| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.998691442030882, | |
| "eval_steps": 500, | |
| "global_step": 477, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.010468463752944255, | |
| "grad_norm": 0.1797010649383585, | |
| "learning_rate": 5.208333333333333e-08, | |
| "logits/chosen": -1.448793649673462, | |
| "logits/rejected": -1.450628638267517, | |
| "logps/chosen": -7.982501983642578, | |
| "logps/rejected": -8.156150817871094, | |
| "loss": -0.0009, | |
| "rewards/accuracies": 0.5062500238418579, | |
| "rewards/chosen": -7.982501983642578, | |
| "rewards/margins": 0.17364946007728577, | |
| "rewards/rejected": -8.156150817871094, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.02093692750588851, | |
| "grad_norm": 0.07416876680191102, | |
| "learning_rate": 1.0416666666666667e-07, | |
| "logits/chosen": -1.4468624591827393, | |
| "logits/rejected": -1.4378924369812012, | |
| "logps/chosen": -8.046422958374023, | |
| "logps/rejected": -7.960408687591553, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 0.4937500059604645, | |
| "rewards/chosen": -8.046422958374023, | |
| "rewards/margins": -0.08601457625627518, | |
| "rewards/rejected": -7.960408687591553, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.031405391258832765, | |
| "grad_norm": 0.1932776998667392, | |
| "learning_rate": 1.5624999999999999e-07, | |
| "logits/chosen": -1.4454925060272217, | |
| "logits/rejected": -1.4418766498565674, | |
| "logps/chosen": -7.853631496429443, | |
| "logps/rejected": -7.868775844573975, | |
| "loss": -0.0011, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -7.853631496429443, | |
| "rewards/margins": 0.01514405943453312, | |
| "rewards/rejected": -7.868775844573975, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.04187385501177702, | |
| "grad_norm": 0.18172311042773612, | |
| "learning_rate": 2.0833333333333333e-07, | |
| "logits/chosen": -1.4386684894561768, | |
| "logits/rejected": -1.4446001052856445, | |
| "logps/chosen": -8.188841819763184, | |
| "logps/rejected": -8.21079158782959, | |
| "loss": 0.0005, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -8.188841819763184, | |
| "rewards/margins": 0.02194919064640999, | |
| "rewards/rejected": -8.21079158782959, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.05234231876472128, | |
| "grad_norm": 0.07565588877379835, | |
| "learning_rate": 2.604166666666667e-07, | |
| "logits/chosen": -1.469273328781128, | |
| "logits/rejected": -1.4595043659210205, | |
| "logps/chosen": -8.117727279663086, | |
| "logps/rejected": -8.016858100891113, | |
| "loss": 0.0008, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -8.117727279663086, | |
| "rewards/margins": -0.1008685976266861, | |
| "rewards/rejected": -8.016858100891113, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.06281078251766553, | |
| "grad_norm": 0.15180659620060438, | |
| "learning_rate": 3.1249999999999997e-07, | |
| "logits/chosen": -1.4331070184707642, | |
| "logits/rejected": -1.4219419956207275, | |
| "logps/chosen": -7.856657981872559, | |
| "logps/rejected": -7.774781703948975, | |
| "loss": 0.001, | |
| "rewards/accuracies": 0.48124998807907104, | |
| "rewards/chosen": -7.856657981872559, | |
| "rewards/margins": -0.08187668025493622, | |
| "rewards/rejected": -7.774781703948975, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.07327924627060979, | |
| "grad_norm": 0.08490698432858583, | |
| "learning_rate": 3.645833333333333e-07, | |
| "logits/chosen": -1.4420219659805298, | |
| "logits/rejected": -1.4201197624206543, | |
| "logps/chosen": -8.129011154174805, | |
| "logps/rejected": -7.994417667388916, | |
| "loss": -0.0007, | |
| "rewards/accuracies": 0.5062500238418579, | |
| "rewards/chosen": -8.129011154174805, | |
| "rewards/margins": -0.13459424674510956, | |
| "rewards/rejected": -7.994417667388916, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.08374771002355404, | |
| "grad_norm": 0.18105138747703975, | |
| "learning_rate": 4.1666666666666667e-07, | |
| "logits/chosen": -1.420586109161377, | |
| "logits/rejected": -1.4364769458770752, | |
| "logps/chosen": -8.086874961853027, | |
| "logps/rejected": -8.243904113769531, | |
| "loss": -0.0007, | |
| "rewards/accuracies": 0.46875, | |
| "rewards/chosen": -8.086874961853027, | |
| "rewards/margins": 0.15702828764915466, | |
| "rewards/rejected": -8.243904113769531, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.0942161737764983, | |
| "grad_norm": 0.05754725019388417, | |
| "learning_rate": 4.6874999999999996e-07, | |
| "logits/chosen": -1.4414559602737427, | |
| "logits/rejected": -1.4433170557022095, | |
| "logps/chosen": -8.102398872375488, | |
| "logps/rejected": -7.88956356048584, | |
| "loss": -0.0001, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -8.102398872375488, | |
| "rewards/margins": -0.21283535659313202, | |
| "rewards/rejected": -7.88956356048584, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.10468463752944256, | |
| "grad_norm": 0.09776323510444927, | |
| "learning_rate": 4.999731868769026e-07, | |
| "logits/chosen": -1.4331969022750854, | |
| "logits/rejected": -1.451460838317871, | |
| "logps/chosen": -8.243871688842773, | |
| "logps/rejected": -8.234269142150879, | |
| "loss": 0.0006, | |
| "rewards/accuracies": 0.48124998807907104, | |
| "rewards/chosen": -8.243871688842773, | |
| "rewards/margins": -0.009602278470993042, | |
| "rewards/rejected": -8.234269142150879, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.11515310128238682, | |
| "grad_norm": 0.06802225811273538, | |
| "learning_rate": 4.996716052911017e-07, | |
| "logits/chosen": -1.4432759284973145, | |
| "logits/rejected": -1.4394512176513672, | |
| "logps/chosen": -8.15147876739502, | |
| "logps/rejected": -8.12590217590332, | |
| "loss": 0.0, | |
| "rewards/accuracies": 0.42500001192092896, | |
| "rewards/chosen": -8.15147876739502, | |
| "rewards/margins": -0.025576595216989517, | |
| "rewards/rejected": -8.12590217590332, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.12562156503533106, | |
| "grad_norm": 0.19637523644104168, | |
| "learning_rate": 4.990353313429303e-07, | |
| "logits/chosen": -1.4439882040023804, | |
| "logits/rejected": -1.4562370777130127, | |
| "logps/chosen": -8.153823852539062, | |
| "logps/rejected": -8.392339706420898, | |
| "loss": -0.0006, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": -8.153823852539062, | |
| "rewards/margins": 0.23851680755615234, | |
| "rewards/rejected": -8.392339706420898, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.1360900287882753, | |
| "grad_norm": 0.15858568302533665, | |
| "learning_rate": 4.980652179769217e-07, | |
| "logits/chosen": -1.3945095539093018, | |
| "logits/rejected": -1.4153038263320923, | |
| "logps/chosen": -7.983679294586182, | |
| "logps/rejected": -8.382383346557617, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": -7.983679294586182, | |
| "rewards/margins": 0.3987043499946594, | |
| "rewards/rejected": -8.382383346557617, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.14655849254121958, | |
| "grad_norm": 0.449952417527854, | |
| "learning_rate": 4.967625656594781e-07, | |
| "logits/chosen": -1.3980720043182373, | |
| "logits/rejected": -1.3917102813720703, | |
| "logps/chosen": -8.066298484802246, | |
| "logps/rejected": -8.08208179473877, | |
| "loss": 0.0015, | |
| "rewards/accuracies": 0.5062500238418579, | |
| "rewards/chosen": -8.066298484802246, | |
| "rewards/margins": 0.015783464536070824, | |
| "rewards/rejected": -8.08208179473877, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.15702695629416383, | |
| "grad_norm": 0.0694833097923889, | |
| "learning_rate": 4.951291206355559e-07, | |
| "logits/chosen": -1.4114166498184204, | |
| "logits/rejected": -1.4182840585708618, | |
| "logps/chosen": -8.104999542236328, | |
| "logps/rejected": -8.062032699584961, | |
| "loss": -0.0, | |
| "rewards/accuracies": 0.45625001192092896, | |
| "rewards/chosen": -8.104999542236328, | |
| "rewards/margins": -0.042965833097696304, | |
| "rewards/rejected": -8.062032699584961, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.16749542004710807, | |
| "grad_norm": 0.13732894835039064, | |
| "learning_rate": 4.93167072587771e-07, | |
| "logits/chosen": -1.416013479232788, | |
| "logits/rejected": -1.4303759336471558, | |
| "logps/chosen": -8.252126693725586, | |
| "logps/rejected": -8.272165298461914, | |
| "loss": -0.0007, | |
| "rewards/accuracies": 0.4437499940395355, | |
| "rewards/chosen": -8.252126693725586, | |
| "rewards/margins": 0.020038936287164688, | |
| "rewards/rejected": -8.272165298461914, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.17796388380005235, | |
| "grad_norm": 0.13841613125213173, | |
| "learning_rate": 4.908790517010636e-07, | |
| "logits/chosen": -1.4297425746917725, | |
| "logits/rejected": -1.4293019771575928, | |
| "logps/chosen": -8.420356750488281, | |
| "logps/rejected": -8.493918418884277, | |
| "loss": 0.0, | |
| "rewards/accuracies": 0.48124998807907104, | |
| "rewards/chosen": -8.420356750488281, | |
| "rewards/margins": 0.07356245070695877, | |
| "rewards/rejected": -8.493918418884277, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.1884323475529966, | |
| "grad_norm": 0.26065093480528173, | |
| "learning_rate": 4.882681251368548e-07, | |
| "logits/chosen": -1.411726713180542, | |
| "logits/rejected": -1.4245941638946533, | |
| "logps/chosen": -8.424099922180176, | |
| "logps/rejected": -8.504185676574707, | |
| "loss": -0.0001, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": -8.424099922180176, | |
| "rewards/margins": 0.08008553832769394, | |
| "rewards/rejected": -8.504185676574707, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.19890081130594087, | |
| "grad_norm": 0.13775483175577152, | |
| "learning_rate": 4.853377929214243e-07, | |
| "logits/chosen": -1.4114550352096558, | |
| "logits/rejected": -1.4067610502243042, | |
| "logps/chosen": -8.460356712341309, | |
| "logps/rejected": -8.213418960571289, | |
| "loss": 0.0002, | |
| "rewards/accuracies": 0.3812499940395355, | |
| "rewards/chosen": -8.460356712341309, | |
| "rewards/margins": -0.24693819880485535, | |
| "rewards/rejected": -8.213418960571289, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.2093692750588851, | |
| "grad_norm": 0.0669599374264611, | |
| "learning_rate": 4.820919832540181e-07, | |
| "logits/chosen": -1.4511619806289673, | |
| "logits/rejected": -1.4595460891723633, | |
| "logps/chosen": -8.38967514038086, | |
| "logps/rejected": -8.327149391174316, | |
| "loss": 0.0005, | |
| "rewards/accuracies": 0.4625000059604645, | |
| "rewards/chosen": -8.38967514038086, | |
| "rewards/margins": -0.06252529472112656, | |
| "rewards/rejected": -8.327149391174316, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.21983773881182936, | |
| "grad_norm": 0.03744091461066133, | |
| "learning_rate": 4.785350472409791e-07, | |
| "logits/chosen": -1.4548447132110596, | |
| "logits/rejected": -1.4479488134384155, | |
| "logps/chosen": -8.238606452941895, | |
| "logps/rejected": -8.342194557189941, | |
| "loss": 0.0003, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": -8.238606452941895, | |
| "rewards/margins": 0.10358722507953644, | |
| "rewards/rejected": -8.342194557189941, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.23030620256477363, | |
| "grad_norm": 0.2325346467677191, | |
| "learning_rate": 4.7467175306295647e-07, | |
| "logits/chosen": -1.4924818277359009, | |
| "logits/rejected": -1.4839593172073364, | |
| "logps/chosen": -8.65635871887207, | |
| "logps/rejected": -8.75800895690918, | |
| "loss": -0.0001, | |
| "rewards/accuracies": 0.5562499761581421, | |
| "rewards/chosen": -8.65635871887207, | |
| "rewards/margins": 0.10164938122034073, | |
| "rewards/rejected": -8.75800895690918, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.24077466631771788, | |
| "grad_norm": 0.07013048443895094, | |
| "learning_rate": 4.70507279583015e-07, | |
| "logits/chosen": -1.4494998455047607, | |
| "logits/rejected": -1.4506022930145264, | |
| "logps/chosen": -8.537219047546387, | |
| "logps/rejected": -8.694546699523926, | |
| "loss": 0.0002, | |
| "rewards/accuracies": 0.53125, | |
| "rewards/chosen": -8.537219047546387, | |
| "rewards/margins": 0.1573261320590973, | |
| "rewards/rejected": -8.694546699523926, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.2512431300706621, | |
| "grad_norm": 0.13941104681218558, | |
| "learning_rate": 4.6604720940421207e-07, | |
| "logits/chosen": -1.4503843784332275, | |
| "logits/rejected": -1.4492030143737793, | |
| "logps/chosen": -8.896661758422852, | |
| "logps/rejected": -8.842092514038086, | |
| "loss": 0.0006, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -8.896661758422852, | |
| "rewards/margins": -0.054569344967603683, | |
| "rewards/rejected": -8.842092514038086, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.26171159382360637, | |
| "grad_norm": 0.01588215105344238, | |
| "learning_rate": 4.612975213859487e-07, | |
| "logits/chosen": -1.5376265048980713, | |
| "logits/rejected": -1.5326207876205444, | |
| "logps/chosen": -9.327282905578613, | |
| "logps/rejected": -9.20351505279541, | |
| "loss": 0.0004, | |
| "rewards/accuracies": 0.46875, | |
| "rewards/chosen": -9.327282905578613, | |
| "rewards/margins": -0.12376775592565536, | |
| "rewards/rejected": -9.20351505279541, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.2721800575765506, | |
| "grad_norm": 0.060048076521496456, | |
| "learning_rate": 4.5626458262912735e-07, | |
| "logits/chosen": -1.560910701751709, | |
| "logits/rejected": -1.5665909051895142, | |
| "logps/chosen": -9.537469863891602, | |
| "logps/rejected": -9.556791305541992, | |
| "loss": -0.0001, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -9.537469863891602, | |
| "rewards/margins": 0.019321158528327942, | |
| "rewards/rejected": -9.556791305541992, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.2826485213294949, | |
| "grad_norm": 0.010696845716364521, | |
| "learning_rate": 4.5095513994085974e-07, | |
| "logits/chosen": -1.5800793170928955, | |
| "logits/rejected": -1.582044243812561, | |
| "logps/chosen": -9.630374908447266, | |
| "logps/rejected": -9.887577056884766, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 0.543749988079071, | |
| "rewards/chosen": -9.630374908447266, | |
| "rewards/margins": 0.2572011351585388, | |
| "rewards/rejected": -9.887577056884766, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.29311698508243916, | |
| "grad_norm": 0.03319847693750231, | |
| "learning_rate": 4.453763107901675e-07, | |
| "logits/chosen": -1.622972846031189, | |
| "logits/rejected": -1.6185715198516846, | |
| "logps/chosen": -9.712789535522461, | |
| "logps/rejected": -9.67750358581543, | |
| "loss": 0.0, | |
| "rewards/accuracies": 0.44999998807907104, | |
| "rewards/chosen": -9.712789535522461, | |
| "rewards/margins": -0.03528591990470886, | |
| "rewards/rejected": -9.67750358581543, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.3035854488353834, | |
| "grad_norm": 0.03281163505651298, | |
| "learning_rate": 4.395355737667985e-07, | |
| "logits/chosen": -1.6432979106903076, | |
| "logits/rejected": -1.6470788717269897, | |
| "logps/chosen": -9.864433288574219, | |
| "logps/rejected": -9.815264701843262, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 0.518750011920929, | |
| "rewards/chosen": -9.864433288574219, | |
| "rewards/margins": -0.049169473350048065, | |
| "rewards/rejected": -9.815264701843262, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.31405391258832765, | |
| "grad_norm": 0.023094845909487956, | |
| "learning_rate": 4.3344075855595097e-07, | |
| "logits/chosen": -1.6595284938812256, | |
| "logits/rejected": -1.669385552406311, | |
| "logps/chosen": -9.924379348754883, | |
| "logps/rejected": -9.930808067321777, | |
| "loss": -0.0003, | |
| "rewards/accuracies": 0.41874998807907104, | |
| "rewards/chosen": -9.924379348754883, | |
| "rewards/margins": 0.006428956985473633, | |
| "rewards/rejected": -9.930808067321777, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.3245223763412719, | |
| "grad_norm": 0.1908026987463115, | |
| "learning_rate": 4.271000354423425e-07, | |
| "logits/chosen": -1.6287696361541748, | |
| "logits/rejected": -1.6056560277938843, | |
| "logps/chosen": -9.800788879394531, | |
| "logps/rejected": -9.611404418945312, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 0.4312500059604645, | |
| "rewards/chosen": -9.800788879394531, | |
| "rewards/margins": -0.1893850713968277, | |
| "rewards/rejected": -9.611404418945312, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.33499084009421615, | |
| "grad_norm": 0.03202775302677848, | |
| "learning_rate": 4.2052190435769554e-07, | |
| "logits/chosen": -1.6098477840423584, | |
| "logits/rejected": -1.6122195720672607, | |
| "logps/chosen": -9.803987503051758, | |
| "logps/rejected": -9.737637519836426, | |
| "loss": 0.0, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": -9.803987503051758, | |
| "rewards/margins": -0.06634993851184845, | |
| "rewards/rejected": -9.737637519836426, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.34545930384716045, | |
| "grad_norm": 0.07232776561687705, | |
| "learning_rate": 4.137151834863213e-07, | |
| "logits/chosen": -1.65714430809021, | |
| "logits/rejected": -1.6471843719482422, | |
| "logps/chosen": -9.869422912597656, | |
| "logps/rejected": -9.760724067687988, | |
| "loss": 0.0003, | |
| "rewards/accuracies": 0.4625000059604645, | |
| "rewards/chosen": -9.869422912597656, | |
| "rewards/margins": -0.10870015621185303, | |
| "rewards/rejected": -9.760724067687988, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.3559277676001047, | |
| "grad_norm": 0.024726012832997274, | |
| "learning_rate": 4.0668899744407567e-07, | |
| "logits/chosen": -1.6514545679092407, | |
| "logits/rejected": -1.6379756927490234, | |
| "logps/chosen": -10.327203750610352, | |
| "logps/rejected": -10.077325820922852, | |
| "loss": 0.0, | |
| "rewards/accuracies": 0.4312500059604645, | |
| "rewards/chosen": -10.327203750610352, | |
| "rewards/margins": -0.24987690150737762, | |
| "rewards/rejected": -10.077325820922852, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.36639623135304894, | |
| "grad_norm": 0.15695534318195675, | |
| "learning_rate": 3.994527650465352e-07, | |
| "logits/chosen": -1.6916990280151367, | |
| "logits/rejected": -1.6754090785980225, | |
| "logps/chosen": -10.47531795501709, | |
| "logps/rejected": -10.233348846435547, | |
| "loss": -0.0001, | |
| "rewards/accuracies": 0.4000000059604645, | |
| "rewards/chosen": -10.47531795501709, | |
| "rewards/margins": -0.24196887016296387, | |
| "rewards/rejected": -10.233348846435547, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.3768646951059932, | |
| "grad_norm": 0.055831382164290344, | |
| "learning_rate": 3.920161866827889e-07, | |
| "logits/chosen": -1.7216554880142212, | |
| "logits/rejected": -1.7197192907333374, | |
| "logps/chosen": -10.651076316833496, | |
| "logps/rejected": -10.578369140625, | |
| "loss": -0.0, | |
| "rewards/accuracies": 0.4625000059604645, | |
| "rewards/chosen": -10.651076316833496, | |
| "rewards/margins": -0.07270796597003937, | |
