|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 1.0, |
|
"eval_steps": 100, |
|
"global_step": 391, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.0, |
|
"grad_norm": 421.7748930851862, |
|
"learning_rate": 2.5e-09, |
|
"logits/chosen": -4.623842239379883, |
|
"logits/rejected": -4.85917854309082, |
|
"logps/chosen": -239.31422424316406, |
|
"logps/rejected": -207.56365966796875, |
|
"loss": 0.693, |
|
"rewards/accuracies": 0.0, |
|
"rewards/chosen": 0.0, |
|
"rewards/margins": 0.0, |
|
"rewards/rejected": 0.0, |
|
"step": 1 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"grad_norm": 395.00460150231794, |
|
"learning_rate": 2.5e-08, |
|
"logits/chosen": -4.33427619934082, |
|
"logits/rejected": -4.64423942565918, |
|
"logps/chosen": -265.1573791503906, |
|
"logps/rejected": -215.7504425048828, |
|
"loss": 0.693, |
|
"rewards/accuracies": 0.4444444477558136, |
|
"rewards/chosen": -0.0050499276258051395, |
|
"rewards/margins": -0.0008213710389100015, |
|
"rewards/rejected": -0.004228556528687477, |
|
"step": 10 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"grad_norm": 398.86324560087934, |
|
"learning_rate": 5e-08, |
|
"logits/chosen": -4.5085954666137695, |
|
"logits/rejected": -4.7428812980651855, |
|
"logps/chosen": -267.77325439453125, |
|
"logps/rejected": -216.7412567138672, |
|
"loss": 0.6821, |
|
"rewards/accuracies": 0.643750011920929, |
|
"rewards/chosen": 0.016667194664478302, |
|
"rewards/margins": 0.029861312359571457, |
|
"rewards/rejected": -0.01319411676377058, |
|
"step": 20 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"grad_norm": 388.1898348935921, |
|
"learning_rate": 7.5e-08, |
|
"logits/chosen": -4.594970703125, |
|
"logits/rejected": -4.776912689208984, |
|
"logps/chosen": -257.5873107910156, |
|
"logps/rejected": -215.4453582763672, |
|
"loss": 0.6195, |
|
"rewards/accuracies": 0.7875000238418579, |
|
"rewards/chosen": 0.09455005824565887, |
|
"rewards/margins": 0.18580853939056396, |
|
"rewards/rejected": -0.09125850349664688, |
|
"step": 30 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"grad_norm": 267.0875315068977, |
|
"learning_rate": 1e-07, |
|
"logits/chosen": -4.637397766113281, |
|
"logits/rejected": -4.731802940368652, |
|
"logps/chosen": -249.911865234375, |
|
"logps/rejected": -222.6023712158203, |
|
"loss": 0.4813, |
|
"rewards/accuracies": 0.8374999761581421, |
|
"rewards/chosen": 0.36873525381088257, |
|
"rewards/margins": 0.661862313747406, |
|
"rewards/rejected": -0.29312705993652344, |
|
"step": 40 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"grad_norm": 193.9339508971275, |
|
"learning_rate": 9.979985922607475e-08, |
|
"logits/chosen": -4.531577110290527, |
|
"logits/rejected": -4.7472710609436035, |
|
"logps/chosen": -265.0107116699219, |
|
"logps/rejected": -231.0427703857422, |
|
"loss": 0.3721, |
|
"rewards/accuracies": 0.893750011920929, |
|
"rewards/chosen": 0.6958638429641724, |
|
"rewards/margins": 1.2129887342453003, |
|
"rewards/rejected": -0.5171248316764832, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"grad_norm": 197.66681943180416, |
|
"learning_rate": 9.92010391574745e-08, |
|
"logits/chosen": -4.6342668533325195, |
|
"logits/rejected": -4.783431053161621, |
|
"logps/chosen": -230.5199432373047, |
|
"logps/rejected": -224.8899383544922, |
|
"loss": 0.281, |
|
"rewards/accuracies": 0.875, |
|
"rewards/chosen": 1.0602517127990723, |
|
"rewards/margins": 2.0471043586730957, |
|
"rewards/rejected": -0.9868523478507996, |
|
"step": 60 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"grad_norm": 162.90653777538404, |
|
"learning_rate": 9.820833372667812e-08, |
|
"logits/chosen": -4.519268989562988, |
|
"logits/rejected": -4.774823188781738, |
|
"logps/chosen": -240.6663818359375, |
|
"logps/rejected": -230.677490234375, |
|
"loss": 0.2583, |
|
"rewards/accuracies": 0.9312499761581421, |
|
"rewards/chosen": 1.247557520866394, |
|
"rewards/margins": 2.539060592651367, |
|
"rewards/rejected": -1.2915031909942627, |
|
"step": 70 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"grad_norm": 139.86229845766394, |
|
"learning_rate": 9.682969016701356e-08, |
|
"logits/chosen": -4.514620780944824, |
|
"logits/rejected": -4.760144233703613, |
|
"logps/chosen": -247.26364135742188, |
|
"logps/rejected": -244.28866577148438, |
|
"loss": 0.2248, |
|
"rewards/accuracies": 0.8999999761581421, |
|
"rewards/chosen": 1.3956347703933716, |
|
"rewards/margins": 2.7926278114318848, |
|
"rewards/rejected": -1.3969930410385132, |
|
"step": 80 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"grad_norm": 208.01416582057018, |
|
"learning_rate": 9.507614539004081e-08, |
|
"logits/chosen": -4.604180812835693, |
|
"logits/rejected": -4.849738597869873, |
|
"logps/chosen": -236.37899780273438, |
|
"logps/rejected": -221.75564575195312, |
|
"loss": 0.2139, |
|
"rewards/accuracies": 0.9312499761581421, |
|
"rewards/chosen": 1.4833731651306152, |
|
"rewards/margins": 3.3759377002716064, |
|
"rewards/rejected": -1.8925644159317017, |
|
"step": 90 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"grad_norm": 168.83388514254116, |
|
"learning_rate": 9.296173762811083e-08, |
|
"logits/chosen": -4.470773220062256, |
|
"logits/rejected": -4.786902904510498, |
|
"logps/chosen": -240.86477661132812, |
|
"logps/rejected": -247.6592559814453, |
|
"loss": 0.216, |
|
"rewards/accuracies": 0.893750011920929, |
|
"rewards/chosen": 1.5853464603424072, |
|
"rewards/margins": 3.6403656005859375, |
|
"rewards/rejected": -2.0550191402435303, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"eval_logits/chosen": -4.555809497833252, |
|
"eval_logits/rejected": -4.771756172180176, |
|
"eval_logps/chosen": -391.56207275390625, |
|
"eval_logps/rejected": -513.7159423828125, |
|
"eval_loss": 1.1906155347824097, |
|
"eval_rewards/accuracies": 0.3984375, |
|
"eval_rewards/chosen": -0.10876215249300003, |
|
"eval_rewards/margins": -0.575933575630188, |
|
"eval_rewards/rejected": 0.46717149019241333, |
|
"eval_runtime": 97.8557, |
|
"eval_samples_per_second": 20.438, |
|
"eval_steps_per_second": 0.327, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.28, |
|
"grad_norm": 158.8626687509829, |
|
"learning_rate": 9.050339404945832e-08, |
|
"logits/chosen": -4.515485763549805, |
|
"logits/rejected": -4.795956134796143, |
|
"logps/chosen": -232.4200439453125, |
|
"logps/rejected": -236.90969848632812, |
|
"loss": 0.1985, |
|
"rewards/accuracies": 0.925000011920929, |
|
"rewards/chosen": 1.6765697002410889, |
|
"rewards/margins": 3.698953628540039, |
|
"rewards/rejected": -2.02238392829895, |
|
"step": 110 |
|
}, |
|
{ |
|
"epoch": 0.31, |
|
"grad_norm": 135.50801896928647, |
|
"learning_rate": 8.77207952455395e-08, |
|
"logits/chosen": -4.454637050628662, |
|
"logits/rejected": -4.722542762756348, |
|
"logps/chosen": -256.7890930175781, |
|
"logps/rejected": -248.82601928710938, |
|
"loss": 0.1975, |
|
"rewards/accuracies": 0.9624999761581421, |
|
"rewards/chosen": 1.9200481176376343, |
|
"rewards/margins": 3.9417896270751953, |
|
"rewards/rejected": -2.0217413902282715, |
|
"step": 120 |
|
}, |
|
{ |
|
"epoch": 0.33, |
|
"grad_norm": 154.49491996909583, |
|
"learning_rate": 8.463621767547997e-08, |
|
"logits/chosen": -4.527877330780029, |
|
"logits/rejected": -4.81399393081665, |
|
"logps/chosen": -241.6123809814453, |
|
"logps/rejected": -239.6664276123047, |
|
"loss": 0.1993, |
|
"rewards/accuracies": 0.8999999761581421, |
|
"rewards/chosen": 1.683153748512268, |
|
"rewards/margins": 4.116929054260254, |
|
"rewards/rejected": -2.4337756633758545, |
|
"step": 130 |
|
}, |
|
{ |
|
"epoch": 0.36, |
|
"grad_norm": 161.00485250742125, |
|
"learning_rate": 8.127435532896387e-08, |
|
"logits/chosen": -4.5556840896606445, |
|
"logits/rejected": -4.849120140075684, |
|
"logps/chosen": -268.4659118652344, |
|
"logps/rejected": -258.8057556152344, |
|
"loss": 0.1874, |
|
"rewards/accuracies": 0.925000011920929, |
|
"rewards/chosen": 1.5729143619537354, |
|
"rewards/margins": 4.236142158508301, |
|
"rewards/rejected": -2.6632275581359863, |
|
"step": 140 |
|
}, |
|
{ |
|
"epoch": 0.38, |
|
"grad_norm": 157.59104792116975, |
|
"learning_rate": 7.766212203526569e-08, |
|
"logits/chosen": -4.545467853546143, |
|
"logits/rejected": -4.822617053985596, |
|
"logps/chosen": -235.0808563232422, |
|
"logps/rejected": -246.62606811523438, |
|
"loss": 0.1878, |
|
"rewards/accuracies": 0.949999988079071, |
|
"rewards/chosen": 1.7337404489517212, |
|
"rewards/margins": 4.484419822692871, |
|
"rewards/rejected": -2.7506790161132812, |
|
"step": 150 |
|
}, |
|
{ |
|
"epoch": 0.41, |
|
"grad_norm": 153.9528654470227, |
|
"learning_rate": 7.382843600106538e-08, |
|
"logits/chosen": -4.591027736663818, |
|
"logits/rejected": -4.801583290100098, |
|
"logps/chosen": -234.0863494873047, |
|
"logps/rejected": -241.58761596679688, |
|
"loss": 0.1687, |
|
"rewards/accuracies": 0.9375, |
|
"rewards/chosen": 1.7370154857635498, |
|
"rewards/margins": 4.334869861602783, |
|
"rewards/rejected": -2.5978541374206543, |
|
"step": 160 |
|
}, |
|
{ |
|
"epoch": 0.43, |
|
"grad_norm": 134.86505617626943, |
|
"learning_rate": 6.980398830195784e-08, |
|
"logits/chosen": -4.457674980163574, |
|
"logits/rejected": -4.7707014083862305, |
|
"logps/chosen": -239.03787231445312, |
|
"logps/rejected": -248.50393676757812, |
|
"loss": 0.1647, |
|
"rewards/accuracies": 0.9437500238418579, |
|
"rewards/chosen": 2.203636407852173, |
|
"rewards/margins": 5.073732852935791, |
|
"rewards/rejected": -2.870095729827881, |
|
"step": 170 |
|
}, |
|
{ |
|
"epoch": 0.46, |
|
"grad_norm": 134.91302963185322, |
|
"learning_rate": 6.562099718102787e-08, |
|
"logits/chosen": -4.556507110595703, |
|
"logits/rejected": -4.814910411834717, |
|
"logps/chosen": -217.7527618408203, |
|
"logps/rejected": -222.974365234375, |
|
"loss": 0.1885, |
|
"rewards/accuracies": 0.9375, |
|
"rewards/chosen": 1.9110908508300781, |
|
"rewards/margins": 4.606502532958984, |
|
"rewards/rejected": -2.695411443710327, |
|
"step": 180 |
|
}, |
|
{ |
|
"epoch": 0.49, |
|
"grad_norm": 164.38739923975325, |
|
"learning_rate": 6.131295012148612e-08, |
|
"logits/chosen": -4.534274101257324, |
|
"logits/rejected": -4.6962666511535645, |
|
"logps/chosen": -241.34304809570312, |
|
"logps/rejected": -259.50860595703125, |
|
"loss": 0.1854, |
|
"rewards/accuracies": 0.8999999761581421, |
|
"rewards/chosen": 1.9549798965454102, |
|
"rewards/margins": 4.389232158660889, |
|
"rewards/rejected": -2.434252977371216, |
|
"step": 190 |
|
}, |
|
{ |
|
"epoch": 0.51, |
|
"grad_norm": 145.61398569600198, |
|
"learning_rate": 5.691433575823665e-08, |
|
"logits/chosen": -4.528763771057129, |
|
"logits/rejected": -4.724316596984863, |
|
"logps/chosen": -234.06503295898438, |
|
"logps/rejected": -241.2750701904297, |
|
"loss": 0.1925, |
|
"rewards/accuracies": 0.875, |
|
"rewards/chosen": 1.8074201345443726, |
|
"rewards/margins": 4.400271892547607, |
|
"rewards/rejected": -2.5928521156311035, |
|
"step": 200 |
|
}, |
|
{ |
|
"epoch": 0.51, |
|
"eval_logits/chosen": -4.5165863037109375, |
|
"eval_logits/rejected": -4.742655277252197, |
|
"eval_logps/chosen": -394.47015380859375, |
|
"eval_logps/rejected": -515.5682983398438, |
|
"eval_loss": 1.3562896251678467, |
|
"eval_rewards/accuracies": 0.375, |
|
"eval_rewards/chosen": -0.39956870675086975, |
|
"eval_rewards/margins": -0.6815085411071777, |
|
"eval_rewards/rejected": 0.28193986415863037, |
|
"eval_runtime": 97.991, |
|
"eval_samples_per_second": 20.41, |
|
"eval_steps_per_second": 0.327, |
|
"step": 200 |
|
}, |
|
{ |
|
"epoch": 0.54, |
|
"grad_norm": 126.33823591196861, |
|
"learning_rate": 5.2460367774593905e-08, |
|
"logits/chosen": -4.592769622802734, |
|
"logits/rejected": -4.851358413696289, |
|
"logps/chosen": -244.64755249023438, |
|
"logps/rejected": -259.65142822265625, |
|
"loss": 0.1489, |
|
"rewards/accuracies": 0.9437500238418579, |
|
"rewards/chosen": 2.1237542629241943, |
|
"rewards/margins": 5.224299907684326, |
|
"rewards/rejected": -3.10054612159729, |
|
"step": 210 |
|
}, |
|
{ |
|
"epoch": 0.56, |
|
"grad_norm": 139.63020748735394, |
|
"learning_rate": 4.798670299452925e-08, |
|
"logits/chosen": -4.435277462005615, |
|
"logits/rejected": -4.771773338317871, |
|
"logps/chosen": -242.7268524169922, |
|
"logps/rejected": -258.1542053222656, |
|
"loss": 0.1754, |
|
"rewards/accuracies": 0.9312499761581421, |
|
"rewards/chosen": 2.0084481239318848, |
|
"rewards/margins": 5.356553077697754, |
|
"rewards/rejected": -3.348104953765869, |
|
"step": 220 |
|
}, |
|
{ |
|
"epoch": 0.59, |
|
"grad_norm": 146.4580429105822, |
|
"learning_rate": 4.3529155927297226e-08, |
|
"logits/chosen": -4.511857986450195, |
|
"logits/rejected": -4.845773696899414, |
|
"logps/chosen": -241.4152374267578, |
|
"logps/rejected": -257.61944580078125, |
|
"loss": 0.1815, |
|
"rewards/accuracies": 0.9312499761581421, |
|
"rewards/chosen": 1.9553873538970947, |
|
"rewards/margins": 5.346222877502441, |
|
"rewards/rejected": -3.390836000442505, |
|
"step": 230 |
|
}, |
|
{ |
|
"epoch": 0.61, |
|
"grad_norm": 120.36905618473185, |
|
"learning_rate": 3.9123412049691636e-08, |
|
"logits/chosen": -4.488260269165039, |
|
"logits/rejected": -4.767331123352051, |
|
"logps/chosen": -252.67538452148438, |
|
"logps/rejected": -255.2183837890625, |
|
"loss": 0.1687, |
|
"rewards/accuracies": 0.9375, |
|
"rewards/chosen": 2.014777421951294, |
|
"rewards/margins": 5.443254470825195, |
|
"rewards/rejected": -3.4284775257110596, |
|
"step": 240 |
|
}, |
|
{ |
|
"epoch": 0.64, |
|
"grad_norm": 143.11323504040695, |
|
"learning_rate": 3.480474212128766e-08, |
|
"logits/chosen": -4.618422031402588, |
|
"logits/rejected": -4.878515243530273, |
|
"logps/chosen": -231.21737670898438, |
|
"logps/rejected": -236.189208984375, |
|
"loss": 0.1614, |
|
"rewards/accuracies": 0.949999988079071, |
|
"rewards/chosen": 1.7484108209609985, |
|
"rewards/margins": 4.606410503387451, |
|
"rewards/rejected": -2.857999324798584, |
|
"step": 250 |
|
}, |
|
{ |
|
"epoch": 0.66, |
|
"grad_norm": 193.1124896787253, |
|
"learning_rate": 3.060771981975726e-08, |
|
"logits/chosen": -4.48922872543335, |
|
"logits/rejected": -4.790060997009277, |
|
"logps/chosen": -234.9071502685547, |
|
"logps/rejected": -258.86639404296875, |
|
"loss": 0.1653, |
|
"rewards/accuracies": 0.949999988079071, |
|
"rewards/chosen": 1.9141381978988647, |
|
"rewards/margins": 5.825932025909424, |
|
"rewards/rejected": -3.9117939472198486, |
|
"step": 260 |
|
}, |
|
{ |
|
"epoch": 0.69, |
|
"grad_norm": 194.5027032928207, |
|
"learning_rate": 2.6565944956764818e-08, |
|
"logits/chosen": -4.580440044403076, |
|
"logits/rejected": -4.814294815063477, |
|
"logps/chosen": -242.62802124023438, |
|
"logps/rejected": -251.90444946289062, |
|
"loss": 0.1739, |
|
"rewards/accuracies": 0.9375, |
|
"rewards/chosen": 1.84052312374115, |
|
"rewards/margins": 5.5284624099731445, |
|
"rewards/rejected": -3.687939405441284, |
|
"step": 270 |
|
}, |
|
{ |
|
"epoch": 0.72, |
|
"grad_norm": 134.789590992645, |
|
"learning_rate": 2.2711774490274766e-08, |
|
"logits/chosen": -4.531357765197754, |
|
"logits/rejected": -4.778365135192871, |
|
"logps/chosen": -244.4646453857422, |
|
"logps/rejected": -277.8075866699219, |
|
"loss": 0.148, |
|
"rewards/accuracies": 0.96875, |
|
"rewards/chosen": 2.004220485687256, |
|
"rewards/margins": 5.491039276123047, |
|
"rewards/rejected": -3.48681902885437, |
|
"step": 280 |
|
}, |
|
{ |
|
"epoch": 0.74, |
|
"grad_norm": 281.0472172318649, |
|
"learning_rate": 1.9076063486687256e-08, |
|
"logits/chosen": -4.403564453125, |
|
"logits/rejected": -4.759608268737793, |
|
"logps/chosen": -251.33419799804688, |
|
"logps/rejected": -248.35458374023438, |
|
"loss": 0.1633, |
|
"rewards/accuracies": 0.925000011920929, |
|
"rewards/chosen": 2.1274170875549316, |
|
"rewards/margins": 5.399934768676758, |
|
"rewards/rejected": -3.272517681121826, |
|
"step": 290 |
|
}, |
|
{ |
|
"epoch": 0.77, |
|
"grad_norm": 153.20971602377742, |
|
"learning_rate": 1.5687918106563324e-08, |
|
"logits/chosen": -4.5167622566223145, |
|
"logits/rejected": -4.735228061676025, |
|
"logps/chosen": -232.205322265625, |
|
"logps/rejected": -250.81884765625, |
|
"loss": 0.179, |
|
"rewards/accuracies": 0.8812500238418579, |
|
"rewards/chosen": 2.1202826499938965, |
|
"rewards/margins": 5.605620861053467, |
|
"rewards/rejected": -3.4853386878967285, |
|
"step": 300 |
|
}, |
|
{ |
|
"epoch": 0.77, |
|
"eval_logits/chosen": -4.521942615509033, |
|
"eval_logits/rejected": -4.748414039611816, |
|
"eval_logps/chosen": -397.169189453125, |
|
"eval_logps/rejected": -518.8555297851562, |
|
"eval_loss": 1.3649907112121582, |
|
"eval_rewards/accuracies": 0.3828125, |
|
"eval_rewards/chosen": -0.6694751977920532, |
|
"eval_rewards/margins": -0.6226927638053894, |
|
"eval_rewards/rejected": -0.04678243398666382, |
|
"eval_runtime": 97.9822, |
|
"eval_samples_per_second": 20.412, |
|
"eval_steps_per_second": 0.327, |
|
"step": 300 |
|
}, |
|
{ |
|
"epoch": 0.79, |
|
"grad_norm": 149.9412754602278, |
|
"learning_rate": 1.257446259144494e-08, |
|
"logits/chosen": -4.436882972717285, |
|
"logits/rejected": -4.769654750823975, |
|
"logps/chosen": -239.6269073486328, |
|
"logps/rejected": -259.68951416015625, |
|
"loss": 0.1671, |
|
"rewards/accuracies": 0.9437500238418579, |
|
"rewards/chosen": 2.1512722969055176, |
|
"rewards/margins": 5.90042781829834, |
|
"rewards/rejected": -3.7491557598114014, |
|
"step": 310 |
|
}, |
|
{ |
|
"epoch": 0.82, |
|
"grad_norm": 162.84992879153398, |
|
"learning_rate": 9.760622117187234e-09, |
|
"logits/chosen": -4.495532512664795, |
|
"logits/rejected": -4.839716911315918, |
|
"logps/chosen": -226.40927124023438, |
|
"logps/rejected": -241.6173553466797, |
|
"loss": 0.1706, |
|
"rewards/accuracies": 0.9312499761581421, |
|
"rewards/chosen": 1.8755052089691162, |
|
"rewards/margins": 5.392090797424316, |
|
"rewards/rejected": -3.5165855884552, |
|
"step": 320 |
|
}, |
|
{ |
|
"epoch": 0.84, |
|
"grad_norm": 140.15174845446813, |
|
"learning_rate": 7.2689232521989885e-09, |
|
"logits/chosen": -4.450541019439697, |
|
"logits/rejected": -4.755129814147949, |
|
"logps/chosen": -248.2396240234375, |
|
"logps/rejected": -268.4003601074219, |
|
"loss": 0.159, |
|
"rewards/accuracies": 0.918749988079071, |
|
"rewards/chosen": 1.9909223318099976, |
|
"rewards/margins": 5.5609917640686035, |
|
"rewards/rejected": -3.5700690746307373, |
|
"step": 330 |
|
}, |
|
{ |
|
"epoch": 0.87, |
|
"grad_norm": 126.08085826498804, |
|
"learning_rate": 5.119313618049309e-09, |
|
"logits/chosen": -4.463766574859619, |
|
"logits/rejected": -4.806758403778076, |
|
"logps/chosen": -251.0575714111328, |
|
"logps/rejected": -241.99609375, |
|
"loss": 0.1558, |
|
"rewards/accuracies": 0.9312499761581421, |
|
"rewards/chosen": 2.308844804763794, |
|
"rewards/margins": 5.830248832702637, |
|
"rewards/rejected": -3.5214037895202637, |
|
"step": 340 |
|
}, |
|
{ |
|
"epoch": 0.9, |
|
"grad_norm": 122.26451804586765, |
|
"learning_rate": 3.3290021961708158e-09, |
|
"logits/chosen": -4.487333297729492, |
|
"logits/rejected": -4.672482967376709, |
|
"logps/chosen": -237.46456909179688, |
|
"logps/rejected": -258.9417419433594, |
|
"loss": 0.1724, |
|
"rewards/accuracies": 0.925000011920929, |
|
"rewards/chosen": 1.8006614446640015, |
|
"rewards/margins": 4.9178361892700195, |
|
"rewards/rejected": -3.117175340652466, |
|
"step": 350 |
|
}, |
|
{ |
|
"epoch": 0.92, |
|
"grad_norm": 144.944818423744, |
|
"learning_rate": 1.9123215591052013e-09, |
|
"logits/chosen": -4.482793807983398, |
|
"logits/rejected": -4.7083001136779785, |
|
"logps/chosen": -243.4811248779297, |
|
"logps/rejected": -260.84857177734375, |
|
"loss": 0.1742, |
|
"rewards/accuracies": 0.918749988079071, |
|
"rewards/chosen": 1.8153793811798096, |
|
"rewards/margins": 5.153982162475586, |
|
"rewards/rejected": -3.3386025428771973, |
|
"step": 360 |
|
}, |
|
{ |
|
"epoch": 0.95, |
|
"grad_norm": 132.15693545315082, |
|
"learning_rate": 8.806131292167618e-10, |
|
"logits/chosen": -4.498129367828369, |
|
"logits/rejected": -4.671444892883301, |
|
"logps/chosen": -237.79995727539062, |
|
"logps/rejected": -269.40106201171875, |
|
"loss": 0.1751, |
|
"rewards/accuracies": 0.887499988079071, |
|
"rewards/chosen": 1.9021259546279907, |
|
"rewards/margins": 5.157528877258301, |
|
"rewards/rejected": -3.2554030418395996, |
|
"step": 370 |
|
}, |
|
{ |
|
"epoch": 0.97, |
|
"grad_norm": 150.94805422584076, |
|
"learning_rate": 2.4213638345040867e-10, |
|
"logits/chosen": -4.600647926330566, |
|
"logits/rejected": -4.883923053741455, |
|
"logps/chosen": -242.57833862304688, |
|
"logps/rejected": -258.09075927734375, |
|
"loss": 0.1659, |
|
"rewards/accuracies": 0.925000011920929, |
|
"rewards/chosen": 1.9504365921020508, |
|
"rewards/margins": 5.70705509185791, |
|
"rewards/rejected": -3.7566189765930176, |
|
"step": 380 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"grad_norm": 216.65022888736644, |
|
"learning_rate": 2.0027310073833516e-12, |
|
"logits/chosen": -4.585878372192383, |
|
"logits/rejected": -4.851449966430664, |
|
"logps/chosen": -237.5348663330078, |
|
"logps/rejected": -255.65542602539062, |
|
"loss": 0.1632, |
|
"rewards/accuracies": 0.9312499761581421, |
|
"rewards/chosen": 2.213772773742676, |
|
"rewards/margins": 5.727804183959961, |
|
"rewards/rejected": -3.514031171798706, |
|
"step": 390 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"step": 391, |
|
"total_flos": 0.0, |
|
"train_loss": 0.23296360492401416, |
|
"train_runtime": 6150.4195, |
|
"train_samples_per_second": 8.13, |
|
"train_steps_per_second": 0.064 |
|
} |
|
], |
|
"logging_steps": 10, |
|
"max_steps": 391, |
|
"num_input_tokens_seen": 0, |
|
"num_train_epochs": 1, |
|
"save_steps": 100, |
|
"total_flos": 0.0, |
|
"train_batch_size": 8, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|