| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.0, | |
| "eval_steps": 500, | |
| "global_step": 500, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 6.6666666666666675e-06, | |
| "logits/chosen": -3.1526219844818115, | |
| "logits/rejected": -3.3119924068450928, | |
| "logps/chosen": -18.28135108947754, | |
| "logps/rejected": -33.52398681640625, | |
| "loss": 0.6997, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -0.006796836853027344, | |
| "rewards/margins": -0.012901116162538528, | |
| "rewards/rejected": 0.006104278843849897, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 6.666666666666667e-05, | |
| "logits/chosen": -3.1073851585388184, | |
| "logits/rejected": -3.090308666229248, | |
| "logps/chosen": -20.141502380371094, | |
| "logps/rejected": -18.037580490112305, | |
| "loss": 0.694, | |
| "rewards/accuracies": 0.0833333358168602, | |
| "rewards/chosen": 0.001263597165234387, | |
| "rewards/margins": -0.001611270010471344, | |
| "rewards/rejected": 0.002874867059290409, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.00013333333333333334, | |
| "logits/chosen": -3.0630269050598145, | |
| "logits/rejected": -3.1416983604431152, | |
| "logps/chosen": -31.93638038635254, | |
| "logps/rejected": -42.507789611816406, | |
| "loss": 0.6916, | |
| "rewards/accuracies": 0.2750000059604645, | |
| "rewards/chosen": 0.010843334719538689, | |
| "rewards/margins": 0.003288193140178919, | |
| "rewards/rejected": 0.0075551411136984825, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0002, | |
| "logits/chosen": -3.148637294769287, | |
| "logits/rejected": -3.150296211242676, | |
| "logps/chosen": -22.95195770263672, | |
| "logps/rejected": -23.612133026123047, | |
| "loss": 0.6974, | |
| "rewards/accuracies": 0.15000000596046448, | |
| "rewards/chosen": -0.002520971465855837, | |
| "rewards/margins": -0.0078018950298428535, | |
| "rewards/rejected": 0.005280924029648304, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0002666666666666667, | |
| "logits/chosen": -3.1271812915802, | |
| "logits/rejected": -3.0903429985046387, | |
| "logps/chosen": -34.57979965209961, | |
| "logps/rejected": -27.37040138244629, | |
| "loss": 0.6925, | |
| "rewards/accuracies": 0.125, | |
| "rewards/chosen": 0.03781440109014511, | |
| "rewards/margins": 0.006826506461948156, | |
| "rewards/rejected": 0.030987894162535667, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0003333333333333333, | |
| "logits/chosen": -3.0824079513549805, | |
| "logits/rejected": -3.0999526977539062, | |
| "logps/chosen": -29.16314697265625, | |
| "logps/rejected": -30.843231201171875, | |
| "loss": 0.7057, | |
| "rewards/accuracies": 0.25, | |
| "rewards/chosen": 0.08534860610961914, | |
| "rewards/margins": -0.013902002945542336, | |
| "rewards/rejected": 0.09925060719251633, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.0004, | |
| "logits/chosen": -3.1495282649993896, | |
| "logits/rejected": -3.184638500213623, | |
| "logps/chosen": -19.72355079650879, | |
| "logps/rejected": -28.213886260986328, | |
| "loss": 0.7213, | |
| "rewards/accuracies": 0.07500000298023224, | |
| "rewards/chosen": -0.05634387582540512, | |
| "rewards/margins": -0.043291497975587845, | |
| "rewards/rejected": -0.01305237878113985, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.00046666666666666666, | |
| "logits/chosen": -3.04826021194458, | |
| "logits/rejected": -3.005667209625244, | |
| "logps/chosen": -25.625635147094727, | |
| "logps/rejected": -24.768199920654297, | |
| "loss": 0.7002, | |
| "rewards/accuracies": 0.20000000298023224, | |
| "rewards/chosen": -0.06301303952932358, | |
| "rewards/margins": -0.00712633365765214, | |
| "rewards/rejected": -0.055886708199977875, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0005333333333333334, | |
| "logits/chosen": -3.048518419265747, | |
| "logits/rejected": -3.061340808868408, | |
| "logps/chosen": -32.414894104003906, | |
| "logps/rejected": -34.773399353027344, | |
| "loss": 0.7835, | |
| "rewards/accuracies": 0.32499998807907104, | |