| "rewards/rejected": -10.578369140625, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.38733315885893743, | |
| "grad_norm": 0.005192915052039036, | |
| "learning_rate": 3.8438923131177237e-07, | |
| "logits/chosen": -1.7085365056991577, | |
| "logits/rejected": -1.695673942565918, | |
| "logps/chosen": -10.687432289123535, | |
| "logps/rejected": -10.502934455871582, | |
| "loss": 0.0, | |
| "rewards/accuracies": 0.45625001192092896, | |
| "rewards/chosen": -10.687432289123535, | |
| "rewards/margins": -0.18449781835079193, | |
| "rewards/rejected": -10.502934455871582, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.39780162261188173, | |
| "grad_norm": 0.00386703700543382, | |
| "learning_rate": 3.765821230985757e-07, | |
| "logits/chosen": -1.7206989526748657, | |
| "logits/rejected": -1.7167381048202515, | |
| "logps/chosen": -10.415846824645996, | |
| "logps/rejected": -10.514742851257324, | |
| "loss": -0.0001, | |
| "rewards/accuracies": 0.48124998807907104, | |
| "rewards/chosen": -10.415846824645996, | |
| "rewards/margins": 0.09889563173055649, | |
| "rewards/rejected": -10.514742851257324, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.408270086364826, | |
| "grad_norm": 0.018906277546872338, | |
| "learning_rate": 3.6860532770864005e-07, | |
| "logits/chosen": -1.6922369003295898, | |
| "logits/rejected": -1.681626558303833, | |
| "logps/chosen": -10.679250717163086, | |
| "logps/rejected": -10.37684154510498, | |
| "loss": 0.0, | |
| "rewards/accuracies": 0.44999998807907104, | |
| "rewards/chosen": -10.679250717163086, | |
| "rewards/margins": -0.3024083971977234, | |
| "rewards/rejected": -10.37684154510498, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.4187385501177702, | |
| "grad_norm": 0.04569629831070709, | |
| "learning_rate": 3.604695382782159e-07, | |
| "logits/chosen": -1.7120405435562134, | |
| "logits/rejected": -1.7134020328521729, | |
| "logps/chosen": -10.352312088012695, | |
| "logps/rejected": -10.341361999511719, | |
| "loss": 0.0, | |
| "rewards/accuracies": 0.44999998807907104, | |
| "rewards/chosen": -10.352312088012695, | |
| "rewards/margins": -0.010949483141303062, | |
| "rewards/rejected": -10.341361999511719, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.42920701387071447, | |
| "grad_norm": 0.01589336857340258, | |
| "learning_rate": 3.5218566107988867e-07, | |
| "logits/chosen": -1.6780574321746826, | |
| "logits/rejected": -1.6641277074813843, | |
| "logps/chosen": -10.600610733032227, | |
| "logps/rejected": -10.596931457519531, | |
| "loss": -0.0, | |
| "rewards/accuracies": 0.46875, | |
| "rewards/chosen": -10.600610733032227, | |
| "rewards/margins": -0.003679597284644842, | |
| "rewards/rejected": -10.596931457519531, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.4396754776236587, | |
| "grad_norm": 0.017049353993959408, | |
| "learning_rate": 3.4376480090239047e-07, | |
| "logits/chosen": -1.6822540760040283, | |
| "logits/rejected": -1.6784160137176514, | |
| "logps/chosen": -10.497976303100586, | |
| "logps/rejected": -10.553048133850098, | |
| "loss": -0.0, | |
| "rewards/accuracies": 0.44999998807907104, | |
| "rewards/chosen": -10.497976303100586, | |
| "rewards/margins": 0.0550723597407341, | |
| "rewards/rejected": -10.553048133850098, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.45014394137660296, | |
| "grad_norm": 0.2539481968464961, | |
| "learning_rate": 3.3521824616429284e-07, | |
| "logits/chosen": -1.6765304803848267, | |
| "logits/rejected": -1.68255615234375, | |
| "logps/chosen": -10.369840621948242, | |
| "logps/rejected": -10.547749519348145, | |
| "loss": -0.0002, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": -10.369840621948242, | |
| "rewards/margins": 0.17790831625461578, | |
| "rewards/rejected": -10.547749519348145, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.46061240512954726, | |
| "grad_norm": 0.03490868784910135, | |
| "learning_rate": 3.265574537815398e-07, | |
| "logits/chosen": -1.7063719034194946, | |
| "logits/rejected": -1.702691674232483, | |
| "logps/chosen": -10.227754592895508, | |
| "logps/rejected": -10.175545692443848, | |
| "loss": -0.0, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": -10.227754592895508, | |
| "rewards/margins": -0.05220963433384895, | |
| "rewards/rejected": -10.175545692443848, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.4710808688824915, | |
| "grad_norm": 0.02089161978800978, | |
| "learning_rate": 3.1779403380910425e-07, | |
| "logits/chosen": -1.6891885995864868, | |
| "logits/rejected": -1.6823409795761108, | |
| "logps/chosen": -9.965743064880371, | |
| "logps/rejected": -10.085700035095215, | |
| "loss": -0.0001, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": -9.965743064880371, | |
| "rewards/margins": 0.11995597183704376, | |
| "rewards/rejected": -10.085700035095215, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.48154933263543576, | |
| "grad_norm": 0.042927151818770604, | |
| "learning_rate": 3.0893973387735683e-07, | |
| "logits/chosen": -1.6752849817276, | |
| "logits/rejected": -1.6569786071777344, | |
| "logps/chosen": -10.218798637390137, | |
| "logps/rejected": -10.10997200012207, | |
| "loss": -0.0004, | |
| "rewards/accuracies": 0.44999998807907104, | |
| "rewards/chosen": -10.218798637390137, | |
| "rewards/margins": -0.1088268980383873, | |
| "rewards/rejected": -10.10997200012207, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.49201779638838, | |
| "grad_norm": 0.055950224699650504, | |
| "learning_rate": 3.000064234440111e-07, | |
| "logits/chosen": -1.6750266551971436, | |
| "logits/rejected": -1.6845659017562866, | |
| "logps/chosen": -9.966501235961914, | |
| "logps/rejected": -10.05066967010498, | |
| "loss": -0.0, | |
| "rewards/accuracies": 0.48124998807907104, | |
| "rewards/chosen": -9.966501235961914, | |
| "rewards/margins": 0.08416947722434998, | |
| "rewards/rejected": -10.05066967010498, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.5024862601413242, | |
| "grad_norm": 0.10458741539165252, | |
| "learning_rate": 2.910060778827554e-07, | |
| "logits/chosen": -1.6746679544448853, | |
| "logits/rejected": -1.6765581369400024, | |
| "logps/chosen": -10.034064292907715, | |
| "logps/rejected": -9.960668563842773, | |
| "loss": -0.0003, | |
| "rewards/accuracies": 0.4437499940395355, | |
| "rewards/chosen": -10.034064292907715, | |
