| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.9994666666666666, | |
| "eval_steps": 500, | |
| "global_step": 937, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 5.319148936170213e-08, | |
| "logits/chosen": 0.06159799173474312, | |
| "logits/rejected": 0.2503473162651062, | |
| "logps/chosen": -348.4295349121094, | |
| "logps/rejected": -269.23321533203125, | |
| "loss": 0.4013, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.319148936170213e-07, | |
| "logits/chosen": 0.050236668437719345, | |
| "logits/rejected": 0.23283202946186066, | |
| "logps/chosen": -357.2872314453125, | |
| "logps/rejected": -284.6590270996094, | |
| "loss": 0.3533, | |
| "rewards/accuracies": 0.3958333432674408, | |
| "rewards/chosen": -9.200325621350203e-06, | |
| "rewards/margins": -3.7092313505127095e-06, | |
| "rewards/rejected": -5.491095635079546e-06, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.0638297872340427e-06, | |
| "logits/chosen": 0.07863714545965195, | |
| "logits/rejected": 0.17622847855091095, | |
| "logps/chosen": -305.1004638671875, | |
| "logps/rejected": -284.56573486328125, | |
| "loss": 0.3668, | |
| "rewards/accuracies": 0.46875, | |
| "rewards/chosen": -3.269618537160568e-05, | |
| "rewards/margins": -7.148426084313542e-05, | |
| "rewards/rejected": 3.878808274748735e-05, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.595744680851064e-06, | |
| "logits/chosen": 0.10170497745275497, | |
| "logits/rejected": 0.2674924433231354, | |
| "logps/chosen": -318.61309814453125, | |
| "logps/rejected": -289.11529541015625, | |
| "loss": 0.3582, | |
| "rewards/accuracies": 0.44999998807907104, | |
| "rewards/chosen": -2.7896196570509346e-06, | |
| "rewards/margins": -0.00010077792830998078, | |
| "rewards/rejected": 9.798832616070285e-05, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 2.1276595744680853e-06, | |
| "logits/chosen": 0.09615897387266159, | |
| "logits/rejected": 0.2696045935153961, | |
| "logps/chosen": -371.0738220214844, | |
| "logps/rejected": -285.8421325683594, | |
| "loss": 0.3338, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": 4.211660780129023e-05, | |
| "rewards/margins": 0.0002511069178581238, | |
| "rewards/rejected": -0.00020899029914289713, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.6595744680851065e-06, | |
| "logits/chosen": 0.1058117151260376, | |
| "logits/rejected": 0.2665451169013977, | |
| "logps/chosen": -365.52001953125, | |
| "logps/rejected": -291.03778076171875, | |
| "loss": 0.3245, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": 0.0004745240439660847, | |
| "rewards/margins": 0.0008677773294039071, | |
| "rewards/rejected": -0.00039325328543782234, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 3.191489361702128e-06, | |
| "logits/chosen": 0.07869451493024826, | |
| "logits/rejected": 0.2602466642856598, | |
| "logps/chosen": -347.3607482910156, | |
| "logps/rejected": -290.67364501953125, | |
| "loss": 0.3335, | |
| "rewards/accuracies": 0.731249988079071, | |
| "rewards/chosen": 0.0010413106065243483, | |
| "rewards/margins": 0.0015250012511387467, | |
| "rewards/rejected": -0.00048369058640673757, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 3.723404255319149e-06, | |
| "logits/chosen": 0.12400275468826294, | |
| "logits/rejected": 0.26277071237564087, | |
| "logps/chosen": -336.801025390625, | |
| "logps/rejected": -295.81439208984375, | |
| "loss": 0.3541, | |
| "rewards/accuracies": 0.668749988079071, | |
| "rewards/chosen": 0.0017572289798408747, | |
| "rewards/margins": 0.002515274565666914, | |
| "rewards/rejected": -0.0007580459350720048, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.255319148936171e-06, | |
| "logits/chosen": 0.06505224108695984, | |
| "logits/rejected": 0.18910573422908783, | |
| "logps/chosen": -312.43511962890625, | |
| "logps/rejected": -274.01617431640625, | |
| "loss": 0.3528, | |
| "rewards/accuracies": 0.6812499761581421, | |
| "rewards/chosen": 0.003332150634378195, | |
| "rewards/margins": 0.004707558546215296, | |
| "rewards/rejected": -0.0013754077954217792, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.787234042553192e-06, | |
| "logits/chosen": 0.09570185840129852, | |
| "logits/rejected": 0.292137086391449, | |
| "logps/chosen": -320.4757080078125, | |
| "logps/rejected": -282.39453125, | |
| "loss": 0.3233, | |
| "rewards/accuracies": 0.706250011920929, | |
| "rewards/chosen": 0.005724198184907436, | |
| "rewards/margins": 0.00849069096148014, | |
| "rewards/rejected": -0.002766493707895279, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.999375059004058e-06, | |
| "logits/chosen": 0.11051477491855621, | |
| "logits/rejected": 0.2542819380760193, | |
| "logps/chosen": -341.4449768066406, | |
| "logps/rejected": -303.7613220214844, | |
| "loss": 0.2954, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": 0.006659829057753086, | |
| "rewards/margins": 0.017101887613534927, | |
| "rewards/rejected": -0.010442061349749565, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.9955571065548795e-06, | |
| "logits/chosen": 0.09644671529531479, | |
| "logits/rejected": 0.26117372512817383, | |
| "logps/chosen": -347.67059326171875, | |
| "logps/rejected": -339.35205078125, | |