| "rewards/chosen": -0.23218408226966858, | |
| "rewards/margins": -0.03052915260195732, | |
| "rewards/rejected": -0.20165491104125977, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.0006, | |
| "logits/chosen": -3.035808563232422, | |
| "logits/rejected": -3.1053929328918457, | |
| "logps/chosen": -31.509021759033203, | |
| "logps/rejected": -45.353553771972656, | |
| "loss": 0.7051, | |
| "rewards/accuracies": 0.30000001192092896, | |
| "rewards/chosen": -0.3204951286315918, | |
| "rewards/margins": 0.283873975276947, | |
| "rewards/rejected": -0.604369044303894, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.0006666666666666666, | |
| "logits/chosen": -2.959228992462158, | |
| "logits/rejected": -2.98266339302063, | |
| "logps/chosen": -51.024803161621094, | |
| "logps/rejected": -54.91625213623047, | |
| "loss": 1.0868, | |
| "rewards/accuracies": 0.20000000298023224, | |
| "rewards/chosen": -1.835883378982544, | |
| "rewards/margins": 0.3176426589488983, | |
| "rewards/rejected": -2.1535260677337646, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.0007333333333333333, | |
| "logits/chosen": -3.0634117126464844, | |
| "logits/rejected": -3.0850846767425537, | |
| "logps/chosen": -23.597423553466797, | |
| "logps/rejected": -26.58676528930664, | |
| "loss": 0.7629, | |
| "rewards/accuracies": 0.20000000298023224, | |
| "rewards/chosen": -0.2058214694261551, | |
| "rewards/margins": -0.05166854336857796, | |
| "rewards/rejected": -0.15415294468402863, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.0008, | |
| "logits/chosen": -3.1232190132141113, | |
| "logits/rejected": -3.1285691261291504, | |
| "logps/chosen": -20.47592544555664, | |
| "logps/rejected": -22.593311309814453, | |
| "loss": 0.7341, | |
| "rewards/accuracies": 0.15000000596046448, | |
| "rewards/chosen": -0.46454209089279175, | |
| "rewards/margins": -0.023245975375175476, | |
| "rewards/rejected": -0.44129619002342224, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 0.0008666666666666667, | |
| "logits/chosen": -2.975956678390503, | |
| "logits/rejected": -3.027247190475464, | |
| "logps/chosen": -34.07396697998047, | |
| "logps/rejected": -42.46125793457031, | |
| "loss": 0.723, | |
| "rewards/accuracies": 0.2750000059604645, | |
| "rewards/chosen": -0.4881093502044678, | |
| "rewards/margins": 0.3038038909435272, | |
| "rewards/rejected": -0.7919132113456726, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.0009333333333333333, | |
| "logits/chosen": -2.981985569000244, | |
| "logits/rejected": -2.9559457302093506, | |
| "logps/chosen": -33.598899841308594, | |
| "logps/rejected": -40.866451263427734, | |
| "loss": 0.7877, | |
| "rewards/accuracies": 0.2750000059604645, | |
| "rewards/chosen": -0.35516494512557983, | |
| "rewards/margins": 0.4116950035095215, | |
| "rewards/rejected": -0.7668598890304565, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.001, | |
| "logits/chosen": -3.05903959274292, | |
| "logits/rejected": -3.0611279010772705, | |
| "logps/chosen": -20.234691619873047, | |
| "logps/rejected": -19.169904708862305, | |
| "loss": 0.9163, | |
| "rewards/accuracies": 0.10000000149011612, | |
| "rewards/chosen": -0.36899399757385254, | |
| "rewards/margins": -0.08086968958377838, | |
| "rewards/rejected": -0.28812432289123535, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 0.0009723756906077348, | |
| "logits/chosen": -2.9612436294555664, | |
| "logits/rejected": -2.9223554134368896, | |
| "logps/chosen": -32.5883674621582, | |
| "logps/rejected": -25.370834350585938, | |
| "loss": 0.7856, | |
| "rewards/accuracies": 0.17499999701976776, | |
| "rewards/chosen": -0.18723489344120026, | |
| "rewards/margins": -0.10649768263101578, | |
| "rewards/rejected": -0.08073721826076508, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 0.0009447513812154696, | |
| "logits/chosen": -2.9623398780822754, | |
| "logits/rejected": -2.914522171020508, | |
| "logps/chosen": -39.70682907104492, | |
| "logps/rejected": -33.20659637451172, | |
| "loss": 0.898, | |
| "rewards/accuracies": 0.20000000298023224, | |
| "rewards/chosen": -0.4028944969177246, | |
| "rewards/margins": -0.04350559413433075, | |
| "rewards/rejected": -0.35938888788223267, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 0.0009171270718232044, | |
| "logits/chosen": -3.0877394676208496, | |
| "logits/rejected": -3.092603921890259, | |
| "logps/chosen": -43.16739273071289, | |
| "logps/rejected": -47.593746185302734, | |
| "loss": 1.2201, | |
| "rewards/accuracies": 0.17499999701976776, | |
| "rewards/chosen": -2.5319814682006836, | |
| "rewards/margins": -0.10082467645406723, | |
| "rewards/rejected": -2.431157112121582, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 0.0008895027624309392, | |
| "logits/chosen": -2.09025239944458, | |
| "logits/rejected": -2.0903568267822266, | |
| "logps/chosen": -112.7010269165039, | |
| "logps/rejected": -105.31596374511719, | |
| "loss": 2.9619, | |
| "rewards/accuracies": 0.20000000298023224, | |
| "rewards/chosen": -7.883659362792969, | |
| "rewards/margins": -0.7001466751098633, | |
| "rewards/rejected": -7.1835126876831055, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 0.0008618784530386741, | |
| "logits/chosen": -1.8135408163070679, | |
| "logits/rejected": -1.8160464763641357, | |
| "logps/chosen": -75.82559967041016, | |
| "logps/rejected": -65.90226745605469, | |
| "loss": 2.555, | |
| "rewards/accuracies": 0.17499999701976776, | |
| "rewards/chosen": -5.0599846839904785, | |
| "rewards/margins": -0.6209059953689575, | |
| "rewards/rejected": -4.439078330993652, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 0.0008342541436464089, | |
| "logits/chosen": -2.6365649700164795, | |
| "logits/rejected": -2.633017063140869, | |
| "logps/chosen": -73.8572998046875, | |
| "logps/rejected": -115.65068054199219, | |
| "loss": 1.3204, | |
| "rewards/accuracies": 0.3499999940395355, | |
| "rewards/chosen": -4.614927291870117, | |
| "rewards/margins": 3.1844677925109863, | |
| "rewards/rejected": -7.799394130706787, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 0.0008066298342541437, | |
| "logits/chosen": -2.338550329208374, | |
| "logits/rejected": -2.337129831314087, | |
| "logps/chosen": -75.24410247802734, | |
| "logps/rejected": -89.40665435791016, | |
| "loss": 1.6189, | |
| "rewards/accuracies": 0.20000000298023224, | |
| "rewards/chosen": -5.084308624267578, | |
| "rewards/margins": 0.9201302528381348, | |
| "rewards/rejected": -6.004438877105713, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 0.0007790055248618785, | |
| "logits/chosen": -2.328997850418091, | |
| "logits/rejected": -2.326946496963501, | |
| "logps/chosen": -62.543540954589844, | |
| "logps/rejected": -84.85903930664062, | |
| "loss": 1.3971, | |
| "rewards/accuracies": 0.22499999403953552, | |
| "rewards/chosen": -3.8174233436584473, | |
| "rewards/margins": 1.793341875076294, | |
| "rewards/rejected": -5.610764503479004, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 0.0007513812154696133, | |
| "logits/chosen": -2.794644832611084, | |
| "logits/rejected": -2.7910995483398438, | |
| "logps/chosen": -58.40827560424805, | |
| "logps/rejected": -79.7098617553711, | |
| "loss": 1.2398, | |
| "rewards/accuracies": 0.25, | |
| "rewards/chosen": -3.556912660598755, | |
| "rewards/margins": 1.693927526473999, | |
| "rewards/rejected": -5.250839710235596, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 0.0007237569060773481, | |
| "logits/chosen": -2.703537702560425, | |
| "logits/rejected": -2.7038962841033936, | |
| "logps/chosen": -40.56044006347656, | |
| "logps/rejected": -43.81157684326172, | |
| "loss": 1.3514, | |
| "rewards/accuracies": 0.10000000149011612, | |
| "rewards/chosen": -2.4380924701690674, | |
| "rewards/margins": 0.2092890739440918, | |
| "rewards/rejected": -2.647381544113159, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 0.0006961325966850829, | |
| "logits/chosen": -2.870227098464966, | |
| "logits/rejected": -2.891268253326416, | |
