| { |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 19.547511312217196, |
| "eval_steps": 500, |
| "global_step": 540, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.36199095022624433, |
| "grad_norm": 0.5034113526344299, |
| "learning_rate": 9.259259259259259e-07, |
| "logits/chosen": 1.8096959590911865, |
| "logits/rejected": 1.7158682346343994, |
| "logps/chosen": -106.03636169433594, |
| "logps/rejected": -84.16211700439453, |
| "loss": 0.695, |
| "rewards/accuracies": 0.3499999940395355, |
| "rewards/chosen": -0.00430720392614603, |
| "rewards/margins": 0.00014101080887485296, |
| "rewards/rejected": -0.004448213614523411, |
| "step": 10 |
| }, |
| { |
| "epoch": 0.7239819004524887, |
| "grad_norm": 0.4689481854438782, |
| "learning_rate": 1.8518518518518519e-06, |
| "logits/chosen": 1.6904691457748413, |
| "logits/rejected": 1.6791242361068726, |
| "logps/chosen": -82.1568603515625, |
| "logps/rejected": -81.06039428710938, |
| "loss": 0.6942, |
| "rewards/accuracies": 0.5375000238418579, |
| "rewards/chosen": -0.002815738320350647, |
| "rewards/margins": 0.005291810259222984, |
| "rewards/rejected": -0.008107547648251057, |
| "step": 20 |
| }, |
| { |
| "epoch": 1.085972850678733, |
| "grad_norm": 0.5041429996490479, |
| "learning_rate": 2.7777777777777783e-06, |
| "logits/chosen": 1.877396583557129, |
| "logits/rejected": 1.9337213039398193, |
| "logps/chosen": -76.86979675292969, |
| "logps/rejected": -86.98431396484375, |
| "loss": 0.6937, |
| "rewards/accuracies": 0.48750001192092896, |
| "rewards/chosen": -0.002632059855386615, |
| "rewards/margins": -0.00260960147716105, |
| "rewards/rejected": -2.245856376248412e-05, |
| "step": 30 |
| }, |
| { |
| "epoch": 1.4479638009049773, |
| "grad_norm": 0.49786925315856934, |
| "learning_rate": 3.7037037037037037e-06, |
| "logits/chosen": 1.622865915298462, |
| "logits/rejected": 1.649457573890686, |
| "logps/chosen": -77.10165405273438, |
| "logps/rejected": -75.56939697265625, |
| "loss": 0.6927, |
| "rewards/accuracies": 0.5249999761581421, |
| "rewards/chosen": -0.005956621374934912, |
| "rewards/margins": -0.0038306519854813814, |
| "rewards/rejected": -0.0021259686909615993, |
| "step": 40 |
| }, |
| { |
| "epoch": 1.8099547511312217, |
| "grad_norm": 0.6152529120445251, |
| "learning_rate": 4.62962962962963e-06, |
| "logits/chosen": 1.7763454914093018, |
| "logits/rejected": 1.8556989431381226, |
| "logps/chosen": -84.42935943603516, |
| "logps/rejected": -97.08312225341797, |
| "loss": 0.6937, |
| "rewards/accuracies": 0.5625, |
| "rewards/chosen": 0.008006314747035503, |
| "rewards/margins": 0.005445700138807297, |
| "rewards/rejected": 0.0025606155395507812, |
| "step": 50 |
| }, |
| { |
| "epoch": 2.171945701357466, |
| "grad_norm": 0.539294958114624, |
| "learning_rate": 4.998119881260576e-06, |
| "logits/chosen": 1.7218296527862549, |
| "logits/rejected": 1.759447455406189, |
| "logps/chosen": -71.67170715332031, |
| "logps/rejected": -86.7840805053711, |
| "loss": 0.6935, |
| "rewards/accuracies": 0.4625000059604645, |
| "rewards/chosen": 0.0036853079218417406, |
| "rewards/margins": -0.002186889760196209, |
| "rewards/rejected": 0.005872196517884731, |
| "step": 60 |
| }, |
| { |
| "epoch": 2.5339366515837103, |
| "grad_norm": 0.5369167327880859, |
| "learning_rate": 4.9866405060165044e-06, |
| "logits/chosen": 1.7708947658538818, |
| "logits/rejected": 1.8614221811294556, |
| "logps/chosen": -86.73070526123047, |
| "logps/rejected": -83.09151458740234, |
| "loss": 0.6909, |
| "rewards/accuracies": 0.5, |
| "rewards/chosen": 0.008487144485116005, |
| "rewards/margins": 0.004935932345688343, |
| "rewards/rejected": 0.0035512119065970182, |
| "step": 70 |
| }, |
| { |
| "epoch": 2.8959276018099547, |
| "grad_norm": 0.5544038414955139, |
| "learning_rate": 4.964774158361991e-06, |
| "logits/chosen": 1.772223711013794, |
| "logits/rejected": 1.834602952003479, |
| "logps/chosen": -69.32654571533203, |
| "logps/rejected": -77.55828094482422, |
| "loss": 0.6943, |
| "rewards/accuracies": 0.5249999761581421, |
| "rewards/chosen": -0.012612676247954369, |
| "rewards/margins": -0.011703734286129475, |
| "rewards/rejected": -0.0009089424274861813, |
| "step": 80 |
| }, |
| { |
| "epoch": 3.257918552036199, |
| "grad_norm": 0.5990457534790039, |
| "learning_rate": 4.93261217644956e-06, |
| "logits/chosen": 1.8591846227645874, |
| "logits/rejected": 1.8804810047149658, |
| "logps/chosen": -79.4209213256836, |
| "logps/rejected": -76.4483642578125, |
| "loss": 0.6927, |
| "rewards/accuracies": 0.4749999940395355, |
| "rewards/chosen": -0.0016643519047647715, |
| "rewards/margins": 0.006901158951222897, |
| "rewards/rejected": -0.008565512485802174, |
| "step": 90 |
| }, |
| { |
| "epoch": 3.6199095022624435, |
| "grad_norm": 0.6173596978187561, |
| "learning_rate": 4.8902889044347e-06, |
| "logits/chosen": 1.777044653892517, |
| "logits/rejected": 1.7735064029693604, |
| "logps/chosen": -85.6143569946289, |
| "logps/rejected": -79.8615493774414, |
| "loss": 0.6894, |
| "rewards/accuracies": 0.5375000238418579, |
| "rewards/chosen": -0.0019477702444419265, |
| "rewards/margins": 0.002171102212741971, |
| "rewards/rejected": -0.004118871875107288, |
| "step": 100 |
| }, |
| { |
| "epoch": 3.981900452488688, |
| "grad_norm": 0.6671907901763916, |
| "learning_rate": 4.837981131305475e-06, |
| "logits/chosen": 1.6935539245605469, |
| "logits/rejected": 1.7035915851593018, |
| "logps/chosen": -83.69059753417969, |
| "logps/rejected": -89.67396545410156, |
| "loss": 0.6907, |
| "rewards/accuracies": 0.48750001192092896, |
| "rewards/chosen": -0.005695090629160404, |
| "rewards/margins": 0.0027313486207276583, |
| "rewards/rejected": -0.00842643715441227, |
| "step": 110 |
| }, |
| { |
| "epoch": 4.343891402714932, |
| "grad_norm": 0.9499974250793457, |
| "learning_rate": 4.775907352415367e-06, |
| "logits/chosen": 1.801300287246704, |
| "logits/rejected": 1.8080482482910156, |
| "logps/chosen": -92.2596206665039, |
| "logps/rejected": -85.3951416015625, |
| "loss": 0.6863, |
| "rewards/accuracies": 0.637499988079071, |
| "rewards/chosen": 0.000912027433514595, |
| "rewards/margins": 0.017921410501003265, |
| "rewards/rejected": -0.01700938306748867, |
| "step": 120 |
| }, |
| { |
| "epoch": 4.705882352941177, |
| "grad_norm": 0.6524670124053955, |
| "learning_rate": 4.70432685680402e-06, |
| "logits/chosen": 1.8659807443618774, |
| "logits/rejected": 1.8858706951141357, |
| "logps/chosen": -78.38592529296875, |
| "logps/rejected": -81.4170913696289, |
| "loss": 0.6873, |
| "rewards/accuracies": 0.5375000238418579, |
| "rewards/chosen": 0.002340498147532344, |
| "rewards/margins": 0.014832529239356518, |
| "rewards/rejected": -0.01249203272163868, |
| "step": 130 |
| }, |
| { |
| "epoch": 5.067873303167421, |
| "grad_norm": 0.7461792826652527, |
| "learning_rate": 4.623538644118244e-06, |
| "logits/chosen": 1.7999374866485596, |
| "logits/rejected": 1.869183897972107, |
| "logps/chosen": -71.37133026123047, |
| "logps/rejected": -83.36344909667969, |
| "loss": 0.6829, |
| "rewards/accuracies": 0.612500011920929, |
| "rewards/chosen": -0.013370734639465809, |
| "rewards/margins": 0.013819178566336632, |
| "rewards/rejected": -0.027189914137125015, |
| "step": 140 |
| }, |
| { |
| "epoch": 5.429864253393665, |
| "grad_norm": 0.6743089556694031, |
| "learning_rate": 4.533880175657419e-06, |
| "logits/chosen": 1.8166606426239014, |
| "logits/rejected": 1.7869322299957275, |
| "logps/chosen": -87.78181457519531, |
| "logps/rejected": -71.32334899902344, |
| "loss": 0.6758, |
| "rewards/accuracies": 0.6499999761581421, |
| "rewards/chosen": -0.0059204003773629665, |
| "rewards/margins": 0.03691873326897621, |
| "rewards/rejected": -0.04283912852406502, |
| "step": 150 |
| }, |
| { |
| "epoch": 5.791855203619909, |
| "grad_norm": 0.7719553112983704, |
| "learning_rate": 4.435725964760331e-06, |
| "logits/chosen": 1.7498859167099, |
| "logits/rejected": 1.7752033472061157, |
| "logps/chosen": -72.18727111816406, |
| "logps/rejected": -84.50582122802734, |
| "loss": 0.6786, |
| "rewards/accuracies": 0.675000011920929, |
| "rewards/chosen": -0.023137167096138, |
| "rewards/margins": 0.042744893580675125, |
| "rewards/rejected": -0.06588206440210342, |
| "step": 160 |
| }, |
| { |
| "epoch": 6.153846153846154, |
| "grad_norm": 0.8010989427566528, |
| "learning_rate": 4.329486012421531e-06, |
| "logits/chosen": 1.7785524129867554, |
| "logits/rejected": 1.770937204360962, |
| "logps/chosen": -87.91519165039062, |
| "logps/rejected": -86.76083374023438, |
| "loss": 0.6689, |
| "rewards/accuracies": 0.6499999761581421, |
| "rewards/chosen": -0.019223090261220932, |
| "rewards/margins": 0.03557150438427925, |
| "rewards/rejected": -0.054794590920209885, |
| "step": 170 |
| }, |
| { |
| "epoch": 6.515837104072398, |
| "grad_norm": 0.9732075929641724, |
| "learning_rate": 4.215604094671835e-06, |
| "logits/chosen": 1.76633620262146, |
| "logits/rejected": 1.782987356185913, |
| "logps/chosen": -85.80155944824219, |
| "logps/rejected": -85.45477294921875, |
| "loss": 0.6648, |
| "rewards/accuracies": 0.675000011920929, |
| "rewards/chosen": -0.009860477410256863, |
| "rewards/margins": 0.0789995938539505, |
| "rewards/rejected": -0.08886007964611053, |
| "step": 180 |
| }, |
| { |
| "epoch": 6.877828054298643, |
| "grad_norm": 0.8777466416358948, |
| "learning_rate": 4.094555908876765e-06, |
| "logits/chosen": 1.8455078601837158, |
| "logits/rejected": 1.871913194656372, |
| "logps/chosen": -71.23600006103516, |
| "logps/rejected": -71.44808197021484, |
| "loss": 0.6606, |
| "rewards/accuracies": 0.612500011920929, |
| "rewards/chosen": -0.04135243222117424, |
| "rewards/margins": 0.06673365831375122, |
| "rewards/rejected": -0.10808608680963516, |
| "step": 190 |
| }, |
| { |
| "epoch": 7.239819004524887, |
| "grad_norm": 0.8959658741950989, |
| "learning_rate": 3.966847086696045e-06, |
| "logits/chosen": 1.8061046600341797, |
| "logits/rejected": 1.8204586505889893, |
| "logps/chosen": -80.91191864013672, |
| "logps/rejected": -73.68770599365234, |
| "loss": 0.6511, |
| "rewards/accuracies": 0.7875000238418579, |
| "rewards/chosen": -0.037331439554691315, |
| "rewards/margins": 0.11111048609018326, |
| "rewards/rejected": -0.14844192564487457, |
| "step": 200 |
| }, |
| { |
| "epoch": 7.601809954751131, |
| "grad_norm": 0.9700360298156738, |
| "learning_rate": 3.833011082004229e-06, |
| "logits/chosen": 1.7680559158325195, |
| "logits/rejected": 1.8058135509490967, |
| "logps/chosen": -77.65849304199219, |
| "logps/rejected": -78.11155700683594, |
| "loss": 0.645, |
| "rewards/accuracies": 0.7250000238418579, |
| "rewards/chosen": -0.0448901504278183, |
| "rewards/margins": 0.12249104678630829, |
| "rewards/rejected": -0.1673811972141266, |
| "step": 210 |
| }, |
| { |
| "epoch": 7.963800904977376, |
| "grad_norm": 1.070133924484253, |
| "learning_rate": 3.693606942594873e-06, |
| "logits/chosen": 1.7242870330810547, |
| "logits/rejected": 1.7760801315307617, |
| "logps/chosen": -79.4949722290039, |
| "logps/rejected": -84.17713928222656, |
| "loss": 0.6516, |
| "rewards/accuracies": 0.699999988079071, |
| "rewards/chosen": -0.06297416985034943, |
| "rewards/margins": 0.1389354169368744, |
| "rewards/rejected": -0.20190958678722382, |
| "step": 220 |
| }, |
| { |
| "epoch": 8.32579185520362, |
| "grad_norm": 0.880223274230957, |
| "learning_rate": 3.549216974976073e-06, |
| "logits/chosen": 1.8455660343170166, |
| "logits/rejected": 1.8497081995010376, |
| "logps/chosen": -82.68794250488281, |
| "logps/rejected": -87.30531311035156, |
| "loss": 0.6396, |
| "rewards/accuracies": 0.6875, |
| "rewards/chosen": -0.09025909751653671, |
| "rewards/margins": 0.1426868587732315, |
| "rewards/rejected": -0.23294594883918762, |
| "step": 230 |
| }, |
| { |
| "epoch": 8.687782805429864, |
| "grad_norm": 0.9315020442008972, |
| "learning_rate": 3.400444312011776e-06, |
| "logits/chosen": 1.7252687215805054, |
| "logits/rejected": 1.7745378017425537, |
| "logps/chosen": -76.34024047851562, |
| "logps/rejected": -89.45391845703125, |
| "loss": 0.6325, |
| "rewards/accuracies": 0.737500011920929, |
| "rewards/chosen": -0.07824709266424179, |
| "rewards/margins": 0.14033600687980652, |
| "rewards/rejected": -0.21858307719230652, |
| "step": 240 |
| }, |
| { |
| "epoch": 9.049773755656108, |
| "grad_norm": 0.9806249141693115, |
| "learning_rate": 3.2479103935691047e-06, |
| "logits/chosen": 1.7673581838607788, |
| "logits/rejected": 1.8124672174453735, |
| "logps/chosen": -91.80342102050781, |
| "logps/rejected": -91.12406921386719, |
| "loss": 0.6314, |
| "rewards/accuracies": 0.75, |
| "rewards/chosen": -0.10384390503168106, |
| "rewards/margins": 0.189823716878891, |
| "rewards/rejected": -0.29366764426231384, |
| "step": 250 |
| }, |
| { |
| "epoch": 9.411764705882353, |
| "grad_norm": 1.0357781648635864, |
| "learning_rate": 3.092252370695298e-06, |
| "logits/chosen": 1.7498366832733154, |
| "logits/rejected": 1.6984941959381104, |
| "logps/chosen": -95.309326171875, |
| "logps/rejected": -80.66096496582031, |
| "loss": 0.6213, |
| "rewards/accuracies": 0.6875, |
| "rewards/chosen": -0.14446459710597992, |
| "rewards/margins": 0.1758357733488083, |
| "rewards/rejected": -0.3203004002571106, |
| "step": 260 |
| }, |
| { |
| "epoch": 9.773755656108598, |
| "grad_norm": 0.9333145022392273, |
| "learning_rate": 2.9341204441673267e-06, |
| "logits/chosen": 1.8672996759414673, |
| "logits/rejected": 1.7658369541168213, |
| "logps/chosen": -104.98448181152344, |
| "logps/rejected": -81.6786880493164, |
| "loss": 0.6131, |
| "rewards/accuracies": 0.7875000238418579, |
| "rewards/chosen": -0.05903264880180359, |
| "rewards/margins": 0.22136883437633514, |
| "rewards/rejected": -0.28040146827697754, |
| "step": 270 |
| }, |
| { |
| "epoch": 10.135746606334841, |
| "grad_norm": 0.9343512654304504, |
| "learning_rate": 2.7741751485313295e-06, |
| "logits/chosen": 1.7365798950195312, |
| "logits/rejected": 1.748958945274353, |
| "logps/chosen": -84.08332824707031, |
| "logps/rejected": -87.89321899414062, |
| "loss": 0.6121, |
| "rewards/accuracies": 0.6875, |
| "rewards/chosen": -0.13087712228298187, |
| "rewards/margins": 0.16090884804725647, |
| "rewards/rejected": -0.29178598523139954, |
| "step": 280 |
| }, |
| { |
| "epoch": 10.497737556561086, |
| "grad_norm": 0.9784368872642517, |
| "learning_rate": 2.6130845929767662e-06, |
| "logits/chosen": 1.7028934955596924, |
| "logits/rejected": 1.6997534036636353, |
| "logps/chosen": -86.77079010009766, |
| "logps/rejected": -85.89974975585938, |
| "loss": 0.612, |
| "rewards/accuracies": 0.7250000238418579, |
| "rewards/chosen": -0.10491044819355011, |
| "rewards/margins": 0.19683709740638733, |
| "rewards/rejected": -0.30174753069877625, |
| "step": 290 |
| }, |
| { |
| "epoch": 10.85972850678733, |
| "grad_norm": 1.030173659324646, |
| "learning_rate": 2.4515216705704396e-06, |
| "logits/chosen": 1.8057117462158203, |
| "logits/rejected": 1.8746700286865234, |
| "logps/chosen": -74.07008361816406, |
| "logps/rejected": -89.6046142578125, |
| "loss": 0.5997, |
| "rewards/accuracies": 0.8125, |
| "rewards/chosen": -0.058038175106048584, |
| "rewards/margins": 0.2527470886707306, |
| "rewards/rejected": -0.3107852637767792, |
| "step": 300 |
| }, |
| { |
| "epoch": 11.221719457013574, |
| "grad_norm": 1.0045686960220337, |
| "learning_rate": 2.290161247507733e-06, |
| "logits/chosen": 1.7842658758163452, |
| "logits/rejected": 1.83527410030365, |
| "logps/chosen": -85.32085418701172, |
| "logps/rejected": -91.45645904541016, |
| "loss": 0.5967, |
| "rewards/accuracies": 0.800000011920929, |
| "rewards/chosen": -0.1375085413455963, |
| "rewards/margins": 0.2444867342710495, |
| "rewards/rejected": -0.3819952607154846, |
| "step": 310 |
| }, |
| { |
| "epoch": 11.583710407239819, |
| "grad_norm": 0.967028796672821, |
| "learning_rate": 2.129677344121879e-06, |
| "logits/chosen": 1.8002763986587524, |
| "logits/rejected": 1.7979824542999268, |
| "logps/chosen": -97.67089080810547, |
| "logps/rejected": -79.14571380615234, |
| "loss": 0.5852, |
| "rewards/accuracies": 0.762499988079071, |
| "rewards/chosen": -0.1535075604915619, |
| "rewards/margins": 0.18517914414405823, |
| "rewards/rejected": -0.3386867046356201, |
| "step": 320 |
| }, |
| { |
| "epoch": 11.945701357466064, |
| "grad_norm": 1.017149806022644, |
| "learning_rate": 1.970740319426474e-06, |
| "logits/chosen": 1.8270378112792969, |
| "logits/rejected": 1.8545118570327759, |
| "logps/chosen": -74.19200897216797, |
| "logps/rejected": -86.13984680175781, |
| "loss": 0.6015, |
| "rewards/accuracies": 0.7749999761581421, |
| "rewards/chosen": -0.1458810567855835, |
| "rewards/margins": 0.20589685440063477, |
| "rewards/rejected": -0.35177791118621826, |
| "step": 330 |
| }, |
| { |
| "epoch": 12.307692307692308, |
| "grad_norm": 0.9695628881454468, |
| "learning_rate": 1.8140140709517467e-06, |
| "logits/chosen": 1.745535135269165, |
| "logits/rejected": 1.7830098867416382, |
| "logps/chosen": -84.69477081298828, |
| "logps/rejected": -82.7412109375, |
| "loss": 0.5882, |
| "rewards/accuracies": 0.762499988079071, |
| "rewards/chosen": -0.12116970866918564, |
| "rewards/margins": 0.25774163007736206, |
| "rewards/rejected": -0.3789113759994507, |
| "step": 340 |
| }, |
| { |
| "epoch": 12.669683257918551, |
| "grad_norm": 1.103960394859314, |
| "learning_rate": 1.6601532615711452e-06, |
| "logits/chosen": 1.777008295059204, |
| "logits/rejected": 1.6892492771148682, |
| "logps/chosen": -100.1700210571289, |
| "logps/rejected": -90.1813735961914, |
| "loss": 0.5802, |
| "rewards/accuracies": 0.7749999761581421, |
| "rewards/chosen": -0.10893051326274872, |
| "rewards/margins": 0.29678279161453247, |
| "rewards/rejected": -0.4057133197784424, |
| "step": 350 |
| }, |
| { |
| "epoch": 13.031674208144796, |
| "grad_norm": 1.0589226484298706, |
| "learning_rate": 1.509800584902108e-06, |
| "logits/chosen": 1.7005256414413452, |
| "logits/rejected": 1.7957611083984375, |
| "logps/chosen": -79.087158203125, |
| "logps/rejected": -102.69563293457031, |
| "loss": 0.5814, |
| "rewards/accuracies": 0.862500011920929, |
| "rewards/chosen": -0.13767623901367188, |
| "rewards/margins": 0.3307909071445465, |
| "rewards/rejected": -0.4684671461582184, |
| "step": 360 |
| }, |
| { |
| "epoch": 13.393665158371041, |
| "grad_norm": 1.056600570678711, |
| "learning_rate": 1.3635840807037487e-06, |
| "logits/chosen": 1.6859004497528076, |
| "logits/rejected": 1.7192614078521729, |
| "logps/chosen": -96.27987670898438, |
| "logps/rejected": -96.15126037597656, |
| "loss": 0.5703, |
| "rewards/accuracies": 0.8500000238418579, |
| "rewards/chosen": -0.16748157143592834, |
| "rewards/margins": 0.38242021203041077, |
| "rewards/rejected": -0.5499017834663391, |
| "step": 370 |
| }, |
| { |
| "epoch": 13.755656108597286, |
| "grad_norm": 2.783137321472168, |
| "learning_rate": 1.2221145114853172e-06, |
| "logits/chosen": 1.8700748682022095, |
| "logits/rejected": 1.886313796043396, |
| "logps/chosen": -82.22146606445312, |
| "logps/rejected": -79.1581802368164, |
| "loss": 0.5791, |
| "rewards/accuracies": 0.8125, |
| "rewards/chosen": -0.16777488589286804, |
| "rewards/margins": 0.2746582329273224, |
| "rewards/rejected": -0.44243311882019043, |
| "step": 380 |
| }, |
| { |
| "epoch": 14.117647058823529, |
| "grad_norm": 1.137703776359558, |
| "learning_rate": 1.085982811283654e-06, |
| "logits/chosen": 1.7121073007583618, |
| "logits/rejected": 1.74045729637146, |
| "logps/chosen": -81.7417221069336, |
| "logps/rejected": -82.1438980102539, |
| "loss": 0.5792, |
| "rewards/accuracies": 0.762499988079071, |
| "rewards/chosen": -0.21657264232635498, |
| "rewards/margins": 0.23677067458629608, |
| "rewards/rejected": -0.45334330201148987, |
| "step": 390 |
| }, |
| { |
| "epoch": 14.479638009049774, |
| "grad_norm": 1.115824580192566, |
| "learning_rate": 9.557576172663577e-07, |
| "logits/chosen": 1.782957673072815, |
| "logits/rejected": 1.740311861038208, |
| "logps/chosen": -95.57044982910156, |
| "logps/rejected": -87.47151184082031, |
| "loss": 0.5648, |
| "rewards/accuracies": 0.800000011920929, |
| "rewards/chosen": -0.13624358177185059, |
| "rewards/margins": 0.3296685218811035, |
| "rewards/rejected": -0.4659121632575989, |
| "step": 400 |
| }, |
| { |
| "epoch": 14.841628959276019, |
| "grad_norm": 1.0241665840148926, |
| "learning_rate": 8.319828944714508e-07, |
| "logits/chosen": 1.8231852054595947, |
| "logits/rejected": 1.7309335470199585, |
| "logps/chosen": -90.45216369628906, |
| "logps/rejected": -84.18362426757812, |
| "loss": 0.5692, |
| "rewards/accuracies": 0.800000011920929, |
| "rewards/chosen": -0.21606135368347168, |
| "rewards/margins": 0.3016027808189392, |
| "rewards/rejected": -0.5176641345024109, |
| "step": 410 |
| }, |
| { |
| "epoch": 15.203619909502262, |
| "grad_norm": 1.0937505960464478, |
| "learning_rate": 7.151756636052529e-07, |
| "logits/chosen": 1.6469671726226807, |
| "logits/rejected": 1.755052924156189, |
| "logps/chosen": -69.20260620117188, |
| "logps/rejected": -91.29935455322266, |
| "loss": 0.5813, |
| "rewards/accuracies": 0.762499988079071, |
| "rewards/chosen": -0.16654367744922638, |
| "rewards/margins": 0.29352378845214844, |
| "rewards/rejected": -0.4600674510002136, |
| "step": 420 |
| }, |
| { |
| "epoch": 15.565610859728507, |
| "grad_norm": 1.189382553100586, |
| "learning_rate": 6.058238413897052e-07, |
| "logits/chosen": 1.7920663356781006, |
| "logits/rejected": 1.767615556716919, |
| "logps/chosen": -83.10585021972656, |
| "logps/rejected": -87.55345916748047, |
| "loss": 0.5717, |
| "rewards/accuracies": 0.8500000238418579, |
| "rewards/chosen": -0.18599426746368408, |
| "rewards/margins": 0.30218419432640076, |
| "rewards/rejected": -0.48817843198776245, |
| "step": 430 |
| }, |
| { |
| "epoch": 15.927601809954751, |
| "grad_norm": 1.0084723234176636, |
| "learning_rate": 5.043842024802675e-07, |
| "logits/chosen": 1.6690113544464111, |
| "logits/rejected": 1.7104164361953735, |
| "logps/chosen": -77.92027282714844, |
| "logps/rejected": -88.28632354736328, |
| "loss": 0.5616, |
| "rewards/accuracies": 0.824999988079071, |
| "rewards/chosen": -0.1644773930311203, |
| "rewards/margins": 0.31829944252967834, |
| "rewards/rejected": -0.48277679085731506, |
| "step": 440 |
| }, |
| { |
| "epoch": 16.289592760180994, |
| "grad_norm": 1.099560022354126, |
| "learning_rate": 4.1128047146765936e-07, |
| "logits/chosen": 1.777606725692749, |
| "logits/rejected": 1.7520103454589844, |
| "logps/chosen": -92.19166564941406, |
| "logps/rejected": -74.17485809326172, |
| "loss": 0.5653, |
| "rewards/accuracies": 0.800000011920929, |
| "rewards/chosen": -0.1612701714038849, |
| "rewards/margins": 0.32633036375045776, |
| "rewards/rejected": -0.4876004755496979, |
| "step": 450 |
| }, |
| { |
| "epoch": 16.65158371040724, |
| "grad_norm": 1.0390541553497314, |
| "learning_rate": 3.269015529333805e-07, |
| "logits/chosen": 1.6575822830200195, |
| "logits/rejected": 1.785485863685608, |
| "logps/chosen": -81.9493179321289, |
| "logps/rejected": -101.3016128540039, |
| "loss": 0.5579, |
| "rewards/accuracies": 0.8125, |
| "rewards/chosen": -0.20868897438049316, |
| "rewards/margins": 0.3946114182472229, |
| "rewards/rejected": -0.6033004522323608, |
| "step": 460 |
| }, |
| { |
| "epoch": 17.013574660633484, |
| "grad_norm": 1.1404516696929932, |
| "learning_rate": 2.515999069522676e-07, |
| "logits/chosen": 1.6690752506256104, |
| "logits/rejected": 1.6541496515274048, |
| "logps/chosen": -88.44367980957031, |
| "logps/rejected": -71.12745666503906, |
| "loss": 0.5675, |
| "rewards/accuracies": 0.824999988079071, |
| "rewards/chosen": -0.19705724716186523, |
| "rewards/margins": 0.26251187920570374, |
| "rewards/rejected": -0.45956912636756897, |
| "step": 470 |
| }, |
| { |
| "epoch": 17.375565610859727, |
| "grad_norm": 1.175963282585144, |
| "learning_rate": 1.8569007682777417e-07, |
| "logits/chosen": 1.6609585285186768, |
| "logits/rejected": 1.6719191074371338, |
| "logps/chosen": -94.07273864746094, |
| "logps/rejected": -76.59893035888672, |
| "loss": 0.5678, |
| "rewards/accuracies": 0.7749999761581421, |
| "rewards/chosen": -0.1740456521511078, |
| "rewards/margins": 0.3117365837097168, |
| "rewards/rejected": -0.4857822358608246, |
| "step": 480 |
| }, |
| { |
| "epoch": 17.737556561085974, |
| "grad_norm": 1.2133598327636719, |
| "learning_rate": 1.2944737520980883e-07, |
| "logits/chosen": 1.7220537662506104, |
| "logits/rejected": 1.7643390893936157, |
| "logps/chosen": -83.6407699584961, |
| "logps/rejected": -81.10734558105469, |
| "loss": 0.5613, |
| "rewards/accuracies": 0.862500011920929, |
| "rewards/chosen": -0.13465282320976257, |
| "rewards/margins": 0.4027688503265381, |
| "rewards/rejected": -0.5374216437339783, |
| "step": 490 |
| }, |
| { |
| "epoch": 18.099547511312217, |
| "grad_norm": 1.1120809316635132, |
| "learning_rate": 8.310673408334496e-08, |
| "logits/chosen": 1.5700793266296387, |
| "logits/rejected": 1.5635223388671875, |
| "logps/chosen": -86.44955444335938, |
| "logps/rejected": -86.97091674804688, |
| "loss": 0.5626, |
| "rewards/accuracies": 0.8374999761581421, |
| "rewards/chosen": -0.17045333981513977, |
| "rewards/margins": 0.29403942823410034, |
| "rewards/rejected": -0.46449270844459534, |
| "step": 500 |
| }, |
| { |
| "epoch": 18.46153846153846, |
| "grad_norm": 1.177876353263855, |
| "learning_rate": 4.6861723431538273e-08, |
| "logits/chosen": 1.5987281799316406, |
| "logits/rejected": 1.7516323328018188, |
| "logps/chosen": -79.38053131103516, |
| "logps/rejected": -97.10954284667969, |
| "loss": 0.5634, |
| "rewards/accuracies": 0.887499988079071, |
| "rewards/chosen": -0.1326012909412384, |
| "rewards/margins": 0.40437060594558716, |
| "rewards/rejected": -0.5369717478752136, |
| "step": 510 |
| }, |
| { |
| "epoch": 18.823529411764707, |
| "grad_norm": 1.2317314147949219, |
| "learning_rate": 2.0863742672497244e-08, |
| "logits/chosen": 1.71554434299469, |
| "logits/rejected": 1.740006446838379, |
| "logps/chosen": -83.19001007080078, |
| "logps/rejected": -78.78974151611328, |
| "loss": 0.573, |
| "rewards/accuracies": 0.75, |
| "rewards/chosen": -0.23421506583690643, |
| "rewards/margins": 0.25640812516212463, |
| "rewards/rejected": -0.4906231760978699, |
| "step": 520 |
| }, |
| { |
| "epoch": 19.18552036199095, |
| "grad_norm": 1.118938684463501, |
| "learning_rate": 5.221388247169945e-09, |
| "logits/chosen": 1.7030197381973267, |
| "logits/rejected": 1.7396942377090454, |
| "logps/chosen": -86.79795837402344, |
| "logps/rejected": -97.96685791015625, |
| "loss": 0.5571, |
| "rewards/accuracies": 0.75, |
| "rewards/chosen": -0.17865614593029022, |
| "rewards/margins": 0.3226766586303711, |
| "rewards/rejected": -0.5013328790664673, |
| "step": 530 |
| }, |
| { |
| "epoch": 19.547511312217196, |
| "grad_norm": 1.1492843627929688, |
| "learning_rate": 0.0, |
| "logits/chosen": 1.6407601833343506, |
| "logits/rejected": 1.645991325378418, |
| "logps/chosen": -74.029296875, |
| "logps/rejected": -82.86942291259766, |
| "loss": 0.5595, |
| "rewards/accuracies": 0.762499988079071, |
| "rewards/chosen": -0.228031724691391, |
| "rewards/margins": 0.28408557176589966, |
| "rewards/rejected": -0.5121172666549683, |
| "step": 540 |
| }, |
| { |
| "epoch": 19.547511312217196, |
| "step": 540, |
| "total_flos": 1.848581334679683e+18, |
| "train_loss": 0.6237324387938888, |
| "train_runtime": 3964.2678, |
| "train_samples_per_second": 8.915, |
| "train_steps_per_second": 0.136 |
| } |
| ], |
| "logging_steps": 10, |
| "max_steps": 540, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 20, |
| "save_steps": 50, |
| "stateful_callbacks": { |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": true, |
| "should_training_stop": true |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 1.848581334679683e+18, |
| "train_batch_size": 1, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