| "loss": 0.3068, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -0.005562308244407177, | |
| "rewards/margins": 0.025120720267295837, | |
| "rewards/rejected": -0.03068302944302559, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.9882736864879e-06, | |
| "logits/chosen": 0.1643160879611969, | |
| "logits/rejected": 0.33352339267730713, | |
| "logps/chosen": -393.2250061035156, | |
| "logps/rejected": -346.75457763671875, | |
| "loss": 0.3154, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.026022305712103844, | |
| "rewards/margins": 0.04141021519899368, | |
| "rewards/rejected": -0.06743250787258148, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.977534912960124e-06, | |
| "logits/chosen": 0.14933325350284576, | |
| "logits/rejected": 0.36655381321907043, | |
| "logps/chosen": -425.4112243652344, | |
| "logps/rejected": -428.3099670410156, | |
| "loss": 0.2788, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.07423840463161469, | |
| "rewards/margins": 0.06771700084209442, | |
| "rewards/rejected": -0.1419553905725479, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.963355698422092e-06, | |
| "logits/chosen": 0.20979711413383484, | |
| "logits/rejected": 0.2947987914085388, | |
| "logps/chosen": -455.8238830566406, | |
| "logps/rejected": -478.470703125, | |
| "loss": 0.2889, | |
| "rewards/accuracies": 0.6812499761581421, | |
| "rewards/chosen": -0.11944756656885147, | |
| "rewards/margins": 0.08143159002065659, | |
| "rewards/rejected": -0.20087914168834686, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.945755732909625e-06, | |
| "logits/chosen": 0.22968748211860657, | |
| "logits/rejected": 0.3435000777244568, | |
| "logps/chosen": -494.1380920410156, | |
| "logps/rejected": -557.603271484375, | |
| "loss": 0.2274, | |
| "rewards/accuracies": 0.78125, | |
| "rewards/chosen": -0.15503188967704773, | |
| "rewards/margins": 0.12964466214179993, | |
| "rewards/rejected": -0.28467655181884766, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.924759456701167e-06, | |
| "logits/chosen": 0.2052285373210907, | |
| "logits/rejected": 0.3779992461204529, | |
| "logps/chosen": -555.717529296875, | |
| "logps/rejected": -618.7412719726562, | |
| "loss": 0.2945, | |
| "rewards/accuracies": 0.668749988079071, | |
| "rewards/chosen": -0.22048434615135193, | |
| "rewards/margins": 0.12210027873516083, | |
| "rewards/rejected": -0.34258466958999634, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.900396026378671e-06, | |
| "logits/chosen": 0.19485141336917877, | |
| "logits/rejected": 0.35038071870803833, | |
| "logps/chosen": -564.8143310546875, | |
| "logps/rejected": -614.8881225585938, | |
| "loss": 0.2459, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.19514423608779907, | |
| "rewards/margins": 0.13954147696495056, | |
| "rewards/rejected": -0.33468568325042725, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.872699274339169e-06, | |
| "logits/chosen": 0.26888132095336914, | |
| "logits/rejected": 0.38210389018058777, | |
| "logps/chosen": -542.71875, | |
| "logps/rejected": -610.9818115234375, | |
| "loss": 0.2668, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.20494337379932404, | |
| "rewards/margins": 0.13571330904960632, | |
| "rewards/rejected": -0.34065669775009155, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.8417076618132434e-06, | |
| "logits/chosen": 0.27069562673568726, | |
| "logits/rejected": 0.3432765603065491, | |
| "logps/chosen": -552.3577270507812, | |
| "logps/rejected": -633.7227783203125, | |
| "loss": 0.2295, | |
| "rewards/accuracies": 0.7437499761581421, | |
| "rewards/chosen": -0.20474235713481903, | |
| "rewards/margins": 0.14836545288562775, | |
| "rewards/rejected": -0.35310783982276917, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.807464225455655e-06, | |
| "logits/chosen": 0.2959613800048828, | |
| "logits/rejected": 0.4290657639503479, | |
| "logps/chosen": -607.9371337890625, | |
| "logps/rejected": -658.5770263671875, | |
| "loss": 0.2489, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -0.22546908259391785, | |
| "rewards/margins": 0.14855849742889404, | |
| "rewards/rejected": -0.3740275502204895, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.770016517582283e-06, | |
| "logits/chosen": 0.315132200717926, | |
| "logits/rejected": 0.43885716795921326, | |
| "logps/chosen": -636.2320556640625, | |
| "logps/rejected": -777.607421875, | |
| "loss": 0.2412, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -0.28788623213768005, | |
| "rewards/margins": 0.1952207386493683, | |
| "rewards/rejected": -0.4831070005893707, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.7294165401363616e-06, | |
| "logits/chosen": 0.27698999643325806, | |
| "logits/rejected": 0.468730628490448, | |
| "logps/chosen": -641.8867797851562, | |
| "logps/rejected": -759.454833984375, | |
| "loss": 0.2575, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -0.26833561062812805, | |
| "rewards/margins": 0.1899726390838623, | |
| "rewards/rejected": -0.45830821990966797, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.68572067247573e-06, | |
| "logits/chosen": 0.2822573781013489, | |
| "logits/rejected": 0.42682987451553345, | |
| "logps/chosen": -578.1446533203125, | |
| "logps/rejected": -648.0927734375, | |
| "loss": 0.2443, | |