| "logps/chosen": -67.85215759277344, | |
| "logps/rejected": -91.36506652832031, | |
| "loss": 1.2686, | |
| "rewards/accuracies": 0.2750000059604645, | |
| "rewards/chosen": -3.6993489265441895, | |
| "rewards/margins": 1.7562892436981201, | |
| "rewards/rejected": -5.455638408660889, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 0.0006685082872928176, | |
| "logits/chosen": -3.1669762134552, | |
| "logits/rejected": -3.1949028968811035, | |
| "logps/chosen": -30.239299774169922, | |
| "logps/rejected": -42.69363021850586, | |
| "loss": 0.7342, | |
| "rewards/accuracies": 0.20000000298023224, | |
| "rewards/chosen": -0.33911675214767456, | |
| "rewards/margins": 0.5873344540596008, | |
| "rewards/rejected": -0.9264512062072754, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 0.0006408839779005525, | |
| "logits/chosen": -3.065441131591797, | |
| "logits/rejected": -3.0655035972595215, | |
| "logps/chosen": -39.723567962646484, | |
| "logps/rejected": -53.760826110839844, | |
| "loss": 1.3162, | |
| "rewards/accuracies": 0.15000000596046448, | |
| "rewards/chosen": -2.2118256092071533, | |
| "rewards/margins": 0.6879772543907166, | |
| "rewards/rejected": -2.8998026847839355, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 0.0006132596685082873, | |
| "logits/chosen": -2.6939263343811035, | |
| "logits/rejected": -2.6940252780914307, | |
| "logps/chosen": -60.013755798339844, | |
| "logps/rejected": -58.038536071777344, | |
| "loss": 2.1225, | |
| "rewards/accuracies": 0.20000000298023224, | |
| "rewards/chosen": -3.602609634399414, | |
| "rewards/margins": 0.10532107204198837, | |
| "rewards/rejected": -3.70793080329895, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 0.000585635359116022, | |
| "logits/chosen": -2.695570468902588, | |
| "logits/rejected": -2.6920197010040283, | |
| "logps/chosen": -46.42293930053711, | |
| "logps/rejected": -44.63296127319336, | |
| "loss": 0.9542, | |
| "rewards/accuracies": 0.17499999701976776, | |
| "rewards/chosen": -2.1268885135650635, | |
| "rewards/margins": 0.09368989616632462, | |
| "rewards/rejected": -2.22057843208313, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 0.000558011049723757, | |
| "logits/chosen": -3.1045467853546143, | |
| "logits/rejected": -3.1222219467163086, | |
| "logps/chosen": -22.447757720947266, | |
| "logps/rejected": -27.05214500427246, | |
| "loss": 0.7383, | |
| "rewards/accuracies": 0.15000000596046448, | |
| "rewards/chosen": -0.1806814968585968, | |
| "rewards/margins": 0.14409010112285614, | |
| "rewards/rejected": -0.32477161288261414, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 0.0005303867403314917, | |
| "logits/chosen": -3.1156842708587646, | |
| "logits/rejected": -3.130932569503784, | |
| "logps/chosen": -21.35702133178711, | |
| "logps/rejected": -20.4589900970459, | |
| "loss": 0.7337, | |
| "rewards/accuracies": 0.20000000298023224, | |
| "rewards/chosen": -0.29580003023147583, | |
| "rewards/margins": 0.08602263033390045, | |
| "rewards/rejected": -0.3818226456642151, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 0.0005027624309392266, | |
| "logits/chosen": -3.050220012664795, | |
| "logits/rejected": -3.046596050262451, | |
| "logps/chosen": -24.79575538635254, | |
| "logps/rejected": -33.02082443237305, | |
| "loss": 0.6601, | |
| "rewards/accuracies": 0.25, | |
| "rewards/chosen": -0.31610769033432007, | |
| "rewards/margins": 0.4491572976112366, | |
| "rewards/rejected": -0.7652650475502014, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 0.00047513812154696136, | |
| "logits/chosen": -3.1225409507751465, | |
| "logits/rejected": -3.142671823501587, | |
| "logps/chosen": -34.958404541015625, | |
| "logps/rejected": -37.874855041503906, | |
| "loss": 0.6053, | |
| "rewards/accuracies": 0.30000001192092896, | |
| "rewards/chosen": -0.23902547359466553, | |
| "rewards/margins": 0.42017507553100586, | |
| "rewards/rejected": -0.6592004895210266, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 0.00044751381215469617, | |
| "logits/chosen": -3.3166356086730957, | |
| "logits/rejected": -3.271238327026367, | |
| "logps/chosen": -41.72985076904297, | |
| "logps/rejected": -43.41400909423828, | |
| "loss": 0.7479, | |
| "rewards/accuracies": 0.30000001192092896, | |
| "rewards/chosen": -0.3358752131462097, | |
| "rewards/margins": 0.2891044020652771, | |
| "rewards/rejected": -0.6249796152114868, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 0.0004198895027624309, | |
| "logits/chosen": -3.2012417316436768, | |
| "logits/rejected": -3.2214763164520264, | |
| "logps/chosen": -29.948162078857422, | |
| "logps/rejected": -34.02383041381836, | |
| "loss": 0.7083, | |
| "rewards/accuracies": 0.15000000596046448, | |
| "rewards/chosen": -0.20552043616771698, | |
| "rewards/margins": -0.0057243406772613525, | |
| "rewards/rejected": -0.19979611039161682, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 0.00039226519337016573, | |
| "logits/chosen": -3.217780590057373, | |
| "logits/rejected": -3.2134361267089844, | |
| "logps/chosen": -16.843505859375, | |
| "logps/rejected": -17.496295928955078, | |
| "loss": 0.763, | |
| "rewards/accuracies": 0.10000000149011612, | |
| "rewards/chosen": -0.05251544713973999, | |
| "rewards/margins": -0.043300654739141464, | |
| "rewards/rejected": -0.009214771911501884, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 0.0003646408839779006, | |
| "logits/chosen": -3.168470859527588, | |
| "logits/rejected": -3.2095096111297607, | |
| "logps/chosen": -31.317157745361328, | |
| "logps/rejected": -40.99824523925781, | |
| "loss": 0.6099, | |
| "rewards/accuracies": 0.3499999940395355, | |
| "rewards/chosen": -0.06307787448167801, | |
| "rewards/margins": 0.3778603971004486, | |
| "rewards/rejected": -0.4409382939338684, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 0.0003370165745856354, | |
| "logits/chosen": -3.1807992458343506, | |
| "logits/rejected": -3.1729633808135986, | |
| "logps/chosen": -25.59493064880371, | |
| "logps/rejected": -24.889175415039062, | |
| "loss": 0.6768, | |
| "rewards/accuracies": 0.20000000298023224, | |
| "rewards/chosen": -0.03841208666563034, | |
| "rewards/margins": 0.2406325787305832, | |
| "rewards/rejected": -0.2790446877479553, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 0.00030939226519337016, | |
| "logits/chosen": -3.1939032077789307, | |
| "logits/rejected": -3.1782355308532715, | |
| "logps/chosen": -14.640347480773926, | |
| "logps/rejected": -13.197868347167969, | |
| "loss": 0.6587, | |
| "rewards/accuracies": 0.15000000596046448, | |
| "rewards/chosen": -0.013051311485469341, | |
| "rewards/margins": 0.11761553585529327, | |
| "rewards/rejected": -0.1306668370962143, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 0.00028176795580110497, | |
| "logits/chosen": -3.2121384143829346, | |
| "logits/rejected": -3.242738723754883, | |
| "logps/chosen": -26.492828369140625, | |
| "logps/rejected": -33.63092041015625, | |
| "loss": 0.6547, | |
| "rewards/accuracies": 0.2750000059604645, | |
| "rewards/chosen": -0.15410800278186798, | |
| "rewards/margins": 0.2064083367586136, | |
| "rewards/rejected": -0.36051633954048157, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 0.0002541436464088398, | |
| "logits/chosen": -3.217862367630005, | |
| "logits/rejected": -3.2220897674560547, | |
| "logps/chosen": -45.029319763183594, | |
| "logps/rejected": -50.58649444580078, | |
| "loss": 0.7074, | |
| "rewards/accuracies": 0.30000001192092896, | |
| "rewards/chosen": -0.21466748416423798, | |
| "rewards/margins": 0.43003931641578674, | |
| "rewards/rejected": -0.644706666469574, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 0.0002265193370165746, | |
| "logits/chosen": -3.2857704162597656, | |
| "logits/rejected": -3.285273313522339, | |
| "logps/chosen": -24.04534912109375, | |
| "logps/rejected": -25.24020767211914, | |
| "loss": 0.7055, | |
| "rewards/accuracies": 0.22499999403953552, | |
| "rewards/chosen": -0.05270111560821533, | |
| "rewards/margins": 0.19839158654212952, | |
| "rewards/rejected": -0.25109270215034485, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 0.0001988950276243094, | |
| "logits/chosen": -3.1307530403137207, | |
| "logits/rejected": -3.097612142562866, | |
| "logps/chosen": -25.230710983276367, | |
| "logps/rejected": -27.142175674438477, | |
| "loss": 0.6265, | |
| "rewards/accuracies": 0.2750000059604645, | |
| "rewards/chosen": 0.06528893858194351, | |
| "rewards/margins": 0.3360690474510193, | |
| "rewards/rejected": -0.27078011631965637, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 0.0001712707182320442, | |
| "logits/chosen": -3.210901975631714, | |
| "logits/rejected": -3.250978469848633, | |
| "logps/chosen": -16.752582550048828, | |
| "logps/rejected": -30.28323745727539, | |
| "loss": 0.5541, | |
| "rewards/accuracies": 0.30000001192092896, | |
| "rewards/chosen": 0.05518122762441635, | |
| "rewards/margins": 0.479708731174469, | |
| "rewards/rejected": -0.42452749609947205, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 0.000143646408839779, | |
| "logits/chosen": -3.048260450363159, | |
| "logits/rejected": -3.1042888164520264, | |
| "logps/chosen": -33.8494758605957, | |
| "logps/rejected": -49.50830841064453, | |
| "loss": 0.5998, | |
| "rewards/accuracies": 0.3499999940395355, | |
| "rewards/chosen": 0.0009274661424569786, | |
| "rewards/margins": 0.6367529630661011, | |
| "rewards/rejected": -0.635825514793396, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 0.0001160220994475138, | |
| "logits/chosen": -3.272062301635742, | |
| "logits/rejected": -3.2841033935546875, | |
| "logps/chosen": -16.145034790039062, | |
| "logps/rejected": -19.171789169311523, | |
| "loss": 0.6134, | |
| "rewards/accuracies": 0.22499999403953552, | |
| "rewards/chosen": 0.10014114528894424, | |
| "rewards/margins": 0.250404953956604, | |
| "rewards/rejected": -0.15026383101940155, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 8.839779005524861e-05, | |
| "logits/chosen": -3.1270499229431152, | |
| "logits/rejected": -3.1859707832336426, | |
| "logps/chosen": -28.680988311767578, | |
| "logps/rejected": -39.859764099121094, | |
| "loss": 0.6342, | |
| "rewards/accuracies": 0.32499998807907104, | |
| "rewards/chosen": -0.053285278379917145, | |
| "rewards/margins": 0.39891186356544495, | |
| "rewards/rejected": -0.4521971344947815, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 6.0773480662983424e-05, | |
| "logits/chosen": -3.140641212463379, | |
| "logits/rejected": -3.090985059738159, | |
| "logps/chosen": -32.661651611328125, | |
| "logps/rejected": -32.10502624511719, | |
| "loss": 0.7772, | |
| "rewards/accuracies": 0.20000000298023224, | |
| "rewards/chosen": -0.3610732853412628, | |
| "rewards/margins": 0.07261800020933151, | |
| "rewards/rejected": -0.43369120359420776, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 3.3149171270718233e-05, | |
| "logits/chosen": -3.1711134910583496, | |
| "logits/rejected": -3.190059185028076, | |
| "logps/chosen": -36.2880973815918, | |
| "logps/rejected": -48.17897415161133, | |
| "loss": 0.613, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.25595012307167053, | |
| "rewards/margins": 0.6426823139190674, | |
| "rewards/rejected": -0.8986324071884155, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_logits/chosen": -3.225074052810669, | |
| "eval_logits/rejected": -3.2351696491241455, | |
| "eval_logps/chosen": -28.331165313720703, | |
| "eval_logps/rejected": -31.33060073852539, | |
| "eval_loss": 0.7142000794410706, | |
| "eval_rewards/accuracies": 0.22200000286102295, | |
| "eval_rewards/chosen": -0.15271247923374176, | |
| "eval_rewards/margins": 0.16096290946006775, | |
| "eval_rewards/rejected": -0.3136754035949707, | |
| "eval_runtime": 411.5707, | |
| "eval_samples_per_second": 2.43, | |
| "eval_steps_per_second": 0.304, | |
| "step": 500 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 512, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 500, | |
| "total_flos": 0.0, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |