{ "best_metric": null, "best_model_checkpoint": null, "epoch": 0.9984301412872841, "eval_steps": 100, "global_step": 477, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0020931449502878076, "grad_norm": 0.035678211599588394, "learning_rate": 6.2499999999999995e-06, "logits/chosen": -1.1021082401275635, "logits/rejected": -0.9669773578643799, "logps/chosen": -392.66217041015625, "logps/rejected": -388.5433654785156, "loss": 0.6931, "rewards/accuracies": 0.0, "rewards/chosen": 0.0, "rewards/margins": 0.0, "rewards/rejected": 0.0, "step": 1 }, { "epoch": 0.020931449502878074, "grad_norm": 0.03580914065241814, "learning_rate": 6.25e-05, "logits/chosen": -1.0458121299743652, "logits/rejected": -1.1097700595855713, "logps/chosen": -340.9582214355469, "logps/rejected": -299.7181396484375, "loss": 0.6929, "rewards/accuracies": 0.4444444477558136, "rewards/chosen": 0.0002490293118171394, "rewards/margins": 0.0008229945669882, "rewards/rejected": -0.0005739651969633996, "step": 10 }, { "epoch": 0.04186289900575615, "grad_norm": 0.037438321858644485, "learning_rate": 0.000125, "logits/chosen": -0.9009532928466797, "logits/rejected": -1.0806249380111694, "logps/chosen": -355.1332092285156, "logps/rejected": -306.08709716796875, "loss": 0.6915, "rewards/accuracies": 0.5249999761581421, "rewards/chosen": -0.0018794331699609756, "rewards/margins": 0.002060088561847806, "rewards/rejected": -0.003939521964639425, "step": 20 }, { "epoch": 0.06279434850863422, "grad_norm": 0.05021581053733826, "learning_rate": 0.00018749999999999998, "logits/chosen": -0.9453525543212891, "logits/rejected": -1.0644036531448364, "logps/chosen": -315.98760986328125, "logps/rejected": -268.6424865722656, "loss": 0.6852, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -0.01840055361390114, "rewards/margins": 0.018759142607450485, "rewards/rejected": -0.03715970367193222, "step": 30 }, { "epoch": 0.0837257980115123, "grad_norm": 0.08909210562705994, "learning_rate": 0.00025, "logits/chosen": -1.0098339319229126, "logits/rejected": -1.0776870250701904, "logps/chosen": -317.6771240234375, "logps/rejected": -308.1802673339844, "loss": 0.6584, "rewards/accuracies": 0.625, "rewards/chosen": -0.16074401140213013, "rewards/margins": 0.09125810116529465, "rewards/rejected": -0.25200212001800537, "step": 40 }, { "epoch": 0.10465724751439037, "grad_norm": 0.1175214871764183, "learning_rate": 0.0002999839121261415, "logits/chosen": -1.0516436100006104, "logits/rejected": -1.1754628419876099, "logps/chosen": -393.256103515625, "logps/rejected": -375.66510009765625, "loss": 0.6346, "rewards/accuracies": 0.706250011920929, "rewards/chosen": -0.458698570728302, "rewards/margins": 0.20651093125343323, "rewards/rejected": -0.6652094125747681, "step": 50 }, { "epoch": 0.12558869701726844, "grad_norm": 0.1383933126926422, "learning_rate": 0.00029942119880575817, "logits/chosen": -1.0281826257705688, "logits/rejected": -1.0935122966766357, "logps/chosen": -332.78228759765625, "logps/rejected": -345.4633483886719, "loss": 0.6227, "rewards/accuracies": 0.75, "rewards/chosen": -0.49291595816612244, "rewards/margins": 0.23161530494689941, "rewards/rejected": -0.7245312929153442, "step": 60 }, { "epoch": 0.14652014652014653, "grad_norm": 0.1275259107351303, "learning_rate": 0.00029805753939568687, "logits/chosen": -1.4444881677627563, "logits/rejected": -1.4646410942077637, "logps/chosen": -412.0874938964844, "logps/rejected": -426.487548828125, "loss": 0.6048, "rewards/accuracies": 0.668749988079071, "rewards/chosen": -0.7971259951591492, "rewards/margins": 0.35752782225608826, "rewards/rejected": -1.154653787612915, "step": 70 }, { "epoch": 0.1674515960230246, "grad_norm": 0.17714321613311768, "learning_rate": 0.00029590024355266256, "logits/chosen": -1.1742874383926392, "logits/rejected": -1.4253867864608765, "logps/chosen": -479.925537109375, "logps/rejected": -425.94598388671875, "loss": 0.583, "rewards/accuracies": 0.731249988079071, "rewards/chosen": -1.072460412979126, "rewards/margins": 0.462944358587265, "rewards/rejected": -1.5354046821594238, "step": 80 }, { "epoch": 0.18838304552590268, "grad_norm": 0.2510909140110016, "learning_rate": 0.0002929608750821129, "logits/chosen": -1.7680635452270508, "logits/rejected": -1.7849355936050415, "logps/chosen": -501.5375061035156, "logps/rejected": -569.551513671875, "loss": 0.5604, "rewards/accuracies": 0.706250011920929, "rewards/chosen": -2.1197433471679688, "rewards/margins": 0.7176006436347961, "rewards/rejected": -2.83734393119812, "step": 90 }, { "epoch": 0.20931449502878074, "grad_norm": 0.1974615901708603, "learning_rate": 0.0002892551899524109, "logits/chosen": -1.4390876293182373, "logits/rejected": -1.5163460969924927, "logps/chosen": -541.5751953125, "logps/rejected": -558.0751953125, "loss": 0.5629, "rewards/accuracies": 0.762499988079071, "rewards/chosen": -1.9338293075561523, "rewards/margins": 0.575296938419342, "rewards/rejected": -2.5091261863708496, "step": 100 }, { "epoch": 0.20931449502878074, "eval_logits/chosen": -1.6571975946426392, "eval_logits/rejected": -1.6835596561431885, "eval_logps/chosen": -536.3665771484375, "eval_logps/rejected": -581.395751953125, "eval_loss": 0.5765172243118286, "eval_rewards/accuracies": 0.72265625, "eval_rewards/chosen": -2.2362735271453857, "eval_rewards/margins": 0.5152010917663574, "eval_rewards/rejected": -2.751474618911743, "eval_runtime": 35.9544, "eval_samples_per_second": 55.626, "eval_steps_per_second": 0.89, "step": 100 }, { "epoch": 0.2302459445316588, "grad_norm": 0.20197245478630066, "learning_rate": 0.00028480305183777385, "logits/chosen": -1.446223497390747, "logits/rejected": -1.5271519422531128, "logps/chosen": -602.8939208984375, "logps/rejected": -629.87158203125, "loss": 0.5644, "rewards/accuracies": 0.706250011920929, "rewards/chosen": -2.7250099182128906, "rewards/margins": 0.5918947458267212, "rewards/rejected": -3.3169047832489014, "step": 110 }, { "epoch": 0.25117739403453687, "grad_norm": 0.2141677737236023, "learning_rate": 0.00027962832564252724, "logits/chosen": -1.6662397384643555, "logits/rejected": -1.6905275583267212, "logps/chosen": -632.6597900390625, "logps/rejected": -688.1976318359375, "loss": 0.5477, "rewards/accuracies": 0.7437499761581421, "rewards/chosen": -3.3453495502471924, "rewards/margins": 0.6758114695549011, "rewards/rejected": -4.021161079406738, "step": 120 }, { "epoch": 0.272108843537415, "grad_norm": 0.20801568031311035, "learning_rate": 0.0002737587495774764, "logits/chosen": -1.4505422115325928, "logits/rejected": -1.4989721775054932, "logps/chosen": -658.4010009765625, "logps/rejected": -700.3973388671875, "loss": 0.5112, "rewards/accuracies": 0.7124999761581421, "rewards/chosen": -3.428065776824951, "rewards/margins": 0.6822351217269897, "rewards/rejected": -4.1103010177612305, "step": 130 }, { "epoch": 0.29304029304029305, "grad_norm": 0.24513058364391327, "learning_rate": 0.0002672257864741005, "logits/chosen": -1.4723621606826782, "logits/rejected": -1.5368088483810425, "logps/chosen": -734.0245361328125, "logps/rejected": -741.5623779296875, "loss": 0.5449, "rewards/accuracies": 0.7250000238418579, "rewards/chosen": -3.7274997234344482, "rewards/margins": 0.7957648634910583, "rewards/rejected": -4.5232648849487305, "step": 140 }, { "epoch": 0.3139717425431711, "grad_norm": 0.2503284513950348, "learning_rate": 0.0002600644551335706, "logits/chosen": -1.2388180494308472, "logits/rejected": -1.3000439405441284, "logps/chosen": -629.1339111328125, "logps/rejected": -636.8096923828125, "loss": 0.5479, "rewards/accuracies": 0.699999988079071, "rewards/chosen": -3.0619120597839355, "rewards/margins": 0.5719773173332214, "rewards/rejected": -3.6338894367218018, "step": 150 }, { "epoch": 0.3349031920460492, "grad_norm": 0.18707025051116943, "learning_rate": 0.0002523131426146173, "logits/chosen": -1.1968213319778442, "logits/rejected": -1.3508607149124146, "logps/chosen": -642.1495971679688, "logps/rejected": -680.3194580078125, "loss": 0.5573, "rewards/accuracies": 0.7124999761581421, "rewards/chosen": -3.2288098335266113, "rewards/margins": 0.7885423898696899, "rewards/rejected": -4.017352104187012, "step": 160 }, { "epoch": 0.35583464154892724, "grad_norm": 0.18305683135986328, "learning_rate": 0.00024401339846644537, "logits/chosen": -1.2790342569351196, "logits/rejected": -1.3211687803268433, "logps/chosen": -597.0249633789062, "logps/rejected": -608.4508666992188, "loss": 0.5409, "rewards/accuracies": 0.706250011920929, "rewards/chosen": -3.0255608558654785, "rewards/margins": 0.49673914909362793, "rewards/rejected": -3.5222995281219482, "step": 170 }, { "epoch": 0.37676609105180536, "grad_norm": 0.2515067458152771, "learning_rate": 0.00023520971200967334, "logits/chosen": -1.1840099096298218, "logits/rejected": -1.2452869415283203, "logps/chosen": -684.8968505859375, "logps/rejected": -719.9307861328125, "loss": 0.5217, "rewards/accuracies": 0.75, "rewards/chosen": -3.799819231033325, "rewards/margins": 0.6900444030761719, "rewards/rejected": -4.489863872528076, "step": 180 }, { "epoch": 0.3976975405546834, "grad_norm": 0.21723517775535583, "learning_rate": 0.00022594927385914543, "logits/chosen": -1.2460777759552002, "logits/rejected": -1.2459148168563843, "logps/chosen": -649.2003784179688, "logps/rejected": -708.8577880859375, "loss": 0.5298, "rewards/accuracies": 0.6499999761581421, "rewards/chosen": -3.549588441848755, "rewards/margins": 0.6792875528335571, "rewards/rejected": -4.22887659072876, "step": 190 }, { "epoch": 0.4186289900575615, "grad_norm": 0.26195383071899414, "learning_rate": 0.0002162817229669295, "logits/chosen": -0.8848621249198914, "logits/rejected": -0.8319551348686218, "logps/chosen": -649.4385986328125, "logps/rejected": -711.3386840820312, "loss": 0.5499, "rewards/accuracies": 0.65625, "rewards/chosen": -3.6400089263916016, "rewards/margins": 0.5325465798377991, "rewards/rejected": -4.172555923461914, "step": 200 }, { "epoch": 0.4186289900575615, "eval_logits/chosen": -0.8333719968795776, "eval_logits/rejected": -0.8208783268928528, "eval_logps/chosen": -679.553466796875, "eval_logps/rejected": -753.2846069335938, "eval_loss": 0.5309607982635498, "eval_rewards/accuracies": 0.7734375, "eval_rewards/chosen": -3.6681408882141113, "eval_rewards/margins": 0.8022220730781555, "eval_rewards/rejected": -4.470362663269043, "eval_runtime": 34.0707, "eval_samples_per_second": 58.702, "eval_steps_per_second": 0.939, "step": 200 }, { "epoch": 0.43956043956043955, "grad_norm": 0.19068197906017303, "learning_rate": 0.00020625888054143427, "logits/chosen": -0.8465880155563354, "logits/rejected": -0.9894304275512695, "logps/chosen": -744.5973510742188, "logps/rejected": -742.8700561523438, "loss": 0.5433, "rewards/accuracies": 0.737500011920929, "rewards/chosen": -3.855426788330078, "rewards/margins": 0.6753937005996704, "rewards/rejected": -4.530820369720459, "step": 210 }, { "epoch": 0.4604918890633176, "grad_norm": 0.22149965167045593, "learning_rate": 0.00019593447226892385, "logits/chosen": -0.9477745294570923, "logits/rejected": -0.9140194058418274, "logps/chosen": -622.2196044921875, "logps/rejected": -685.7794189453125, "loss": 0.535, "rewards/accuracies": 0.737500011920929, "rewards/chosen": -3.5976402759552, "rewards/margins": 0.6984301805496216, "rewards/rejected": -4.296070575714111, "step": 220 }, { "epoch": 0.48142333856619574, "grad_norm": 0.26226335763931274, "learning_rate": 0.0001853638403264141, "logits/chosen": -0.8337217569351196, "logits/rejected": -0.8363305330276489, "logps/chosen": -682.59423828125, "logps/rejected": -737.8418579101562, "loss": 0.5457, "rewards/accuracies": 0.7562500238418579, "rewards/chosen": -3.968369722366333, "rewards/margins": 0.8683279156684875, "rewards/rejected": -4.836697578430176, "step": 230 }, { "epoch": 0.5023547880690737, "grad_norm": 0.21782232820987701, "learning_rate": 0.00017460364672965327, "logits/chosen": -0.799470841884613, "logits/rejected": -0.8606585264205933, "logps/chosen": -692.9552001953125, "logps/rejected": -750.71484375, "loss": 0.5022, "rewards/accuracies": 0.7124999761581421, "rewards/chosen": -3.8266777992248535, "rewards/margins": 0.792169451713562, "rewards/rejected": -4.618846893310547, "step": 240 }, { "epoch": 0.5232862375719518, "grad_norm": 0.2562369406223297, "learning_rate": 0.00016371156960634018, "logits/chosen": -1.0665327310562134, "logits/rejected": -1.1944453716278076, "logps/chosen": -768.0906982421875, "logps/rejected": -832.09521484375, "loss": 0.5107, "rewards/accuracies": 0.8125, "rewards/chosen": -4.486889839172363, "rewards/margins": 1.1490213871002197, "rewards/rejected": -5.635911464691162, "step": 250 }, { "epoch": 0.54421768707483, "grad_norm": 0.29225146770477295, "learning_rate": 0.0001527459940226516, "logits/chosen": -1.1020123958587646, "logits/rejected": -1.1401180028915405, "logps/chosen": -767.971435546875, "logps/rejected": -831.6994018554688, "loss": 0.5209, "rewards/accuracies": 0.6812499761581421, "rewards/chosen": -4.771060943603516, "rewards/margins": 0.8642056584358215, "rewards/rejected": -5.635266304016113, "step": 260 }, { "epoch": 0.565149136577708, "grad_norm": 0.1788558065891266, "learning_rate": 0.00014176569902035088, "logits/chosen": -1.176230549812317, "logits/rejected": -1.214536428451538, "logps/chosen": -739.462646484375, "logps/rejected": -794.5394287109375, "loss": 0.5172, "rewards/accuracies": 0.7437499761581421, "rewards/chosen": -4.040095329284668, "rewards/margins": 0.9585875272750854, "rewards/rejected": -4.998682975769043, "step": 270 }, { "epoch": 0.5860805860805861, "grad_norm": 0.22155936062335968, "learning_rate": 0.0001308295425420593, "logits/chosen": -1.1204885244369507, "logits/rejected": -1.2108111381530762, "logps/chosen": -661.9423217773438, "logps/rejected": -689.7301635742188, "loss": 0.5431, "rewards/accuracies": 0.668749988079071, "rewards/chosen": -3.6794509887695312, "rewards/margins": 0.6488171219825745, "rewards/rejected": -4.328268527984619, "step": 280 }, { "epoch": 0.6070120355834642, "grad_norm": 0.2569100558757782, "learning_rate": 0.00011999614593359334, "logits/chosen": -1.0885214805603027, "logits/rejected": -1.1296148300170898, "logps/chosen": -734.2669067382812, "logps/rejected": -823.91015625, "loss": 0.5135, "rewards/accuracies": 0.768750011920929, "rewards/chosen": -4.625601291656494, "rewards/margins": 0.7508729100227356, "rewards/rejected": -5.376473903656006, "step": 290 }, { "epoch": 0.6279434850863422, "grad_norm": 0.20601074397563934, "learning_rate": 0.00010932357971453744, "logits/chosen": -1.036828875541687, "logits/rejected": -1.1336538791656494, "logps/chosen": -893.9344482421875, "logps/rejected": -916.6799926757812, "loss": 0.511, "rewards/accuracies": 0.7562500238418579, "rewards/chosen": -5.284546852111816, "rewards/margins": 0.8582801818847656, "rewards/rejected": -6.142827033996582, "step": 300 }, { "epoch": 0.6279434850863422, "eval_logits/chosen": -1.115662693977356, "eval_logits/rejected": -1.1394394636154175, "eval_logps/chosen": -794.273193359375, "eval_logps/rejected": -883.09912109375, "eval_loss": 0.5225423574447632, "eval_rewards/accuracies": 0.76953125, "eval_rewards/chosen": -4.815339088439941, "eval_rewards/margins": 0.953169584274292, "eval_rewards/rejected": -5.768507957458496, "eval_runtime": 34.0232, "eval_samples_per_second": 58.783, "eval_steps_per_second": 0.941, "step": 300 }, { "epoch": 0.6488749345892203, "grad_norm": 0.2347152978181839, "learning_rate": 9.886905230142432e-05, "logits/chosen": -1.124932050704956, "logits/rejected": -1.1792018413543701, "logps/chosen": -800.5477905273438, "logps/rejected": -829.0084228515625, "loss": 0.5016, "rewards/accuracies": 0.7875000238418579, "rewards/chosen": -4.595987796783447, "rewards/margins": 0.9579511880874634, "rewards/rejected": -5.553939342498779, "step": 310 }, { "epoch": 0.6698063840920984, "grad_norm": 0.22556880116462708, "learning_rate": 8.868860335206678e-05, "logits/chosen": -0.987607479095459, "logits/rejected": -1.0762406587600708, "logps/chosen": -740.9945068359375, "logps/rejected": -809.6924438476562, "loss": 0.49, "rewards/accuracies": 0.78125, "rewards/chosen": -4.5622124671936035, "rewards/margins": 1.0737111568450928, "rewards/rejected": -5.635923385620117, "step": 320 }, { "epoch": 0.6907378335949764, "grad_norm": 0.22182781994342804, "learning_rate": 7.883680337481599e-05, "logits/chosen": -1.090471863746643, "logits/rejected": -1.1103293895721436, "logps/chosen": -724.054931640625, "logps/rejected": -794.3676147460938, "loss": 0.5071, "rewards/accuracies": 0.7437499761581421, "rewards/chosen": -4.166729927062988, "rewards/margins": 0.8365222215652466, "rewards/rejected": -5.003252029418945, "step": 330 }, { "epoch": 0.7116692830978545, "grad_norm": 0.280301034450531, "learning_rate": 6.936646121293653e-05, "logits/chosen": -1.1425997018814087, "logits/rejected": -1.207086205482483, "logps/chosen": -778.4783935546875, "logps/rejected": -811.2172241210938, "loss": 0.5069, "rewards/accuracies": 0.7437499761581421, "rewards/chosen": -4.307219505310059, "rewards/margins": 0.8611894845962524, "rewards/rejected": -5.168410301208496, "step": 340 }, { "epoch": 0.7326007326007326, "grad_norm": 0.2707894444465637, "learning_rate": 6.032834097207889e-05, "logits/chosen": -1.2494275569915771, "logits/rejected": -1.364678144454956, "logps/chosen": -731.4861450195312, "logps/rejected": -781.0096435546875, "loss": 0.4841, "rewards/accuracies": 0.793749988079071, "rewards/chosen": -4.36992073059082, "rewards/margins": 0.9430965185165405, "rewards/rejected": -5.313016891479492, "step": 350 }, { "epoch": 0.7535321821036107, "grad_norm": 0.17782525718212128, "learning_rate": 5.1770889908207245e-05, "logits/chosen": -1.1481292247772217, "logits/rejected": -1.296864628791809, "logps/chosen": -804.0589599609375, "logps/rejected": -836.4563598632812, "loss": 0.5281, "rewards/accuracies": 0.7749999761581421, "rewards/chosen": -4.434665203094482, "rewards/margins": 0.9478410482406616, "rewards/rejected": -5.382506370544434, "step": 360 }, { "epoch": 0.7744636316064888, "grad_norm": 0.19104325771331787, "learning_rate": 4.373997873459449e-05, "logits/chosen": -1.1908161640167236, "logits/rejected": -1.3131165504455566, "logps/chosen": -732.2794189453125, "logps/rejected": -775.1005859375, "loss": 0.5126, "rewards/accuracies": 0.75, "rewards/chosen": -4.367339134216309, "rewards/margins": 0.8501862287521362, "rewards/rejected": -5.217526435852051, "step": 370 }, { "epoch": 0.7953950811093669, "grad_norm": 0.2322331815958023, "learning_rate": 3.6278655739920866e-05, "logits/chosen": -1.2175922393798828, "logits/rejected": -1.2287733554840088, "logps/chosen": -763.4591674804688, "logps/rejected": -873.5496215820312, "loss": 0.5018, "rewards/accuracies": 0.699999988079071, "rewards/chosen": -4.82603120803833, "rewards/margins": 0.9868711233139038, "rewards/rejected": -5.812902927398682, "step": 380 }, { "epoch": 0.8163265306122449, "grad_norm": 0.2882665693759918, "learning_rate": 2.942691603548416e-05, "logits/chosen": -1.0723416805267334, "logits/rejected": -1.1688863039016724, "logps/chosen": -854.5564575195312, "logps/rejected": -905.0885009765625, "loss": 0.5087, "rewards/accuracies": 0.7250000238418579, "rewards/chosen": -4.80784273147583, "rewards/margins": 0.8358656167984009, "rewards/rejected": -5.643708229064941, "step": 390 }, { "epoch": 0.837257980115123, "grad_norm": 0.2635945677757263, "learning_rate": 2.3221487168430806e-05, "logits/chosen": -1.185377836227417, "logits/rejected": -1.252471685409546, "logps/chosen": -734.5894775390625, "logps/rejected": -798.04296875, "loss": 0.508, "rewards/accuracies": 0.7437499761581421, "rewards/chosen": -4.4496073722839355, "rewards/margins": 0.9138796925544739, "rewards/rejected": -5.363487720489502, "step": 400 }, { "epoch": 0.837257980115123, "eval_logits/chosen": -1.1871331930160522, "eval_logits/rejected": -1.2109087705612183, "eval_logps/chosen": -747.29638671875, "eval_logps/rejected": -842.17724609375, "eval_loss": 0.5157404541969299, "eval_rewards/accuracies": 0.76953125, "eval_rewards/chosen": -4.345571041107178, "eval_rewards/margins": 1.0137183666229248, "eval_rewards/rejected": -5.359289646148682, "eval_runtime": 34.0708, "eval_samples_per_second": 58.701, "eval_steps_per_second": 0.939, "step": 400 }, { "epoch": 0.858189429618001, "grad_norm": 0.22161860764026642, "learning_rate": 1.7695632250191e-05, "logits/chosen": -0.9348050951957703, "logits/rejected": -1.0523372888565063, "logps/chosen": -794.3521118164062, "logps/rejected": -829.3026123046875, "loss": 0.511, "rewards/accuracies": 0.762499988079071, "rewards/chosen": -4.382437705993652, "rewards/margins": 0.990971565246582, "rewards/rejected": -5.373409271240234, "step": 410 }, { "epoch": 0.8791208791208791, "grad_norm": 0.2439228594303131, "learning_rate": 1.2878971655412513e-05, "logits/chosen": -1.1185680627822876, "logits/rejected": -1.1273977756500244, "logps/chosen": -739.8529052734375, "logps/rejected": -834.0125122070312, "loss": 0.4955, "rewards/accuracies": 0.737500011920929, "rewards/chosen": -4.466958045959473, "rewards/margins": 0.9562638401985168, "rewards/rejected": -5.423221588134766, "step": 420 }, { "epoch": 0.9000523286237572, "grad_norm": 0.2282486855983734, "learning_rate": 8.79732424714541e-06, "logits/chosen": -1.2293227910995483, "logits/rejected": -1.3410972356796265, "logps/chosen": -786.1281127929688, "logps/rejected": -832.2101440429688, "loss": 0.4959, "rewards/accuracies": 0.737500011920929, "rewards/chosen": -4.519133567810059, "rewards/margins": 0.8781270980834961, "rewards/rejected": -5.397259712219238, "step": 430 }, { "epoch": 0.9209837781266352, "grad_norm": 0.2305663526058197, "learning_rate": 5.472568979361852e-06, "logits/chosen": -1.0845152139663696, "logits/rejected": -1.1872012615203857, "logps/chosen": -724.5098266601562, "logps/rejected": -755.2883911132812, "loss": 0.519, "rewards/accuracies": 0.737500011920929, "rewards/chosen": -4.300398349761963, "rewards/margins": 0.8024924397468567, "rewards/rejected": -5.102890968322754, "step": 440 }, { "epoch": 0.9419152276295133, "grad_norm": 0.22703427076339722, "learning_rate": 2.922527618666465e-06, "logits/chosen": -1.2591861486434937, "logits/rejected": -1.3150806427001953, "logps/chosen": -736.691650390625, "logps/rejected": -812.67626953125, "loss": 0.4926, "rewards/accuracies": 0.7562500238418579, "rewards/chosen": -4.421994686126709, "rewards/margins": 0.9103468060493469, "rewards/rejected": -5.33234167098999, "step": 450 }, { "epoch": 0.9628466771323915, "grad_norm": 0.25431472063064575, "learning_rate": 1.1608692138469379e-06, "logits/chosen": -1.1412612199783325, "logits/rejected": -1.286179542541504, "logps/chosen": -751.5174560546875, "logps/rejected": -791.9276733398438, "loss": 0.5078, "rewards/accuracies": 0.762499988079071, "rewards/chosen": -4.383217811584473, "rewards/margins": 0.9881366491317749, "rewards/rejected": -5.371354579925537, "step": 460 }, { "epoch": 0.9837781266352695, "grad_norm": 0.2301347702741623, "learning_rate": 1.9703682533901977e-07, "logits/chosen": -1.1379841566085815, "logits/rejected": -1.2217354774475098, "logps/chosen": -767.3609619140625, "logps/rejected": -820.5641479492188, "loss": 0.4915, "rewards/accuracies": 0.7437499761581421, "rewards/chosen": -4.3214111328125, "rewards/margins": 0.8934388160705566, "rewards/rejected": -5.214850425720215, "step": 470 }, { "epoch": 0.9984301412872841, "step": 477, "total_flos": 0.0, "train_loss": 0.5423707397229016, "train_runtime": 2137.6738, "train_samples_per_second": 28.598, "train_steps_per_second": 0.223 } ], "logging_steps": 10, "max_steps": 477, "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": 4, "trial_name": null, "trial_params": null }