| "rewards/accuracies": 0.78125, | |
| "rewards/chosen": -0.19278240203857422, | |
| "rewards/margins": 0.18077878654003143, | |
| "rewards/rejected": -0.37356117367744446, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.638989593081364e-06, | |
| "logits/chosen": 0.3316207230091095, | |
| "logits/rejected": 0.42460235953330994, | |
| "logps/chosen": -508.28924560546875, | |
| "logps/rejected": -627.9178466796875, | |
| "loss": 0.2394, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.18461547791957855, | |
| "rewards/margins": 0.17850570380687714, | |
| "rewards/rejected": -0.3631211817264557, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.5892881952959015e-06, | |
| "logits/chosen": 0.34463202953338623, | |
| "logits/rejected": 0.4213656485080719, | |
| "logps/chosen": -602.0831909179688, | |
| "logps/rejected": -723.1624145507812, | |
| "loss": 0.2229, | |
| "rewards/accuracies": 0.768750011920929, | |
| "rewards/chosen": -0.22979848086833954, | |
| "rewards/margins": 0.1861880123615265, | |
| "rewards/rejected": -0.41598647832870483, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.536685497209182e-06, | |
| "logits/chosen": 0.2841508984565735, | |
| "logits/rejected": 0.4096981883049011, | |
| "logps/chosen": -649.3954467773438, | |
| "logps/rejected": -714.6544189453125, | |
| "loss": 0.263, | |
| "rewards/accuracies": 0.706250011920929, | |
| "rewards/chosen": -0.26575830578804016, | |
| "rewards/margins": 0.15293100476264954, | |
| "rewards/rejected": -0.4186893105506897, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.481254545815943e-06, | |
| "logits/chosen": 0.33302605152130127, | |
| "logits/rejected": 0.4252694249153137, | |
| "logps/chosen": -652.48828125, | |
| "logps/rejected": -734.8269653320312, | |
| "loss": 0.2375, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": -0.2703024744987488, | |
| "rewards/margins": 0.16575393080711365, | |
| "rewards/rejected": -0.43605637550354004, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.42307231557875e-06, | |
| "logits/chosen": 0.2720186710357666, | |
| "logits/rejected": 0.4189075827598572, | |
| "logps/chosen": -606.5438232421875, | |
| "logps/rejected": -711.4382934570312, | |
| "loss": 0.234, | |
| "rewards/accuracies": 0.78125, | |
| "rewards/chosen": -0.2561277151107788, | |
| "rewards/margins": 0.1838252693414688, | |
| "rewards/rejected": -0.4399529993534088, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.3622196015370305e-06, | |
| "logits/chosen": 0.2856540381908417, | |
| "logits/rejected": 0.46640047430992126, | |
| "logps/chosen": -581.0446166992188, | |
| "logps/rejected": -699.5716552734375, | |
| "loss": 0.2034, | |
| "rewards/accuracies": 0.7562500238418579, | |
| "rewards/chosen": -0.22450003027915955, | |
| "rewards/margins": 0.187502920627594, | |
| "rewards/rejected": -0.41200295090675354, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.298780907110648e-06, | |
| "logits/chosen": 0.30591830611228943, | |
| "logits/rejected": 0.4262414574623108, | |
| "logps/chosen": -547.7359619140625, | |
| "logps/rejected": -688.0494995117188, | |
| "loss": 0.2329, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -0.2417764961719513, | |
| "rewards/margins": 0.18638503551483154, | |
| "rewards/rejected": -0.42816153168678284, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.23284432675381e-06, | |
| "logits/chosen": 0.2821730375289917, | |
| "logits/rejected": 0.4414825439453125, | |
| "logps/chosen": -625.08984375, | |
| "logps/rejected": -736.9857177734375, | |
| "loss": 0.2392, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.29913753271102905, | |
| "rewards/margins": 0.1723407357931137, | |
| "rewards/rejected": -0.4714782238006592, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.164501423622277e-06, | |
| "logits/chosen": 0.3146980404853821, | |
| "logits/rejected": 0.5349381566047668, | |
| "logps/chosen": -667.1527099609375, | |
| "logps/rejected": -809.9267578125, | |
| "loss": 0.2347, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -0.3215456008911133, | |
| "rewards/margins": 0.2190614491701126, | |
| "rewards/rejected": -0.5406070947647095, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.0938471024237355e-06, | |
| "logits/chosen": 0.3459940552711487, | |
| "logits/rejected": 0.5182437300682068, | |
| "logps/chosen": -632.4589233398438, | |
| "logps/rejected": -797.7081298828125, | |
| "loss": 0.1764, | |
| "rewards/accuracies": 0.862500011920929, | |
| "rewards/chosen": -0.2630694508552551, | |
| "rewards/margins": 0.2305426150560379, | |
| "rewards/rejected": -0.49361199140548706, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.020979477627907e-06, | |
| "logits/chosen": 0.3081711232662201, | |
| "logits/rejected": 0.3925454020500183, | |
| "logps/chosen": -592.3121948242188, | |
| "logps/rejected": -695.0133666992188, | |
| "loss": 0.2528, | |
| "rewards/accuracies": 0.706250011920929, | |
| "rewards/chosen": -0.2648671865463257, | |
| "rewards/margins": 0.17098167538642883, | |
| "rewards/rejected": -0.4358488619327545, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 3.9459997372194105e-06, | |
| "logits/chosen": 0.3076136112213135, | |
| "logits/rejected": 0.38813453912734985, | |
| "logps/chosen": -642.5826416015625, | |
| "logps/rejected": -815.0631103515625, | |
| "loss": 0.1942, | |
| "rewards/accuracies": 0.7562500238418579, | |
| "rewards/chosen": -0.31470799446105957, | |
| "rewards/margins": 0.20841769874095917, | |
| "rewards/rejected": -0.5231257677078247, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 3.869012002182573e-06, | |
| "logits/chosen": 0.2800385057926178, | |
| "logits/rejected": 0.45763272047042847, | |
| "logps/chosen": -611.0481567382812, | |
| "logps/rejected": -790.9743041992188, | |
| "loss": 0.1966, | |
| "rewards/accuracies": 0.7562500238418579, | |
| "rewards/chosen": -0.275074303150177, | |
| "rewards/margins": 0.22857530415058136, | |
| "rewards/rejected": -0.5036496520042419, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 3.7901231819133104e-06, | |
| "logits/chosen": 0.2838382422924042, | |
| "logits/rejected": 0.4253064692020416, | |
| "logps/chosen": -663.5225830078125, | |
| "logps/rejected": -790.7160034179688, | |
| "loss": 0.2077, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -0.28215542435646057, | |
| "rewards/margins": 0.2109585702419281, | |
| "rewards/rejected": -0.49311399459838867, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 3.709442825758875e-06, | |
| "logits/chosen": 0.32005783915519714, | |
| "logits/rejected": 0.42395836114883423, | |
| "logps/chosen": -589.5636596679688, | |
| "logps/rejected": -722.5950317382812, | |
| "loss": 0.2375, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -0.24356071650981903, | |
| "rewards/margins": 0.18706892430782318, | |
| "rewards/rejected": -0.4306296408176422, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 3.6270829708916113e-06, | |
| "logits/chosen": 0.318645715713501, | |
| "logits/rejected": 0.4325946271419525, | |
| "logps/chosen": -611.2227783203125, | |
| "logps/rejected": -759.0226440429688, | |
| "loss": 0.2232, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -0.25262197852134705, | |
| "rewards/margins": 0.19418981671333313, | |
| "rewards/rejected": -0.44681182503700256, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 3.543157986727991e-06, | |
| "logits/chosen": 0.27690261602401733, | |
| "logits/rejected": 0.4690285325050354, | |
| "logps/chosen": -600.1328125, | |
| "logps/rejected": -722.5906982421875, | |
| "loss": 0.2403, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -0.2763115167617798, | |
| "rewards/margins": 0.16976875066757202, | |
| "rewards/rejected": -0.4460802674293518, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 3.4577844161089614e-06, | |
| "logits/chosen": 0.2978890836238861, | |
| "logits/rejected": 0.5168454647064209, | |
| "logps/chosen": -631.8878173828125, | |
| "logps/rejected": -754.962646484375, | |
| "loss": 0.2448, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -0.3049059808254242, | |
| "rewards/margins": 0.16670657694339752, | |
| "rewards/rejected": -0.4716125428676605, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 3.3710808134621577e-06, | |
| "logits/chosen": 0.31090012192726135, | |
| "logits/rejected": 0.4086151123046875, | |
| "logps/chosen": -639.5802612304688, | |
| "logps/rejected": -715.4155883789062, | |
| "loss": 0.2345, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.29435452818870544, | |
| "rewards/margins": 0.1740979552268982, | |
| "rewards/rejected": -0.46845245361328125, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 3.2831675801707126e-06, | |
| "logits/chosen": 0.29356813430786133, | |
| "logits/rejected": 0.45901551842689514, | |
| "logps/chosen": -648.76806640625, | |
| "logps/rejected": -793.9247436523438, | |
| "loss": 0.195, | |
| "rewards/accuracies": 0.7437499761581421, | |
| "rewards/chosen": -0.30719709396362305, | |
| "rewards/margins": 0.2049371302127838, | |
| "rewards/rejected": -0.5121342539787292, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 3.194166797377289e-06, | |
| "logits/chosen": 0.35314708948135376, | |
| "logits/rejected": 0.4861513078212738, | |
| "logps/chosen": -669.8665771484375, | |
| "logps/rejected": -808.8753051757812, | |
| "loss": 0.2386, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.34487032890319824, | |
| "rewards/margins": 0.20408621430397034, | |
| "rewards/rejected": -0.548956573009491, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 3.104202056455501e-06, | |
| "logits/chosen": 0.3531484007835388, | |
| "logits/rejected": 0.5174172520637512, | |
| "logps/chosen": -657.7909545898438, | |
| "logps/rejected": -806.8094482421875, | |
| "loss": 0.2039, | |
| "rewards/accuracies": 0.78125, | |
| "rewards/chosen": -0.3043491244316101, | |
| "rewards/margins": 0.21579626202583313, | |
| "rewards/rejected": -0.5201453566551208, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 3.013398287384144e-06, | |
| "logits/chosen": 0.29440468549728394, | |
| "logits/rejected": 0.4629596769809723, | |
| "logps/chosen": -613.6539916992188, | |
| "logps/rejected": -718.9657592773438, | |
| "loss": 0.2514, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.28395384550094604, | |
| "rewards/margins": 0.17435352504253387, | |
| "rewards/rejected": -0.4583073556423187, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.9218815852625717e-06, | |
| "logits/chosen": 0.26766303181648254, | |
| "logits/rejected": 0.4210576117038727, | |
| "logps/chosen": -611.0421142578125, | |
| "logps/rejected": -752.10546875, | |
| "loss": 0.2478, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -0.2685486972332001, | |
| "rewards/margins": 0.20156879723072052, | |
| "rewards/rejected": -0.4701175093650818, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 2.829779035208113e-06, | |
| "logits/chosen": 0.3156279921531677, | |
| "logits/rejected": 0.4234594702720642, | |
| "logps/chosen": -583.9177856445312, | |
| "logps/rejected": -695.6286010742188, | |
| "loss": 0.2687, | |
| "rewards/accuracies": 0.731249988079071, | |
| "rewards/chosen": -0.2727667987346649, | |
| "rewards/margins": 0.16848285496234894, | |
| "rewards/rejected": -0.44124966859817505, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 2.737218535878705e-06, | |
| "logits/chosen": 0.2484460324048996, | |
| "logits/rejected": 0.3799341320991516, | |
| "logps/chosen": -582.5774536132812, | |
| "logps/rejected": -692.3199462890625, | |
| "loss": 0.2314, | |
| "rewards/accuracies": 0.731249988079071, | |
| "rewards/chosen": -0.23775920271873474, | |
| "rewards/margins": 0.1904359757900238, | |
| "rewards/rejected": -0.42819517850875854, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 2.64432862186579e-06, | |
| "logits/chosen": 0.31521672010421753, | |
| "logits/rejected": 0.4720017910003662, | |
| "logps/chosen": -617.0424194335938, | |
| "logps/rejected": -763.6742553710938, | |
| "loss": 0.2106, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": -0.2587849795818329, | |
| "rewards/margins": 0.21287491917610168, | |
| "rewards/rejected": -0.4716598391532898, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 2.551238285204126e-06, | |
| "logits/chosen": 0.23266558349132538, | |
| "logits/rejected": 0.4128551483154297, | |
| "logps/chosen": -645.7574462890625, | |
| "logps/rejected": -775.3682861328125, | |
| "loss": 0.2031, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -0.26859793066978455, | |
| "rewards/margins": 0.21920350193977356, | |
| "rewards/rejected": -0.4878014624118805, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 2.4580767962463688e-06, | |
| "logits/chosen": 0.2809138298034668, | |
| "logits/rejected": 0.36528998613357544, | |
| "logps/chosen": -635.5758666992188, | |
| "logps/rejected": -790.5833129882812, | |
| "loss": 0.2095, | |
| "rewards/accuracies": 0.768750011920929, | |
| "rewards/chosen": -0.29754263162612915, | |
| "rewards/margins": 0.1984301209449768, | |
| "rewards/rejected": -0.49597278237342834, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 2.3649735241511546e-06, | |
| "logits/chosen": 0.27885499596595764, | |
| "logits/rejected": 0.4202384352684021, | |
| "logps/chosen": -633.392333984375, | |
| "logps/rejected": -738.7381591796875, | |
| "loss": 0.2417, | |
| "rewards/accuracies": 0.6812499761581421, | |
| "rewards/chosen": -0.296310693025589, | |
| "rewards/margins": 0.18409204483032227, | |
| "rewards/rejected": -0.48040276765823364, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 2.2720577572339914e-06, | |
| "logits/chosen": 0.2572694420814514, | |
| "logits/rejected": 0.3829544484615326, | |
| "logps/chosen": -661.2225341796875, | |
| "logps/rejected": -775.2647094726562, | |
| "loss": 0.2215, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.3194531500339508, | |
| "rewards/margins": 0.1850832998752594, | |
| "rewards/rejected": -0.5045364499092102, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.1794585234303995e-06, | |
| "logits/chosen": 0.22785460948944092, | |
| "logits/rejected": 0.39559316635131836, | |
| "logps/chosen": -616.7025146484375, | |
| "logps/rejected": -784.1016235351562, | |
| "loss": 0.2179, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.28892993927001953, | |
| "rewards/margins": 0.21685433387756348, | |
| "rewards/rejected": -0.505784273147583, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 2.0873044111206407e-06, | |
| "logits/chosen": 0.24686650931835175, | |
| "logits/rejected": 0.39195549488067627, | |
| "logps/chosen": -679.55859375, | |
| "logps/rejected": -784.8621215820312, | |
| "loss": 0.2101, | |
| "rewards/accuracies": 0.768750011920929, | |
| "rewards/chosen": -0.3024412989616394, | |
| "rewards/margins": 0.1977556049823761, | |
| "rewards/rejected": -0.5001968741416931, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.9957233905648293e-06, | |
| "logits/chosen": 0.2292439192533493, | |
| "logits/rejected": 0.4420854151248932, | |
| "logps/chosen": -635.6842041015625, | |
| "logps/rejected": -802.7564697265625, | |
| "loss": 0.2054, | |
| "rewards/accuracies": 0.7437499761581421, | |
| "rewards/chosen": -0.2987775206565857, | |
| "rewards/margins": 0.22222964465618134, | |
| "rewards/rejected": -0.5210071206092834, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.904842636196402e-06, | |
| "logits/chosen": 0.24736735224723816, | |
| "logits/rejected": 0.4444567561149597, | |
| "logps/chosen": -658.0628662109375, | |
| "logps/rejected": -821.857421875, | |
| "loss": 0.2104, | |
| "rewards/accuracies": 0.768750011920929, | |
| "rewards/chosen": -0.3125050663948059, | |
| "rewards/margins": 0.21667668223381042, | |
| "rewards/rejected": -0.5291817784309387, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.814788350020726e-06, | |
| "logits/chosen": 0.18406185507774353, | |
| "logits/rejected": 0.3239056468009949, | |
| "logps/chosen": -618.2083129882812, | |
| "logps/rejected": -770.7822875976562, | |
| "loss": 0.2304, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -0.2990264892578125, | |
| "rewards/margins": 0.19497379660606384, | |
| "rewards/rejected": -0.49400028586387634, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.725685586364051e-06, | |
| "logits/chosen": 0.23465320467948914, | |
| "logits/rejected": 0.41701340675354004, | |
| "logps/chosen": -603.99609375, | |
| "logps/rejected": -717.4134521484375, | |
| "loss": 0.2277, | |
| "rewards/accuracies": 0.6937500238418579, | |
| "rewards/chosen": -0.274019718170166, | |
| "rewards/margins": 0.18845617771148682, | |
| "rewards/rejected": -0.4624759256839752, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.6376580782162172e-06, | |
| "logits/chosen": 0.2198990285396576, | |
| "logits/rejected": 0.3981742560863495, | |
| "logps/chosen": -660.04931640625, | |
| "logps/rejected": -776.05322265625, | |
| "loss": 0.2166, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.275198370218277, | |
| "rewards/margins": 0.21421167254447937, | |
| "rewards/rejected": -0.4894101023674011, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.550828065408227e-06, | |
| "logits/chosen": 0.21209344267845154, | |
| "logits/rejected": 0.35057374835014343, | |
| "logps/chosen": -608.8699951171875, | |
| "logps/rejected": -748.3526611328125, | |
| "loss": 0.2356, | |
| "rewards/accuracies": 0.706250011920929, | |
| "rewards/chosen": -0.30509600043296814, | |
| "rewards/margins": 0.183022603392601, | |
| "rewards/rejected": -0.48811858892440796, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.4653161248633053e-06, | |
| "logits/chosen": 0.1554277241230011, | |
| "logits/rejected": 0.38564786314964294, | |
| "logps/chosen": -655.8323974609375, | |
| "logps/rejected": -804.4343872070312, | |
| "loss": 0.2157, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -0.29179584980010986, | |
| "rewards/margins": 0.22734172642230988, | |
| "rewards/rejected": -0.5191375613212585, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.381241003157162e-06, | |
| "logits/chosen": 0.21835489571094513, | |
| "logits/rejected": 0.365634948015213, | |
| "logps/chosen": -661.9124145507812, | |
| "logps/rejected": -783.0175170898438, | |
| "loss": 0.1961, | |
| "rewards/accuracies": 0.7437499761581421, | |
| "rewards/chosen": -0.3171085715293884, | |
| "rewards/margins": 0.20071883499622345, | |
| "rewards/rejected": -0.5178273916244507, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.298719451619979e-06, | |
| "logits/chosen": 0.23657183349132538, | |
| "logits/rejected": 0.3743918836116791, | |
| "logps/chosen": -680.4862670898438, | |
| "logps/rejected": -835.0526123046875, | |
| "loss": 0.2212, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -0.3263990879058838, | |
| "rewards/margins": 0.22416400909423828, | |
| "rewards/rejected": -0.5505631566047668, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.2178660642091036e-06, | |
| "logits/chosen": 0.2182171642780304, | |
| "logits/rejected": 0.32706716656684875, | |
| "logps/chosen": -675.8248901367188, | |
| "logps/rejected": -836.1906127929688, | |
| "loss": 0.2204, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.3406047821044922, | |
| "rewards/margins": 0.20144791901111603, | |
| "rewards/rejected": -0.5420526266098022, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.1387931183775821e-06, | |
| "logits/chosen": 0.21965551376342773, | |
| "logits/rejected": 0.377861887216568, | |
| "logps/chosen": -645.6212158203125, | |
| "logps/rejected": -850.5048828125, | |
| "loss": 0.2011, | |
| "rewards/accuracies": 0.7562500238418579, | |
| "rewards/chosen": -0.32531431317329407, | |
| "rewards/margins": 0.23981444537639618, | |
| "rewards/rejected": -0.565128743648529, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.061610419159532e-06, | |
| "logits/chosen": 0.23912313580513, | |
| "logits/rejected": 0.33457493782043457, | |
| "logps/chosen": -652.5158081054688, | |
| "logps/rejected": -804.5699462890625, | |
| "loss": 0.2, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.3123088479042053, | |
| "rewards/margins": 0.22432692348957062, | |
| "rewards/rejected": -0.5366358160972595, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 9.864251466888364e-07, | |
| "logits/chosen": 0.22238676249980927, | |
| "logits/rejected": 0.4501045346260071, | |
| "logps/chosen": -702.1354370117188, | |
| "logps/rejected": -854.2493896484375, | |
| "loss": 0.2429, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -0.3496148884296417, | |
| "rewards/margins": 0.21850232779979706, | |
| "rewards/rejected": -0.5681171417236328, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 9.133417073629288e-07, | |
| "logits/chosen": 0.21312983334064484, | |
| "logits/rejected": 0.3885112702846527, | |
| "logps/chosen": -668.33447265625, | |
| "logps/rejected": -807.5436401367188, | |
| "loss": 0.236, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.3556881546974182, | |
| "rewards/margins": 0.1771022081375122, | |
| "rewards/rejected": -0.5327903628349304, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 8.424615888583332e-07, | |
| "logits/chosen": 0.17015187442302704, | |
| "logits/rejected": 0.3787848949432373, | |
| "logps/chosen": -694.0350341796875, | |
| "logps/rejected": -877.3869018554688, | |
| "loss": 0.1718, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -0.32832252979278564, | |
| "rewards/margins": 0.2480977475643158, | |
| "rewards/rejected": -0.5764203071594238, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 7.738832191993092e-07, | |
| "logits/chosen": 0.18861600756645203, | |
| "logits/rejected": 0.3507705330848694, | |
| "logps/chosen": -677.43701171875, | |
| "logps/rejected": -824.4730224609375, | |
| "loss": 0.1952, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": -0.3222949504852295, | |
| "rewards/margins": 0.23219437897205353, | |
| "rewards/rejected": -0.5544893145561218, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 7.077018300752917e-07, | |
| "logits/chosen": 0.22210793197155, | |
| "logits/rejected": 0.3462095856666565, | |
| "logps/chosen": -642.8484497070312, | |
| "logps/rejected": -801.2487182617188, | |
| "loss": 0.2144, | |
| "rewards/accuracies": 0.7562500238418579, | |
| "rewards/chosen": -0.2975847125053406, | |
| "rewards/margins": 0.23604163527488708, | |
| "rewards/rejected": -0.53362637758255, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 6.440093245969342e-07, | |
| "logits/chosen": 0.18037883937358856, | |
| "logits/rejected": 0.32701772451400757, | |
| "logps/chosen": -619.3107299804688, | |
| "logps/rejected": -796.5120849609375, | |
| "loss": 0.2121, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.2894679307937622, | |
| "rewards/margins": 0.22074396908283234, | |
| "rewards/rejected": -0.5102118849754333, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 5.828941496744075e-07, | |
| "logits/chosen": 0.21488244831562042, | |
| "logits/rejected": 0.3539666533470154, | |
| "logps/chosen": -646.1163330078125, | |
| "logps/rejected": -821.9992065429688, | |
| "loss": 0.1931, | |
| "rewards/accuracies": 0.78125, | |
| "rewards/chosen": -0.30917179584503174, | |
| "rewards/margins": 0.22842054069042206, | |
| "rewards/rejected": -0.537592351436615, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 5.244411731951671e-07, | |
| "logits/chosen": 0.2123112678527832, | |
| "logits/rejected": 0.38105306029319763, | |
| "logps/chosen": -718.8529663085938, | |
| "logps/rejected": -841.3015747070312, | |
| "loss": 0.2265, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": -0.3423205316066742, | |
| "rewards/margins": 0.21526296436786652, | |
| "rewards/rejected": -0.5575835108757019, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 4.6873156617173594e-07, | |
| "logits/chosen": 0.2072085589170456, | |
| "logits/rejected": 0.33443737030029297, | |
| "logps/chosen": -687.7586669921875, | |
| "logps/rejected": -815.7947998046875, | |
| "loss": 0.2148, | |
| "rewards/accuracies": 0.731249988079071, | |
| "rewards/chosen": -0.3153381943702698, | |
| "rewards/margins": 0.20979492366313934, | |
| "rewards/rejected": -0.5251331329345703, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 4.1584269002318653e-07, | |
| "logits/chosen": 0.22315290570259094, | |
| "logits/rejected": 0.39601173996925354, | |
| "logps/chosen": -641.2395629882812, | |
| "logps/rejected": -768.5919799804688, | |
| "loss": 0.2332, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -0.2931545674800873, | |
| "rewards/margins": 0.20355018973350525, | |
| "rewards/rejected": -0.49670475721359253, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 3.658479891468258e-07, | |
| "logits/chosen": 0.23425737023353577, | |
| "logits/rejected": 0.3700588047504425, | |
| "logps/chosen": -647.1505737304688, | |
| "logps/rejected": -796.8721313476562, | |
| "loss": 0.2232, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -0.3021949827671051, | |
| "rewards/margins": 0.2227589339017868, | |
| "rewards/rejected": -0.5249539017677307, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 3.18816888929272e-07, | |
| "logits/chosen": 0.21941573917865753, | |
| "logits/rejected": 0.38576945662498474, | |
| "logps/chosen": -681.2139282226562, | |
| "logps/rejected": -785.1143188476562, | |
| "loss": 0.2456, | |
| "rewards/accuracies": 0.65625, | |
| "rewards/chosen": -0.3359334468841553, | |
| "rewards/margins": 0.15878985822200775, | |
| "rewards/rejected": -0.4947233200073242, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 2.748146993385484e-07, | |
| "logits/chosen": 0.21103620529174805, | |
| "logits/rejected": 0.37812119722366333, | |
| "logps/chosen": -675.7420654296875, | |
| "logps/rejected": -835.7942504882812, | |
| "loss": 0.2342, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -0.3283590078353882, | |
| "rewards/margins": 0.1955871880054474, | |
| "rewards/rejected": -0.5239461660385132, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 2.3390252423108077e-07, | |
| "logits/chosen": 0.18820640444755554, | |
| "logits/rejected": 0.3453311324119568, | |
| "logps/chosen": -699.3671264648438, | |
| "logps/rejected": -808.0126342773438, | |
| "loss": 0.2185, | |
| "rewards/accuracies": 0.7437499761581421, | |
| "rewards/chosen": -0.32744988799095154, | |
| "rewards/margins": 0.19639787077903748, | |
| "rewards/rejected": -0.523847758769989, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.961371764995243e-07, | |
| "logits/chosen": 0.1902572214603424, | |
| "logits/rejected": 0.3874988257884979, | |
| "logps/chosen": -704.2755737304688, | |
| "logps/rejected": -798.5086669921875, | |
| "loss": 0.2272, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.3353065848350525, | |
| "rewards/margins": 0.1846623420715332, | |
| "rewards/rejected": -0.5199689269065857, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.61571099179261e-07, | |
| "logits/chosen": 0.2346924990415573, | |
| "logits/rejected": 0.35877925157546997, | |
| "logps/chosen": -631.2687377929688, | |
| "logps/rejected": -800.9693603515625, | |
| "loss": 0.2343, | |
| "rewards/accuracies": 0.7562500238418579, | |
| "rewards/chosen": -0.31699609756469727, | |
| "rewards/margins": 0.20049229264259338, | |
| "rewards/rejected": -0.517488420009613, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.3025229262312367e-07, | |
| "logits/chosen": 0.22632627189159393, | |
| "logits/rejected": 0.4042627811431885, | |
| "logps/chosen": -647.9192504882812, | |
| "logps/rejected": -805.1476440429688, | |
| "loss": 0.1767, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": -0.29563456773757935, | |
| "rewards/margins": 0.23449036478996277, | |
| "rewards/rejected": -0.5301249623298645, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.0222424784546853e-07, | |
| "logits/chosen": 0.18376657366752625, | |
| "logits/rejected": 0.2582804560661316, | |
| "logps/chosen": -614.8178100585938, | |
| "logps/rejected": -754.7501220703125, | |
| "loss": 0.2084, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.27827420830726624, | |
| "rewards/margins": 0.21241740882396698, | |
| "rewards/rejected": -0.49069157242774963, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 7.752588612816553e-08, | |
| "logits/chosen": 0.2309151142835617, | |
| "logits/rejected": 0.32766053080558777, | |
| "logps/chosen": -618.4566650390625, | |
| "logps/rejected": -769.9896850585938, | |
| "loss": 0.2188, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.2959611415863037, | |
| "rewards/margins": 0.19887039065361023, | |
| "rewards/rejected": -0.49483156204223633, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 5.619150497236991e-08, | |
| "logits/chosen": 0.24140353500843048, | |
| "logits/rejected": 0.29053181409835815, | |
| "logps/chosen": -647.7692260742188, | |
| "logps/rejected": -806.162109375, | |
| "loss": 0.2179, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.31588608026504517, | |
| "rewards/margins": 0.21005694568157196, | |
| "rewards/rejected": -0.5259430408477783, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 3.825073047112743e-08, | |
| "logits/chosen": 0.21158334612846375, | |
| "logits/rejected": 0.31529536843299866, | |
| "logps/chosen": -656.6387939453125, | |
| "logps/rejected": -829.8762817382812, | |
| "loss": 0.2066, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -0.31754207611083984, | |
| "rewards/margins": 0.2269022911787033, | |
| "rewards/rejected": -0.5444443821907043, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 2.372847616895685e-08, | |
| "logits/chosen": 0.18535788357257843, | |
| "logits/rejected": 0.3380703330039978, | |
| "logps/chosen": -669.4324340820312, | |
| "logps/rejected": -809.5096435546875, | |
| "loss": 0.2213, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -0.33157414197921753, | |
| "rewards/margins": 0.20540854334831238, | |
| "rewards/rejected": -0.5369827151298523, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.264490846553279e-08, | |
| "logits/chosen": 0.21935626864433289, | |
| "logits/rejected": 0.3299446702003479, | |
| "logps/chosen": -643.43310546875, | |
| "logps/rejected": -753.556884765625, | |
| "loss": 0.2527, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.3096220791339874, | |
| "rewards/margins": 0.17697684466838837, | |
| "rewards/rejected": -0.4865988790988922, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 5.015418611516165e-09, | |
| "logits/chosen": 0.20347245037555695, | |
| "logits/rejected": 0.37708914279937744, | |
| "logps/chosen": -650.4703979492188, | |
| "logps/rejected": -769.6436767578125, | |
| "loss": 0.2711, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": -0.3225279450416565, | |
| "rewards/margins": 0.16174957156181335, | |
| "rewards/rejected": -0.48427754640579224, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 8.506013354186993e-10, | |
| "logits/chosen": 0.22470831871032715, | |
| "logits/rejected": 0.3196450173854828, | |
| "logps/chosen": -647.2747192382812, | |
| "logps/rejected": -821.5018310546875, | |
| "loss": 0.2047, | |
| "rewards/accuracies": 0.7562500238418579, | |
| "rewards/chosen": -0.3270157277584076, | |
| "rewards/margins": 0.21042361855506897, | |
| "rewards/rejected": -0.5374393463134766, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 937, | |
| "total_flos": 0.0, | |
| "train_loss": 0.2409534176996449, | |
| "train_runtime": 8445.1148, | |
| "train_samples_per_second": 1.776, | |
| "train_steps_per_second": 0.111 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 937, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 100, | |
| "total_flos": 0.0, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |