{ "best_metric": null, "best_model_checkpoint": null, "epoch": 1.0, "eval_steps": 50, "global_step": 436, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.022935779816513763, "grad_norm": 6.666838198680302, "learning_rate": 1.1363636363636363e-07, "logits/chosen": -2.6193127632141113, "logits/rejected": -2.5524516105651855, "logps/chosen": -265.4483337402344, "logps/rejected": -236.09951782226562, "loss": 0.6931, "rewards/accuracies": 0.35624998807907104, "rewards/chosen": -2.520665293559432e-05, "rewards/margins": -0.0003298574883956462, "rewards/rejected": 0.00030465080635622144, "step": 10 }, { "epoch": 0.045871559633027525, "grad_norm": 6.256434066153107, "learning_rate": 2.2727272727272726e-07, "logits/chosen": -2.6579253673553467, "logits/rejected": -2.5760178565979004, "logps/chosen": -298.82305908203125, "logps/rejected": -274.31829833984375, "loss": 0.6922, "rewards/accuracies": 0.612500011920929, "rewards/chosen": -0.0003779557882808149, "rewards/margins": 0.0017056630458682775, "rewards/rejected": -0.0020836188923567533, "step": 20 }, { "epoch": 0.06880733944954129, "grad_norm": 6.014004575787888, "learning_rate": 3.4090909090909085e-07, "logits/chosen": -2.675771713256836, "logits/rejected": -2.6020450592041016, "logps/chosen": -290.2829895019531, "logps/rejected": -234.431396484375, "loss": 0.6883, "rewards/accuracies": 0.6937500238418579, "rewards/chosen": 0.01049796212464571, "rewards/margins": 0.01499337237328291, "rewards/rejected": -0.004495412111282349, "step": 30 }, { "epoch": 0.09174311926605505, "grad_norm": 6.545633202821474, "learning_rate": 4.545454545454545e-07, "logits/chosen": -2.6612305641174316, "logits/rejected": -2.6114158630371094, "logps/chosen": -280.93109130859375, "logps/rejected": -267.76605224609375, "loss": 0.6776, "rewards/accuracies": 0.668749988079071, "rewards/chosen": 0.04238061234354973, "rewards/margins": 0.04211103171110153, "rewards/rejected": 0.00026957987574860454, "step": 40 }, { "epoch": 0.11467889908256881, "grad_norm": 7.0518705334996215, "learning_rate": 4.997110275491701e-07, "logits/chosen": -2.6194958686828613, "logits/rejected": -2.6111502647399902, "logps/chosen": -294.24957275390625, "logps/rejected": -305.2260437011719, "loss": 0.6607, "rewards/accuracies": 0.706250011920929, "rewards/chosen": 0.020016254857182503, "rewards/margins": 0.08043506741523743, "rewards/rejected": -0.060418806970119476, "step": 50 }, { "epoch": 0.11467889908256881, "eval_logits/chosen": -2.5735373497009277, "eval_logits/rejected": -2.4922757148742676, "eval_logps/chosen": -285.52752685546875, "eval_logps/rejected": -261.0398254394531, "eval_loss": 0.6447442770004272, "eval_rewards/accuracies": 0.6896551847457886, "eval_rewards/chosen": -0.004374094773083925, "eval_rewards/margins": 0.140799880027771, "eval_rewards/rejected": -0.14517399668693542, "eval_runtime": 94.492, "eval_samples_per_second": 19.24, "eval_steps_per_second": 0.307, "step": 50 }, { "epoch": 0.13761467889908258, "grad_norm": 10.342302388199876, "learning_rate": 4.979475034558115e-07, "logits/chosen": -2.559204578399658, "logits/rejected": -2.5000882148742676, "logps/chosen": -296.8924255371094, "logps/rejected": -279.6778869628906, "loss": 0.6362, "rewards/accuracies": 0.668749988079071, "rewards/chosen": -0.07407740503549576, "rewards/margins": 0.14713598787784576, "rewards/rejected": -0.22121338546276093, "step": 60 }, { "epoch": 0.16055045871559634, "grad_norm": 11.57375336560156, "learning_rate": 4.945923025551788e-07, "logits/chosen": -2.439746141433716, "logits/rejected": -2.39266037940979, "logps/chosen": -341.58990478515625, "logps/rejected": -305.513427734375, "loss": 0.614, "rewards/accuracies": 0.706250011920929, "rewards/chosen": -0.14018920063972473, "rewards/margins": 0.3008686304092407, "rewards/rejected": -0.44105786085128784, "step": 70 }, { "epoch": 0.1834862385321101, "grad_norm": 13.101676053286978, "learning_rate": 4.896669632591651e-07, "logits/chosen": -2.333707332611084, "logits/rejected": -2.2206273078918457, "logps/chosen": -311.94189453125, "logps/rejected": -312.6253662109375, "loss": 0.5993, "rewards/accuracies": 0.7250000238418579, "rewards/chosen": -0.2609206438064575, "rewards/margins": 0.33877214789390564, "rewards/rejected": -0.5996928215026855, "step": 80 }, { "epoch": 0.20642201834862386, "grad_norm": 18.464805323764754, "learning_rate": 4.832031033425662e-07, "logits/chosen": -1.2291042804718018, "logits/rejected": -1.0986605882644653, "logps/chosen": -349.7447814941406, "logps/rejected": -355.8329162597656, "loss": 0.5841, "rewards/accuracies": 0.6812499761581421, "rewards/chosen": -0.6215134859085083, "rewards/margins": 0.4688030779361725, "rewards/rejected": -1.0903165340423584, "step": 90 }, { "epoch": 0.22935779816513763, "grad_norm": 18.44194365148455, "learning_rate": 4.752422169756047e-07, "logits/chosen": -0.5803453326225281, "logits/rejected": -0.4585650861263275, "logps/chosen": -330.8556823730469, "logps/rejected": -383.45281982421875, "loss": 0.5616, "rewards/accuracies": 0.706250011920929, "rewards/chosen": -0.5843100547790527, "rewards/margins": 0.5309332609176636, "rewards/rejected": -1.1152434349060059, "step": 100 }, { "epoch": 0.22935779816513763, "eval_logits/chosen": -0.241001695394516, "eval_logits/rejected": 0.20754042267799377, "eval_logps/chosen": -370.36248779296875, "eval_logps/rejected": -404.2408142089844, "eval_loss": 0.5463992357254028, "eval_rewards/accuracies": 0.6853448152542114, "eval_rewards/chosen": -0.8527240753173828, "eval_rewards/margins": 0.7244595289230347, "eval_rewards/rejected": -1.5771836042404175, "eval_runtime": 94.4096, "eval_samples_per_second": 19.257, "eval_steps_per_second": 0.307, "step": 100 }, { "epoch": 0.25229357798165136, "grad_norm": 20.029871948259476, "learning_rate": 4.658354083558188e-07, "logits/chosen": -0.10621030628681183, "logits/rejected": 0.371669203042984, "logps/chosen": -369.4322204589844, "logps/rejected": -397.5771179199219, "loss": 0.5451, "rewards/accuracies": 0.71875, "rewards/chosen": -1.0424408912658691, "rewards/margins": 0.5959984660148621, "rewards/rejected": -1.638439416885376, "step": 110 }, { "epoch": 0.27522935779816515, "grad_norm": 25.325536766276052, "learning_rate": 4.550430636492389e-07, "logits/chosen": 0.43674445152282715, "logits/rejected": 0.7346484065055847, "logps/chosen": -387.1654968261719, "logps/rejected": -426.51898193359375, "loss": 0.5604, "rewards/accuracies": 0.6875, "rewards/chosen": -1.1134614944458008, "rewards/margins": 0.6370891332626343, "rewards/rejected": -1.7505505084991455, "step": 120 }, { "epoch": 0.2981651376146789, "grad_norm": 25.1538794496433, "learning_rate": 4.429344633468004e-07, "logits/chosen": 0.7902742624282837, "logits/rejected": 1.2213919162750244, "logps/chosen": -351.0482482910156, "logps/rejected": -406.3229675292969, "loss": 0.5509, "rewards/accuracies": 0.699999988079071, "rewards/chosen": -0.9486812353134155, "rewards/margins": 0.6884225606918335, "rewards/rejected": -1.6371036767959595, "step": 130 }, { "epoch": 0.3211009174311927, "grad_norm": 21.483029077159443, "learning_rate": 4.2958733752443187e-07, "logits/chosen": 0.7886204123497009, "logits/rejected": 1.4716787338256836, "logps/chosen": -352.66986083984375, "logps/rejected": -371.2418212890625, "loss": 0.5467, "rewards/accuracies": 0.731249988079071, "rewards/chosen": -0.764846920967102, "rewards/margins": 0.704605758190155, "rewards/rejected": -1.4694526195526123, "step": 140 }, { "epoch": 0.3440366972477064, "grad_norm": 21.063627942952767, "learning_rate": 4.150873668617898e-07, "logits/chosen": 1.314632534980774, "logits/rejected": 2.00125789642334, "logps/chosen": -385.40447998046875, "logps/rejected": -437.3175354003906, "loss": 0.5333, "rewards/accuracies": 0.71875, "rewards/chosen": -1.1886088848114014, "rewards/margins": 0.8526908755302429, "rewards/rejected": -2.04129958152771, "step": 150 }, { "epoch": 0.3440366972477064, "eval_logits/chosen": 0.5890395045280457, "eval_logits/rejected": 1.4808454513549805, "eval_logps/chosen": -385.3254699707031, "eval_logps/rejected": -434.7274169921875, "eval_loss": 0.5195103883743286, "eval_rewards/accuracies": 0.7112069129943848, "eval_rewards/chosen": -1.0023537874221802, "eval_rewards/margins": 0.8796959519386292, "eval_rewards/rejected": -1.8820499181747437, "eval_runtime": 94.3575, "eval_samples_per_second": 19.267, "eval_steps_per_second": 0.307, "step": 150 }, { "epoch": 0.3669724770642202, "grad_norm": 27.196294375049977, "learning_rate": 3.9952763262280397e-07, "logits/chosen": 0.8305681943893433, "logits/rejected": 1.4138782024383545, "logps/chosen": -414.48675537109375, "logps/rejected": -488.97027587890625, "loss": 0.5169, "rewards/accuracies": 0.7124999761581421, "rewards/chosen": -1.179356336593628, "rewards/margins": 0.8506600260734558, "rewards/rejected": -2.0300164222717285, "step": 160 }, { "epoch": 0.38990825688073394, "grad_norm": 20.448285631169824, "learning_rate": 3.8300801912883414e-07, "logits/chosen": 1.4580880403518677, "logits/rejected": 1.959507703781128, "logps/chosen": -411.11505126953125, "logps/rejected": -499.7334899902344, "loss": 0.5017, "rewards/accuracies": 0.6937500238418579, "rewards/chosen": -1.395141363143921, "rewards/margins": 0.9015046954154968, "rewards/rejected": -2.2966461181640625, "step": 170 }, { "epoch": 0.41284403669724773, "grad_norm": 19.388105918914484, "learning_rate": 3.6563457256020884e-07, "logits/chosen": 1.2716890573501587, "logits/rejected": 2.1560723781585693, "logps/chosen": -418.70068359375, "logps/rejected": -435.9356994628906, "loss": 0.5267, "rewards/accuracies": 0.731249988079071, "rewards/chosen": -1.1534276008605957, "rewards/margins": 0.8209800720214844, "rewards/rejected": -1.9744075536727905, "step": 180 }, { "epoch": 0.43577981651376146, "grad_norm": 24.752350050203567, "learning_rate": 3.475188202022617e-07, "logits/chosen": 1.6485626697540283, "logits/rejected": 2.1269421577453613, "logps/chosen": -365.346435546875, "logps/rejected": -476.385009765625, "loss": 0.5115, "rewards/accuracies": 0.762499988079071, "rewards/chosen": -1.0723192691802979, "rewards/margins": 1.048086404800415, "rewards/rejected": -2.120405673980713, "step": 190 }, { "epoch": 0.45871559633027525, "grad_norm": 20.591094290068703, "learning_rate": 3.287770545059052e-07, "logits/chosen": 1.790414810180664, "logits/rejected": 2.4998342990875244, "logps/chosen": -400.47991943359375, "logps/rejected": -456.0809020996094, "loss": 0.5219, "rewards/accuracies": 0.71875, "rewards/chosen": -1.221982717514038, "rewards/margins": 0.900266170501709, "rewards/rejected": -2.122248888015747, "step": 200 }, { "epoch": 0.45871559633027525, "eval_logits/chosen": 1.4256442785263062, "eval_logits/rejected": 2.4260053634643555, "eval_logps/chosen": -392.2837829589844, "eval_logps/rejected": -451.93536376953125, "eval_loss": 0.5009589195251465, "eval_rewards/accuracies": 0.732758641242981, "eval_rewards/chosen": -1.0719367265701294, "eval_rewards/margins": 0.9821925759315491, "eval_rewards/rejected": -2.0541296005249023, "eval_runtime": 96.4986, "eval_samples_per_second": 18.84, "eval_steps_per_second": 0.301, "step": 200 }, { "epoch": 0.481651376146789, "grad_norm": 21.25864225686741, "learning_rate": 3.0952958655864954e-07, "logits/chosen": 1.83803391456604, "logits/rejected": 2.399480104446411, "logps/chosen": -394.57037353515625, "logps/rejected": -468.97021484375, "loss": 0.5006, "rewards/accuracies": 0.75, "rewards/chosen": -1.212659478187561, "rewards/margins": 0.9336206316947937, "rewards/rejected": -2.14628005027771, "step": 210 }, { "epoch": 0.5045871559633027, "grad_norm": 24.95309197907959, "learning_rate": 2.898999737583448e-07, "logits/chosen": 1.8371950387954712, "logits/rejected": 2.8440494537353516, "logps/chosen": -448.06109619140625, "logps/rejected": -526.2056884765625, "loss": 0.491, "rewards/accuracies": 0.8062499761581421, "rewards/chosen": -1.2538080215454102, "rewards/margins": 1.158546805381775, "rewards/rejected": -2.4123549461364746, "step": 220 }, { "epoch": 0.5275229357798165, "grad_norm": 24.874258979226397, "learning_rate": 2.7001422664752333e-07, "logits/chosen": 1.71869695186615, "logits/rejected": 2.2633819580078125, "logps/chosen": -376.9873046875, "logps/rejected": -471.9263610839844, "loss": 0.5124, "rewards/accuracies": 0.731249988079071, "rewards/chosen": -1.0904380083084106, "rewards/margins": 0.9726268649101257, "rewards/rejected": -2.0630650520324707, "step": 230 }, { "epoch": 0.5504587155963303, "grad_norm": 18.86072207414666, "learning_rate": 2.5e-07, "logits/chosen": 1.5722936391830444, "logits/rejected": 2.684537410736084, "logps/chosen": -418.7516174316406, "logps/rejected": -480.7166442871094, "loss": 0.5197, "rewards/accuracies": 0.7124999761581421, "rewards/chosen": -1.3121980428695679, "rewards/margins": 0.8814684152603149, "rewards/rejected": -2.193666696548462, "step": 240 }, { "epoch": 0.573394495412844, "grad_norm": 19.046011375519186, "learning_rate": 2.2998577335247667e-07, "logits/chosen": 1.6455481052398682, "logits/rejected": 2.8547589778900146, "logps/chosen": -438.38238525390625, "logps/rejected": -492.9747619628906, "loss": 0.5007, "rewards/accuracies": 0.7562500238418579, "rewards/chosen": -1.3458855152130127, "rewards/margins": 0.9794231653213501, "rewards/rejected": -2.3253085613250732, "step": 250 }, { "epoch": 0.573394495412844, "eval_logits/chosen": 1.4527463912963867, "eval_logits/rejected": 2.673773765563965, "eval_logps/chosen": -408.2994384765625, "eval_logps/rejected": -479.4298095703125, "eval_loss": 0.4917016923427582, "eval_rewards/accuracies": 0.7241379022598267, "eval_rewards/chosen": -1.232093095779419, "eval_rewards/margins": 1.096980333328247, "eval_rewards/rejected": -2.329073429107666, "eval_runtime": 95.5197, "eval_samples_per_second": 19.033, "eval_steps_per_second": 0.304, "step": 250 }, { "epoch": 0.5963302752293578, "grad_norm": 19.810153184401507, "learning_rate": 2.1010002624165524e-07, "logits/chosen": 1.6456598043441772, "logits/rejected": 2.469041347503662, "logps/chosen": -406.33642578125, "logps/rejected": -532.65869140625, "loss": 0.4916, "rewards/accuracies": 0.800000011920929, "rewards/chosen": -1.356218934059143, "rewards/margins": 1.226430892944336, "rewards/rejected": -2.5826497077941895, "step": 260 }, { "epoch": 0.6192660550458715, "grad_norm": 24.810334002988427, "learning_rate": 1.9047041344135043e-07, "logits/chosen": 2.157437801361084, "logits/rejected": 2.676581621170044, "logps/chosen": -414.63427734375, "logps/rejected": -517.6594848632812, "loss": 0.4963, "rewards/accuracies": 0.7562500238418579, "rewards/chosen": -1.4857197999954224, "rewards/margins": 1.172404408454895, "rewards/rejected": -2.6581242084503174, "step": 270 }, { "epoch": 0.6422018348623854, "grad_norm": 22.618579944183196, "learning_rate": 1.7122294549409482e-07, "logits/chosen": 1.9851901531219482, "logits/rejected": 2.9032249450683594, "logps/chosen": -424.8306579589844, "logps/rejected": -531.9037475585938, "loss": 0.5145, "rewards/accuracies": 0.800000011920929, "rewards/chosen": -1.5183179378509521, "rewards/margins": 1.1627616882324219, "rewards/rejected": -2.681079387664795, "step": 280 }, { "epoch": 0.6651376146788991, "grad_norm": 29.728057206179926, "learning_rate": 1.524811797977383e-07, "logits/chosen": 1.9606460332870483, "logits/rejected": 3.0971603393554688, "logps/chosen": -424.40155029296875, "logps/rejected": -510.2555236816406, "loss": 0.4931, "rewards/accuracies": 0.7437499761581421, "rewards/chosen": -1.3667072057724, "rewards/margins": 1.1966798305511475, "rewards/rejected": -2.563387155532837, "step": 290 }, { "epoch": 0.6880733944954128, "grad_norm": 32.56443619783417, "learning_rate": 1.3436542743979125e-07, "logits/chosen": 1.548921823501587, "logits/rejected": 2.2454886436462402, "logps/chosen": -435.3851623535156, "logps/rejected": -480.7173767089844, "loss": 0.5109, "rewards/accuracies": 0.762499988079071, "rewards/chosen": -1.2943581342697144, "rewards/margins": 0.9242169260978699, "rewards/rejected": -2.2185752391815186, "step": 300 }, { "epoch": 0.6880733944954128, "eval_logits/chosen": 1.5762183666229248, "eval_logits/rejected": 2.8883702754974365, "eval_logps/chosen": -418.6533508300781, "eval_logps/rejected": -496.99908447265625, "eval_loss": 0.48780226707458496, "eval_rewards/accuracies": 0.7284482717514038, "eval_rewards/chosen": -1.3356326818466187, "eval_rewards/margins": 1.1691336631774902, "eval_rewards/rejected": -2.5047662258148193, "eval_runtime": 94.5476, "eval_samples_per_second": 19.228, "eval_steps_per_second": 0.307, "step": 300 }, { "epoch": 0.7110091743119266, "grad_norm": 28.8657602827159, "learning_rate": 1.1699198087116588e-07, "logits/chosen": 1.6565430164337158, "logits/rejected": 2.6408467292785645, "logps/chosen": -422.91943359375, "logps/rejected": -511.9986267089844, "loss": 0.5225, "rewards/accuracies": 0.7562500238418579, "rewards/chosen": -1.449542760848999, "rewards/margins": 1.0059467554092407, "rewards/rejected": -2.4554896354675293, "step": 310 }, { "epoch": 0.7339449541284404, "grad_norm": 28.01306501480046, "learning_rate": 1.00472367377196e-07, "logits/chosen": 2.107029914855957, "logits/rejected": 3.2613253593444824, "logps/chosen": -415.86041259765625, "logps/rejected": -515.1480712890625, "loss": 0.4977, "rewards/accuracies": 0.762499988079071, "rewards/chosen": -1.4162843227386475, "rewards/margins": 1.390293836593628, "rewards/rejected": -2.8065781593322754, "step": 320 }, { "epoch": 0.7568807339449541, "grad_norm": 23.410627824071433, "learning_rate": 8.49126331382102e-08, "logits/chosen": 2.485326051712036, "logits/rejected": 3.1795496940612793, "logps/chosen": -442.5235290527344, "logps/rejected": -500.9856872558594, "loss": 0.4944, "rewards/accuracies": 0.6875, "rewards/chosen": -1.68001389503479, "rewards/margins": 0.8588422536849976, "rewards/rejected": -2.538856029510498, "step": 330 }, { "epoch": 0.7798165137614679, "grad_norm": 36.6413379461555, "learning_rate": 7.041266247556812e-08, "logits/chosen": 2.2517406940460205, "logits/rejected": 2.871661901473999, "logps/chosen": -440.378662109375, "logps/rejected": -506.90625, "loss": 0.4867, "rewards/accuracies": 0.793749988079071, "rewards/chosen": -1.5238606929779053, "rewards/margins": 0.9453619718551636, "rewards/rejected": -2.4692225456237793, "step": 340 }, { "epoch": 0.8027522935779816, "grad_norm": 32.4776553295204, "learning_rate": 5.706553665319955e-08, "logits/chosen": 2.2310845851898193, "logits/rejected": 3.158881187438965, "logps/chosen": -426.0604553222656, "logps/rejected": -490.872314453125, "loss": 0.5063, "rewards/accuracies": 0.7437499761581421, "rewards/chosen": -1.5156935453414917, "rewards/margins": 1.016414761543274, "rewards/rejected": -2.5321083068847656, "step": 350 }, { "epoch": 0.8027522935779816, "eval_logits/chosen": 2.1698544025421143, "eval_logits/rejected": 3.3468618392944336, "eval_logps/chosen": -433.7904052734375, "eval_logps/rejected": -514.8548583984375, "eval_loss": 0.4814243018627167, "eval_rewards/accuracies": 0.7370689511299133, "eval_rewards/chosen": -1.487002968788147, "eval_rewards/margins": 1.1963213682174683, "eval_rewards/rejected": -2.683324098587036, "eval_runtime": 95.0017, "eval_samples_per_second": 19.136, "eval_steps_per_second": 0.305, "step": 350 }, { "epoch": 0.8256880733944955, "grad_norm": 22.48584624269231, "learning_rate": 4.4956936350761005e-08, "logits/chosen": 2.1727540493011475, "logits/rejected": 2.8084352016448975, "logps/chosen": -391.9521179199219, "logps/rejected": -506.19482421875, "loss": 0.4802, "rewards/accuracies": 0.7124999761581421, "rewards/chosen": -1.3963208198547363, "rewards/margins": 1.0802565813064575, "rewards/rejected": -2.4765772819519043, "step": 360 }, { "epoch": 0.8486238532110092, "grad_norm": 26.981108661605717, "learning_rate": 3.416459164418123e-08, "logits/chosen": 1.7840194702148438, "logits/rejected": 2.861525535583496, "logps/chosen": -447.0101013183594, "logps/rejected": -527.9534912109375, "loss": 0.4861, "rewards/accuracies": 0.7562500238418579, "rewards/chosen": -1.4742560386657715, "rewards/margins": 1.179532527923584, "rewards/rejected": -2.6537883281707764, "step": 370 }, { "epoch": 0.8715596330275229, "grad_norm": 32.47929591425802, "learning_rate": 2.475778302439524e-08, "logits/chosen": 2.0354526042938232, "logits/rejected": 3.16379976272583, "logps/chosen": -440.89593505859375, "logps/rejected": -530.92919921875, "loss": 0.4928, "rewards/accuracies": 0.762499988079071, "rewards/chosen": -1.4451792240142822, "rewards/margins": 1.287550687789917, "rewards/rejected": -2.732729911804199, "step": 380 }, { "epoch": 0.8944954128440367, "grad_norm": 33.85969586257753, "learning_rate": 1.6796896657433805e-08, "logits/chosen": 2.4787354469299316, "logits/rejected": 3.341116428375244, "logps/chosen": -407.4842834472656, "logps/rejected": -490.75555419921875, "loss": 0.503, "rewards/accuracies": 0.75, "rewards/chosen": -1.5552680492401123, "rewards/margins": 1.0592255592346191, "rewards/rejected": -2.6144936084747314, "step": 390 }, { "epoch": 0.9174311926605505, "grad_norm": 23.099468591696, "learning_rate": 1.0333036740834855e-08, "logits/chosen": 2.7448534965515137, "logits/rejected": 3.38175630569458, "logps/chosen": -370.7669372558594, "logps/rejected": -485.293701171875, "loss": 0.4936, "rewards/accuracies": 0.762499988079071, "rewards/chosen": -1.4620729684829712, "rewards/margins": 1.098707914352417, "rewards/rejected": -2.5607810020446777, "step": 400 }, { "epoch": 0.9174311926605505, "eval_logits/chosen": 2.333301305770874, "eval_logits/rejected": 3.4838266372680664, "eval_logps/chosen": -443.5758361816406, "eval_logps/rejected": -526.9686279296875, "eval_loss": 0.4815457761287689, "eval_rewards/accuracies": 0.732758641242981, "eval_rewards/chosen": -1.5848578214645386, "eval_rewards/margins": 1.2196037769317627, "eval_rewards/rejected": -2.8044614791870117, "eval_runtime": 95.9268, "eval_samples_per_second": 18.952, "eval_steps_per_second": 0.302, "step": 400 }, { "epoch": 0.9403669724770642, "grad_norm": 24.65582545776771, "learning_rate": 5.4076974448211685e-09, "logits/chosen": 2.678440809249878, "logits/rejected": 3.7608656883239746, "logps/chosen": -434.31890869140625, "logps/rejected": -525.4384765625, "loss": 0.4901, "rewards/accuracies": 0.737500011920929, "rewards/chosen": -1.633670449256897, "rewards/margins": 1.2402476072311401, "rewards/rejected": -2.873918056488037, "step": 410 }, { "epoch": 0.963302752293578, "grad_norm": 24.35589696145217, "learning_rate": 2.052496544188487e-09, "logits/chosen": 2.744503974914551, "logits/rejected": 3.5964274406433105, "logps/chosen": -424.82373046875, "logps/rejected": -535.3244018554688, "loss": 0.4891, "rewards/accuracies": 0.768750011920929, "rewards/chosen": -1.6538474559783936, "rewards/margins": 1.2406232357025146, "rewards/rejected": -2.894470691680908, "step": 420 }, { "epoch": 0.9862385321100917, "grad_norm": 24.120958058828865, "learning_rate": 2.889724508297886e-10, "logits/chosen": 2.3434717655181885, "logits/rejected": 3.6963977813720703, "logps/chosen": -459.7005310058594, "logps/rejected": -510.67413330078125, "loss": 0.4816, "rewards/accuracies": 0.731249988079071, "rewards/chosen": -1.5652509927749634, "rewards/margins": 1.1011250019073486, "rewards/rejected": -2.6663756370544434, "step": 430 }, { "epoch": 1.0, "step": 436, "total_flos": 0.0, "train_loss": 0.5380998532706445, "train_runtime": 11735.5354, "train_samples_per_second": 4.751, "train_steps_per_second": 0.037 } ], "logging_steps": 10, "max_steps": 436, "num_input_tokens_seen": 0, "num_train_epochs": 1, "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 }