| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.9982238010657194, | |
| "eval_steps": 100, | |
| "global_step": 281, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "grad_norm": 17.4636091361885, | |
| "learning_rate": 1.7241379310344825e-08, | |
| "logits/chosen": -1.180719256401062, | |
| "logits/rejected": -0.36153754591941833, | |
| "logps/chosen": -227.51303100585938, | |
| "logps/rejected": -271.6326599121094, | |
| "loss": 0.6931, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 15.871649969398254, | |
| "learning_rate": 1.7241379310344828e-07, | |
| "logits/chosen": -1.6364216804504395, | |
| "logits/rejected": -0.7718086242675781, | |
| "logps/chosen": -188.4293670654297, | |
| "logps/rejected": -273.3389587402344, | |
| "loss": 0.693, | |
| "rewards/accuracies": 0.4930555522441864, | |
| "rewards/chosen": 0.00015968189109116793, | |
| "rewards/margins": 0.0003999065957032144, | |
| "rewards/rejected": -0.00024022474826779217, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 15.802939122761929, | |
| "learning_rate": 3.4482758620689656e-07, | |
| "logits/chosen": -0.9365239143371582, | |
| "logits/rejected": -0.7781739234924316, | |
| "logps/chosen": -264.11224365234375, | |
| "logps/rejected": -268.46124267578125, | |
| "loss": 0.691, | |
| "rewards/accuracies": 0.668749988079071, | |
| "rewards/chosen": 0.0019152642926201224, | |
| "rewards/margins": 0.003832753049209714, | |
| "rewards/rejected": -0.0019174888730049133, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 16.050018865575062, | |
| "learning_rate": 4.999805731202437e-07, | |
| "logits/chosen": -1.2832940816879272, | |
| "logits/rejected": -0.9560123682022095, | |
| "logps/chosen": -223.49111938476562, | |
| "logps/rejected": -384.7261657714844, | |
| "loss": 0.6818, | |
| "rewards/accuracies": 0.8062499761581421, | |
| "rewards/chosen": 0.009979133494198322, | |
| "rewards/margins": 0.02327062375843525, | |
| "rewards/rejected": -0.013291488401591778, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "grad_norm": 16.173498483987025, | |
| "learning_rate": 4.976529986032632e-07, | |
| "logits/chosen": -1.3671165704727173, | |
| "logits/rejected": -1.1606197357177734, | |
| "logps/chosen": -226.5165557861328, | |
| "logps/rejected": -420.215576171875, | |
| "loss": 0.6557, | |
| "rewards/accuracies": 0.793749988079071, | |
| "rewards/chosen": -0.014291724190115929, | |
| "rewards/margins": 0.0719335526227951, | |
| "rewards/rejected": -0.08622527867555618, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "grad_norm": 18.73697306009316, | |
| "learning_rate": 4.91481456572267e-07, | |
| "logits/chosen": -1.0613763332366943, | |
| "logits/rejected": -0.8452272415161133, | |
| "logps/chosen": -228.3408966064453, | |
| "logps/rejected": -345.06158447265625, | |
| "loss": 0.5945, | |
| "rewards/accuracies": 0.8187500238418579, | |
| "rewards/chosen": -0.04819232225418091, | |
| "rewards/margins": 0.2763238251209259, | |
| "rewards/rejected": -0.3245161473751068, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "grad_norm": 36.9064937358061, | |
| "learning_rate": 4.815617391525771e-07, | |
| "logits/chosen": -1.2162449359893799, | |
| "logits/rejected": -0.5397701263427734, | |
| "logps/chosen": -247.07247924804688, | |
| "logps/rejected": -348.9381408691406, | |
| "loss": 0.525, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -0.17551271617412567, | |
| "rewards/margins": 0.47526684403419495, | |
| "rewards/rejected": -0.6507795453071594, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "grad_norm": 21.124245684409534, | |
| "learning_rate": 4.680478160991513e-07, | |
| "logits/chosen": -1.0888421535491943, | |
| "logits/rejected": -0.7150756120681763, | |
| "logps/chosen": -268.48040771484375, | |
| "logps/rejected": -540.1458129882812, | |
| "loss": 0.4559, | |
| "rewards/accuracies": 0.856249988079071, | |
| "rewards/chosen": -0.25963330268859863, | |
| "rewards/margins": 2.0888125896453857, | |
| "rewards/rejected": -2.3484463691711426, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "grad_norm": 19.8677514066161, | |
| "learning_rate": 4.511494449416671e-07, | |
| "logits/chosen": -1.0766880512237549, | |
| "logits/rejected": -0.03768174722790718, | |
| "logps/chosen": -283.8450622558594, | |
| "logps/rejected": -504.466552734375, | |
| "loss": 0.417, | |
| "rewards/accuracies": 0.862500011920929, | |
| "rewards/chosen": -0.32856112718582153, | |
| "rewards/margins": 1.6246074438095093, | |
| "rewards/rejected": -1.953168511390686, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "grad_norm": 20.978509774304264, | |
| "learning_rate": 4.3112891521481815e-07, | |
| "logits/chosen": -0.8445385098457336, | |
| "logits/rejected": -0.18864837288856506, | |
| "logps/chosen": -287.674560546875, | |
| "logps/rejected": -594.6647338867188, | |
| "loss": 0.3829, | |
| "rewards/accuracies": 0.9125000238418579, | |
| "rewards/chosen": -0.3922039568424225, | |
| "rewards/margins": 2.068591833114624, | |
| "rewards/rejected": -2.4607958793640137, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "grad_norm": 24.770830802497013, | |
| "learning_rate": 4.0829697730853494e-07, | |
| "logits/chosen": -0.6940481662750244, | |
| "logits/rejected": 0.17405104637145996, | |
| "logps/chosen": -258.2374572753906, | |
| "logps/rejected": -551.6319580078125, | |
| "loss": 0.3643, | |
| "rewards/accuracies": 0.9312499761581421, | |
| "rewards/chosen": -0.2993431091308594, | |
| "rewards/margins": 2.340083360671997, | |
| "rewards/rejected": -2.6394264698028564, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "eval_logits/chosen": -1.748563289642334, | |
| "eval_logits/rejected": 0.045369070023298264, | |
| "eval_logps/chosen": -284.61224365234375, | |
| "eval_logps/rejected": -259.5780334472656, | |
| "eval_loss": 0.5978485941886902, | |
| "eval_rewards/accuracies": 0.7593283653259277, | |
| "eval_rewards/chosen": -0.37915343046188354, | |
| "eval_rewards/margins": 0.3338019549846649, | |
| "eval_rewards/rejected": -0.7129553556442261, | |
| "eval_runtime": 219.8632, | |
| "eval_samples_per_second": 9.742, | |
| "eval_steps_per_second": 0.305, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "grad_norm": 26.461913652670617, | |
| "learning_rate": 3.8300801912883414e-07, | |
| "logits/chosen": -0.9723485112190247, | |
| "logits/rejected": 0.37491127848625183, | |
| "logps/chosen": -249.15823364257812, | |
| "logps/rejected": -532.2054443359375, | |
| "loss": 0.3442, | |
| "rewards/accuracies": 0.8999999761581421, | |
| "rewards/chosen": -0.35393720865249634, | |
| "rewards/margins": 2.0555851459503174, | |
| "rewards/rejected": -2.40952205657959, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "grad_norm": 26.7014694652173, | |
| "learning_rate": 3.5565456543517485e-07, | |
| "logits/chosen": -0.09540247917175293, | |
| "logits/rejected": 0.513046383857727, | |
| "logps/chosen": -261.6147155761719, | |
| "logps/rejected": -637.6137084960938, | |
| "loss": 0.306, | |
| "rewards/accuracies": 0.9125000238418579, | |
| "rewards/chosen": -0.2667621970176697, | |
| "rewards/margins": 3.23994779586792, | |
| "rewards/rejected": -3.5067100524902344, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "grad_norm": 68.5297134578811, | |
| "learning_rate": 3.266611852333336e-07, | |
| "logits/chosen": -0.16894316673278809, | |
| "logits/rejected": 0.41675764322280884, | |
| "logps/chosen": -267.2029724121094, | |
| "logps/rejected": -793.5310668945312, | |
| "loss": 0.3386, | |
| "rewards/accuracies": 0.90625, | |
| "rewards/chosen": -0.3927966356277466, | |
| "rewards/margins": 4.087599754333496, | |
| "rewards/rejected": -4.480396270751953, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "grad_norm": 24.995403128308034, | |
| "learning_rate": 2.964779017907287e-07, | |
| "logits/chosen": -0.3116651177406311, | |
| "logits/rejected": 0.7034340500831604, | |
| "logps/chosen": -248.13546752929688, | |
| "logps/rejected": -628.6119384765625, | |
| "loss": 0.2752, | |
| "rewards/accuracies": 0.9437500238418579, | |
| "rewards/chosen": -0.2110290229320526, | |
| "rewards/margins": 3.0275955200195312, | |
| "rewards/rejected": -3.2386245727539062, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "grad_norm": 21.04952742243863, | |
| "learning_rate": 2.6557320756121306e-07, | |
| "logits/chosen": -0.4465761184692383, | |
| "logits/rejected": 0.5097697973251343, | |
| "logps/chosen": -253.65927124023438, | |
| "logps/rejected": -561.6199951171875, | |
| "loss": 0.2543, | |
| "rewards/accuracies": 0.925000011920929, | |
| "rewards/chosen": -0.2355656623840332, | |
| "rewards/margins": 2.4431042671203613, | |
| "rewards/rejected": -2.6786696910858154, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "grad_norm": 36.34356588459772, | |
| "learning_rate": 2.3442679243878697e-07, | |
| "logits/chosen": -0.053225208073854446, | |
| "logits/rejected": 0.7030835747718811, | |
| "logps/chosen": -290.45831298828125, | |
| "logps/rejected": -679.306640625, | |
| "loss": 0.254, | |
| "rewards/accuracies": 0.956250011920929, | |
| "rewards/chosen": -0.2719694674015045, | |
| "rewards/margins": 3.4167423248291016, | |
| "rewards/rejected": -3.6887118816375732, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "grad_norm": 19.597521304398125, | |
| "learning_rate": 2.0352209820927135e-07, | |
| "logits/chosen": -0.07039432227611542, | |
| "logits/rejected": 1.1628978252410889, | |
| "logps/chosen": -268.78289794921875, | |
| "logps/rejected": -704.7100830078125, | |
| "loss": 0.2347, | |
| "rewards/accuracies": 0.9375, | |
| "rewards/chosen": -0.25404995679855347, | |
| "rewards/margins": 3.9025752544403076, | |
| "rewards/rejected": -4.156625747680664, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "grad_norm": 33.193096952517436, | |
| "learning_rate": 1.7333881476666646e-07, | |
| "logits/chosen": -0.859085738658905, | |
| "logits/rejected": 0.6180144548416138, | |
| "logps/chosen": -228.8545379638672, | |
| "logps/rejected": -779.0535888671875, | |
| "loss": 0.2314, | |
| "rewards/accuracies": 0.949999988079071, | |
| "rewards/chosen": -0.3073621392250061, | |
| "rewards/margins": 4.101855278015137, | |
| "rewards/rejected": -4.409217834472656, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "grad_norm": 23.061813637320956, | |
| "learning_rate": 1.4434543456482518e-07, | |
| "logits/chosen": -0.6410714387893677, | |
| "logits/rejected": 0.8705042004585266, | |
| "logps/chosen": -246.0118408203125, | |
| "logps/rejected": -682.171142578125, | |
| "loss": 0.2376, | |
| "rewards/accuracies": 0.9312499761581421, | |
| "rewards/chosen": -0.35869041085243225, | |
| "rewards/margins": 3.1796650886535645, | |
| "rewards/rejected": -3.538356065750122, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "grad_norm": 26.215462134933507, | |
| "learning_rate": 1.1699198087116588e-07, | |
| "logits/chosen": -0.2712032198905945, | |
| "logits/rejected": 0.8643509149551392, | |
| "logps/chosen": -240.9992218017578, | |
| "logps/rejected": -750.1979370117188, | |
| "loss": 0.2041, | |
| "rewards/accuracies": 0.949999988079071, | |
| "rewards/chosen": -0.3355063796043396, | |
| "rewards/margins": 4.0295023918151855, | |
| "rewards/rejected": -4.365009307861328, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "eval_logits/chosen": -1.433498740196228, | |
| "eval_logits/rejected": 0.8359671235084534, | |
| "eval_logps/chosen": -334.36553955078125, | |
| "eval_logps/rejected": -351.18157958984375, | |
| "eval_loss": 0.5962842702865601, | |
| "eval_rewards/accuracies": 0.7108209133148193, | |
| "eval_rewards/chosen": -0.8766867518424988, | |
| "eval_rewards/margins": 0.7523040771484375, | |
| "eval_rewards/rejected": -1.628990888595581, | |
| "eval_runtime": 219.6386, | |
| "eval_samples_per_second": 9.752, | |
| "eval_steps_per_second": 0.305, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "grad_norm": 20.46821959291358, | |
| "learning_rate": 9.170302269146507e-08, | |
| "logits/chosen": -0.14280591905117035, | |
| "logits/rejected": 0.9915373921394348, | |
| "logps/chosen": -273.5321960449219, | |
| "logps/rejected": -798.1834716796875, | |
| "loss": 0.1891, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -0.2832937240600586, | |
| "rewards/margins": 4.233273506164551, | |
| "rewards/rejected": -4.516566753387451, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "grad_norm": 25.672453750455183, | |
| "learning_rate": 6.887108478518183e-08, | |
| "logits/chosen": -0.6617056131362915, | |
| "logits/rejected": 1.1209099292755127, | |
| "logps/chosen": -266.068359375, | |
| "logps/rejected": -728.5145263671875, | |
| "loss": 0.2108, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": -0.27059105038642883, | |
| "rewards/margins": 4.039263725280762, | |
| "rewards/rejected": -4.309854984283447, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "grad_norm": 36.27184670290087, | |
| "learning_rate": 4.885055505833291e-08, | |
| "logits/chosen": -0.10255080461502075, | |
| "logits/rejected": 0.778251051902771, | |
| "logps/chosen": -260.8437194824219, | |
| "logps/rejected": -754.1353149414062, | |
| "loss": 0.2016, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -0.3063124120235443, | |
| "rewards/margins": 3.9367871284484863, | |
| "rewards/rejected": -4.243099689483643, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "grad_norm": 31.207116090159346, | |
| "learning_rate": 3.1952183900848665e-08, | |
| "logits/chosen": -0.4079841673374176, | |
| "logits/rejected": 0.5633789896965027, | |
| "logps/chosen": -238.26962280273438, | |
| "logps/rejected": -706.0747680664062, | |
| "loss": 0.2235, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -0.2822808027267456, | |
| "rewards/margins": 3.6717209815979004, | |
| "rewards/rejected": -3.9540016651153564, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "grad_norm": 24.232889457842543, | |
| "learning_rate": 1.8438260847422837e-08, | |
| "logits/chosen": -0.11115667968988419, | |
| "logits/rejected": 1.0082677602767944, | |
| "logps/chosen": -271.56109619140625, | |
| "logps/rejected": -707.6219482421875, | |
| "loss": 0.2049, | |
| "rewards/accuracies": 0.96875, | |
| "rewards/chosen": -0.2982483506202698, | |
| "rewards/margins": 3.9781322479248047, | |
| "rewards/rejected": -4.2763800621032715, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "grad_norm": 24.71147868673447, | |
| "learning_rate": 8.518543427732949e-09, | |
| "logits/chosen": -0.2919185161590576, | |
| "logits/rejected": 0.8239177465438843, | |
| "logps/chosen": -285.7502136230469, | |
| "logps/rejected": -721.8952026367188, | |
| "loss": 0.2005, | |
| "rewards/accuracies": 0.956250011920929, | |
| "rewards/chosen": -0.3373405635356903, | |
| "rewards/margins": 4.360747337341309, | |
| "rewards/rejected": -4.6980881690979, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "grad_norm": 32.225622448127446, | |
| "learning_rate": 2.3470013967367976e-09, | |
| "logits/chosen": 0.019734883680939674, | |
| "logits/rejected": 0.908920168876648, | |
| "logps/chosen": -254.5631866455078, | |
| "logps/rejected": -786.376708984375, | |
| "loss": 0.1909, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -0.3218054175376892, | |
| "rewards/margins": 4.211289405822754, | |
| "rewards/rejected": -4.533095359802246, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "grad_norm": 19.533422797206217, | |
| "learning_rate": 1.9426879756284655e-11, | |
| "logits/chosen": -0.17949172854423523, | |
| "logits/rejected": 0.9797495007514954, | |
| "logps/chosen": -278.2919921875, | |
| "logps/rejected": -752.1105346679688, | |
| "loss": 0.173, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -0.28086012601852417, | |
| "rewards/margins": 4.271262168884277, | |
| "rewards/rejected": -4.552122116088867, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 281, | |
| "total_flos": 0.0, | |
| "train_loss": 0.34737620712175066, | |
| "train_runtime": 4632.3521, | |
| "train_samples_per_second": 3.885, | |
| "train_steps_per_second": 0.061 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 281, | |
| "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 | |
| } | |