.6489715576172, |
| "loss": 0.0054, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -1.9467172622680664, |
| "rewards/margins": 11.003195762634277, |
| "rewards/rejected": -12.949913024902344, |
| "step": 80 |
| }, |
| { |
| "epoch": 0.36324786324786323, |
| "grad_norm": 0.032958984375, |
| "learning_rate": 1.5208840622972272e-05, |
| "logits/chosen": -2.0241971015930176, |
| "logits/rejected": -1.8603122234344482, |
| "logps/chosen": -59.0450553894043, |
| "logps/rejected": -208.4755401611328, |
| "loss": 0.023, |
| "rewards/accuracies": 0.9958333969116211, |
| "rewards/chosen": -1.7697633504867554, |
| "rewards/margins": 11.253450393676758, |
| "rewards/rejected": -13.023214340209961, |
| "step": 85 |
| }, |
| { |
| "epoch": 0.38461538461538464, |
| "grad_norm": 1.203125, |
| "learning_rate": 1.5176255816844948e-05, |
| "logits/chosen": -1.9871257543563843, |
| "logits/rejected": -1.795332670211792, |
| "logps/chosen": -50.853111267089844, |
| "logps/rejected": -205.7463836669922, |
| "loss": 0.0063, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -1.1833254098892212, |
| "rewards/margins": 11.639432907104492, |
| "rewards/rejected": -12.82275676727295, |
| "step": 90 |
| }, |
| { |
| "epoch": 0.405982905982906, |
| "grad_norm": 0.59765625, |
| "learning_rate": 1.5140579051014502e-05, |
| "logits/chosen": -1.9575443267822266, |
| "logits/rejected": -1.730380654335022, |
| "logps/chosen": -55.5305290222168, |
| "logps/rejected": -221.403564453125, |
| "loss": 0.0028, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -1.54853355884552, |
| "rewards/margins": 12.512056350708008, |
| "rewards/rejected": -14.060590744018555, |
| "step": 95 |
| }, |
| { |
| "epoch": 0.42735042735042733, |
| "grad_norm": 0.107421875, |
| "learning_rate": 1.5101830111275334e-05, |
| "logits/chosen": -1.9054193496704102, |
| "logits/rejected": -1.7034189701080322, |
| "logps/chosen": -59.12446975708008, |
| "logps/rejected": -213.0598907470703, |
| "loss": 0.0117, |
| "rewards/accuracies": 0.9958333969116211, |
| "rewards/chosen": -1.9219242334365845, |
| "rewards/margins": 11.585506439208984, |
| "rewards/rejected": -13.507431030273438, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.44871794871794873, |
| "grad_norm": 3.484375, |
| "learning_rate": 1.5060030487203004e-05, |
| "logits/chosen": -1.84870183467865, |
| "logits/rejected": -1.5906012058258057, |
| "logps/chosen": -61.794334411621094, |
| "logps/rejected": -229.24472045898438, |
| "loss": 0.0058, |
| "rewards/accuracies": 0.9958333969116211, |
| "rewards/chosen": -2.169433116912842, |
| "rewards/margins": 12.846372604370117, |
| "rewards/rejected": -15.015806198120117, |
| "step": 105 |
| }, |
| { |
| "epoch": 0.4700854700854701, |
| "grad_norm": 0.0888671875, |
| "learning_rate": 1.501520336023643e-05, |
| "logits/chosen": -1.8036330938339233, |
| "logits/rejected": -1.4892421960830688, |
| "logps/chosen": -63.50853729248047, |
| "logps/rejected": -225.1219940185547, |
| "loss": 0.0011, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.3367650508880615, |
| "rewards/margins": 12.470510482788086, |
| "rewards/rejected": -14.807273864746094, |
| "step": 110 |
| }, |
| { |
| "epoch": 0.49145299145299143, |
| "grad_norm": 2.203125, |
| "learning_rate": 1.4967373590821828e-05, |
| "logits/chosen": -1.77732253074646, |
| "logits/rejected": -1.4749488830566406, |
| "logps/chosen": -69.56774139404297, |
| "logps/rejected": -229.52078247070312, |
| "loss": 0.0192, |
| "rewards/accuracies": 0.9916666746139526, |
| "rewards/chosen": -2.8969478607177734, |
| "rewards/margins": 12.228403091430664, |
| "rewards/rejected": -15.125350952148438, |
| "step": 115 |
| }, |
| { |
| "epoch": 0.5128205128205128, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 1.491656770462546e-05, |
| "logits/chosen": -1.6314146518707275, |
| "logits/rejected": -1.2666215896606445, |
| "logps/chosen": -61.828941345214844, |
| "logps/rejected": -231.4860076904297, |
| "loss": 0.0026, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.1970794200897217, |
| "rewards/margins": 13.084007263183594, |
| "rewards/rejected": -15.281087875366211, |
| "step": 120 |
| }, |
| { |
| "epoch": 0.5341880341880342, |
| "grad_norm": 9.5625, |
| "learning_rate": 1.4862813877822923e-05, |
| "logits/chosen": -1.695638656616211, |
| "logits/rejected": -1.3275301456451416, |
| "logps/chosen": -59.98823165893555, |
| "logps/rejected": -228.39599609375, |
| "loss": 0.0141, |
| "rewards/accuracies": 0.9916666746139526, |
| "rewards/chosen": -1.9660316705703735, |
| "rewards/margins": 13.168081283569336, |
| "rewards/rejected": -15.134114265441895, |
| "step": 125 |
| }, |
| { |
| "epoch": 0.5555555555555556, |
| "grad_norm": 0.2021484375, |
| "learning_rate": 1.4806141921473063e-05, |
| "logits/chosen": -1.7461655139923096, |
| "logits/rejected": -1.3385170698165894, |
| "logps/chosen": -57.775604248046875, |
| "logps/rejected": -232.9607391357422, |
| "loss": 0.006, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -1.762024164199829, |
| "rewards/margins": 13.665751457214355, |
| "rewards/rejected": -15.427775382995605, |
| "step": 130 |
| }, |
| { |
| "epoch": 0.5769230769230769, |
| "grad_norm": 0.10791015625, |
| "learning_rate": 1.4746583264985202e-05, |
| "logits/chosen": -1.7675241231918335, |
| "logits/rejected": -1.349183201789856, |
| "logps/chosen": -59.05058670043945, |
| "logps/rejected": -236.23599243164062, |
| "loss": 0.0032, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -1.866996169090271, |
| "rewards/margins": 13.846403121948242, |
| "rewards/rejected": -15.713399887084961, |
| "step": 135 |
| }, |
| { |
| "epoch": 0.5982905982905983, |
| "grad_norm": 0.029052734375, |
| "learning_rate": 1.468417093868888e-05, |
| "logits/chosen": -1.7844130992889404, |
| "logits/rejected": -1.393346905708313, |
| "logps/chosen": -55.1547966003418, |
| "logps/rejected": -230.82485961914062, |
| "loss": 0.0039, |
| "rewards/accuracies": 0.9958333969116211, |
| "rewards/chosen": -1.6007912158966064, |
| "rewards/margins": 13.634336471557617, |
| "rewards/rejected": -15.235128402709961, |
| "step": 140 |
| }, |
| { |
| "epoch": 0.6196581196581197, |
| "grad_norm": 0.10888671875, |
| "learning_rate": 1.4618939555515721e-05, |
| "logits/chosen": -1.736817717552185, |
| "logits/rejected": -1.3590004444122314, |
| "logps/chosen": -57.30821990966797, |
| "logps/rejected": -228.35568237304688, |
| "loss": 0.0029, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -1.7177435159683228, |
| "rewards/margins": 13.364781379699707, |
| "rewards/rejected": -15.08252239227295, |
| "step": 145 |
| }, |
| { |
| "epoch": 0.6410256410256411, |
| "grad_norm": 0.0126953125, |
| "learning_rate": 1.455092529180363e-05, |
| "logits/chosen": -1.767240285873413, |
| "logits/rejected": -1.3727061748504639, |
| "logps/chosen": -54.07440185546875, |
| "logps/rejected": -230.85598754882812, |
| "loss": 0.0004, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -1.538040041923523, |
| "rewards/margins": 13.841748237609863, |
| "rewards/rejected": -15.379789352416992, |
| "step": 150 |
| }, |
| { |
| "epoch": 0.6623931623931624, |
| "grad_norm": 0.296875, |
| "learning_rate": 1.4480165867233946e-05, |
| "logits/chosen": -1.7429969310760498, |
| "logits/rejected": -1.3700876235961914, |
| "logps/chosen": -58.81968307495117, |
| "logps/rejected": -237.78775024414062, |
| "loss": 0.005, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -1.9107345342636108, |
| "rewards/margins": 13.975476264953613, |
| "rewards/rejected": -15.886209487915039, |
| "step": 155 |
| }, |
| { |
| "epoch": 0.6837606837606838, |
| "grad_norm": 3.71875, |
| "learning_rate": 1.440670052391267e-05, |
| "logits/chosen": -1.7467654943466187, |
| "logits/rejected": -1.333251714706421, |
| "logps/chosen": -55.909812927246094, |
| "logps/rejected": -228.398681640625, |
| "loss": 0.006, |
| "rewards/accuracies": 0.9958333969116211, |
| "rewards/chosen": -1.737717866897583, |
| "rewards/margins": 13.500417709350586, |
| "rewards/rejected": -15.238134384155273, |
| "step": 160 |
| }, |
| { |
| "epoch": 0.7051282051282052, |
| "grad_norm": 2.0625, |
| "learning_rate": 1.4330570004607398e-05, |
| "logits/chosen": -1.8055540323257446, |
| "logits/rejected": -1.365281343460083, |
| "logps/chosen": -56.362159729003906, |
| "logps/rejected": -241.47616577148438, |
| "loss": 0.0033, |
| "rewards/accuracies": 0.9958333969116211, |
| "rewards/chosen": -1.7291532754898071, |
| "rewards/margins": 14.622825622558594, |
| "rewards/rejected": -16.351978302001953, |
| "step": 165 |
| }, |
| { |
| "epoch": 0.7264957264957265, |
| "grad_norm": 0.71875, |
| "learning_rate": 1.4251816530151986e-05, |
| "logits/chosen": -1.7655203342437744, |
| "logits/rejected": -1.3694696426391602, |
| "logps/chosen": -59.91992950439453, |
| "logps/rejected": -240.9622039794922, |
| "loss": 0.0035, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -1.9651933908462524, |
| "rewards/margins": 14.319523811340332, |
| "rewards/rejected": -16.284717559814453, |
| "step": 170 |
| }, |
| { |
| "epoch": 0.7478632478632479, |
| "grad_norm": 0.228515625, |
| "learning_rate": 1.4170483776031526e-05, |
| "logits/chosen": -1.705065369606018, |
| "logits/rejected": -1.2570855617523193, |
| "logps/chosen": -59.033119201660156, |
| "logps/rejected": -244.8512725830078, |
| "loss": 0.0004, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -1.9400125741958618, |
| "rewards/margins": 14.696588516235352, |
| "rewards/rejected": -16.636600494384766, |
| "step": 175 |
| }, |
| { |
| "epoch": 0.7692307692307693, |
| "grad_norm": 0.048095703125, |
| "learning_rate": 1.4086616848160574e-05, |
| "logits/chosen": -1.6960264444351196, |
| "logits/rejected": -1.2389576435089111, |
| "logps/chosen": -62.79465866088867, |
| "logps/rejected": -241.3025360107422, |
| "loss": 0.0004, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.3021397590637207, |
| "rewards/margins": 14.0509614944458, |
| "rewards/rejected": -16.353099822998047, |
| "step": 180 |
| }, |
| { |
| "epoch": 0.7905982905982906, |
| "grad_norm": 0.087890625, |
| "learning_rate": 1.4000262257868096e-05, |
| "logits/chosen": -1.665123701095581, |
| "logits/rejected": -1.1488758325576782, |
| "logps/chosen": -61.19834518432617, |
| "logps/rejected": -244.506103515625, |
| "loss": 0.0027, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.189235210418701, |
| "rewards/margins": 14.604774475097656, |
| "rewards/rejected": -16.794010162353516, |
| "step": 185 |
| }, |
| { |
| "epoch": 0.811965811965812, |
| "grad_norm": 0.921875, |
| "learning_rate": 1.3911467896102994e-05, |
| "logits/chosen": -1.6246988773345947, |
| "logits/rejected": -1.1719400882720947, |
| "logps/chosen": -58.097412109375, |
| "logps/rejected": -242.1889190673828, |
| "loss": 0.0049, |
| "rewards/accuracies": 0.9958333969116211, |
| "rewards/chosen": -1.8747377395629883, |
| "rewards/margins": 14.691726684570312, |
| "rewards/rejected": -16.566465377807617, |
| "step": 190 |
| }, |
| { |
| "epoch": 0.8333333333333334, |
| "grad_norm": 0.216796875, |
| "learning_rate": 1.3820283006874503e-05, |
| "logits/chosen": -1.6263649463653564, |
| "logits/rejected": -1.1733753681182861, |
| "logps/chosen": -64.73831939697266, |
| "logps/rejected": -246.49752807617188, |
| "loss": 0.0091, |
| "rewards/accuracies": 0.9958333969116211, |
| "rewards/chosen": -2.265516996383667, |
| "rewards/margins": 14.604884147644043, |
| "rewards/rejected": -16.87040138244629, |
| "step": 195 |
| }, |
| { |
| "epoch": 0.8547008547008547, |
| "grad_norm": 0.1875, |
| "learning_rate": 1.372675815994221e-05, |
| "logits/chosen": -1.5483165979385376, |
| "logits/rejected": -1.0878021717071533, |
| "logps/chosen": -52.65502166748047, |
| "logps/rejected": -248.62643432617188, |
| "loss": 0.0008, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -1.5096328258514404, |
| "rewards/margins": 15.69578742980957, |
| "rewards/rejected": -17.205419540405273, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.8760683760683761, |
| "grad_norm": 9.5, |
| "learning_rate": 1.3630945222770829e-05, |
| "logits/chosen": -1.5609447956085205, |
| "logits/rejected": -1.0415008068084717, |
| "logps/chosen": -58.0806884765625, |
| "logps/rejected": -255.28414916992188, |
| "loss": 0.0182, |
| "rewards/accuracies": 0.987500011920929, |
| "rewards/chosen": -1.7902355194091797, |
| "rewards/margins": 15.908655166625977, |
| "rewards/rejected": -17.69889259338379, |
| "step": 205 |
| }, |
| { |
| "epoch": 0.8974358974358975, |
| "grad_norm": 1.6484375, |
| "learning_rate": 1.3532897331765301e-05, |
| "logits/chosen": -1.5497121810913086, |
| "logits/rejected": -1.0044095516204834, |
| "logps/chosen": -59.89592742919922, |
| "logps/rejected": -257.2408752441406, |
| "loss": 0.0162, |
| "rewards/accuracies": 0.9958333969116211, |
| "rewards/chosen": -1.991355538368225, |
| "rewards/margins": 15.942468643188477, |
| "rewards/rejected": -17.933826446533203, |
| "step": 210 |
| }, |
| { |
| "epoch": 0.9188034188034188, |
| "grad_norm": 0.1884765625, |
| "learning_rate": 1.3432668862802134e-05, |
| "logits/chosen": -1.4721449613571167, |
| "logits/rejected": -0.9380281567573547, |
| "logps/chosen": -59.00041961669922, |
| "logps/rejected": -255.982421875, |
| "loss": 0.0003, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -1.8558237552642822, |
| "rewards/margins": 15.840028762817383, |
| "rewards/rejected": -17.695852279663086, |
| "step": 215 |
| }, |
| { |
| "epoch": 0.9401709401709402, |
| "grad_norm": 0.1318359375, |
| "learning_rate": 1.3330315401073371e-05, |
| "logits/chosen": -1.4841619729995728, |
| "logits/rejected": -0.9569031596183777, |
| "logps/chosen": -60.60542678833008, |
| "logps/rejected": -249.15353393554688, |
| "loss": 0.0015, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -1.9671761989593506, |
| "rewards/margins": 15.184603691101074, |
| "rewards/rejected": -17.151779174804688, |
| "step": 220 |
| }, |
| { |
| "epoch": 0.9615384615384616, |
| "grad_norm": 9.8125, |
| "learning_rate": 1.3225893710259887e-05, |
| "logits/chosen": -1.1225636005401611, |
| "logits/rejected": -0.5225231051445007, |
| "logps/chosen": -63.623390197753906, |
| "logps/rejected": -252.66281127929688, |
| "loss": 0.0095, |
| "rewards/accuracies": 0.9958333969116211, |
| "rewards/chosen": -2.4430124759674072, |
| "rewards/margins": 15.111320495605469, |
| "rewards/rejected": -17.554332733154297, |
| "step": 225 |
| }, |
| { |
| "epoch": 0.9829059829059829, |
| "grad_norm": 12.875, |
| "learning_rate": 1.3119461701051105e-05, |
| "logits/chosen": -1.2154924869537354, |
| "logits/rejected": -0.7296081781387329, |
| "logps/chosen": -65.45369720458984, |
| "logps/rejected": -248.72622680664062, |
| "loss": 0.0116, |
| "rewards/accuracies": 0.9916666746139526, |
| "rewards/chosen": -2.7806317806243896, |
| "rewards/margins": 14.623510360717773, |
| "rewards/rejected": -17.40414047241211, |
| "step": 230 |
| }, |
| { |
| "epoch": 0.9957264957264957, |
| "eval_logits/chosen": -1.554829716682434, |
| "eval_logits/rejected": -0.9467828869819641, |
| "eval_logps/chosen": -59.39779281616211, |
| "eval_logps/rejected": -258.1591491699219, |
| "eval_loss": 0.0002499581314623356, |
| "eval_rewards/accuracies": 1.0, |
| "eval_rewards/chosen": -1.995406150817871, |
| "eval_rewards/margins": 16.018821716308594, |
| "eval_rewards/rejected": -18.01422882080078, |
| "eval_runtime": 9.7352, |
| "eval_samples_per_second": 20.544, |
| "eval_steps_per_second": 20.544, |
| "step": 233 |
| }, |
| { |
| "epoch": 1.0042735042735043, |
| "grad_norm": 0.030029296875, |
| "learning_rate": 1.3011078399028605e-05, |
| "logits/chosen": -1.5108505487442017, |
| "logits/rejected": -0.9396005868911743, |
| "logps/chosen": -60.448936462402344, |
| "logps/rejected": -262.9476013183594, |
| "loss": 0.0035, |
| "rewards/accuracies": 0.9958333969116211, |
| "rewards/chosen": -2.0872457027435303, |
| "rewards/margins": 16.35396957397461, |
| "rewards/rejected": -18.44121742248535, |
| "step": 235 |
| }, |
| { |
| "epoch": 1.0256410256410255, |
| "grad_norm": 0.1025390625, |
| "learning_rate": 1.2900803911931431e-05, |
| "logits/chosen": -1.5595149993896484, |
| "logits/rejected": -0.9638253450393677, |
| "logps/chosen": -56.65764617919922, |
| "logps/rejected": -258.0482482910156, |
| "loss": 0.0001, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -1.6903537511825562, |
| "rewards/margins": 16.261157989501953, |
| "rewards/rejected": -17.95151138305664, |
| "step": 240 |
| }, |
| { |
| "epoch": 1.047008547008547, |
| "grad_norm": 0.022705078125, |
| "learning_rate": 1.2788699396321252e-05, |
| "logits/chosen": -1.520930290222168, |
| "logits/rejected": -0.9344191551208496, |
| "logps/chosen": -54.35309982299805, |
| "logps/rejected": -249.8235321044922, |
| "loss": 0.0002, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -1.5775222778320312, |
| "rewards/margins": 15.838638305664062, |
| "rewards/rejected": -17.416160583496094, |
| "step": 245 |
| }, |
| { |
| "epoch": 1.0683760683760684, |
| "grad_norm": 0.00031280517578125, |
| "learning_rate": 1.2674827023665853e-05, |
| "logits/chosen": -1.5657716989517212, |
| "logits/rejected": -0.9992179870605469, |
| "logps/chosen": -57.8154411315918, |
| "logps/rejected": -263.98992919921875, |
| "loss": 0.0004, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -1.8247381448745728, |
| "rewards/margins": 16.810840606689453, |
| "rewards/rejected": -18.635578155517578, |
| "step": 250 |
| }, |
| { |
| "epoch": 1.0897435897435896, |
| "grad_norm": 0.1005859375, |
| "learning_rate": 1.255924994585978e-05, |
| "logits/chosen": -1.532013177871704, |
| "logits/rejected": -0.9263051152229309, |
| "logps/chosen": -60.24767303466797, |
| "logps/rejected": -260.04595947265625, |
| "loss": 0.0001, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.0023863315582275, |
| "rewards/margins": 16.219375610351562, |
| "rewards/rejected": -18.221763610839844, |
| "step": 255 |
| }, |
| { |
| "epoch": 1.1111111111111112, |
| "grad_norm": 0.0054931640625, |
| "learning_rate": 1.2442032260201255e-05, |
| "logits/chosen": -1.5195871591567993, |
| "logits/rejected": -0.9636222124099731, |
| "logps/chosen": -59.14739990234375, |
| "logps/rejected": -256.98883056640625, |
| "loss": 0.0001, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -1.9601608514785767, |
| "rewards/margins": 15.914080619812012, |
| "rewards/rejected": -17.874242782592773, |
| "step": 260 |
| }, |
| { |
| "epoch": 1.1324786324786325, |
| "grad_norm": 0.283203125, |
| "learning_rate": 1.2323238973844796e-05, |
| "logits/chosen": -1.5553245544433594, |
| "logits/rejected": -0.9653908014297485, |
| "logps/chosen": -59.334739685058594, |
| "logps/rejected": -266.408935546875, |
| "loss": 0.0002, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -1.896758794784546, |
| "rewards/margins": 16.70488739013672, |
| "rewards/rejected": -18.601646423339844, |
| "step": 265 |
| }, |
| { |
| "epoch": 1.1538461538461537, |
| "grad_norm": 0.0145263671875, |
| "learning_rate": 1.2202935967749212e-05, |
| "logits/chosen": -1.5117428302764893, |
| "logits/rejected": -0.9256662130355835, |
| "logps/chosen": -62.41327667236328, |
| "logps/rejected": -261.29559326171875, |
| "loss": 0.0006, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.1235432624816895, |
| "rewards/margins": 16.034034729003906, |
| "rewards/rejected": -18.15757942199707, |
| "step": 270 |
| }, |
| { |
| "epoch": 1.1752136752136753, |
| "grad_norm": 0.16796875, |
| "learning_rate": 1.2081189960141038e-05, |
| "logits/chosen": -1.5283492803573608, |
| "logits/rejected": -0.9296523332595825, |
| "logps/chosen": -57.735595703125, |
| "logps/rejected": -260.36627197265625, |
| "loss": 0.0049, |
| "rewards/accuracies": 0.9958332777023315, |
| "rewards/chosen": -1.738776445388794, |
| "rewards/margins": 16.458282470703125, |
| "rewards/rejected": -18.19706153869629, |
| "step": 275 |
| }, |
| { |
| "epoch": 1.1965811965811965, |
| "grad_norm": 0.51171875, |
| "learning_rate": 1.1958068469513604e-05, |
| "logits/chosen": -1.5641071796417236, |
| "logits/rejected": -0.977310299873352, |
| "logps/chosen": -59.65924072265625, |
| "logps/rejected": -269.66253662109375, |
| "loss": 0.0029, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -1.9156348705291748, |
| "rewards/margins": 17.137256622314453, |
| "rewards/rejected": -19.05289077758789, |
| "step": 280 |
| }, |
| { |
| "epoch": 1.217948717948718, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 1.1833639777182316e-05, |
| "logits/chosen": -1.5035396814346313, |
| "logits/rejected": -0.8975561857223511, |
| "logps/chosen": -54.24858856201172, |
| "logps/rejected": -262.6181945800781, |
| "loss": 0.0001, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -1.5752283334732056, |
| "rewards/margins": 16.97848129272461, |
| "rewards/rejected": -18.5537109375, |
| "step": 285 |
| }, |
| { |
| "epoch": 1.2393162393162394, |
| "grad_norm": 0.025390625, |
| "learning_rate": 1.170797288941685e-05, |
| "logits/chosen": -1.4596306085586548, |
| "logits/rejected": -0.8935238718986511, |
| "logps/chosen": -60.20256805419922, |
| "logps/rejected": -270.41259765625, |
| "loss": 0.001, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.0154104232788086, |
| "rewards/margins": 16.990856170654297, |
| "rewards/rejected": -19.00626564025879, |
| "step": 290 |
| }, |
| { |
| "epoch": 1.2606837606837606, |
| "grad_norm": 0.443359375, |
| "learning_rate": 1.1581137499171342e-05, |
| "logits/chosen": -1.478342890739441, |
| "logits/rejected": -0.8745730519294739, |
| "logps/chosen": -61.747718811035156, |
| "logps/rejected": -260.48358154296875, |
| "loss": 0.0006, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.1923012733459473, |
| "rewards/margins": 16.068017959594727, |
| "rewards/rejected": -18.260318756103516, |
| "step": 295 |
| }, |
| { |
| "epoch": 1.282051282051282, |
| "grad_norm": 0.0164794921875, |
| "learning_rate": 1.145320394743371e-05, |
| "logits/chosen": -1.439082384109497, |
| "logits/rejected": -0.8525334596633911, |
| "logps/chosen": -59.48680877685547, |
| "logps/rejected": -256.1888122558594, |
| "loss": 0.0002, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -1.9997012615203857, |
| "rewards/margins": 16.03377914428711, |
| "rewards/rejected": -18.033479690551758, |
| "step": 300 |
| }, |
| { |
| "epoch": 1.3034188034188035, |
| "grad_norm": 0.004180908203125, |
| "learning_rate": 1.1324243184215622e-05, |
| "logits/chosen": -1.415179967880249, |
| "logits/rejected": -0.811850368976593, |
| "logps/chosen": -60.232139587402344, |
| "logps/rejected": -271.9323425292969, |
| "loss": 0.0, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.1755707263946533, |
| "rewards/margins": 17.213939666748047, |
| "rewards/rejected": -19.389511108398438, |
| "step": 305 |
| }, |
| { |
| "epoch": 1.3247863247863247, |
| "grad_norm": 0.0079345703125, |
| "learning_rate": 1.1194326729204686e-05, |
| "logits/chosen": -1.4362504482269287, |
| "logits/rejected": -0.8141258358955383, |
| "logps/chosen": -60.56780242919922, |
| "logps/rejected": -269.37628173828125, |
| "loss": 0.0002, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.1700046062469482, |
| "rewards/margins": 17.103158950805664, |
| "rewards/rejected": -19.273160934448242, |
| "step": 310 |
| }, |
| { |
| "epoch": 1.3461538461538463, |
| "grad_norm": 0.0264892578125, |
| "learning_rate": 1.1063526632100717e-05, |
| "logits/chosen": -1.4126843214035034, |
| "logits/rejected": -0.8487581014633179, |
| "logps/chosen": -63.435699462890625, |
| "logps/rejected": -263.54461669921875, |
| "loss": 0.0035, |
| "rewards/accuracies": 0.9958333969116211, |
| "rewards/chosen": -2.3221802711486816, |
| "rewards/margins": 16.341337203979492, |
| "rewards/rejected": -18.66351890563965, |
| "step": 315 |
| }, |
| { |
| "epoch": 1.3675213675213675, |
| "grad_norm": 0.01202392578125, |
| "learning_rate": 1.0931915432658055e-05, |
| "logits/chosen": -1.3911561965942383, |
| "logits/rejected": -0.7923445701599121, |
| "logps/chosen": -63.30759811401367, |
| "logps/rejected": -268.752685546875, |
| "loss": 0.0001, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.365898609161377, |
| "rewards/margins": 16.713449478149414, |
| "rewards/rejected": -19.079347610473633, |
| "step": 320 |
| }, |
| { |
| "epoch": 1.3888888888888888, |
| "grad_norm": 0.01446533203125, |
| "learning_rate": 1.0799566120456133e-05, |
| "logits/chosen": -1.3831051588058472, |
| "logits/rejected": -0.7522771954536438, |
| "logps/chosen": -64.623046875, |
| "logps/rejected": -268.64984130859375, |
| "loss": 0.0002, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.4689812660217285, |
| "rewards/margins": 16.626056671142578, |
| "rewards/rejected": -19.09503746032715, |
| "step": 325 |
| }, |
| { |
| "epoch": 1.4102564102564101, |
| "grad_norm": 0.01055908203125, |
| "learning_rate": 1.066655209442054e-05, |
| "logits/chosen": -1.4041249752044678, |
| "logits/rejected": -0.7933310866355896, |
| "logps/chosen": -65.8058853149414, |
| "logps/rejected": -272.6845397949219, |
| "loss": 0.0021, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.579711675643921, |
| "rewards/margins": 16.771991729736328, |
| "rewards/rejected": -19.351699829101562, |
| "step": 330 |
| }, |
| { |
| "epoch": 1.4316239316239316, |
| "grad_norm": 0.0155029296875, |
| "learning_rate": 1.0532947122117101e-05, |
| "logits/chosen": -1.3893976211547852, |
| "logits/rejected": -0.7593523263931274, |
| "logps/chosen": -65.74365234375, |
| "logps/rejected": -271.69537353515625, |
| "loss": 0.0, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.5910518169403076, |
| "rewards/margins": 16.82119369506836, |
| "rewards/rejected": -19.41224479675293, |
| "step": 335 |
| }, |
| { |
| "epoch": 1.452991452991453, |
| "grad_norm": 0.0167236328125, |
| "learning_rate": 1.0398825298841499e-05, |
| "logits/chosen": -1.3964606523513794, |
| "logits/rejected": -0.8304880261421204, |
| "logps/chosen": -69.43986511230469, |
| "logps/rejected": -275.15960693359375, |
| "loss": 0.0003, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.830099582672119, |
| "rewards/margins": 16.80275535583496, |
| "rewards/rejected": -19.632854461669922, |
| "step": 340 |
| }, |
| { |
| "epoch": 1.4743589743589745, |
| "grad_norm": 0.0263671875, |
| "learning_rate": 1.0264261006527144e-05, |
| "logits/chosen": -1.4159036874771118, |
| "logits/rejected": -0.8477792739868164, |
| "logps/chosen": -66.49871063232422, |
| "logps/rejected": -265.2044372558594, |
| "loss": 0.004, |
| "rewards/accuracies": 0.9958333969116211, |
| "rewards/chosen": -2.641693592071533, |
| "rewards/margins": 16.258649826049805, |
| "rewards/rejected": -18.900344848632812, |
| "step": 345 |
| }, |
| { |
| "epoch": 1.4957264957264957, |
| "grad_norm": 0.4140625, |
| "learning_rate": 1.0129328872494075e-05, |
| "logits/chosen": -1.4811489582061768, |
| "logits/rejected": -0.9099611043930054, |
| "logps/chosen": -66.62818908691406, |
| "logps/rejected": -271.5672912597656, |
| "loss": 0.003, |
| "rewards/accuracies": 0.9958333969116211, |
| "rewards/chosen": -2.69557523727417, |
| "rewards/margins": 16.735515594482422, |
| "rewards/rejected": -19.431089401245117, |
| "step": 350 |
| }, |
| { |
| "epoch": 1.517094017094017, |
| "grad_norm": 0.134765625, |
| "learning_rate": 9.994103728061786e-06, |
| "logits/chosen": -1.434282660484314, |
| "logits/rejected": -0.8590477705001831, |
| "logps/chosen": -68.98523712158203, |
| "logps/rejected": -271.1317138671875, |
| "loss": 0.0003, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.879164934158325, |
| "rewards/margins": 16.430065155029297, |
| "rewards/rejected": -19.30923080444336, |
| "step": 355 |
| }, |
| { |
| "epoch": 1.5384615384615383, |
| "grad_norm": 0.00104522705078125, |
| "learning_rate": 9.858660567048902e-06, |
| "logits/chosen": -1.4451745748519897, |
| "logits/rejected": -0.8094388842582703, |
| "logps/chosen": -71.38107299804688, |
| "logps/rejected": -284.3163757324219, |
| "loss": 0.0, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -3.1282448768615723, |
| "rewards/margins": 17.431547164916992, |
| "rewards/rejected": -20.559795379638672, |
| "step": 360 |
| }, |
| { |
| "epoch": 1.5598290598290598, |
| "grad_norm": 0.06591796875, |
| "learning_rate": 9.72307450418274e-06, |
| "logits/chosen": -1.4680107831954956, |
| "logits/rejected": -0.8780597448348999, |
| "logps/chosen": -66.4084701538086, |
| "logps/rejected": -269.15948486328125, |
| "loss": 0.0004, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.635270357131958, |
| "rewards/margins": 16.57504653930664, |
| "rewards/rejected": -19.210315704345703, |
| "step": 365 |
| }, |
| { |
| "epoch": 1.5811965811965814, |
| "grad_norm": 1.921875, |
| "learning_rate": 9.587420733441835e-06, |
| "logits/chosen": -1.484803557395935, |
| "logits/rejected": -0.9129883646965027, |
| "logps/chosen": -63.17878341674805, |
| "logps/rejected": -272.344482421875, |
| "loss": 0.0006, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.336841106414795, |
| "rewards/margins": 17.08325958251953, |
| "rewards/rejected": -19.420101165771484, |
| "step": 370 |
| }, |
| { |
| "epoch": 1.6025641025641026, |
| "grad_norm": 0.000766754150390625, |
| "learning_rate": 9.45177448635447e-06, |
| "logits/chosen": -1.4960134029388428, |
| "logits/rejected": -0.8223481178283691, |
| "logps/chosen": -63.178985595703125, |
| "logps/rejected": -274.9540100097656, |
| "loss": 0.0, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.362715482711792, |
| "rewards/margins": 17.505586624145508, |
| "rewards/rejected": -19.868301391601562, |
| "step": 375 |
| }, |
| { |
| "epoch": 1.623931623931624, |
| "grad_norm": 0.046630859375, |
| "learning_rate": 9.316210990276434e-06, |
| "logits/chosen": -1.425362467765808, |
| "logits/rejected": -0.851107120513916, |
| "logps/chosen": -62.67730712890625, |
| "logps/rejected": -263.357666015625, |
| "loss": 0.0001, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.2725672721862793, |
| "rewards/margins": 16.371471405029297, |
| "rewards/rejected": -18.644039154052734, |
| "step": 380 |
| }, |
| { |
| "epoch": 1.6452991452991452, |
| "grad_norm": 0.0045166015625, |
| "learning_rate": 9.18080542667105e-06, |
| "logits/chosen": -1.4633301496505737, |
| "logits/rejected": -0.8226639628410339, |
| "logps/chosen": -63.28871536254883, |
| "logps/rejected": -280.1250915527344, |
| "loss": 0.0001, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.2893106937408447, |
| "rewards/margins": 17.715099334716797, |
| "rewards/rejected": -20.004411697387695, |
| "step": 385 |
| }, |
| { |
| "epoch": 1.6666666666666665, |
| "grad_norm": 0.004486083984375, |
| "learning_rate": 9.045632889414686e-06, |
| "logits/chosen": -1.451174020767212, |
| "logits/rejected": -0.8335480690002441, |
| "logps/chosen": -61.83076095581055, |
| "logps/rejected": -272.6547546386719, |
| "loss": 0.0001, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.2272868156433105, |
| "rewards/margins": 17.232776641845703, |
| "rewards/rejected": -19.460063934326172, |
| "step": 390 |
| }, |
| { |
| "epoch": 1.688034188034188, |
| "grad_norm": 0.01153564453125, |
| "learning_rate": 8.910768343150828e-06, |
| "logits/chosen": -1.4321696758270264, |
| "logits/rejected": -0.8404842615127563, |
| "logps/chosen": -64.93434143066406, |
| "logps/rejected": -273.5433044433594, |
| "loss": 0.0001, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.5962464809417725, |
| "rewards/margins": 17.047863006591797, |
| "rewards/rejected": -19.64410972595215, |
| "step": 395 |
| }, |
| { |
| "epoch": 1.7094017094017095, |
| "grad_norm": 0.006378173828125, |
| "learning_rate": 8.77628658171581e-06, |
| "logits/chosen": -1.4662706851959229, |
| "logits/rejected": -0.8858024477958679, |
| "logps/chosen": -63.3156852722168, |
| "logps/rejected": -268.3216552734375, |
| "loss": 0.0003, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.346360683441162, |
| "rewards/margins": 16.638729095458984, |
| "rewards/rejected": -18.98509407043457, |
| "step": 400 |
| }, |
| { |
| "epoch": 1.7307692307692308, |
| "grad_norm": 0.033203125, |
| "learning_rate": 8.642262186659298e-06, |
| "logits/chosen": -1.4258873462677002, |
| "logits/rejected": -0.8420296907424927, |
| "logps/chosen": -64.47160339355469, |
| "logps/rejected": -271.00213623046875, |
| "loss": 0.0002, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.43169903755188, |
| "rewards/margins": 16.784656524658203, |
| "rewards/rejected": -19.21635627746582, |
| "step": 405 |
| }, |
| { |
| "epoch": 1.7521367521367521, |
| "grad_norm": 0.023681640625, |
| "learning_rate": 8.508769485882487e-06, |
| "logits/chosen": -1.4450222253799438, |
| "logits/rejected": -0.8718585968017578, |
| "logps/chosen": -69.8866195678711, |
| "logps/rejected": -272.61322021484375, |
| "loss": 0.0022, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.834794759750366, |
| "rewards/margins": 16.60897445678711, |
| "rewards/rejected": -19.443769454956055, |
| "step": 410 |
| }, |
| { |
| "epoch": 1.7735042735042734, |
| "grad_norm": 0.703125, |
| "learning_rate": 8.375882512416969e-06, |
| "logits/chosen": -1.4202231168746948, |
| "logits/rejected": -0.7998637557029724, |
| "logps/chosen": -67.1268310546875, |
| "logps/rejected": -275.6379089355469, |
| "loss": 0.0006, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.6624515056610107, |
| "rewards/margins": 17.090303421020508, |
| "rewards/rejected": -19.752756118774414, |
| "step": 415 |
| }, |
| { |
| "epoch": 1.7948717948717947, |
| "grad_norm": 0.0458984375, |
| "learning_rate": 8.243674963367137e-06, |
| "logits/chosen": -1.420807957649231, |
| "logits/rejected": -0.8041313886642456, |
| "logps/chosen": -68.99574279785156, |
| "logps/rejected": -272.53936767578125, |
| "loss": 0.0033, |
| "rewards/accuracies": 0.9958333969116211, |
| "rewards/chosen": -2.8813891410827637, |
| "rewards/margins": 16.42134666442871, |
| "rewards/rejected": -19.302734375, |
| "step": 420 |
| }, |
| { |
| "epoch": 1.8162393162393162, |
| "grad_norm": 0.000850677490234375, |
| "learning_rate": 8.11222015903888e-06, |
| "logits/chosen": -1.4625945091247559, |
| "logits/rejected": -0.8097749948501587, |
| "logps/chosen": -67.76467895507812, |
| "logps/rejected": -278.23931884765625, |
| "loss": 0.0032, |
| "rewards/accuracies": 0.9958333969116211, |
| "rewards/chosen": -2.830801010131836, |
| "rewards/margins": 17.239028930664062, |
| "rewards/rejected": -20.0698299407959, |
| "step": 425 |
| }, |
| { |
| "epoch": 1.8376068376068377, |
| "grad_norm": 0.03515625, |
| "learning_rate": 7.981591002277265e-06, |
| "logits/chosen": -1.4442012310028076, |
| "logits/rejected": -0.8179518580436707, |
| "logps/chosen": -69.31373596191406, |
| "logps/rejected": -273.0030822753906, |
| "loss": 0.0002, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.9278433322906494, |
| "rewards/margins": 16.736169815063477, |
| "rewards/rejected": -19.66401481628418, |
| "step": 430 |
| }, |
| { |
| "epoch": 1.858974358974359, |
| "grad_norm": 0.01116943359375, |
| "learning_rate": 7.851859938035712e-06, |
| "logits/chosen": -1.4216701984405518, |
| "logits/rejected": -0.7942286729812622, |
| "logps/chosen": -71.70731353759766, |
| "logps/rejected": -282.1054382324219, |
| "loss": 0.0001, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -3.118105888366699, |
| "rewards/margins": 17.2371768951416, |
| "rewards/rejected": -20.355281829833984, |
| "step": 435 |
| }, |
| { |
| "epoch": 1.8803418803418803, |
| "grad_norm": 0.0279541015625, |
| "learning_rate": 7.723098913199118e-06, |
| "logits/chosen": -1.4586832523345947, |
| "logits/rejected": -0.8305650949478149, |
| "logps/chosen": -67.39842224121094, |
| "logps/rejected": -269.7090148925781, |
| "loss": 0.0001, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.8383772373199463, |
| "rewards/margins": 16.61200714111328, |
| "rewards/rejected": -19.450382232666016, |
| "step": 440 |
| }, |
| { |
| "epoch": 1.9017094017094016, |
| "grad_norm": 0.0419921875, |
| "learning_rate": 7.595379336683204e-06, |
| "logits/chosen": -1.433342456817627, |
| "logits/rejected": -0.7530536651611328, |
| "logps/chosen": -65.9086685180664, |
| "logps/rejected": -276.2018127441406, |
| "loss": 0.0, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.7526674270629883, |
| "rewards/margins": 17.266189575195312, |
| "rewards/rejected": -20.018856048583984, |
| "step": 445 |
| }, |
| { |
| "epoch": 1.9230769230769231, |
| "grad_norm": 0.00433349609375, |
| "learning_rate": 7.468772039832218e-06, |
| "logits/chosen": -1.4057344198226929, |
| "logits/rejected": -0.7420554757118225, |
| "logps/chosen": -61.406890869140625, |
| "logps/rejected": -270.76568603515625, |
| "loss": 0.0, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.308377742767334, |
| "rewards/margins": 17.16427230834961, |
| "rewards/rejected": -19.472646713256836, |
| "step": 450 |
| }, |
| { |
| "epoch": 1.9444444444444444, |
| "grad_norm": 0.00634765625, |
| "learning_rate": 7.3433472371369404e-06, |
| "logits/chosen": -1.4746873378753662, |
| "logits/rejected": -0.8472849726676941, |
| "logps/chosen": -66.808837890625, |
| "logps/rejected": -274.1222839355469, |
| "loss": 0.0001, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.7300994396209717, |
| "rewards/margins": 16.893524169921875, |
| "rewards/rejected": -19.623624801635742, |
| "step": 455 |
| }, |
| { |
| "epoch": 1.965811965811966, |
| "grad_norm": 0.00092315673828125, |
| "learning_rate": 7.219174487294784e-06, |
| "logits/chosen": -1.4693877696990967, |
| "logits/rejected": -0.856170654296875, |
| "logps/chosen": -64.80795288085938, |
| "logps/rejected": -276.7772216796875, |
| "loss": 0.0001, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.508439779281616, |
| "rewards/margins": 17.357221603393555, |
| "rewards/rejected": -19.865665435791016, |
| "step": 460 |
| }, |
| { |
| "epoch": 1.9871794871794872, |
| "grad_norm": 0.0234375, |
| "learning_rate": 7.0963226546336e-06, |
| "logits/chosen": -1.475427269935608, |
| "logits/rejected": -0.8358310461044312, |
| "logps/chosen": -66.6759262084961, |
| "logps/rejected": -272.0386962890625, |
| "loss": 0.0001, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.7268948554992676, |
| "rewards/margins": 16.760303497314453, |
| "rewards/rejected": -19.487197875976562, |
| "step": 465 |
| }, |
| { |
| "epoch": 1.9914529914529915, |
| "eval_logits/chosen": -1.4994289875030518, |
| "eval_logits/rejected": -0.8360934257507324, |
| "eval_logps/chosen": -67.28651428222656, |
| "eval_logps/rejected": -275.5657043457031, |
| "eval_loss": 0.00012571073602885008, |
| "eval_rewards/accuracies": 1.0, |
| "eval_rewards/chosen": -2.7842776775360107, |
| "eval_rewards/margins": 16.970603942871094, |
| "eval_rewards/rejected": -19.754880905151367, |
| "eval_runtime": 9.583, |
| "eval_samples_per_second": 20.87, |
| "eval_steps_per_second": 20.87, |
| "step": 466 |
| }, |
| { |
| "epoch": 2.0085470085470085, |
| "grad_norm": 0.080078125, |
| "learning_rate": 6.974859870920561e-06, |
| "logits/chosen": -1.3924753665924072, |
| "logits/rejected": -0.8071160316467285, |
| "logps/chosen": -68.68719482421875, |
| "logps/rejected": -270.14801025390625, |
| "loss": 0.0, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.911695957183838, |
| "rewards/margins": 16.461864471435547, |
| "rewards/rejected": -19.37356185913086, |
| "step": 470 |
| }, |
| { |
| "epoch": 2.02991452991453, |
| "grad_norm": 0.0012359619140625, |
| "learning_rate": 6.8548534975773135e-06, |
| "logits/chosen": -1.4488104581832886, |
| "logits/rejected": -0.82325679063797, |
| "logps/chosen": -69.13402557373047, |
| "logps/rejected": -275.86407470703125, |
| "loss": 0.0001, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.8428947925567627, |
| "rewards/margins": 16.885896682739258, |
| "rewards/rejected": -19.728792190551758, |
| "step": 475 |
| }, |
| { |
| "epoch": 2.051282051282051, |
| "grad_norm": 0.279296875, |
| "learning_rate": 6.736370088322359e-06, |
| "logits/chosen": -1.4575817584991455, |
| "logits/rejected": -0.8154885172843933, |
| "logps/chosen": -66.7756576538086, |
| "logps/rejected": -272.9770812988281, |
| "loss": 0.0006, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.7416322231292725, |
| "rewards/margins": 16.960206985473633, |
| "rewards/rejected": -19.701839447021484, |
| "step": 480 |
| }, |
| { |
| "epoch": 2.072649572649573, |
| "grad_norm": 0.00994873046875, |
| "learning_rate": 6.619475352261356e-06, |
| "logits/chosen": -1.4201619625091553, |
| "logits/rejected": -0.8027874827384949, |
| "logps/chosen": -72.34085845947266, |
| "logps/rejected": -276.4122009277344, |
| "loss": 0.0, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -3.0328619480133057, |
| "rewards/margins": 16.768281936645508, |
| "rewards/rejected": -19.801143646240234, |
| "step": 485 |
| }, |
| { |
| "epoch": 2.094017094017094, |
| "grad_norm": 0.0030975341796875, |
| "learning_rate": 6.504234117445857e-06, |
| "logits/chosen": -1.4316365718841553, |
| "logits/rejected": -0.8082255125045776, |
| "logps/chosen": -67.41413879394531, |
| "logps/rejected": -272.8454284667969, |
| "loss": 0.0001, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.8370206356048584, |
| "rewards/margins": 16.798355102539062, |
| "rewards/rejected": -19.635372161865234, |
| "step": 490 |
| }, |
| { |
| "epoch": 2.1153846153846154, |
| "grad_norm": 0.012939453125, |
| "learning_rate": 6.39071029492065e-06, |
| "logits/chosen": -1.4187885522842407, |
| "logits/rejected": -0.7696577906608582, |
| "logps/chosen": -66.7591781616211, |
| "logps/rejected": -271.50927734375, |
| "loss": 0.0001, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.810885190963745, |
| "rewards/margins": 16.76914405822754, |
| "rewards/rejected": -19.580028533935547, |
| "step": 495 |
| }, |
| { |
| "epoch": 2.1367521367521367, |
| "grad_norm": 0.125, |
| "learning_rate": 6.2789668432796535e-06, |
| "logits/chosen": -1.4300968647003174, |
| "logits/rejected": -0.8352675437927246, |
| "logps/chosen": -67.97314453125, |
| "logps/rejected": -273.3833312988281, |
| "loss": 0.0001, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.8787360191345215, |
| "rewards/margins": 16.89468002319336, |
| "rewards/rejected": -19.773418426513672, |
| "step": 500 |
| }, |
| { |
| "epoch": 2.1367521367521367, |
| "eval_logits/chosen": -1.4944664239883423, |
| "eval_logits/rejected": -0.8317099213600159, |
| "eval_logps/chosen": -67.55956268310547, |
| "eval_logps/rejected": -276.07452392578125, |
| "eval_loss": 0.0001331747480435297, |
| "eval_rewards/accuracies": 1.0, |
| "eval_rewards/chosen": -2.811582565307617, |
| "eval_rewards/margins": 16.994182586669922, |
| "eval_rewards/rejected": -19.805763244628906, |
| "eval_runtime": 9.5603, |
| "eval_samples_per_second": 20.92, |
| "eval_steps_per_second": 20.92, |
| "step": 500 |
| }, |
| { |
| "epoch": 2.158119658119658, |
| "grad_norm": 0.016845703125, |
| "learning_rate": 6.16906573375004e-06, |
| "logits/chosen": -1.4502363204956055, |
| "logits/rejected": -0.8533867001533508, |
| "logps/chosen": -67.79924774169922, |
| "logps/rejected": -271.75946044921875, |
| "loss": 0.0002, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.8223788738250732, |
| "rewards/margins": 16.64777374267578, |
| "rewards/rejected": -19.47015380859375, |
| "step": 505 |
| }, |
| { |
| "epoch": 2.1794871794871793, |
| "grad_norm": 0.05908203125, |
| "learning_rate": 6.061067915823923e-06, |
| "logits/chosen": -1.3864041566848755, |
| "logits/rejected": -0.7836092114448547, |
| "logps/chosen": -65.47071838378906, |
| "logps/rejected": -273.322265625, |
| "loss": 0.0, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.629401683807373, |
| "rewards/margins": 16.939075469970703, |
| "rewards/rejected": -19.568477630615234, |
| "step": 510 |
| }, |
| { |
| "epoch": 2.200854700854701, |
| "grad_norm": 0.006103515625, |
| "learning_rate": 5.955033283456711e-06, |
| "logits/chosen": -1.4336755275726318, |
| "logits/rejected": -0.778075098991394, |
| "logps/chosen": -70.93939971923828, |
| "logps/rejected": -283.32598876953125, |
| "loss": 0.0001, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -3.026041030883789, |
| "rewards/margins": 17.39080238342285, |
| "rewards/rejected": -20.41684341430664, |
| "step": 515 |
| }, |
| { |
| "epoch": 2.2222222222222223, |
| "grad_norm": 0.00128936767578125, |
| "learning_rate": 5.8510206418507914e-06, |
| "logits/chosen": -1.4767441749572754, |
| "logits/rejected": -0.8630167245864868, |
| "logps/chosen": -68.52049255371094, |
| "logps/rejected": -288.3775939941406, |
| "loss": 0.0, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.804711103439331, |
| "rewards/margins": 18.02071762084961, |
| "rewards/rejected": -20.825428009033203, |
| "step": 520 |
| }, |
| { |
| "epoch": 2.2435897435897436, |
| "grad_norm": 0.0078125, |
| "learning_rate": 5.749087674843095e-06, |
| "logits/chosen": -1.416854977607727, |
| "logits/rejected": -0.8053888082504272, |
| "logps/chosen": -63.85468673706055, |
| "logps/rejected": -274.10955810546875, |
| "loss": 0.0, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.541356086730957, |
| "rewards/margins": 17.136423110961914, |
| "rewards/rejected": -19.67778205871582, |
| "step": 525 |
| }, |
| { |
| "epoch": 2.264957264957265, |
| "grad_norm": 0.024658203125, |
| "learning_rate": 5.649290912914482e-06, |
| "logits/chosen": -1.4132906198501587, |
| "logits/rejected": -0.8062090873718262, |
| "logps/chosen": -71.18649291992188, |
| "logps/rejected": -287.32562255859375, |
| "loss": 0.0, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.9816391468048096, |
| "rewards/margins": 17.66859245300293, |
| "rewards/rejected": -20.650232315063477, |
| "step": 530 |
| }, |
| { |
| "epoch": 2.286324786324786, |
| "grad_norm": 0.031005859375, |
| "learning_rate": 5.5516857018388144e-06, |
| "logits/chosen": -1.4750016927719116, |
| "logits/rejected": -0.8834052085876465, |
| "logps/chosen": -66.85182189941406, |
| "logps/rejected": -271.5949401855469, |
| "loss": 0.0003, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.7011666297912598, |
| "rewards/margins": 16.737390518188477, |
| "rewards/rejected": -19.438556671142578, |
| "step": 535 |
| }, |
| { |
| "epoch": 2.3076923076923075, |
| "grad_norm": 0.0098876953125, |
| "learning_rate": 5.456326171989005e-06, |
| "logits/chosen": -1.4350497722625732, |
| "logits/rejected": -0.854056715965271, |
| "logps/chosen": -64.83439636230469, |
| "logps/rejected": -287.66961669921875, |
| "loss": 0.0, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.5173258781433105, |
| "rewards/margins": 18.059844970703125, |
| "rewards/rejected": -20.57716941833496, |
| "step": 540 |
| }, |
| { |
| "epoch": 2.3290598290598292, |
| "grad_norm": 0.0036163330078125, |
| "learning_rate": 5.363265208317156e-06, |
| "logits/chosen": -1.4079747200012207, |
| "logits/rejected": -0.8292921781539917, |
| "logps/chosen": -66.03783416748047, |
| "logps/rejected": -271.29266357421875, |
| "loss": 0.0, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.6700127124786377, |
| "rewards/margins": 16.861469268798828, |
| "rewards/rejected": -19.531482696533203, |
| "step": 545 |
| }, |
| { |
| "epoch": 2.3504273504273505, |
| "grad_norm": 0.0035247802734375, |
| "learning_rate": 5.272554421025347e-06, |
| "logits/chosen": -1.4516395330429077, |
| "logits/rejected": -0.8227602243423462, |
| "logps/chosen": -68.09796142578125, |
| "logps/rejected": -281.10687255859375, |
| "loss": 0.0, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.847126007080078, |
| "rewards/margins": 17.43533706665039, |
| "rewards/rejected": -20.2824649810791, |
| "step": 550 |
| }, |
| { |
| "epoch": 2.371794871794872, |
| "grad_norm": 0.04638671875, |
| "learning_rate": 5.184244116943411e-06, |
| "logits/chosen": -1.4369877576828003, |
| "logits/rejected": -0.8126457333564758, |
| "logps/chosen": -67.11319732666016, |
| "logps/rejected": -276.0461730957031, |
| "loss": 0.003, |
| "rewards/accuracies": 0.9958333969116211, |
| "rewards/chosen": -2.69486665725708, |
| "rewards/margins": 17.134822845458984, |
| "rewards/rejected": -19.829692840576172, |
| "step": 555 |
| }, |
| { |
| "epoch": 2.393162393162393, |
| "grad_norm": 0.013427734375, |
| "learning_rate": 5.098383271629512e-06, |
| "logits/chosen": -1.4552416801452637, |
| "logits/rejected": -0.8165750503540039, |
| "logps/chosen": -67.05858612060547, |
| "logps/rejected": -270.05596923828125, |
| "loss": 0.0001, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.7700178623199463, |
| "rewards/margins": 16.640972137451172, |
| "rewards/rejected": -19.41098976135254, |
| "step": 560 |
| }, |
| { |
| "epoch": 2.4145299145299144, |
| "grad_norm": 0.024658203125, |
| "learning_rate": 5.015019502209056e-06, |
| "logits/chosen": -1.4252703189849854, |
| "logits/rejected": -0.7998120188713074, |
| "logps/chosen": -65.51549530029297, |
| "logps/rejected": -266.0838623046875, |
| "loss": 0.0, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.6941094398498535, |
| "rewards/margins": 16.32393455505371, |
| "rewards/rejected": -19.018043518066406, |
| "step": 565 |
| }, |
| { |
| "epoch": 2.435897435897436, |
| "grad_norm": 0.00130462646484375, |
| "learning_rate": 4.934199040966955e-06, |
| "logits/chosen": -1.473314642906189, |
| "logits/rejected": -0.847906231880188, |
| "logps/chosen": -66.51988220214844, |
| "logps/rejected": -269.98150634765625, |
| "loss": 0.0001, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.734851360321045, |
| "rewards/margins": 16.677799224853516, |
| "rewards/rejected": -19.412649154663086, |
| "step": 570 |
| }, |
| { |
| "epoch": 2.4572649572649574, |
| "grad_norm": 0.0120849609375, |
| "learning_rate": 4.855966709707881e-06, |
| "logits/chosen": -1.4144232273101807, |
| "logits/rejected": -0.8331915736198425, |
| "logps/chosen": -70.74285888671875, |
| "logps/rejected": -275.6701354980469, |
| "loss": 0.003, |
| "rewards/accuracies": 0.9958333969116211, |
| "rewards/chosen": -3.0315637588500977, |
| "rewards/margins": 16.809673309326172, |
| "rewards/rejected": -19.841238021850586, |
| "step": 575 |
| }, |
| { |
| "epoch": 2.4786324786324787, |
| "grad_norm": 0.035888671875, |
| "learning_rate": 4.780365894898799e-06, |
| "logits/chosen": -1.446194052696228, |
| "logits/rejected": -0.8259027600288391, |
| "logps/chosen": -67.32369995117188, |
| "logps/rejected": -276.65728759765625, |
| "loss": 0.0, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.730370044708252, |
| "rewards/margins": 17.11151123046875, |
| "rewards/rejected": -19.841882705688477, |
| "step": 580 |
| }, |
| { |
| "epoch": 2.5, |
| "grad_norm": 0.00885009765625, |
| "learning_rate": 4.7074385236074684e-06, |
| "logits/chosen": -1.4675248861312866, |
| "logits/rejected": -0.8278427124023438, |
| "logps/chosen": -71.28508758544922, |
| "logps/rejected": -284.6138000488281, |
| "loss": 0.0, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.97928786277771, |
| "rewards/margins": 17.529958724975586, |
| "rewards/rejected": -20.509244918823242, |
| "step": 585 |
| }, |
| { |
| "epoch": 2.5213675213675213, |
| "grad_norm": 0.054931640625, |
| "learning_rate": 4.63722504025034e-06, |
| "logits/chosen": -1.4478528499603271, |
| "logits/rejected": -0.8462594747543335, |
| "logps/chosen": -64.45930480957031, |
| "logps/rejected": -275.44036865234375, |
| "loss": 0.0008, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.4965498447418213, |
| "rewards/margins": 17.28372573852539, |
| "rewards/rejected": -19.780277252197266, |
| "step": 590 |
| }, |
| { |
| "epoch": 2.5427350427350426, |
| "grad_norm": 0.021240234375, |
| "learning_rate": 4.569764384162676e-06, |
| "logits/chosen": -1.4624818563461304, |
| "logits/rejected": -0.8267906308174133, |
| "logps/chosen": -61.47455596923828, |
| "logps/rejected": -273.9830017089844, |
| "loss": 0.0001, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.310239791870117, |
| "rewards/margins": 17.343975067138672, |
| "rewards/rejected": -19.65421485900879, |
| "step": 595 |
| }, |
| { |
| "epoch": 2.564102564102564, |
| "grad_norm": 0.031494140625, |
| "learning_rate": 4.50509396800341e-06, |
| "logits/chosen": -1.4217302799224854, |
| "logits/rejected": -0.7868741750717163, |
| "logps/chosen": -65.87348937988281, |
| "logps/rejected": -277.1817321777344, |
| "loss": 0.0, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.619393825531006, |
| "rewards/margins": 17.432384490966797, |
| "rewards/rejected": -20.051776885986328, |
| "step": 600 |
| }, |
| { |
| "epoch": 2.5854700854700856, |
| "grad_norm": 0.0732421875, |
| "learning_rate": 4.443249657006627e-06, |
| "logits/chosen": -1.4423940181732178, |
| "logits/rejected": -0.7987694144248962, |
| "logps/chosen": -63.83234786987305, |
| "logps/rejected": -280.8373718261719, |
| "loss": 0.0029, |
| "rewards/accuracies": 0.9958333969116211, |
| "rewards/chosen": -2.4912962913513184, |
| "rewards/margins": 17.850605010986328, |
| "rewards/rejected": -20.341899871826172, |
| "step": 605 |
| }, |
| { |
| "epoch": 2.606837606837607, |
| "grad_norm": 0.01336669921875, |
| "learning_rate": 4.384265749091266e-06, |
| "logits/chosen": -1.391326665878296, |
| "logits/rejected": -0.7768113613128662, |
| "logps/chosen": -71.64138793945312, |
| "logps/rejected": -278.91400146484375, |
| "loss": 0.0001, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -3.0774004459381104, |
| "rewards/margins": 17.042530059814453, |
| "rewards/rejected": -20.119930267333984, |
| "step": 610 |
| }, |
| { |
| "epoch": 2.628205128205128, |
| "grad_norm": 0.01239013671875, |
| "learning_rate": 4.328174955840002e-06, |
| "logits/chosen": -1.4435267448425293, |
| "logits/rejected": -0.8103730082511902, |
| "logps/chosen": -62.11974334716797, |
| "logps/rejected": -273.298828125, |
| "loss": 0.0, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.468442440032959, |
| "rewards/margins": 17.258960723876953, |
| "rewards/rejected": -19.727405548095703, |
| "step": 615 |
| }, |
| { |
| "epoch": 2.6495726495726495, |
| "grad_norm": 0.004791259765625, |
| "learning_rate": 4.275008384357902e-06, |
| "logits/chosen": -1.4775636196136475, |
| "logits/rejected": -0.8828374743461609, |
| "logps/chosen": -65.2374267578125, |
| "logps/rejected": -272.7725524902344, |
| "loss": 0.0, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.561352491378784, |
| "rewards/margins": 17.067296981811523, |
| "rewards/rejected": -19.62864875793457, |
| "step": 620 |
| }, |
| { |
| "epoch": 2.6709401709401708, |
| "grad_norm": 0.005157470703125, |
| "learning_rate": 4.224795520020898e-06, |
| "logits/chosen": -1.418306827545166, |
| "logits/rejected": -0.7850183248519897, |
| "logps/chosen": -68.51617431640625, |
| "logps/rejected": -272.66455078125, |
| "loss": 0.0001, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.727930784225464, |
| "rewards/margins": 16.756616592407227, |
| "rewards/rejected": -19.484546661376953, |
| "step": 625 |
| }, |
| { |
| "epoch": 2.6923076923076925, |
| "grad_norm": 0.042724609375, |
| "learning_rate": 4.177564210123634e-06, |
| "logits/chosen": -1.4454832077026367, |
| "logits/rejected": -0.8340616226196289, |
| "logps/chosen": -65.76726531982422, |
| "logps/rejected": -282.2267150878906, |
| "loss": 0.0, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.610382556915283, |
| "rewards/margins": 17.712736129760742, |
| "rewards/rejected": -20.323118209838867, |
| "step": 630 |
| }, |
| { |
| "epoch": 2.713675213675214, |
| "grad_norm": 0.039306640625, |
| "learning_rate": 4.133340648435789e-06, |
| "logits/chosen": -1.4548784494400024, |
| "logits/rejected": -0.8171395063400269, |
| "logps/chosen": -66.9704360961914, |
| "logps/rejected": -277.67156982421875, |
| "loss": 0.0001, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.7575552463531494, |
| "rewards/margins": 17.253223419189453, |
| "rewards/rejected": -20.01078224182129, |
| "step": 635 |
| }, |
| { |
| "epoch": 2.735042735042735, |
| "grad_norm": 0.003448486328125, |
| "learning_rate": 4.092149360675402e-06, |
| "logits/chosen": -1.429237961769104, |
| "logits/rejected": -0.825334906578064, |
| "logps/chosen": -71.72950744628906, |
| "logps/rejected": -282.79864501953125, |
| "loss": 0.0, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.9618561267852783, |
| "rewards/margins": 17.245059967041016, |
| "rewards/rejected": -20.206918716430664, |
| "step": 640 |
| }, |
| { |
| "epoch": 2.7564102564102564, |
| "grad_norm": 0.04833984375, |
| "learning_rate": 4.054013190907282e-06, |
| "logits/chosen": -1.3933217525482178, |
| "logits/rejected": -0.7962290048599243, |
| "logps/chosen": -62.87525177001953, |
| "logps/rejected": -274.01885986328125, |
| "loss": 0.0, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.444106101989746, |
| "rewards/margins": 17.38211441040039, |
| "rewards/rejected": -19.826221466064453, |
| "step": 645 |
| }, |
| { |
| "epoch": 2.7777777777777777, |
| "grad_norm": 0.0032196044921875, |
| "learning_rate": 4.018953288874035e-06, |
| "logits/chosen": -1.4579218626022339, |
| "logits/rejected": -0.8607872724533081, |
| "logps/chosen": -67.1677474975586, |
| "logps/rejected": -278.09576416015625, |
| "loss": 0.0, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.7266063690185547, |
| "rewards/margins": 17.331296920776367, |
| "rewards/rejected": -20.057903289794922, |
| "step": 650 |
| }, |
| { |
| "epoch": 2.799145299145299, |
| "grad_norm": 0.0028533935546875, |
| "learning_rate": 3.9869890982667385e-06, |
| "logits/chosen": -1.4375386238098145, |
| "logits/rejected": -0.7617470026016235, |
| "logps/chosen": -66.19978332519531, |
| "logps/rejected": -279.52581787109375, |
| "loss": 0.0, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.685457706451416, |
| "rewards/margins": 17.297794342041016, |
| "rewards/rejected": -19.983253479003906, |
| "step": 655 |
| }, |
| { |
| "epoch": 2.8205128205128203, |
| "grad_norm": 0.01300048828125, |
| "learning_rate": 3.9581383459417625e-06, |
| "logits/chosen": -1.409898042678833, |
| "logits/rejected": -0.7993089556694031, |
| "logps/chosen": -71.67328643798828, |
| "logps/rejected": -286.96978759765625, |
| "loss": 0.0, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -3.136204242706299, |
| "rewards/margins": 17.608535766601562, |
| "rewards/rejected": -20.744739532470703, |
| "step": 660 |
| }, |
| { |
| "epoch": 2.841880341880342, |
| "grad_norm": 0.029052734375, |
| "learning_rate": 3.932417032089722e-06, |
| "logits/chosen": -1.4576003551483154, |
| "logits/rejected": -0.821110725402832, |
| "logps/chosen": -68.1684799194336, |
| "logps/rejected": -282.56744384765625, |
| "loss": 0.0001, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.7884955406188965, |
| "rewards/margins": 17.563434600830078, |
| "rewards/rejected": -20.351930618286133, |
| "step": 665 |
| }, |
| { |
| "epoch": 2.8632478632478633, |
| "grad_norm": 0.000885009765625, |
| "learning_rate": 3.909839421362017e-06, |
| "logits/chosen": -1.4005993604660034, |
| "logits/rejected": -0.7887626886367798, |
| "logps/chosen": -67.8237075805664, |
| "logps/rejected": -278.14056396484375, |
| "loss": 0.0001, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.829472064971924, |
| "rewards/margins": 17.069793701171875, |
| "rewards/rejected": -19.899267196655273, |
| "step": 670 |
| }, |
| { |
| "epoch": 2.8846153846153846, |
| "grad_norm": 0.048828125, |
| "learning_rate": 3.890418034959871e-06, |
| "logits/chosen": -1.3995271921157837, |
| "logits/rejected": -0.7455986738204956, |
| "logps/chosen": -66.70311737060547, |
| "logps/rejected": -276.72540283203125, |
| "loss": 0.0, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.7374672889709473, |
| "rewards/margins": 17.271923065185547, |
| "rewards/rejected": -20.00939178466797, |
| "step": 675 |
| }, |
| { |
| "epoch": 2.905982905982906, |
| "grad_norm": 0.045166015625, |
| "learning_rate": 3.874163643690263e-06, |
| "logits/chosen": -1.3939011096954346, |
| "logits/rejected": -0.7581735849380493, |
| "logps/chosen": -74.34012603759766, |
| "logps/rejected": -280.92291259765625, |
| "loss": 0.0, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -3.3185927867889404, |
| "rewards/margins": 16.952451705932617, |
| "rewards/rejected": -20.27104377746582, |
| "step": 680 |
| }, |
| { |
| "epoch": 2.9273504273504276, |
| "grad_norm": 0.01312255859375, |
| "learning_rate": 3.861085261992599e-06, |
| "logits/chosen": -1.4342122077941895, |
| "logits/rejected": -0.8670775294303894, |
| "logps/chosen": -70.84014129638672, |
| "logps/rejected": -280.0901794433594, |
| "loss": 0.0, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -3.0605998039245605, |
| "rewards/margins": 17.057954788208008, |
| "rewards/rejected": -20.118555068969727, |
| "step": 685 |
| }, |
| { |
| "epoch": 2.948717948717949, |
| "grad_norm": 0.1376953125, |
| "learning_rate": 3.851190142939442e-06, |
| "logits/chosen": -1.435960054397583, |
| "logits/rejected": -0.8204304575920105, |
| "logps/chosen": -65.02848052978516, |
| "logps/rejected": -277.8736877441406, |
| "loss": 0.0001, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.4977402687072754, |
| "rewards/margins": 17.434282302856445, |
| "rewards/rejected": -19.932022094726562, |
| "step": 690 |
| }, |
| { |
| "epoch": 2.97008547008547, |
| "grad_norm": 0.003662109375, |
| "learning_rate": 3.844483774214069e-06, |
| "logits/chosen": -1.4186488389968872, |
| "logits/rejected": -0.7549436688423157, |
| "logps/chosen": -64.54364776611328, |
| "logps/rejected": -277.66680908203125, |
| "loss": 0.0002, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.5480058193206787, |
| "rewards/margins": 17.503028869628906, |
| "rewards/rejected": -20.0510311126709, |
| "step": 695 |
| }, |
| { |
| "epoch": 2.9871794871794872, |
| "eval_logits/chosen": -1.4855865240097046, |
| "eval_logits/rejected": -0.8208981156349182, |
| "eval_logps/chosen": -67.3394775390625, |
| "eval_logps/rejected": -276.6026306152344, |
| "eval_loss": 0.00011922841804334894, |
| "eval_rewards/accuracies": 1.0, |
| "eval_rewards/chosen": -2.789573907852173, |
| "eval_rewards/margins": 17.068998336791992, |
| "eval_rewards/rejected": -19.858572006225586, |
| "eval_runtime": 9.6169, |
| "eval_samples_per_second": 20.797, |
| "eval_steps_per_second": 20.797, |
| "step": 699 |
| } |
| ], |
| "logging_steps": 5, |
| "max_steps": 702, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 3, |
| "save_steps": 500, |
| "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": 12, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
