{ "best_metric": null, "best_model_checkpoint": null, "epoch": 1.9977349943374858, "eval_steps": 1000, "global_step": 882, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0022650056625141564, "grad_norm": 2.6474389252526, "learning_rate": 5.617977528089887e-09, "logits/chosen": -2.3315391540527344, "logits/rejected": -2.3719687461853027, "logps/chosen": -323.98822021484375, "logps/rejected": -333.302734375, "loss": 0.6931, "rewards/accuracies": 0.0, "rewards/chosen": 0.0, "rewards/margins": 0.0, "rewards/rejected": 0.0, "step": 1 }, { "epoch": 0.022650056625141562, "grad_norm": 2.546226472747536, "learning_rate": 5.617977528089887e-08, "logits/chosen": -2.4010610580444336, "logits/rejected": -2.3566555976867676, "logps/chosen": -317.4289245605469, "logps/rejected": -297.19677734375, "loss": 0.6933, "rewards/accuracies": 0.46875, "rewards/chosen": 0.00016070896526798606, "rewards/margins": -4.6764034777879715e-07, "rewards/rejected": 0.00016117654740810394, "step": 10 }, { "epoch": 0.045300113250283124, "grad_norm": 2.575936847764113, "learning_rate": 1.1235955056179774e-07, "logits/chosen": -2.368352174758911, "logits/rejected": -2.383195638656616, "logps/chosen": -317.442138671875, "logps/rejected": -310.83355712890625, "loss": 0.693, "rewards/accuracies": 0.5062500238418579, "rewards/chosen": 0.0005755432066507638, "rewards/margins": 0.0001175893921754323, "rewards/rejected": 0.00045795380719937384, "step": 20 }, { "epoch": 0.06795016987542468, "grad_norm": 2.9029645784064435, "learning_rate": 1.6853932584269663e-07, "logits/chosen": -2.4023149013519287, "logits/rejected": -2.3384761810302734, "logps/chosen": -324.33111572265625, "logps/rejected": -293.6396179199219, "loss": 0.6924, "rewards/accuracies": 0.59375, "rewards/chosen": 0.0024513269308954477, "rewards/margins": 0.001740488805808127, "rewards/rejected": 0.000710838008671999, "step": 30 }, { "epoch": 0.09060022650056625, "grad_norm": 2.7631508582024225, "learning_rate": 2.2471910112359549e-07, "logits/chosen": -2.366269588470459, "logits/rejected": -2.384061813354492, "logps/chosen": -307.44244384765625, "logps/rejected": -305.51422119140625, "loss": 0.6905, "rewards/accuracies": 0.737500011920929, "rewards/chosen": 0.006844083778560162, "rewards/margins": 0.005638611502945423, "rewards/rejected": 0.0012054723920300603, "step": 40 }, { "epoch": 0.11325028312570781, "grad_norm": 2.4659439994932892, "learning_rate": 2.8089887640449437e-07, "logits/chosen": -2.337625741958618, "logits/rejected": -2.3457820415496826, "logps/chosen": -317.6352844238281, "logps/rejected": -304.02239990234375, "loss": 0.6859, "rewards/accuracies": 0.793749988079071, "rewards/chosen": 0.02052396535873413, "rewards/margins": 0.015680011361837387, "rewards/rejected": 0.004843952599912882, "step": 50 }, { "epoch": 0.13590033975084936, "grad_norm": 2.3504525836204753, "learning_rate": 3.3707865168539325e-07, "logits/chosen": -2.337041139602661, "logits/rejected": -2.3470675945281982, "logps/chosen": -307.0657653808594, "logps/rejected": -308.83709716796875, "loss": 0.6779, "rewards/accuracies": 0.7906249761581421, "rewards/chosen": 0.03320229798555374, "rewards/margins": 0.029773065820336342, "rewards/rejected": 0.003429233329370618, "step": 60 }, { "epoch": 0.15855039637599094, "grad_norm": 2.4436159820736068, "learning_rate": 3.9325842696629214e-07, "logits/chosen": -2.3073842525482178, "logits/rejected": -2.2655694484710693, "logps/chosen": -322.8226623535156, "logps/rejected": -297.44964599609375, "loss": 0.6641, "rewards/accuracies": 0.793749988079071, "rewards/chosen": 0.052168797701597214, "rewards/margins": 0.05918798968195915, "rewards/rejected": -0.007019191049039364, "step": 70 }, { "epoch": 0.1812004530011325, "grad_norm": 2.6664317328149236, "learning_rate": 4.4943820224719097e-07, "logits/chosen": -2.3002970218658447, "logits/rejected": -2.2402286529541016, "logps/chosen": -320.1070861816406, "logps/rejected": -302.884033203125, "loss": 0.6465, "rewards/accuracies": 0.831250011920929, "rewards/chosen": 0.03679288178682327, "rewards/margins": 0.10391112416982651, "rewards/rejected": -0.06711824238300323, "step": 80 }, { "epoch": 0.20385050962627407, "grad_norm": 3.2572528737706152, "learning_rate": 4.999980381634755e-07, "logits/chosen": -2.1139278411865234, "logits/rejected": -2.1478025913238525, "logps/chosen": -333.51519775390625, "logps/rejected": -344.88470458984375, "loss": 0.6106, "rewards/accuracies": 0.831250011920929, "rewards/chosen": -0.061306070536375046, "rewards/margins": 0.20201241970062256, "rewards/rejected": -0.2633184492588043, "step": 90 }, { "epoch": 0.22650056625141562, "grad_norm": 3.441455714530276, "learning_rate": 4.99762655034665e-07, "logits/chosen": -1.9943259954452515, "logits/rejected": -2.025644302368164, "logps/chosen": -347.1192321777344, "logps/rejected": -368.75213623046875, "loss": 0.5747, "rewards/accuracies": 0.815625011920929, "rewards/chosen": -0.25914034247398376, "rewards/margins": 0.2930234670639038, "rewards/rejected": -0.5521637797355652, "step": 100 }, { "epoch": 0.2491506228765572, "grad_norm": 3.756828982195148, "learning_rate": 4.991353278599861e-07, "logits/chosen": -1.9040400981903076, "logits/rejected": -1.8980512619018555, "logps/chosen": -368.0120544433594, "logps/rejected": -400.9908752441406, "loss": 0.5297, "rewards/accuracies": 0.828125, "rewards/chosen": -0.6502432227134705, "rewards/margins": 0.4519161283969879, "rewards/rejected": -1.1021593809127808, "step": 110 }, { "epoch": 0.2718006795016987, "grad_norm": 4.624437428257384, "learning_rate": 4.981170410826531e-07, "logits/chosen": -1.8735542297363281, "logits/rejected": -1.83624267578125, "logps/chosen": -419.53680419921875, "logps/rejected": -448.97528076171875, "loss": 0.4874, "rewards/accuracies": 0.793749988079071, "rewards/chosen": -0.9317169189453125, "rewards/margins": 0.6421880722045898, "rewards/rejected": -1.5739049911499023, "step": 120 }, { "epoch": 0.2944507361268403, "grad_norm": 4.949083580442793, "learning_rate": 4.967093926654972e-07, "logits/chosen": -1.8730270862579346, "logits/rejected": -1.8644899129867554, "logps/chosen": -421.25946044921875, "logps/rejected": -489.58526611328125, "loss": 0.4491, "rewards/accuracies": 0.824999988079071, "rewards/chosen": -0.9310008883476257, "rewards/margins": 0.8616712689399719, "rewards/rejected": -1.7926719188690186, "step": 130 }, { "epoch": 0.3171007927519819, "grad_norm": 4.996031451766477, "learning_rate": 4.949145915833377e-07, "logits/chosen": -1.8482372760772705, "logits/rejected": -1.809913992881775, "logps/chosen": -437.0384826660156, "logps/rejected": -518.9461669921875, "loss": 0.4203, "rewards/accuracies": 0.8374999761581421, "rewards/chosen": -1.2464931011199951, "rewards/margins": 1.060370922088623, "rewards/rejected": -2.306863784790039, "step": 140 }, { "epoch": 0.33975084937712347, "grad_norm": 5.659599672667922, "learning_rate": 4.92735454356513e-07, "logits/chosen": -1.7960946559906006, "logits/rejected": -1.747685194015503, "logps/chosen": -468.9212341308594, "logps/rejected": -572.489013671875, "loss": 0.3953, "rewards/accuracies": 0.831250011920929, "rewards/chosen": -1.327953577041626, "rewards/margins": 1.3009440898895264, "rewards/rejected": -2.6288976669311523, "step": 150 }, { "epoch": 0.362400906002265, "grad_norm": 6.071105217564194, "learning_rate": 4.901754006310109e-07, "logits/chosen": -1.8912861347198486, "logits/rejected": -1.8290300369262695, "logps/chosen": -469.5503845214844, "logps/rejected": -571.6123657226562, "loss": 0.3822, "rewards/accuracies": 0.8374999761581421, "rewards/chosen": -1.3224334716796875, "rewards/margins": 1.3983814716339111, "rewards/rejected": -2.7208149433135986, "step": 160 }, { "epoch": 0.38505096262740657, "grad_norm": 5.706084632255313, "learning_rate": 4.872384478121341e-07, "logits/chosen": -1.7244150638580322, "logits/rejected": -1.7288404703140259, "logps/chosen": -460.2561950683594, "logps/rejected": -626.7166748046875, "loss": 0.3686, "rewards/accuracies": 0.8531249761581421, "rewards/chosen": -1.572575569152832, "rewards/margins": 1.626971960067749, "rewards/rejected": -3.199547290802002, "step": 170 }, { "epoch": 0.40770101925254815, "grad_norm": 6.231205543249148, "learning_rate": 4.839292047601234e-07, "logits/chosen": -1.714738130569458, "logits/rejected": -1.655225157737732, "logps/chosen": -453.4337463378906, "logps/rejected": -580.6221923828125, "loss": 0.3646, "rewards/accuracies": 0.8656250238418579, "rewards/chosen": -1.3668302297592163, "rewards/margins": 1.5435349941253662, "rewards/rejected": -2.910365343093872, "step": 180 }, { "epoch": 0.43035107587768967, "grad_norm": 7.7415354302251185, "learning_rate": 4.80252864557629e-07, "logits/chosen": -1.5334300994873047, "logits/rejected": -1.5589441061019897, "logps/chosen": -473.0502014160156, "logps/rejected": -634.1920776367188, "loss": 0.3501, "rewards/accuracies": 0.8218749761581421, "rewards/chosen": -1.5327621698379517, "rewards/margins": 1.8059412240982056, "rewards/rejected": -3.338703155517578, "step": 190 }, { "epoch": 0.45300113250283125, "grad_norm": 6.247718164321941, "learning_rate": 4.7621519636038314e-07, "logits/chosen": -1.6528962850570679, "logits/rejected": -1.607163429260254, "logps/chosen": -468.12823486328125, "logps/rejected": -652.6515502929688, "loss": 0.3528, "rewards/accuracies": 0.871874988079071, "rewards/chosen": -1.4182040691375732, "rewards/margins": 2.0675435066223145, "rewards/rejected": -3.4857475757598877, "step": 200 }, { "epoch": 0.47565118912797283, "grad_norm": 5.866194755643268, "learning_rate": 4.7182253634385947e-07, "logits/chosen": -1.8014767169952393, "logits/rejected": -1.7056528329849243, "logps/chosen": -465.86004638671875, "logps/rejected": -619.522705078125, "loss": 0.3197, "rewards/accuracies": 0.875, "rewards/chosen": -1.308288335800171, "rewards/margins": 1.8610426187515259, "rewards/rejected": -3.1693310737609863, "step": 210 }, { "epoch": 0.4983012457531144, "grad_norm": 8.540677017846228, "learning_rate": 4.670817777601288e-07, "logits/chosen": -1.6279010772705078, "logits/rejected": -1.611676573753357, "logps/chosen": -486.3113708496094, "logps/rejected": -701.3948974609375, "loss": 0.331, "rewards/accuracies": 0.8843749761581421, "rewards/chosen": -1.751829743385315, "rewards/margins": 2.2083218097686768, "rewards/rejected": -3.960151195526123, "step": 220 }, { "epoch": 0.5209513023782559, "grad_norm": 7.180290608469894, "learning_rate": 4.620003601205124e-07, "logits/chosen": -1.618285894393921, "logits/rejected": -1.5827783346176147, "logps/chosen": -495.9786682128906, "logps/rejected": -671.6422729492188, "loss": 0.3079, "rewards/accuracies": 0.840624988079071, "rewards/chosen": -1.722399353981018, "rewards/margins": 1.888711929321289, "rewards/rejected": -3.6111114025115967, "step": 230 }, { "epoch": 0.5436013590033975, "grad_norm": 7.812256996678949, "learning_rate": 4.565862575210102e-07, "logits/chosen": -1.5925363302230835, "logits/rejected": -1.5041205883026123, "logps/chosen": -467.281494140625, "logps/rejected": -654.7891845703125, "loss": 0.3183, "rewards/accuracies": 0.846875011920929, "rewards/chosen": -1.521933913230896, "rewards/margins": 2.0333094596862793, "rewards/rejected": -3.555243730545044, "step": 240 }, { "epoch": 0.5662514156285391, "grad_norm": 7.655324105498245, "learning_rate": 4.5084796612882256e-07, "logits/chosen": -1.6674625873565674, "logits/rejected": -1.6272201538085938, "logps/chosen": -447.31683349609375, "logps/rejected": -636.6031494140625, "loss": 0.3193, "rewards/accuracies": 0.8531249761581421, "rewards/chosen": -1.515677809715271, "rewards/margins": 1.9939563274383545, "rewards/rejected": -3.509634494781494, "step": 250 }, { "epoch": 0.5889014722536806, "grad_norm": 11.00110804956485, "learning_rate": 4.447944908496042e-07, "logits/chosen": -1.5731537342071533, "logits/rejected": -1.5266633033752441, "logps/chosen": -487.8212890625, "logps/rejected": -684.2847900390625, "loss": 0.294, "rewards/accuracies": 0.8656250238418579, "rewards/chosen": -1.6535365581512451, "rewards/margins": 2.1172165870666504, "rewards/rejected": -3.7707531452178955, "step": 260 }, { "epoch": 0.6115515288788222, "grad_norm": 8.565085184192531, "learning_rate": 4.3843533119637125e-07, "logits/chosen": -1.6604524850845337, "logits/rejected": -1.6133596897125244, "logps/chosen": -491.6846618652344, "logps/rejected": -732.2810668945312, "loss": 0.2903, "rewards/accuracies": 0.8812500238418579, "rewards/chosen": -1.610835075378418, "rewards/margins": 2.487947940826416, "rewards/rejected": -4.098783493041992, "step": 270 }, { "epoch": 0.6342015855039638, "grad_norm": 9.092182327634628, "learning_rate": 4.317804663822391e-07, "logits/chosen": -1.5771735906600952, "logits/rejected": -1.4461592435836792, "logps/chosen": -494.58538818359375, "logps/rejected": -696.6702880859375, "loss": 0.2894, "rewards/accuracies": 0.878125011920929, "rewards/chosen": -1.75591242313385, "rewards/margins": 2.3270115852355957, "rewards/rejected": -4.082923412322998, "step": 280 }, { "epoch": 0.6568516421291053, "grad_norm": 6.8364437103089495, "learning_rate": 4.248403396603817e-07, "logits/chosen": -1.662105917930603, "logits/rejected": -1.6381728649139404, "logps/chosen": -465.8875427246094, "logps/rejected": -674.3609619140625, "loss": 0.2922, "rewards/accuracies": 0.8812500238418579, "rewards/chosen": -1.528600811958313, "rewards/margins": 2.1931700706481934, "rewards/rejected": -3.7217705249786377, "step": 290 }, { "epoch": 0.6795016987542469, "grad_norm": 9.441509895706943, "learning_rate": 4.176258419357899e-07, "logits/chosen": -1.6376762390136719, "logits/rejected": -1.5463725328445435, "logps/chosen": -469.1526794433594, "logps/rejected": -699.0280151367188, "loss": 0.2994, "rewards/accuracies": 0.8843749761581421, "rewards/chosen": -1.5148029327392578, "rewards/margins": 2.2971949577331543, "rewards/rejected": -3.811997890472412, "step": 300 }, { "epoch": 0.7021517553793885, "grad_norm": 9.112495750796707, "learning_rate": 4.101482946745438e-07, "logits/chosen": -1.5513279438018799, "logits/rejected": -1.4326368570327759, "logps/chosen": -492.14154052734375, "logps/rejected": -711.12255859375, "loss": 0.2852, "rewards/accuracies": 0.862500011920929, "rewards/chosen": -1.719791054725647, "rewards/margins": 2.3883697986602783, "rewards/rejected": -4.108160972595215, "step": 310 }, { "epoch": 0.72480181200453, "grad_norm": 8.205308841823761, "learning_rate": 4.024194321374212e-07, "logits/chosen": -1.5063291788101196, "logits/rejected": -1.4347339868545532, "logps/chosen": -481.82080078125, "logps/rejected": -725.50830078125, "loss": 0.2888, "rewards/accuracies": 0.875, "rewards/chosen": -1.73996901512146, "rewards/margins": 2.5435125827789307, "rewards/rejected": -4.283481597900391, "step": 320 }, { "epoch": 0.7474518686296716, "grad_norm": 8.205344156872085, "learning_rate": 3.9445138296572113e-07, "logits/chosen": -1.5704634189605713, "logits/rejected": -1.4724892377853394, "logps/chosen": -482.04742431640625, "logps/rejected": -713.767578125, "loss": 0.2815, "rewards/accuracies": 0.8812500238418579, "rewards/chosen": -1.6026813983917236, "rewards/margins": 2.5262672901153564, "rewards/rejected": -4.128949165344238, "step": 330 }, { "epoch": 0.7701019252548131, "grad_norm": 8.687044357645386, "learning_rate": 3.8625665114819873e-07, "logits/chosen": -1.6331688165664673, "logits/rejected": -1.5526864528656006, "logps/chosen": -502.4422912597656, "logps/rejected": -728.0780029296875, "loss": 0.2967, "rewards/accuracies": 0.871874988079071, "rewards/chosen": -1.7365747690200806, "rewards/margins": 2.4770658016204834, "rewards/rejected": -4.2136406898498535, "step": 340 }, { "epoch": 0.7927519818799547, "grad_norm": 8.427795252063401, "learning_rate": 3.778480963989811e-07, "logits/chosen": -1.598920464515686, "logits/rejected": -1.5631158351898193, "logps/chosen": -502.58648681640625, "logps/rejected": -721.0816650390625, "loss": 0.2818, "rewards/accuracies": 0.875, "rewards/chosen": -1.7856614589691162, "rewards/margins": 2.304830551147461, "rewards/rejected": -4.090491771697998, "step": 350 }, { "epoch": 0.8154020385050963, "grad_norm": 7.260269705482, "learning_rate": 3.6923891397725475e-07, "logits/chosen": -1.5732126235961914, "logits/rejected": -1.4579116106033325, "logps/chosen": -505.892822265625, "logps/rejected": -708.0283813476562, "loss": 0.273, "rewards/accuracies": 0.893750011920929, "rewards/chosen": -1.6417592763900757, "rewards/margins": 2.4231035709381104, "rewards/rejected": -4.0648627281188965, "step": 360 }, { "epoch": 0.8380520951302378, "grad_norm": 10.479677653279909, "learning_rate": 3.604426139803941e-07, "logits/chosen": -1.439273715019226, "logits/rejected": -1.3620045185089111, "logps/chosen": -517.6815795898438, "logps/rejected": -789.8757934570312, "loss": 0.2764, "rewards/accuracies": 0.840624988079071, "rewards/chosen": -1.919734239578247, "rewards/margins": 2.797135591506958, "rewards/rejected": -4.716870307922363, "step": 370 }, { "epoch": 0.8607021517553793, "grad_norm": 7.985186234740201, "learning_rate": 3.5147300014302457e-07, "logits/chosen": -1.559147596359253, "logits/rejected": -1.436856746673584, "logps/chosen": -494.31500244140625, "logps/rejected": -769.4177856445312, "loss": 0.2741, "rewards/accuracies": 0.903124988079071, "rewards/chosen": -1.6577142477035522, "rewards/margins": 2.9566261768341064, "rewards/rejected": -4.614340305328369, "step": 380 }, { "epoch": 0.883352208380521, "grad_norm": 9.216176794426861, "learning_rate": 3.423441481752911e-07, "logits/chosen": -1.5555119514465332, "logits/rejected": -1.4373959302902222, "logps/chosen": -485.8768615722656, "logps/rejected": -722.7447509765625, "loss": 0.2552, "rewards/accuracies": 0.9125000238418579, "rewards/chosen": -1.6954419612884521, "rewards/margins": 2.5930514335632324, "rewards/rejected": -4.2884931564331055, "step": 390 }, { "epoch": 0.9060022650056625, "grad_norm": 11.159180818421824, "learning_rate": 3.3307038367432443e-07, "logits/chosen": -1.3249311447143555, "logits/rejected": -1.2017321586608887, "logps/chosen": -486.0079040527344, "logps/rejected": -734.5474853515625, "loss": 0.2917, "rewards/accuracies": 0.890625, "rewards/chosen": -1.7877933979034424, "rewards/margins": 2.6582465171813965, "rewards/rejected": -4.446040153503418, "step": 400 }, { "epoch": 0.928652321630804, "grad_norm": 9.06036199110745, "learning_rate": 3.2366625964356904e-07, "logits/chosen": -1.3760263919830322, "logits/rejected": -1.157128095626831, "logps/chosen": -521.535888671875, "logps/rejected": -756.6954345703125, "loss": 0.2745, "rewards/accuracies": 0.9156249761581421, "rewards/chosen": -1.8339954614639282, "rewards/margins": 2.690295696258545, "rewards/rejected": -4.524291038513184, "step": 410 }, { "epoch": 0.9513023782559457, "grad_norm": 8.613445152071886, "learning_rate": 3.1414653365524824e-07, "logits/chosen": -1.3222942352294922, "logits/rejected": -1.1949504613876343, "logps/chosen": -496.029296875, "logps/rejected": -751.8590087890625, "loss": 0.268, "rewards/accuracies": 0.903124988079071, "rewards/chosen": -1.8118371963500977, "rewards/margins": 2.6409120559692383, "rewards/rejected": -4.452749252319336, "step": 420 }, { "epoch": 0.9739524348810872, "grad_norm": 10.126507731476897, "learning_rate": 3.0452614469180916e-07, "logits/chosen": -1.3294384479522705, "logits/rejected": -1.1579396724700928, "logps/chosen": -505.3882751464844, "logps/rejected": -744.6370849609375, "loss": 0.2817, "rewards/accuracies": 0.8687499761581421, "rewards/chosen": -1.895179033279419, "rewards/margins": 2.613166332244873, "rewards/rejected": -4.508345603942871, "step": 430 }, { "epoch": 0.9966024915062288, "grad_norm": 8.359546887824873, "learning_rate": 2.948201897026839e-07, "logits/chosen": -1.2789385318756104, "logits/rejected": -1.1186952590942383, "logps/chosen": -518.8313598632812, "logps/rejected": -757.8375244140625, "loss": 0.2713, "rewards/accuracies": 0.8687499761581421, "rewards/chosen": -1.9324840307235718, "rewards/margins": 2.687189817428589, "rewards/rejected": -4.619673728942871, "step": 440 }, { "epoch": 1.0192525481313703, "grad_norm": 9.916277053083935, "learning_rate": 2.8504389991316105e-07, "logits/chosen": -1.2456687688827515, "logits/rejected": -1.116817831993103, "logps/chosen": -517.7054443359375, "logps/rejected": -768.4178466796875, "loss": 0.2485, "rewards/accuracies": 0.903124988079071, "rewards/chosen": -2.0004186630249023, "rewards/margins": 2.688814640045166, "rewards/rejected": -4.689233303070068, "step": 450 }, { "epoch": 1.0419026047565119, "grad_norm": 10.473177616855008, "learning_rate": 2.7521261692254064e-07, "logits/chosen": -1.201494574546814, "logits/rejected": -1.1607218980789185, "logps/chosen": -502.1883850097656, "logps/rejected": -778.163818359375, "loss": 0.2687, "rewards/accuracies": 0.856249988079071, "rewards/chosen": -1.8305113315582275, "rewards/margins": 2.7049150466918945, "rewards/rejected": -4.535426139831543, "step": 460 }, { "epoch": 1.0645526613816534, "grad_norm": 8.161060651228674, "learning_rate": 2.653417686290843e-07, "logits/chosen": -1.3370718955993652, "logits/rejected": -1.1073302030563354, "logps/chosen": -504.89801025390625, "logps/rejected": -745.27099609375, "loss": 0.2668, "rewards/accuracies": 0.8968750238418579, "rewards/chosen": -1.8848193883895874, "rewards/margins": 2.6601107120513916, "rewards/rejected": -4.544930458068848, "step": 470 }, { "epoch": 1.087202718006795, "grad_norm": 9.08329196731907, "learning_rate": 2.5544684501953874e-07, "logits/chosen": -1.3237013816833496, "logits/rejected": -1.1105015277862549, "logps/chosen": -522.7344970703125, "logps/rejected": -795.3836669921875, "loss": 0.2582, "rewards/accuracies": 0.90625, "rewards/chosen": -1.9063637256622314, "rewards/margins": 2.994534492492676, "rewards/rejected": -4.900897979736328, "step": 480 }, { "epoch": 1.1098527746319367, "grad_norm": 9.239751540023942, "learning_rate": 2.455433738612257e-07, "logits/chosen": -1.2477200031280518, "logits/rejected": -1.140298843383789, "logps/chosen": -510.219970703125, "logps/rejected": -790.1013793945312, "loss": 0.2627, "rewards/accuracies": 0.8812500238418579, "rewards/chosen": -1.9776588678359985, "rewards/margins": 2.872147798538208, "rewards/rejected": -4.849806785583496, "step": 490 }, { "epoch": 1.1325028312570782, "grad_norm": 8.063198151261274, "learning_rate": 2.356468963348451e-07, "logits/chosen": -1.2390304803848267, "logits/rejected": -1.0948108434677124, "logps/chosen": -526.1553955078125, "logps/rejected": -783.7991943359375, "loss": 0.2513, "rewards/accuracies": 0.8812500238418579, "rewards/chosen": -1.924363136291504, "rewards/margins": 2.824939250946045, "rewards/rejected": -4.749302864074707, "step": 500 }, { "epoch": 1.1551528878822197, "grad_norm": 10.628914294455955, "learning_rate": 2.257729426462276e-07, "logits/chosen": -1.2929015159606934, "logits/rejected": -1.1904685497283936, "logps/chosen": -500.63934326171875, "logps/rejected": -745.3140869140625, "loss": 0.2745, "rewards/accuracies": 0.878125011920929, "rewards/chosen": -1.9067703485488892, "rewards/margins": 2.5259783267974854, "rewards/rejected": -4.432748317718506, "step": 510 }, { "epoch": 1.1778029445073612, "grad_norm": 8.572156157412216, "learning_rate": 2.1593700765531089e-07, "logits/chosen": -1.3477250337600708, "logits/rejected": -1.147517442703247, "logps/chosen": -497.2181091308594, "logps/rejected": -794.3870849609375, "loss": 0.2556, "rewards/accuracies": 0.8999999761581421, "rewards/chosen": -1.7573410272598267, "rewards/margins": 3.1540889739990234, "rewards/rejected": -4.9114298820495605, "step": 520 }, { "epoch": 1.2004530011325028, "grad_norm": 9.357116476254854, "learning_rate": 2.0615452656058264e-07, "logits/chosen": -1.1733572483062744, "logits/rejected": -1.077141523361206, "logps/chosen": -533.3096923828125, "logps/rejected": -809.8570556640625, "loss": 0.2487, "rewards/accuracies": 0.8968750238418579, "rewards/chosen": -2.1453044414520264, "rewards/margins": 2.8495564460754395, "rewards/rejected": -4.994861602783203, "step": 530 }, { "epoch": 1.2231030577576445, "grad_norm": 10.169756727358006, "learning_rate": 1.964408506771477e-07, "logits/chosen": -1.330486536026001, "logits/rejected": -1.1011441946029663, "logps/chosen": -514.3831787109375, "logps/rejected": -743.8990478515625, "loss": 0.2611, "rewards/accuracies": 0.8999999761581421, "rewards/chosen": -1.9130659103393555, "rewards/margins": 2.6177687644958496, "rewards/rejected": -4.530834674835205, "step": 540 }, { "epoch": 1.245753114382786, "grad_norm": 10.039033828119221, "learning_rate": 1.868112233464317e-07, "logits/chosen": -1.2120153903961182, "logits/rejected": -1.0327503681182861, "logps/chosen": -521.9749755859375, "logps/rejected": -802.6920166015625, "loss": 0.2457, "rewards/accuracies": 0.903124988079071, "rewards/chosen": -1.9761244058609009, "rewards/margins": 2.9549593925476074, "rewards/rejected": -4.931083679199219, "step": 550 }, { "epoch": 1.2684031710079275, "grad_norm": 10.38214991301508, "learning_rate": 1.7728075601532256e-07, "logits/chosen": -1.1885329484939575, "logits/rejected": -0.9879345893859863, "logps/chosen": -522.0006103515625, "logps/rejected": -811.2237548828125, "loss": 0.24, "rewards/accuracies": 0.8812500238418579, "rewards/chosen": -2.0510880947113037, "rewards/margins": 2.9444658756256104, "rewards/rejected": -4.995553970336914, "step": 560 }, { "epoch": 1.291053227633069, "grad_norm": 8.356041939902939, "learning_rate": 1.678644045222913e-07, "logits/chosen": -1.2759056091308594, "logits/rejected": -1.119461178779602, "logps/chosen": -509.73583984375, "logps/rejected": -767.29443359375, "loss": 0.262, "rewards/accuracies": 0.887499988079071, "rewards/chosen": -1.9916547536849976, "rewards/margins": 2.671016216278076, "rewards/rejected": -4.662671089172363, "step": 570 }, { "epoch": 1.3137032842582106, "grad_norm": 8.912049656879141, "learning_rate": 1.585769456277027e-07, "logits/chosen": -1.2154964208602905, "logits/rejected": -1.0624061822891235, "logps/chosen": -511.436767578125, "logps/rejected": -811.2041015625, "loss": 0.2244, "rewards/accuracies": 0.918749988079071, "rewards/chosen": -1.986825704574585, "rewards/margins": 3.094810962677002, "rewards/rejected": -5.081636905670166, "step": 580 }, { "epoch": 1.3363533408833521, "grad_norm": 9.754438029355793, "learning_rate": 1.4943295382514728e-07, "logits/chosen": -1.162213921546936, "logits/rejected": -1.0503453016281128, "logps/chosen": -510.4007263183594, "logps/rejected": -791.3707275390625, "loss": 0.2591, "rewards/accuracies": 0.8843749761581421, "rewards/chosen": -1.8774373531341553, "rewards/margins": 2.9028286933898926, "rewards/rejected": -4.780265808105469, "step": 590 }, { "epoch": 1.3590033975084936, "grad_norm": 10.049386128170934, "learning_rate": 1.404467784701844e-07, "logits/chosen": -1.2329612970352173, "logits/rejected": -1.0405888557434082, "logps/chosen": -513.1143798828125, "logps/rejected": -783.7535400390625, "loss": 0.2439, "rewards/accuracies": 0.921875, "rewards/chosen": -1.8017826080322266, "rewards/margins": 3.078281879425049, "rewards/rejected": -4.880064487457275, "step": 600 }, { "epoch": 1.3816534541336354, "grad_norm": 10.351517688722195, "learning_rate": 1.3163252126238522e-07, "logits/chosen": -1.2502057552337646, "logits/rejected": -1.0325191020965576, "logps/chosen": -521.5572509765625, "logps/rejected": -792.7130737304688, "loss": 0.2364, "rewards/accuracies": 0.8999999761581421, "rewards/chosen": -2.0930302143096924, "rewards/margins": 2.9488086700439453, "rewards/rejected": -5.041838645935059, "step": 610 }, { "epoch": 1.404303510758777, "grad_norm": 10.201760293762261, "learning_rate": 1.2300401411601636e-07, "logits/chosen": -1.2752196788787842, "logits/rejected": -1.0628340244293213, "logps/chosen": -522.7730712890625, "logps/rejected": -773.3001708984375, "loss": 0.2389, "rewards/accuracies": 0.921875, "rewards/chosen": -1.9185714721679688, "rewards/margins": 2.8828213214874268, "rewards/rejected": -4.801392555236816, "step": 620 }, { "epoch": 1.4269535673839184, "grad_norm": 8.835921463685338, "learning_rate": 1.145747974540856e-07, "logits/chosen": -1.2582758665084839, "logits/rejected": -1.0210275650024414, "logps/chosen": -551.0906982421875, "logps/rejected": -785.31982421875, "loss": 0.2562, "rewards/accuracies": 0.8687499761581421, "rewards/chosen": -2.2444424629211426, "rewards/margins": 2.7261741161346436, "rewards/rejected": -4.970616817474365, "step": 630 }, { "epoch": 1.44960362400906, "grad_norm": 12.991915284453638, "learning_rate": 1.0635809895981699e-07, "logits/chosen": -1.2509222030639648, "logits/rejected": -1.0769033432006836, "logps/chosen": -552.90087890625, "logps/rejected": -806.5215454101562, "loss": 0.2494, "rewards/accuracies": 0.887499988079071, "rewards/chosen": -2.2703638076782227, "rewards/margins": 2.754547119140625, "rewards/rejected": -5.024910926818848, "step": 640 }, { "epoch": 1.4722536806342017, "grad_norm": 10.504984052416424, "learning_rate": 9.836681281889678e-08, "logits/chosen": -1.34755277633667, "logits/rejected": -1.1464807987213135, "logps/chosen": -515.4022216796875, "logps/rejected": -784.6029052734375, "loss": 0.2408, "rewards/accuracies": 0.918749988079071, "rewards/chosen": -1.8913652896881104, "rewards/margins": 2.9281973838806152, "rewards/rejected": -4.819562911987305, "step": 650 }, { "epoch": 1.4949037372593432, "grad_norm": 10.125922089796893, "learning_rate": 9.061347948506629e-08, "logits/chosen": -1.2215931415557861, "logits/rejected": -1.1074914932250977, "logps/chosen": -523.8232421875, "logps/rejected": -797.5764770507812, "loss": 0.2312, "rewards/accuracies": 0.878125011920929, "rewards/chosen": -1.9787216186523438, "rewards/margins": 2.8579554557800293, "rewards/rejected": -4.836677551269531, "step": 660 }, { "epoch": 1.5175537938844847, "grad_norm": 13.328368593276167, "learning_rate": 8.311026600081394e-08, "logits/chosen": -1.11947762966156, "logits/rejected": -0.9734872579574585, "logps/chosen": -534.2103271484375, "logps/rejected": -830.9747924804688, "loss": 0.2318, "rewards/accuracies": 0.887499988079071, "rewards/chosen": -2.1272778511047363, "rewards/margins": 3.069934844970703, "rewards/rejected": -5.1972126960754395, "step": 670 }, { "epoch": 1.5402038505096263, "grad_norm": 9.729514763537226, "learning_rate": 7.586894690404894e-08, "logits/chosen": -1.217233657836914, "logits/rejected": -0.9491454362869263, "logps/chosen": -547.2406005859375, "logps/rejected": -827.4415283203125, "loss": 0.2277, "rewards/accuracies": 0.921875, "rewards/chosen": -2.2959611415863037, "rewards/margins": 3.175316095352173, "rewards/rejected": -5.471277713775635, "step": 680 }, { "epoch": 1.5628539071347678, "grad_norm": 9.576592993059183, "learning_rate": 6.890088575071954e-08, "logits/chosen": -1.187366247177124, "logits/rejected": -1.0213922262191772, "logps/chosen": -545.3994750976562, "logps/rejected": -804.3179321289062, "loss": 0.2423, "rewards/accuracies": 0.8843749761581421, "rewards/chosen": -2.1425087451934814, "rewards/margins": 2.924528121948242, "rewards/rejected": -5.067037105560303, "step": 690 }, { "epoch": 1.5855039637599093, "grad_norm": 12.126314956409692, "learning_rate": 6.221701728237008e-08, "logits/chosen": -1.1499079465866089, "logits/rejected": -0.9196650385856628, "logps/chosen": -524.6936645507812, "logps/rejected": -810.4720458984375, "loss": 0.248, "rewards/accuracies": 0.8843749761581421, "rewards/chosen": -2.0638883113861084, "rewards/margins": 3.0613222122192383, "rewards/rejected": -5.125210762023926, "step": 700 }, { "epoch": 1.6081540203850508, "grad_norm": 11.134418963577295, "learning_rate": 5.582783026662349e-08, "logits/chosen": -1.0319141149520874, "logits/rejected": -0.956072986125946, "logps/chosen": -526.4638671875, "logps/rejected": -814.6434326171875, "loss": 0.26, "rewards/accuracies": 0.875, "rewards/chosen": -2.0331358909606934, "rewards/margins": 2.952976703643799, "rewards/rejected": -4.986112117767334, "step": 710 }, { "epoch": 1.6308040770101924, "grad_norm": 9.315625565902986, "learning_rate": 4.9743351037514534e-08, "logits/chosen": -1.129042148590088, "logits/rejected": -0.8932101130485535, "logps/chosen": -548.4468994140625, "logps/rejected": -836.2429809570312, "loss": 0.2368, "rewards/accuracies": 0.890625, "rewards/chosen": -2.1513266563415527, "rewards/margins": 3.1787352561950684, "rewards/rejected": -5.330061912536621, "step": 720 }, { "epoch": 1.6534541336353341, "grad_norm": 9.92739535657428, "learning_rate": 4.397312776150439e-08, "logits/chosen": -1.048117995262146, "logits/rejected": -0.967863917350769, "logps/chosen": -528.8873901367188, "logps/rejected": -818.2623291015625, "loss": 0.2489, "rewards/accuracies": 0.875, "rewards/chosen": -2.2309556007385254, "rewards/margins": 2.885885715484619, "rewards/rejected": -5.116840839385986, "step": 730 }, { "epoch": 1.6761041902604756, "grad_norm": 11.55426000236745, "learning_rate": 3.852621545386761e-08, "logits/chosen": -1.1159693002700806, "logits/rejected": -0.8810169100761414, "logps/chosen": -534.227294921875, "logps/rejected": -810.526611328125, "loss": 0.2553, "rewards/accuracies": 0.8812500238418579, "rewards/chosen": -2.1146936416625977, "rewards/margins": 3.039085626602173, "rewards/rejected": -5.153779029846191, "step": 740 }, { "epoch": 1.6987542468856174, "grad_norm": 11.882365137452437, "learning_rate": 3.341116176896447e-08, "logits/chosen": -1.1606030464172363, "logits/rejected": -0.969256579875946, "logps/chosen": -527.7787475585938, "logps/rejected": -783.6763916015625, "loss": 0.2362, "rewards/accuracies": 0.903124988079071, "rewards/chosen": -2.098135471343994, "rewards/margins": 2.7795679569244385, "rewards/rejected": -4.877703666687012, "step": 750 }, { "epoch": 1.721404303510759, "grad_norm": 11.932846657725108, "learning_rate": 2.863599358669755e-08, "logits/chosen": -1.1670676469802856, "logits/rejected": -0.9956063032150269, "logps/chosen": -504.7257385253906, "logps/rejected": -799.7178955078125, "loss": 0.2366, "rewards/accuracies": 0.918749988079071, "rewards/chosen": -2.0211801528930664, "rewards/margins": 3.071319818496704, "rewards/rejected": -5.09250020980835, "step": 760 }, { "epoch": 1.7440543601359004, "grad_norm": 11.068137379022545, "learning_rate": 2.4208204416201665e-08, "logits/chosen": -1.1682803630828857, "logits/rejected": -1.0027186870574951, "logps/chosen": -526.065673828125, "logps/rejected": -813.3223876953125, "loss": 0.239, "rewards/accuracies": 0.909375011920929, "rewards/chosen": -2.0436623096466064, "rewards/margins": 3.140903949737549, "rewards/rejected": -5.184566020965576, "step": 770 }, { "epoch": 1.766704416761042, "grad_norm": 12.478180187301291, "learning_rate": 2.0134742636534945e-08, "logits/chosen": -1.0638221502304077, "logits/rejected": -0.9631050229072571, "logps/chosen": -504.27154541015625, "logps/rejected": -821.7930908203125, "loss": 0.2399, "rewards/accuracies": 0.9281250238418579, "rewards/chosen": -1.9072799682617188, "rewards/margins": 3.2128396034240723, "rewards/rejected": -5.120119571685791, "step": 780 }, { "epoch": 1.7893544733861835, "grad_norm": 10.797469509942093, "learning_rate": 1.6422000592823642e-08, "logits/chosen": -1.1970932483673096, "logits/rejected": -1.005079984664917, "logps/chosen": -508.31494140625, "logps/rejected": -812.4329223632812, "loss": 0.2332, "rewards/accuracies": 0.940625011920929, "rewards/chosen": -1.9276374578475952, "rewards/margins": 3.216580867767334, "rewards/rejected": -5.144218444824219, "step": 790 }, { "epoch": 1.812004530011325, "grad_norm": 9.201913655606754, "learning_rate": 1.3075804564971981e-08, "logits/chosen": -1.165398120880127, "logits/rejected": -0.946398913860321, "logps/chosen": -515.598876953125, "logps/rejected": -806.0333251953125, "loss": 0.2595, "rewards/accuracies": 0.90625, "rewards/chosen": -2.098996639251709, "rewards/margins": 3.0318830013275146, "rewards/rejected": -5.130879878997803, "step": 800 }, { "epoch": 1.8346545866364665, "grad_norm": 12.171909373979952, "learning_rate": 1.01014056246794e-08, "logits/chosen": -1.14596426486969, "logits/rejected": -0.9247859716415405, "logps/chosen": -521.8255615234375, "logps/rejected": -803.5557861328125, "loss": 0.2352, "rewards/accuracies": 0.918749988079071, "rewards/chosen": -2.031855344772339, "rewards/margins": 3.0747625827789307, "rewards/rejected": -5.1066179275512695, "step": 810 }, { "epoch": 1.857304643261608, "grad_norm": 10.898179869470376, "learning_rate": 7.503471395112065e-09, "logits/chosen": -1.1780989170074463, "logits/rejected": -0.931725025177002, "logps/chosen": -521.3743896484375, "logps/rejected": -795.4444580078125, "loss": 0.2545, "rewards/accuracies": 0.8968750238418579, "rewards/chosen": -2.1411118507385254, "rewards/margins": 2.875601291656494, "rewards/rejected": -5.0167131423950195, "step": 820 }, { "epoch": 1.8799546998867496, "grad_norm": 8.213548605406768, "learning_rate": 5.286078726160548e-09, "logits/chosen": -1.0937467813491821, "logits/rejected": -0.9241981506347656, "logps/chosen": -538.4014282226562, "logps/rejected": -784.3555297851562, "loss": 0.2322, "rewards/accuracies": 0.9156249761581421, "rewards/chosen": -2.020725727081299, "rewards/margins": 2.831569194793701, "rewards/rejected": -4.852294921875, "step": 830 }, { "epoch": 1.9026047565118913, "grad_norm": 11.652554680737152, "learning_rate": 3.4527072967777404e-09, "logits/chosen": -1.1555739641189575, "logits/rejected": -0.9933372735977173, "logps/chosen": -534.9871215820312, "logps/rejected": -795.6361083984375, "loss": 0.2605, "rewards/accuracies": 0.878125011920929, "rewards/chosen": -2.151045083999634, "rewards/margins": 2.7527434825897217, "rewards/rejected": -4.9037885665893555, "step": 840 }, { "epoch": 1.9252548131370328, "grad_norm": 11.308556042176512, "learning_rate": 2.0062341544370143e-09, "logits/chosen": -1.1897794008255005, "logits/rejected": -0.9474241137504578, "logps/chosen": -540.0062255859375, "logps/rejected": -806.772216796875, "loss": 0.2696, "rewards/accuracies": 0.8656250238418579, "rewards/chosen": -2.1642906665802, "rewards/margins": 2.9361448287963867, "rewards/rejected": -5.100435733795166, "step": 850 }, { "epoch": 1.9479048697621744, "grad_norm": 10.396927511013917, "learning_rate": 9.489292002793913e-10, "logits/chosen": -1.1064279079437256, "logits/rejected": -0.8940499424934387, "logps/chosen": -526.8607177734375, "logps/rejected": -789.2191772460938, "loss": 0.2429, "rewards/accuracies": 0.887499988079071, "rewards/chosen": -2.0915961265563965, "rewards/margins": 2.8657567501068115, "rewards/rejected": -4.957352638244629, "step": 860 }, { "epoch": 1.970554926387316, "grad_norm": 8.728874761020926, "learning_rate": 2.824516270348576e-10, "logits/chosen": -1.2010364532470703, "logits/rejected": -1.0010019540786743, "logps/chosen": -522.8780517578125, "logps/rejected": -835.6373901367188, "loss": 0.2245, "rewards/accuracies": 0.9281250238418579, "rewards/chosen": -1.9398930072784424, "rewards/margins": 3.3346550464630127, "rewards/rejected": -5.274548053741455, "step": 870 }, { "epoch": 1.9932049830124576, "grad_norm": 8.736000347816738, "learning_rate": 7.847315307535352e-12, "logits/chosen": -1.1201304197311401, "logits/rejected": -0.9259728193283081, "logps/chosen": -526.8204345703125, "logps/rejected": -830.1511840820312, "loss": 0.2423, "rewards/accuracies": 0.9375, "rewards/chosen": -2.0089852809906006, "rewards/margins": 3.2325141429901123, "rewards/rejected": -5.241499423980713, "step": 880 }, { "epoch": 1.9977349943374858, "step": 882, "total_flos": 0.0, "train_loss": 0.32467431276022984, "train_runtime": 5857.252, "train_samples_per_second": 38.584, "train_steps_per_second": 0.151 } ], "logging_steps": 10, "max_steps": 882, "num_input_tokens_seen": 0, "num_train_epochs": 2, "save_steps": 100, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 0.0, "train_batch_size": 8, "trial_name": null, "trial_params": null }