| "rewards/margins": -0.07339614629745483, | |
| "rewards/rejected": -9.960668563842773, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.5129547238942685, | |
| "grad_norm": 0.14020310226910793, | |
| "learning_rate": 2.8195076242990116e-07, | |
| "logits/chosen": -1.656341314315796, | |
| "logits/rejected": -1.6607738733291626, | |
| "logps/chosen": -9.828956604003906, | |
| "logps/rejected": -9.777771949768066, | |
| "loss": 0.0, | |
| "rewards/accuracies": 0.5062500238418579, | |
| "rewards/chosen": -9.828956604003906, | |
| "rewards/margins": -0.05118449404835701, | |
| "rewards/rejected": -9.777771949768066, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.5234231876472127, | |
| "grad_norm": 0.029712092586844915, | |
| "learning_rate": 2.7285261601056697e-07, | |
| "logits/chosen": -1.651460886001587, | |
| "logits/rejected": -1.6511856317520142, | |
| "logps/chosen": -9.723172187805176, | |
| "logps/rejected": -9.84160327911377, | |
| "loss": 0.0005, | |
| "rewards/accuracies": 0.518750011920929, | |
| "rewards/chosen": -9.723172187805176, | |
| "rewards/margins": 0.11843116581439972, | |
| "rewards/rejected": -9.84160327911377, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.533891651400157, | |
| "grad_norm": 0.03206118746185812, | |
| "learning_rate": 2.6372383496608186e-07, | |
| "logits/chosen": -1.6517183780670166, | |
| "logits/rejected": -1.6432650089263916, | |
| "logps/chosen": -9.812753677368164, | |
| "logps/rejected": -9.902433395385742, | |
| "loss": -0.0, | |
| "rewards/accuracies": 0.518750011920929, | |
| "rewards/chosen": -9.812753677368164, | |
| "rewards/margins": 0.08968041837215424, | |
| "rewards/rejected": -9.902433395385742, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.5443601151531012, | |
| "grad_norm": 0.020964013644010484, | |
| "learning_rate": 2.5457665670441937e-07, | |
| "logits/chosen": -1.592605710029602, | |
| "logits/rejected": -1.5922209024429321, | |
| "logps/chosen": -9.709136962890625, | |
| "logps/rejected": -9.819464683532715, | |
| "loss": 0.0002, | |
| "rewards/accuracies": 0.4937500059604645, | |
| "rewards/chosen": -9.709136962890625, | |
| "rewards/margins": 0.11032851040363312, | |
| "rewards/rejected": -9.819464683532715, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.5548285789060455, | |
| "grad_norm": 0.042268508150251446, | |
| "learning_rate": 2.454233432955807e-07, | |
| "logits/chosen": -1.6564003229141235, | |
| "logits/rejected": -1.6685075759887695, | |
| "logps/chosen": -9.937254905700684, | |
| "logps/rejected": -10.057160377502441, | |
| "loss": -0.0, | |
| "rewards/accuracies": 0.5062500238418579, | |
| "rewards/chosen": -9.937254905700684, | |
| "rewards/margins": 0.11990515887737274, | |
| "rewards/rejected": -10.057160377502441, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.5652970426589898, | |
| "grad_norm": 0.13512185096282692, | |
| "learning_rate": 2.3627616503391812e-07, | |
| "logits/chosen": -1.680994987487793, | |
| "logits/rejected": -1.6468158960342407, | |
| "logps/chosen": -10.322751998901367, | |
| "logps/rejected": -10.258121490478516, | |
| "loss": 0.0003, | |
| "rewards/accuracies": 0.4937500059604645, | |
| "rewards/chosen": -10.322751998901367, | |
| "rewards/margins": -0.06462957710027695, | |
| "rewards/rejected": -10.258121490478516, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.575765506411934, | |
| "grad_norm": 0.012066073597164235, | |
| "learning_rate": 2.2714738398943308e-07, | |
| "logits/chosen": -1.7274888753890991, | |
| "logits/rejected": -1.7199790477752686, | |
| "logps/chosen": -10.117765426635742, | |
| "logps/rejected": -10.104223251342773, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 0.4625000059604645, | |
| "rewards/chosen": -10.117765426635742, | |
| "rewards/margins": -0.013540792278945446, | |
| "rewards/rejected": -10.104223251342773, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.5862339701648783, | |
| "grad_norm": 0.01636207441210717, | |
| "learning_rate": 2.1804923757009882e-07, | |
| "logits/chosen": -1.7347486019134521, | |
| "logits/rejected": -1.7411794662475586, | |
| "logps/chosen": -10.153875350952148, | |
| "logps/rejected": -10.280826568603516, | |
| "loss": -0.0, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": -10.153875350952148, | |
| "rewards/margins": 0.12695105373859406, | |
| "rewards/rejected": -10.280826568603516, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.5967024339178225, | |
| "grad_norm": 0.007389012661156647, | |
| "learning_rate": 2.089939221172446e-07, | |
| "logits/chosen": -1.6744228601455688, | |
| "logits/rejected": -1.6806707382202148, | |
| "logps/chosen": -10.230897903442383, | |
| "logps/rejected": -10.295170783996582, | |
| "loss": -0.0, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": -10.230897903442383, | |
| "rewards/margins": 0.06427150964736938, | |
| "rewards/rejected": -10.295170783996582, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.6071708976707668, | |
| "grad_norm": 0.11180626971526854, | |
| "learning_rate": 1.9999357655598891e-07, | |
| "logits/chosen": -1.715934157371521, | |
| "logits/rejected": -1.7268234491348267, | |
| "logps/chosen": -10.12391471862793, | |
| "logps/rejected": -10.262609481811523, | |
| "loss": -0.0003, | |
| "rewards/accuracies": 0.518750011920929, | |
| "rewards/chosen": -10.12391471862793, | |
| "rewards/margins": 0.13869477808475494, | |
| "rewards/rejected": -10.262609481811523, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.6176393614237111, | |
| "grad_norm": 0.03266198194021972, | |
| "learning_rate": 1.9106026612264315e-07, | |
| "logits/chosen": -1.744179368019104, | |
| "logits/rejected": -1.7408939599990845, | |
| "logps/chosen": -10.210718154907227, | |
| "logps/rejected": -10.17192554473877, | |
| "loss": -0.0, | |
| "rewards/accuracies": 0.5062500238418579, | |
| "rewards/chosen": -10.210718154907227, | |
| "rewards/margins": -0.0387921966612339, | |
| "rewards/rejected": -10.17192554473877, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.6281078251766553, | |
| "grad_norm": 0.043605609853355175, | |
| "learning_rate": 1.8220596619089573e-07, | |
| "logits/chosen": -1.69484543800354, | |
| "logits/rejected": -1.692705512046814, | |
| "logps/chosen": -10.109774589538574, | |
| "logps/rejected": -10.205177307128906, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": -10.109774589538574, | |
| "rewards/margins": 0.09540363401174545, | |
| "rewards/rejected": -10.205177307128906, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.6385762889295996, | |
| "grad_norm": 0.013464112956501361, | |
| "learning_rate": 1.7344254621846017e-07, | |
| "logits/chosen": -1.6985801458358765, | |
| "logits/rejected": -1.7095019817352295, | |
| "logps/chosen": -10.274127960205078, | |
| "logps/rejected": -10.364879608154297, | |
| "loss": 0.0, | |
| "rewards/accuracies": 0.48124998807907104, | |
| "rewards/chosen": -10.274127960205078, | |
| "rewards/margins": 0.09075053036212921, | |
| "rewards/rejected": -10.364879608154297, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.6490447526825438, | |
| "grad_norm": 0.030133040256006938, | |
| "learning_rate": 1.647817538357072e-07, | |
| "logits/chosen": -1.7417259216308594, | |
| "logits/rejected": -1.740393877029419, | |
| "logps/chosen": -10.28361701965332, | |
| "logps/rejected": -10.132238388061523, | |
| "loss": 0.0, | |
| "rewards/accuracies": 0.46875, | |
| "rewards/chosen": -10.28361701965332, | |
| "rewards/margins": -0.15137743949890137, | |
| "rewards/rejected": -10.132238388061523, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.6595132164354881, | |
| "grad_norm": 0.04002742762071502, | |
| "learning_rate": 1.562351990976095e-07, | |
| "logits/chosen": -1.69840407371521, | |
| "logits/rejected": -1.7028906345367432, | |
| "logps/chosen": -10.156418800354004, | |
| "logps/rejected": -10.191594123840332, | |
| "loss": -0.0007, | |
| "rewards/accuracies": 0.48124998807907104, | |
| "rewards/chosen": -10.156418800354004, | |
| "rewards/margins": 0.03517543524503708, | |
| "rewards/rejected": -10.191594123840332, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.6699816801884323, | |
| "grad_norm": 0.021724114416033388, | |
| "learning_rate": 1.478143389201113e-07, | |
| "logits/chosen": -1.7531684637069702, | |
| "logits/rejected": -1.7422311305999756, | |
| "logps/chosen": -10.416919708251953, | |
| "logps/rejected": -10.20901107788086, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 0.48124998807907104, | |
| "rewards/chosen": -10.416919708251953, | |
| "rewards/margins": -0.20790883898735046, | |
| "rewards/rejected": -10.20901107788086, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.6804501439413766, | |
| "grad_norm": 0.08104613852953838, | |
| "learning_rate": 1.3953046172178413e-07, | |
| "logits/chosen": -1.725590467453003, | |
| "logits/rejected": -1.7158119678497314, | |
| "logps/chosen": -10.29697322845459, | |
| "logps/rejected": -10.40684986114502, | |
| "loss": -0.0003, | |
| "rewards/accuracies": 0.4625000059604645, | |
| "rewards/chosen": -10.29697322845459, | |
| "rewards/margins": 0.10987655818462372, | |
| "rewards/rejected": -10.40684986114502, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.6909186076943209, | |
| "grad_norm": 0.04339077417181125, | |
| "learning_rate": 1.3139467229135998e-07, | |
| "logits/chosen": -1.7010434865951538, | |
| "logits/rejected": -1.6965417861938477, | |
| "logps/chosen": -10.263284683227539, | |
| "logps/rejected": -10.193758010864258, | |
| "loss": -0.0, | |
| "rewards/accuracies": 0.48124998807907104, | |
| "rewards/chosen": -10.263284683227539, | |
| "rewards/margins": -0.0695262923836708, | |
| "rewards/rejected": -10.193758010864258, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.7013870714472651, | |
| "grad_norm": 0.019942381418589637, | |
| "learning_rate": 1.2341787690142435e-07, | |
| "logits/chosen": -1.7109758853912354, | |
| "logits/rejected": -1.7116413116455078, | |
| "logps/chosen": -10.025394439697266, | |
| "logps/rejected": -10.113506317138672, | |
| "loss": 0.0, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -10.025394439697266, | |
| "rewards/margins": 0.08811323344707489, | |
| "rewards/rejected": -10.113506317138672, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.7118555352002094, | |
| "grad_norm": 0.030458935372565502, | |
| "learning_rate": 1.1561076868822755e-07, | |
| "logits/chosen": -1.7183024883270264, | |
| "logits/rejected": -1.7164380550384521, | |
| "logps/chosen": -10.216609954833984, | |
| "logps/rejected": -10.225852966308594, | |
| "loss": 0.0, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": -10.216609954833984, | |
| "rewards/margins": 0.00924335140734911, | |
| "rewards/rejected": -10.225852966308594, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.7223239989531536, | |
| "grad_norm": 0.04400215150996701, | |
| "learning_rate": 1.0798381331721107e-07, | |
| "logits/chosen": -1.718132734298706, | |
| "logits/rejected": -1.732581377029419, | |
| "logps/chosen": -10.151108741760254, | |
| "logps/rejected": -10.323533058166504, | |
| "loss": -0.0, | |
| "rewards/accuracies": 0.4937500059604645, | |
| "rewards/chosen": -10.151108741760254, | |
| "rewards/margins": 0.17242416739463806, | |
| "rewards/rejected": -10.323533058166504, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.7327924627060979, | |
| "grad_norm": 0.03515217556644255, | |
| "learning_rate": 1.0054723495346482e-07, | |
| "logits/chosen": -1.6874383687973022, | |
| "logits/rejected": -1.6842788457870483, | |
| "logps/chosen": -10.151719093322754, | |
| "logps/rejected": -10.054476737976074, | |
| "loss": -0.0001, | |
| "rewards/accuracies": 0.4312500059604645, | |
| "rewards/chosen": -10.151719093322754, | |
| "rewards/margins": -0.09724216163158417, | |
| "rewards/rejected": -10.054476737976074, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.7432609264590422, | |
| "grad_norm": 0.021861424316864955, | |
| "learning_rate": 9.331100255592436e-08, | |
| "logits/chosen": -1.7121349573135376, | |
| "logits/rejected": -1.6940410137176514, | |
| "logps/chosen": -10.209026336669922, | |
| "logps/rejected": -9.959612846374512, | |
| "loss": 0.0002, | |
| "rewards/accuracies": 0.4124999940395355, | |
| "rewards/chosen": -10.209026336669922, | |
| "rewards/margins": -0.24941349029541016, | |
| "rewards/rejected": -9.959612846374512, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.7537293902119864, | |
| "grad_norm": 0.013506797744103302, | |
| "learning_rate": 8.628481651367875e-08, | |
| "logits/chosen": -1.6897398233413696, | |
| "logits/rejected": -1.6821842193603516, | |
| "logps/chosen": -9.989233016967773, | |
| "logps/rejected": -10.048975944519043, | |
| "loss": 0.0, | |
| "rewards/accuracies": 0.518750011920929, | |
| "rewards/chosen": -9.989233016967773, | |
| "rewards/margins": 0.05974303558468819, | |
| "rewards/rejected": -10.048975944519043, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.7641978539649307, | |
| "grad_norm": 0.07239679867383533, | |
| "learning_rate": 7.947809564230445e-08, | |
| "logits/chosen": -1.6575393676757812, | |
| "logits/rejected": -1.6490436792373657, | |
| "logps/chosen": -10.096014022827148, | |
| "logps/rejected": -10.024383544921875, | |
| "loss": 0.0, | |
| "rewards/accuracies": 0.4937500059604645, | |
| "rewards/chosen": -10.096014022827148, | |
| "rewards/margins": -0.07163101434707642, | |
| "rewards/rejected": -10.024383544921875, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.7746663177178749, | |
| "grad_norm": 0.08043566497276283, | |
| "learning_rate": 7.289996455765748e-08, | |
| "logits/chosen": -1.730115294456482, | |
| "logits/rejected": -1.7217315435409546, | |
| "logps/chosen": -10.174753189086914, | |
| "logps/rejected": -10.153951644897461, | |
| "loss": 0.0, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": -10.174753189086914, | |
| "rewards/margins": -0.02080065943300724, | |
| "rewards/rejected": -10.153951644897461, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.7851347814708192, | |
| "grad_norm": 0.018294513390755808, | |
| "learning_rate": 6.655924144404906e-08, | |
| "logits/chosen": -1.7276290655136108, | |
| "logits/rejected": -1.728005051612854, | |
| "logps/chosen": -10.183911323547363, | |
| "logps/rejected": -10.19709587097168, | |
| "loss": 0.0, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -10.183911323547363, | |
| "rewards/margins": 0.01318590622395277, | |
| "rewards/rejected": -10.19709587097168, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.7956032452237635, | |
| "grad_norm": 0.006815837156718109, | |
| "learning_rate": 6.046442623320145e-08, | |
| "logits/chosen": -1.7046406269073486, | |
| "logits/rejected": -1.6621993780136108, | |
| "logps/chosen": -10.309165954589844, | |
| "logps/rejected": -10.365669250488281, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": -10.309165954589844, | |
| "rewards/margins": 0.056504249572753906, | |
| "rewards/rejected": -10.365669250488281, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.8060717089767077, | |
| "grad_norm": 0.030388876590904525, | |
| "learning_rate": 5.4623689209832484e-08, | |
| "logits/chosen": -1.7268683910369873, | |
| "logits/rejected": -1.7090175151824951, | |
| "logps/chosen": -10.306918144226074, | |
| "logps/rejected": -10.260132789611816, | |
| "loss": -0.0, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -10.306918144226074, | |
| "rewards/margins": -0.04678579792380333, | |
| "rewards/rejected": -10.260132789611816, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.816540172729652, | |
| "grad_norm": 0.04364016204439969, | |
| "learning_rate": 4.904486005914027e-08, | |
| "logits/chosen": -1.7389265298843384, | |
| "logits/rejected": -1.7257276773452759, | |
| "logps/chosen": -10.114287376403809, | |
| "logps/rejected": -10.103799819946289, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 0.44999998807907104, | |
| "rewards/chosen": -10.114287376403809, | |
| "rewards/margins": -0.010487285442650318, | |
| "rewards/rejected": -10.103799819946289, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.8270086364825961, | |
| "grad_norm": 0.026690398874973438, | |
| "learning_rate": 4.373541737087263e-08, | |
| "logits/chosen": -1.689626693725586, | |
| "logits/rejected": -1.6864984035491943, | |
| "logps/chosen": -9.968095779418945, | |
| "logps/rejected": -9.953658103942871, | |
| "loss": -0.0005, | |
| "rewards/accuracies": 0.5062500238418579, | |
| "rewards/chosen": -9.968095779418945, | |
| "rewards/margins": -0.014436977915465832, | |
| "rewards/rejected": -9.953658103942871, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.8374771002355405, | |
| "grad_norm": 0.023648830133579133, | |
| "learning_rate": 3.8702478614051345e-08, | |
| "logits/chosen": -1.6882154941558838, | |
| "logits/rejected": -1.672609567642212, | |
| "logps/chosen": -10.05548095703125, | |
| "logps/rejected": -9.970457077026367, | |
| "loss": -0.0002, | |
| "rewards/accuracies": 0.48124998807907104, | |
| "rewards/chosen": -10.05548095703125, | |
| "rewards/margins": -0.08502475917339325, | |
| "rewards/rejected": -9.970457077026367, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.8479455639884846, | |
| "grad_norm": 0.11775947165792712, | |
| "learning_rate": 3.3952790595787986e-08, | |
| "logits/chosen": -1.7214962244033813, | |
| "logits/rejected": -1.7271496057510376, | |
| "logps/chosen": -10.209165573120117, | |
| "logps/rejected": -10.225208282470703, | |
| "loss": -0.0002, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": -10.209165573120117, | |
| "rewards/margins": 0.01604318618774414, | |
| "rewards/rejected": -10.225208282470703, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.8584140277414289, | |
| "grad_norm": 0.016736815353629707, | |
| "learning_rate": 2.9492720416985e-08, | |
| "logits/chosen": -1.731414556503296, | |
| "logits/rejected": -1.7266390323638916, | |
| "logps/chosen": -10.035455703735352, | |
| "logps/rejected": -9.934586524963379, | |
| "loss": 0.0002, | |
| "rewards/accuracies": 0.5062500238418579, | |
| "rewards/chosen": -10.035455703735352, | |
| "rewards/margins": -0.10086920112371445, | |
| "rewards/rejected": -9.934586524963379, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.8688824914943732, | |
| "grad_norm": 0.010543453920296524, | |
| "learning_rate": 2.5328246937043525e-08, | |
| "logits/chosen": -1.7460575103759766, | |
| "logits/rejected": -1.7472984790802002, | |
| "logps/chosen": -10.30466079711914, | |
| "logps/rejected": -10.271262168884277, | |
| "loss": 0.0, | |
| "rewards/accuracies": 0.44999998807907104, | |
| "rewards/chosen": -10.30466079711914, | |
| "rewards/margins": -0.033397965133190155, | |
| "rewards/rejected": -10.271262168884277, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.8793509552473174, | |
| "grad_norm": 0.1619070822929026, | |
| "learning_rate": 2.1464952759020856e-08, | |
| "logits/chosen": -1.6861051321029663, | |
| "logits/rejected": -1.6756645441055298, | |
| "logps/chosen": -10.12368392944336, | |
| "logps/rejected": -10.014533996582031, | |
| "loss": 0.0, | |
| "rewards/accuracies": 0.44999998807907104, | |
| "rewards/chosen": -10.12368392944336, | |
| "rewards/margins": -0.10914961993694305, | |
| "rewards/rejected": -10.014533996582031, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.8898194190002617, | |
| "grad_norm": 0.009278538808287691, | |
| "learning_rate": 1.7908016745981856e-08, | |
| "logits/chosen": -1.7117290496826172, | |
| "logits/rejected": -1.7069095373153687, | |
| "logps/chosen": -10.257353782653809, | |
| "logps/rejected": -10.189541816711426, | |
| "loss": 0.0, | |
| "rewards/accuracies": 0.4625000059604645, | |
| "rewards/chosen": -10.257353782653809, | |
| "rewards/margins": -0.06781347841024399, | |
| "rewards/rejected": -10.189541816711426, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.9002878827532059, | |
| "grad_norm": 0.02721516902148881, | |
| "learning_rate": 1.4662207078575684e-08, | |
| "logits/chosen": -1.717145562171936, | |
| "logits/rejected": -1.7124665975570679, | |
| "logps/chosen": -10.076754570007324, | |
| "logps/rejected": -10.053953170776367, | |
| "loss": -0.0002, | |
| "rewards/accuracies": 0.45625001192092896, | |
| "rewards/chosen": -10.076754570007324, | |
| "rewards/margins": -0.022800538688898087, | |
| "rewards/rejected": -10.053953170776367, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.9107563465061502, | |
| "grad_norm": 0.17249350484919276, | |
| "learning_rate": 1.1731874863145142e-08, | |
| "logits/chosen": -1.74248468875885, | |
| "logits/rejected": -1.7478322982788086, | |
| "logps/chosen": -10.22671890258789, | |
| "logps/rejected": -10.209540367126465, | |
| "loss": 0.0007, | |
| "rewards/accuracies": 0.4937500059604645, | |
| "rewards/chosen": -10.22671890258789, | |
| "rewards/margins": -0.017179716378450394, | |
| "rewards/rejected": -10.209540367126465, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.9212248102590945, | |
| "grad_norm": 0.059902683583967914, | |
| "learning_rate": 9.12094829893642e-09, | |
| "logits/chosen": -1.6926848888397217, | |
| "logits/rejected": -1.6890709400177002, | |
| "logps/chosen": -10.345104217529297, | |
| "logps/rejected": -10.262327194213867, | |
| "loss": 0.0004, | |
| "rewards/accuracies": 0.44999998807907104, | |
| "rewards/chosen": -10.345104217529297, | |
| "rewards/margins": -0.08277805149555206, | |
| "rewards/rejected": -10.262327194213867, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.9316932740120387, | |
| "grad_norm": 0.10421328261403418, | |
| "learning_rate": 6.832927412229017e-09, | |
| "logits/chosen": -1.7495588064193726, | |
| "logits/rejected": -1.7407792806625366, | |
| "logps/chosen": -10.188596725463867, | |
| "logps/rejected": -10.2090425491333, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": -10.188596725463867, | |
| "rewards/margins": 0.02044558897614479, | |
| "rewards/rejected": -10.2090425491333, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.942161737764983, | |
| "grad_norm": 0.033309080831972024, | |
| "learning_rate": 4.8708793644441086e-09, | |
| "logits/chosen": -1.6661462783813477, | |
| "logits/rejected": -1.6469885110855103, | |
| "logps/chosen": -10.289316177368164, | |
| "logps/rejected": -10.164963722229004, | |
| "loss": 0.0002, | |
| "rewards/accuracies": 0.44999998807907104, | |
| "rewards/chosen": -10.289316177368164, | |
| "rewards/margins": -0.1243540421128273, | |
| "rewards/rejected": -10.164963722229004, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.9526302015179272, | |
| "grad_norm": 0.1949041816950042, | |
| "learning_rate": 3.2374343405217884e-09, | |
| "logits/chosen": -1.7108314037322998, | |
| "logits/rejected": -1.710608720779419, | |
| "logps/chosen": -10.109529495239258, | |
| "logps/rejected": -10.343156814575195, | |
| "loss": 0.0002, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": -10.109529495239258, | |
| "rewards/margins": 0.2336268424987793, | |
| "rewards/rejected": -10.343156814575195, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.9630986652708715, | |
| "grad_norm": 0.01652601811230786, | |
| "learning_rate": 1.9347820230782295e-09, | |
| "logits/chosen": -1.7258888483047485, | |
| "logits/rejected": -1.728637933731079, | |
| "logps/chosen": -10.354926109313965, | |
| "logps/rejected": -10.41981315612793, | |
| "loss": -0.0, | |
| "rewards/accuracies": 0.48124998807907104, | |
| "rewards/chosen": -10.354926109313965, | |
| "rewards/margins": 0.06488712131977081, | |
| "rewards/rejected": -10.41981315612793, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.9735671290238157, | |
| "grad_norm": 0.12136785382119708, | |
| "learning_rate": 9.64668657069706e-10, | |
| "logits/chosen": -1.7371879816055298, | |
| "logits/rejected": -1.7365795373916626, | |
| "logps/chosen": -10.489349365234375, | |
| "logps/rejected": -10.43460750579834, | |
| "loss": -0.0003, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -10.489349365234375, | |
| "rewards/margins": -0.0547415092587471, | |
| "rewards/rejected": -10.43460750579834, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.98403559277676, | |
| "grad_norm": 0.0469433162766017, | |
| "learning_rate": 3.2839470889836627e-10, | |
| "logits/chosen": -1.7311279773712158, | |
| "logits/rejected": -1.7309757471084595, | |
| "logps/chosen": -10.06026554107666, | |
| "logps/rejected": -10.253448486328125, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": -10.06026554107666, | |
| "rewards/margins": 0.19318366050720215, | |
| "rewards/rejected": -10.253448486328125, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.9945040565297043, | |
| "grad_norm": 0.039418088806606334, | |
| "learning_rate": 2.6813123097352287e-11, | |
| "logits/chosen": -1.7103230953216553, | |
| "logits/rejected": -1.6960973739624023, | |
| "logps/chosen": -10.431560516357422, | |
| "logps/rejected": -10.349306106567383, | |
| "loss": -0.0, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": -10.431560516357422, | |
| "rewards/margins": -0.08225315064191818, | |
| "rewards/rejected": -10.349306106567383, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.998691442030882, | |
| "step": 477, | |
| "total_flos": 0.0, | |
| "train_loss": 1.550099917438415e-05, | |
| "train_runtime": 8183.0083, | |
| "train_samples_per_second": 7.471, | |
| "train_steps_per_second": 0.058 | |
| } | |
| ], | |
| "logging_steps": 5, | |
| "max_steps": 477, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 1000000, | |
| "total_flos": 0.0, | |
| "train_batch_size": 2, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |