| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.0, | |
| "eval_steps": 500, | |
| "global_step": 933, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.003215434083601286, | |
| "grad_norm": 79.87917506057431, | |
| "learning_rate": 5.3191489361702125e-09, | |
| "logits/chosen": -4.125, | |
| "logits/rejected": -3.765625, | |
| "logps/chosen": -132.0, | |
| "logps/rejected": -141.0, | |
| "loss": 0.6914, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.03215434083601286, | |
| "grad_norm": 118.20552774316917, | |
| "learning_rate": 5.3191489361702123e-08, | |
| "logits/chosen": -4.15625, | |
| "logits/rejected": -3.8125, | |
| "logps/chosen": -114.0, | |
| "logps/rejected": -178.0, | |
| "loss": 0.6858, | |
| "rewards/accuracies": 0.2638888955116272, | |
| "rewards/chosen": 0.0086669921875, | |
| "rewards/margins": 0.00555419921875, | |
| "rewards/rejected": 0.00311279296875, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.06430868167202572, | |
| "grad_norm": 66.66640901080964, | |
| "learning_rate": 1.0638297872340425e-07, | |
| "logits/chosen": -4.0, | |
| "logits/rejected": -3.8125, | |
| "logps/chosen": -122.0, | |
| "logps/rejected": -203.0, | |
| "loss": 0.6479, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": 0.05615234375, | |
| "rewards/margins": 0.09423828125, | |
| "rewards/rejected": -0.0380859375, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.09646302250803858, | |
| "grad_norm": 81.48311869513411, | |
| "learning_rate": 1.5957446808510638e-07, | |
| "logits/chosen": -4.03125, | |
| "logits/rejected": -3.84375, | |
| "logps/chosen": -142.0, | |
| "logps/rejected": -224.0, | |
| "loss": 0.5284, | |
| "rewards/accuracies": 0.925000011920929, | |
| "rewards/chosen": 0.1962890625, | |
| "rewards/margins": 0.474609375, | |
| "rewards/rejected": -0.279296875, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.12861736334405144, | |
| "grad_norm": 483.03383426380265, | |
| "learning_rate": 2.127659574468085e-07, | |
| "logits/chosen": -3.921875, | |
| "logits/rejected": -3.859375, | |
| "logps/chosen": -130.0, | |
| "logps/rejected": -196.0, | |
| "loss": 0.4049, | |
| "rewards/accuracies": 0.8999999761581421, | |
| "rewards/chosen": 0.04638671875, | |
| "rewards/margins": 0.9296875, | |
| "rewards/rejected": -0.8828125, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.1607717041800643, | |
| "grad_norm": 521.8067650467351, | |
| "learning_rate": 2.659574468085106e-07, | |
| "logits/chosen": -3.875, | |
| "logits/rejected": -3.90625, | |
| "logps/chosen": -130.0, | |
| "logps/rejected": -214.0, | |
| "loss": 0.4117, | |
| "rewards/accuracies": 0.949999988079071, | |
| "rewards/chosen": 0.259765625, | |
| "rewards/margins": 2.078125, | |
| "rewards/rejected": -1.8203125, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.19292604501607716, | |
| "grad_norm": 122.5164041936752, | |
| "learning_rate": 3.1914893617021275e-07, | |
| "logits/chosen": -3.90625, | |
| "logits/rejected": -3.953125, | |
| "logps/chosen": -116.0, | |
| "logps/rejected": -214.0, | |
| "loss": 0.2465, | |
| "rewards/accuracies": 0.9125000238418579, | |
| "rewards/chosen": -0.248046875, | |
| "rewards/margins": 2.34375, | |
| "rewards/rejected": -2.59375, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.22508038585209003, | |
| "grad_norm": 98.58564731221452, | |
| "learning_rate": 3.7234042553191484e-07, | |
| "logits/chosen": -3.84375, | |
| "logits/rejected": -3.90625, | |
| "logps/chosen": -130.0, | |
| "logps/rejected": -274.0, | |
| "loss": 0.2316, | |
| "rewards/accuracies": 0.925000011920929, | |
| "rewards/chosen": -0.1923828125, | |
| "rewards/margins": 3.359375, | |
| "rewards/rejected": -3.546875, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.2572347266881029, | |
| "grad_norm": 58.684567343268625, | |
| "learning_rate": 4.25531914893617e-07, | |
| "logits/chosen": -3.6875, | |
| "logits/rejected": -3.828125, | |
| "logps/chosen": -144.0, | |
| "logps/rejected": -209.0, | |
| "loss": 0.3018, | |
| "rewards/accuracies": 0.8999999761581421, | |
| "rewards/chosen": -0.50390625, | |
| "rewards/margins": 2.765625, | |
| "rewards/rejected": -3.265625, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.28938906752411575, | |
| "grad_norm": 44.1285015532565, | |
| "learning_rate": 4.787234042553192e-07, | |
| "logits/chosen": -3.484375, | |
| "logits/rejected": -3.65625, | |
| "logps/chosen": -121.0, | |
| "logps/rejected": -198.0, | |
| "loss": 0.3009, | |
| "rewards/accuracies": 0.925000011920929, | |
| "rewards/chosen": -0.181640625, | |
| "rewards/margins": 2.4375, | |
| "rewards/rejected": -2.625, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.3215434083601286, | |
| "grad_norm": 50.263563110645, | |
| "learning_rate": 4.964243146603099e-07, | |
| "logits/chosen": -3.5, | |
| "logits/rejected": -3.609375, | |
| "logps/chosen": -150.0, | |
| "logps/rejected": -222.0, | |
| "loss": 0.2831, | |
| "rewards/accuracies": 0.875, | |
| "rewards/chosen": -4.887580871582031e-05, | |
| "rewards/margins": 3.09375, | |
| "rewards/rejected": -3.09375, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.3536977491961415, | |
| "grad_norm": 77.17996002738214, | |
| "learning_rate": 4.904648390941597e-07, | |
| "logits/chosen": -3.578125, | |
| "logits/rejected": -3.703125, | |
| "logps/chosen": -114.5, | |
| "logps/rejected": -233.0, | |
| "loss": 0.1551, | |
| "rewards/accuracies": 0.949999988079071, | |
| "rewards/chosen": -0.0654296875, | |
| "rewards/margins": 3.828125, | |
| "rewards/rejected": -3.890625, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.3858520900321543, | |
| "grad_norm": 50.629644956690505, | |
| "learning_rate": 4.845053635280095e-07, | |
| "logits/chosen": -3.59375, | |
| "logits/rejected": -3.8125, | |
| "logps/chosen": -120.5, | |
| "logps/rejected": -262.0, | |
| "loss": 0.1768, | |
| "rewards/accuracies": 0.887499988079071, | |
| "rewards/chosen": -0.388671875, | |
| "rewards/margins": 4.15625, | |
| "rewards/rejected": -4.53125, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.4180064308681672, | |
| "grad_norm": 87.75654163889804, | |
| "learning_rate": 4.785458879618593e-07, | |
| "logits/chosen": -3.484375, | |
| "logits/rejected": -3.640625, | |
| "logps/chosen": -153.0, | |
| "logps/rejected": -280.0, | |
| "loss": 0.1777, | |
| "rewards/accuracies": 0.949999988079071, | |
| "rewards/chosen": -0.66796875, | |
| "rewards/margins": 4.25, | |
| "rewards/rejected": -4.90625, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.45016077170418006, | |
| "grad_norm": 48.30101230741645, | |
| "learning_rate": 4.7258641239570916e-07, | |
| "logits/chosen": -3.640625, | |
| "logits/rejected": -3.765625, | |
| "logps/chosen": -119.0, | |
| "logps/rejected": -233.0, | |
| "loss": 0.1792, | |
| "rewards/accuracies": 0.925000011920929, | |
| "rewards/chosen": 0.142578125, | |
| "rewards/margins": 3.859375, | |
| "rewards/rejected": -3.71875, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.48231511254019294, | |
| "grad_norm": 160.46320495134455, | |
| "learning_rate": 4.66626936829559e-07, | |
| "logits/chosen": -3.75, | |
| "logits/rejected": -3.875, | |
| "logps/chosen": -111.0, | |
| "logps/rejected": -248.0, | |
| "loss": 0.3135, | |
| "rewards/accuracies": 0.887499988079071, | |
| "rewards/chosen": 0.1611328125, | |
| "rewards/margins": 4.03125, | |
| "rewards/rejected": -3.875, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.5144694533762058, | |
| "grad_norm": 27.202159849500408, | |
| "learning_rate": 4.606674612634088e-07, | |
| "logits/chosen": -3.5625, | |
| "logits/rejected": -3.765625, | |
| "logps/chosen": -132.0, | |
| "logps/rejected": -260.0, | |
| "loss": 0.1584, | |
| "rewards/accuracies": 0.9375, | |
| "rewards/chosen": 0.181640625, | |
| "rewards/margins": 4.34375, | |
| "rewards/rejected": -4.15625, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.5466237942122186, | |
| "grad_norm": 59.16977770326313, | |
| "learning_rate": 4.547079856972586e-07, | |
| "logits/chosen": -3.734375, | |
| "logits/rejected": -3.796875, | |
| "logps/chosen": -140.0, | |
| "logps/rejected": -241.0, | |
| "loss": 0.1732, | |
| "rewards/accuracies": 0.9125000238418579, | |
| "rewards/chosen": -0.1953125, | |
| "rewards/margins": 3.8125, | |
| "rewards/rejected": -4.0, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.5787781350482315, | |
| "grad_norm": 54.881650170057746, | |
| "learning_rate": 4.487485101311084e-07, | |
| "logits/chosen": -3.625, | |
| "logits/rejected": -3.75, | |
| "logps/chosen": -122.0, | |
| "logps/rejected": -239.0, | |
| "loss": 0.2094, | |
| "rewards/accuracies": 0.9125000238418579, | |
| "rewards/chosen": -0.58203125, | |
| "rewards/margins": 3.734375, | |
| "rewards/rejected": -4.34375, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.6109324758842444, | |
| "grad_norm": 46.393130621023865, | |
| "learning_rate": 4.4278903456495827e-07, | |
| "logits/chosen": -3.5, | |
| "logits/rejected": -3.734375, | |
| "logps/chosen": -105.5, | |
| "logps/rejected": -248.0, | |
| "loss": 0.1894, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": 0.248046875, | |
| "rewards/margins": 4.0, | |
| "rewards/rejected": -3.765625, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.6430868167202572, | |
| "grad_norm": 61.77038241219252, | |
| "learning_rate": 4.368295589988081e-07, | |
| "logits/chosen": -3.546875, | |
| "logits/rejected": -3.609375, | |
| "logps/chosen": -121.5, | |
| "logps/rejected": -246.0, | |
| "loss": 0.2004, | |
| "rewards/accuracies": 0.9375, | |
| "rewards/chosen": 0.43359375, | |
| "rewards/margins": 3.84375, | |
| "rewards/rejected": -3.421875, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.6752411575562701, | |
| "grad_norm": 54.13127427585697, | |
| "learning_rate": 4.308700834326579e-07, | |
| "logits/chosen": -3.546875, | |
| "logits/rejected": -3.8125, | |
| "logps/chosen": -134.0, | |
| "logps/rejected": -258.0, | |
| "loss": 0.1556, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": 0.259765625, | |
| "rewards/margins": 4.46875, | |
| "rewards/rejected": -4.1875, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.707395498392283, | |
| "grad_norm": 39.76866338516343, | |
| "learning_rate": 4.249106078665077e-07, | |
| "logits/chosen": -3.734375, | |
| "logits/rejected": -3.84375, | |
| "logps/chosen": -114.5, | |
| "logps/rejected": -254.0, | |
| "loss": 0.1614, | |
| "rewards/accuracies": 0.9375, | |
| "rewards/chosen": 0.41796875, | |
| "rewards/margins": 4.9375, | |
| "rewards/rejected": -4.53125, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.7395498392282959, | |
| "grad_norm": 19.376631378140395, | |
| "learning_rate": 4.1895113230035757e-07, | |
| "logits/chosen": -3.65625, | |
| "logits/rejected": -3.703125, | |
| "logps/chosen": -125.5, | |
| "logps/rejected": -266.0, | |
| "loss": 0.2333, | |
| "rewards/accuracies": 0.949999988079071, | |
| "rewards/chosen": 0.96484375, | |
| "rewards/margins": 4.5, | |
| "rewards/rejected": -3.515625, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.7717041800643086, | |
| "grad_norm": 38.972752947728395, | |
| "learning_rate": 4.129916567342074e-07, | |
| "logits/chosen": -3.53125, | |
| "logits/rejected": -3.5625, | |
| "logps/chosen": -133.0, | |
| "logps/rejected": -266.0, | |
| "loss": 0.2215, | |
| "rewards/accuracies": 0.949999988079071, | |
| "rewards/chosen": 1.46875, | |
| "rewards/margins": 5.03125, | |
| "rewards/rejected": -3.578125, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.8038585209003215, | |
| "grad_norm": 25.549122278044088, | |
| "learning_rate": 4.070321811680572e-07, | |
| "logits/chosen": -3.546875, | |
| "logits/rejected": -3.671875, | |
| "logps/chosen": -139.0, | |
| "logps/rejected": -238.0, | |
| "loss": 0.1699, | |
| "rewards/accuracies": 0.887499988079071, | |
| "rewards/chosen": -0.59375, | |
| "rewards/margins": 4.46875, | |
| "rewards/rejected": -5.09375, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.8360128617363344, | |
| "grad_norm": 36.7792748916066, | |
| "learning_rate": 4.0107270560190706e-07, | |
| "logits/chosen": -3.640625, | |
| "logits/rejected": -3.84375, | |
| "logps/chosen": -125.0, | |
| "logps/rejected": -248.0, | |
| "loss": 0.1649, | |
| "rewards/accuracies": 0.8999999761581421, | |
| "rewards/chosen": 0.51171875, | |
| "rewards/margins": 4.53125, | |
| "rewards/rejected": -4.03125, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.8681672025723473, | |
| "grad_norm": 10.885169604932974, | |
| "learning_rate": 3.9511323003575687e-07, | |
| "logits/chosen": -3.59375, | |
| "logits/rejected": -3.640625, | |
| "logps/chosen": -145.0, | |
| "logps/rejected": -246.0, | |
| "loss": 0.1556, | |
| "rewards/accuracies": 0.925000011920929, | |
| "rewards/chosen": 0.396484375, | |
| "rewards/margins": 4.53125, | |
| "rewards/rejected": -4.125, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.9003215434083601, | |
| "grad_norm": 33.68141687625864, | |
| "learning_rate": 3.8915375446960663e-07, | |
| "logits/chosen": -3.5625, | |
| "logits/rejected": -3.734375, | |
| "logps/chosen": -132.0, | |
| "logps/rejected": -276.0, | |
| "loss": 0.13, | |
| "rewards/accuracies": 0.949999988079071, | |
| "rewards/chosen": -0.203125, | |
| "rewards/margins": 5.125, | |
| "rewards/rejected": -5.34375, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.932475884244373, | |
| "grad_norm": 38.54727132055094, | |
| "learning_rate": 3.8319427890345644e-07, | |
| "logits/chosen": -3.453125, | |
| "logits/rejected": -3.578125, | |
| "logps/chosen": -136.0, | |
| "logps/rejected": -266.0, | |
| "loss": 0.1626, | |
| "rewards/accuracies": 0.8999999761581421, | |
| "rewards/chosen": -0.1279296875, | |
| "rewards/margins": 4.96875, | |
| "rewards/rejected": -5.09375, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.9646302250803859, | |
| "grad_norm": 33.100977527402115, | |
| "learning_rate": 3.772348033373063e-07, | |
| "logits/chosen": -3.4375, | |
| "logits/rejected": -3.65625, | |
| "logps/chosen": -137.0, | |
| "logps/rejected": -256.0, | |
| "loss": 0.1537, | |
| "rewards/accuracies": 0.9125000238418579, | |
| "rewards/chosen": 0.031982421875, | |
| "rewards/margins": 4.96875, | |
| "rewards/rejected": -4.9375, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.9967845659163987, | |
| "grad_norm": 67.60299631407663, | |
| "learning_rate": 3.712753277711561e-07, | |
| "logits/chosen": -3.484375, | |
| "logits/rejected": -3.703125, | |
| "logps/chosen": -143.0, | |
| "logps/rejected": -296.0, | |
| "loss": 0.1769, | |
| "rewards/accuracies": 0.949999988079071, | |
| "rewards/chosen": 0.10888671875, | |
| "rewards/margins": 5.4375, | |
| "rewards/rejected": -5.34375, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_logits/chosen": -3.78125, | |
| "eval_logits/rejected": -3.78125, | |
| "eval_logps/chosen": -206.0, | |
| "eval_logps/rejected": -234.0, | |
| "eval_loss": 0.5869876742362976, | |
| "eval_rewards/accuracies": 0.7678571343421936, | |
| "eval_rewards/chosen": -2.078125, | |
| "eval_rewards/margins": 2.265625, | |
| "eval_rewards/rejected": -4.34375, | |
| "eval_runtime": 14.2416, | |
| "eval_samples_per_second": 13.973, | |
| "eval_steps_per_second": 0.492, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 1.0289389067524115, | |
| "grad_norm": 17.744370347927422, | |
| "learning_rate": 3.6531585220500593e-07, | |
| "logits/chosen": -3.546875, | |
| "logits/rejected": -3.71875, | |
| "logps/chosen": -104.0, | |
| "logps/rejected": -240.0, | |
| "loss": 0.0588, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": 0.56640625, | |
| "rewards/margins": 6.75, | |
| "rewards/rejected": -6.1875, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 1.0610932475884245, | |
| "grad_norm": 4.160707504041419, | |
| "learning_rate": 3.5935637663885575e-07, | |
| "logits/chosen": -3.484375, | |
| "logits/rejected": -3.65625, | |
| "logps/chosen": -113.0, | |
| "logps/rejected": -286.0, | |
| "loss": 0.0337, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": 1.2265625, | |
| "rewards/margins": 8.4375, | |
| "rewards/rejected": -7.1875, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 1.0932475884244373, | |
| "grad_norm": 17.23951731292116, | |
| "learning_rate": 3.533969010727056e-07, | |
| "logits/chosen": -3.640625, | |
| "logits/rejected": -3.796875, | |
| "logps/chosen": -115.5, | |
| "logps/rejected": -252.0, | |
| "loss": 0.0347, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": 1.1875, | |
| "rewards/margins": 8.375, | |
| "rewards/rejected": -7.1875, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 1.1254019292604502, | |
| "grad_norm": 49.64075213527915, | |
| "learning_rate": 3.474374255065554e-07, | |
| "logits/chosen": -3.40625, | |
| "logits/rejected": -3.5625, | |
| "logps/chosen": -140.0, | |
| "logps/rejected": -300.0, | |
| "loss": 0.0189, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": 1.0390625, | |
| "rewards/margins": 8.625, | |
| "rewards/rejected": -7.625, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 1.157556270096463, | |
| "grad_norm": 7.306932203883141, | |
| "learning_rate": 3.4147794994040524e-07, | |
| "logits/chosen": -3.171875, | |
| "logits/rejected": -3.421875, | |
| "logps/chosen": -109.5, | |
| "logps/rejected": -274.0, | |
| "loss": 0.0278, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 1.4609375, | |
| "rewards/margins": 9.5625, | |
| "rewards/rejected": -8.125, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 1.189710610932476, | |
| "grad_norm": 35.16526862248255, | |
| "learning_rate": 3.3551847437425505e-07, | |
| "logits/chosen": -3.25, | |
| "logits/rejected": -3.421875, | |
| "logps/chosen": -120.5, | |
| "logps/rejected": -282.0, | |
| "loss": 0.0399, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": 0.44140625, | |
| "rewards/margins": 8.8125, | |
| "rewards/rejected": -8.375, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 1.2218649517684887, | |
| "grad_norm": 7.575200512680006, | |
| "learning_rate": 3.295589988081049e-07, | |
| "logits/chosen": -3.359375, | |
| "logits/rejected": -3.515625, | |
| "logps/chosen": -120.0, | |
| "logps/rejected": -300.0, | |
| "loss": 0.0216, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": 0.9140625, | |
| "rewards/margins": 9.3125, | |
| "rewards/rejected": -8.4375, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 1.2540192926045015, | |
| "grad_norm": 0.6452900307805328, | |
| "learning_rate": 3.235995232419547e-07, | |
| "logits/chosen": -3.265625, | |
| "logits/rejected": -3.453125, | |
| "logps/chosen": -132.0, | |
| "logps/rejected": -322.0, | |
| "loss": 0.0475, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": 0.8515625, | |
| "rewards/margins": 10.0, | |
| "rewards/rejected": -9.1875, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 1.2861736334405145, | |
| "grad_norm": 7.420661461533486, | |
| "learning_rate": 3.176400476758045e-07, | |
| "logits/chosen": -3.15625, | |
| "logits/rejected": -3.328125, | |
| "logps/chosen": -105.5, | |
| "logps/rejected": -272.0, | |
| "loss": 0.0335, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": 0.90625, | |
| "rewards/margins": 9.1875, | |
| "rewards/rejected": -8.25, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.3183279742765273, | |
| "grad_norm": 27.92454063116238, | |
| "learning_rate": 3.116805721096543e-07, | |
| "logits/chosen": -3.28125, | |
| "logits/rejected": -3.484375, | |
| "logps/chosen": -129.0, | |
| "logps/rejected": -282.0, | |
| "loss": 0.0376, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": 0.023193359375, | |
| "rewards/margins": 9.0625, | |
| "rewards/rejected": -9.0625, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 1.3504823151125402, | |
| "grad_norm": 48.832852324456454, | |
| "learning_rate": 3.0572109654350416e-07, | |
| "logits/chosen": -3.203125, | |
| "logits/rejected": -3.34375, | |
| "logps/chosen": -134.0, | |
| "logps/rejected": -282.0, | |
| "loss": 0.0261, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": 0.173828125, | |
| "rewards/margins": 9.0625, | |
| "rewards/rejected": -8.875, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 1.382636655948553, | |
| "grad_norm": 65.54781768372352, | |
| "learning_rate": 2.99761620977354e-07, | |
| "logits/chosen": -3.28125, | |
| "logits/rejected": -3.46875, | |
| "logps/chosen": -117.0, | |
| "logps/rejected": -276.0, | |
| "loss": 0.0396, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": 0.494140625, | |
| "rewards/margins": 9.1875, | |
| "rewards/rejected": -8.6875, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 1.414790996784566, | |
| "grad_norm": 25.7310269564794, | |
| "learning_rate": 2.938021454112038e-07, | |
| "logits/chosen": -3.25, | |
| "logits/rejected": -3.484375, | |
| "logps/chosen": -156.0, | |
| "logps/rejected": -316.0, | |
| "loss": 0.0399, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": 0.78515625, | |
| "rewards/margins": 9.125, | |
| "rewards/rejected": -8.375, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 1.4469453376205788, | |
| "grad_norm": 8.455918132650991, | |
| "learning_rate": 2.878426698450536e-07, | |
| "logits/chosen": -3.25, | |
| "logits/rejected": -3.484375, | |
| "logps/chosen": -110.5, | |
| "logps/rejected": -255.0, | |
| "loss": 0.0372, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": 0.6640625, | |
| "rewards/margins": 8.75, | |
| "rewards/rejected": -8.0625, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 1.4790996784565915, | |
| "grad_norm": 23.483398578435676, | |
| "learning_rate": 2.8188319427890346e-07, | |
| "logits/chosen": -3.296875, | |
| "logits/rejected": -3.515625, | |
| "logps/chosen": -115.5, | |
| "logps/rejected": -304.0, | |
| "loss": 0.0359, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": 0.91796875, | |
| "rewards/margins": 9.5, | |
| "rewards/rejected": -8.5625, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 1.5112540192926045, | |
| "grad_norm": 1.560695472946775, | |
| "learning_rate": 2.759237187127533e-07, | |
| "logits/chosen": -3.296875, | |
| "logits/rejected": -3.46875, | |
| "logps/chosen": -103.0, | |
| "logps/rejected": -294.0, | |
| "loss": 0.0296, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": 0.91015625, | |
| "rewards/margins": 10.125, | |
| "rewards/rejected": -9.25, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 1.5434083601286175, | |
| "grad_norm": 4.044291002647325, | |
| "learning_rate": 2.699642431466031e-07, | |
| "logits/chosen": -3.40625, | |
| "logits/rejected": -3.625, | |
| "logps/chosen": -119.0, | |
| "logps/rejected": -292.0, | |
| "loss": 0.0338, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": 0.72265625, | |
| "rewards/margins": 10.125, | |
| "rewards/rejected": -9.4375, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 1.5755627009646302, | |
| "grad_norm": 52.234565040636085, | |
| "learning_rate": 2.640047675804529e-07, | |
| "logits/chosen": -3.421875, | |
| "logits/rejected": -3.53125, | |
| "logps/chosen": -133.0, | |
| "logps/rejected": -296.0, | |
| "loss": 0.0613, | |
| "rewards/accuracies": 0.949999988079071, | |
| "rewards/chosen": 0.9375, | |
| "rewards/margins": 9.3125, | |
| "rewards/rejected": -8.375, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 1.607717041800643, | |
| "grad_norm": 3.80360585724896, | |
| "learning_rate": 2.5804529201430277e-07, | |
| "logits/chosen": -3.34375, | |
| "logits/rejected": -3.578125, | |
| "logps/chosen": -128.0, | |
| "logps/rejected": -300.0, | |
| "loss": 0.1382, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": 0.16796875, | |
| "rewards/margins": 8.9375, | |
| "rewards/rejected": -8.8125, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.639871382636656, | |
| "grad_norm": 2.651977727973715, | |
| "learning_rate": 2.520858164481526e-07, | |
| "logits/chosen": -3.421875, | |
| "logits/rejected": -3.65625, | |
| "logps/chosen": -119.0, | |
| "logps/rejected": -276.0, | |
| "loss": 0.0504, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": 0.37890625, | |
| "rewards/margins": 8.6875, | |
| "rewards/rejected": -8.3125, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 1.6720257234726688, | |
| "grad_norm": 28.125055557089983, | |
| "learning_rate": 2.461263408820024e-07, | |
| "logits/chosen": -3.34375, | |
| "logits/rejected": -3.53125, | |
| "logps/chosen": -100.0, | |
| "logps/rejected": -276.0, | |
| "loss": 0.0269, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": 1.5390625, | |
| "rewards/margins": 10.25, | |
| "rewards/rejected": -8.75, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 1.7041800643086815, | |
| "grad_norm": 20.361634972075212, | |
| "learning_rate": 2.401668653158522e-07, | |
| "logits/chosen": -3.4375, | |
| "logits/rejected": -3.59375, | |
| "logps/chosen": -135.0, | |
| "logps/rejected": -310.0, | |
| "loss": 0.0232, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": 0.75390625, | |
| "rewards/margins": 10.875, | |
| "rewards/rejected": -10.125, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 1.7363344051446945, | |
| "grad_norm": 29.37467125959926, | |
| "learning_rate": 2.3420738974970201e-07, | |
| "logits/chosen": -3.578125, | |
| "logits/rejected": -3.734375, | |
| "logps/chosen": -130.0, | |
| "logps/rejected": -300.0, | |
| "loss": 0.0688, | |
| "rewards/accuracies": 0.8999999761581421, | |
| "rewards/chosen": -0.859375, | |
| "rewards/margins": 9.0625, | |
| "rewards/rejected": -9.9375, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 1.7684887459807075, | |
| "grad_norm": 16.99690727375027, | |
| "learning_rate": 2.2824791418355183e-07, | |
| "logits/chosen": -3.453125, | |
| "logits/rejected": -3.6875, | |
| "logps/chosen": -145.0, | |
| "logps/rejected": -312.0, | |
| "loss": 0.0231, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.048828125, | |
| "rewards/margins": 9.5, | |
| "rewards/rejected": -9.5, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.8006430868167203, | |
| "grad_norm": 49.05866215948682, | |
| "learning_rate": 2.2228843861740164e-07, | |
| "logits/chosen": -3.4375, | |
| "logits/rejected": -3.609375, | |
| "logps/chosen": -126.5, | |
| "logps/rejected": -284.0, | |
| "loss": 0.0162, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 0.453125, | |
| "rewards/margins": 10.0, | |
| "rewards/rejected": -9.5625, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 1.832797427652733, | |
| "grad_norm": 17.76760727253399, | |
| "learning_rate": 2.1632896305125148e-07, | |
| "logits/chosen": -3.515625, | |
| "logits/rejected": -3.71875, | |
| "logps/chosen": -126.5, | |
| "logps/rejected": -298.0, | |
| "loss": 0.0305, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": 0.3125, | |
| "rewards/margins": 10.1875, | |
| "rewards/rejected": -9.875, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 1.864951768488746, | |
| "grad_norm": 6.0736731293937, | |
| "learning_rate": 2.1036948748510132e-07, | |
| "logits/chosen": -3.4375, | |
| "logits/rejected": -3.5625, | |
| "logps/chosen": -127.0, | |
| "logps/rejected": -312.0, | |
| "loss": 0.054, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -0.1064453125, | |
| "rewards/margins": 10.25, | |
| "rewards/rejected": -10.375, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 1.897106109324759, | |
| "grad_norm": 39.17766360963206, | |
| "learning_rate": 2.0441001191895113e-07, | |
| "logits/chosen": -3.59375, | |
| "logits/rejected": -3.796875, | |
| "logps/chosen": -137.0, | |
| "logps/rejected": -300.0, | |
| "loss": 0.0353, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -0.1953125, | |
| "rewards/margins": 9.8125, | |
| "rewards/rejected": -10.0, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 1.9292604501607717, | |
| "grad_norm": 38.14836445184288, | |
| "learning_rate": 1.9845053635280097e-07, | |
| "logits/chosen": -3.34375, | |
| "logits/rejected": -3.53125, | |
| "logps/chosen": -138.0, | |
| "logps/rejected": -324.0, | |
| "loss": 0.0299, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": 0.458984375, | |
| "rewards/margins": 11.0625, | |
| "rewards/rejected": -10.625, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.9614147909967845, | |
| "grad_norm": 55.90659543384472, | |
| "learning_rate": 1.9249106078665075e-07, | |
| "logits/chosen": -3.3125, | |
| "logits/rejected": -3.546875, | |
| "logps/chosen": -113.0, | |
| "logps/rejected": -272.0, | |
| "loss": 0.0322, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": 1.09375, | |
| "rewards/margins": 9.375, | |
| "rewards/rejected": -8.25, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 1.9935691318327975, | |
| "grad_norm": 0.34892216328376285, | |
| "learning_rate": 1.865315852205006e-07, | |
| "logits/chosen": -3.296875, | |
| "logits/rejected": -3.46875, | |
| "logps/chosen": -150.0, | |
| "logps/rejected": -334.0, | |
| "loss": 0.018, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 0.484375, | |
| "rewards/margins": 10.9375, | |
| "rewards/rejected": -10.4375, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_logits/chosen": -3.65625, | |
| "eval_logits/rejected": -3.734375, | |
| "eval_logps/chosen": -226.0, | |
| "eval_logps/rejected": -266.0, | |
| "eval_loss": 0.9209327697753906, | |
| "eval_rewards/accuracies": 0.7142857313156128, | |
| "eval_rewards/chosen": -4.09375, | |
| "eval_rewards/margins": 3.375, | |
| "eval_rewards/rejected": -7.46875, | |
| "eval_runtime": 14.1621, | |
| "eval_samples_per_second": 14.052, | |
| "eval_steps_per_second": 0.494, | |
| "step": 622 | |
| }, | |
| { | |
| "epoch": 2.0257234726688105, | |
| "grad_norm": 1.0084688310317689, | |
| "learning_rate": 1.805721096543504e-07, | |
| "logits/chosen": -3.265625, | |
| "logits/rejected": -3.53125, | |
| "logps/chosen": -138.0, | |
| "logps/rejected": -296.0, | |
| "loss": 0.0133, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 0.82421875, | |
| "rewards/margins": 10.125, | |
| "rewards/rejected": -9.3125, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 2.057877813504823, | |
| "grad_norm": 1.0564688005496572, | |
| "learning_rate": 1.7461263408820024e-07, | |
| "logits/chosen": -3.28125, | |
| "logits/rejected": -3.46875, | |
| "logps/chosen": -123.5, | |
| "logps/rejected": -278.0, | |
| "loss": 0.0128, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": 0.5, | |
| "rewards/margins": 10.0, | |
| "rewards/rejected": -9.5, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 2.090032154340836, | |
| "grad_norm": 2.814703376282905, | |
| "learning_rate": 1.6865315852205006e-07, | |
| "logits/chosen": -3.171875, | |
| "logits/rejected": -3.390625, | |
| "logps/chosen": -122.5, | |
| "logps/rejected": -330.0, | |
| "loss": 0.0096, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 1.4296875, | |
| "rewards/margins": 13.4375, | |
| "rewards/rejected": -12.0, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 2.122186495176849, | |
| "grad_norm": 9.832993297771846, | |
| "learning_rate": 1.626936829558999e-07, | |
| "logits/chosen": -3.25, | |
| "logits/rejected": -3.53125, | |
| "logps/chosen": -138.0, | |
| "logps/rejected": -330.0, | |
| "loss": 0.0111, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 0.1591796875, | |
| "rewards/margins": 12.3125, | |
| "rewards/rejected": -12.1875, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 2.154340836012862, | |
| "grad_norm": 0.21098065783878342, | |
| "learning_rate": 1.5673420738974968e-07, | |
| "logits/chosen": -3.234375, | |
| "logits/rejected": -3.34375, | |
| "logps/chosen": -109.0, | |
| "logps/rejected": -276.0, | |
| "loss": 0.0034, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 1.03125, | |
| "rewards/margins": 12.3125, | |
| "rewards/rejected": -11.3125, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 2.1864951768488745, | |
| "grad_norm": 2.476443587212778, | |
| "learning_rate": 1.5077473182359952e-07, | |
| "logits/chosen": -3.09375, | |
| "logits/rejected": -3.375, | |
| "logps/chosen": -106.5, | |
| "logps/rejected": -320.0, | |
| "loss": 0.0043, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 0.546875, | |
| "rewards/margins": 12.75, | |
| "rewards/rejected": -12.1875, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 2.2186495176848875, | |
| "grad_norm": 11.363132455431685, | |
| "learning_rate": 1.4481525625744933e-07, | |
| "logits/chosen": -3.125, | |
| "logits/rejected": -3.40625, | |
| "logps/chosen": -134.0, | |
| "logps/rejected": -348.0, | |
| "loss": 0.0122, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": 0.7421875, | |
| "rewards/margins": 12.9375, | |
| "rewards/rejected": -12.1875, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 2.2508038585209005, | |
| "grad_norm": 2.5108344224003334, | |
| "learning_rate": 1.3885578069129917e-07, | |
| "logits/chosen": -3.203125, | |
| "logits/rejected": -3.5, | |
| "logps/chosen": -123.5, | |
| "logps/rejected": -318.0, | |
| "loss": 0.0063, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": 0.13671875, | |
| "rewards/margins": 12.5625, | |
| "rewards/rejected": -12.375, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 2.282958199356913, | |
| "grad_norm": 0.16002189769618574, | |
| "learning_rate": 1.3289630512514898e-07, | |
| "logits/chosen": -3.0625, | |
| "logits/rejected": -3.28125, | |
| "logps/chosen": -119.0, | |
| "logps/rejected": -340.0, | |
| "loss": 0.0073, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 0.96484375, | |
| "rewards/margins": 13.3125, | |
| "rewards/rejected": -12.3125, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 2.315112540192926, | |
| "grad_norm": 0.24278926553916283, | |
| "learning_rate": 1.2693682955899882e-07, | |
| "logits/chosen": -3.125, | |
| "logits/rejected": -3.296875, | |
| "logps/chosen": -136.0, | |
| "logps/rejected": -342.0, | |
| "loss": 0.001, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.1201171875, | |
| "rewards/margins": 12.875, | |
| "rewards/rejected": -13.0, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 2.347266881028939, | |
| "grad_norm": 8.462048605522668, | |
| "learning_rate": 1.2097735399284863e-07, | |
| "logits/chosen": -3.15625, | |
| "logits/rejected": -3.328125, | |
| "logps/chosen": -114.5, | |
| "logps/rejected": -298.0, | |
| "loss": 0.0085, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": 0.5546875, | |
| "rewards/margins": 12.125, | |
| "rewards/rejected": -11.5625, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 2.379421221864952, | |
| "grad_norm": 0.2836935848609144, | |
| "learning_rate": 1.1501787842669844e-07, | |
| "logits/chosen": -3.1875, | |
| "logits/rejected": -3.4375, | |
| "logps/chosen": -149.0, | |
| "logps/rejected": -342.0, | |
| "loss": 0.0095, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": 0.388671875, | |
| "rewards/margins": 12.625, | |
| "rewards/rejected": -12.25, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 2.4115755627009645, | |
| "grad_norm": 0.20905746208537096, | |
| "learning_rate": 1.0905840286054827e-07, | |
| "logits/chosen": -3.21875, | |
| "logits/rejected": -3.4375, | |
| "logps/chosen": -144.0, | |
| "logps/rejected": -336.0, | |
| "loss": 0.0024, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 0.66015625, | |
| "rewards/margins": 13.125, | |
| "rewards/rejected": -12.5, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 2.4437299035369775, | |
| "grad_norm": 0.13532730518065741, | |
| "learning_rate": 1.030989272943981e-07, | |
| "logits/chosen": -3.15625, | |
| "logits/rejected": -3.4375, | |
| "logps/chosen": -135.0, | |
| "logps/rejected": -340.0, | |
| "loss": 0.0052, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": 0.07666015625, | |
| "rewards/margins": 12.9375, | |
| "rewards/rejected": -12.875, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 2.4758842443729905, | |
| "grad_norm": 1.0359164426855658, | |
| "learning_rate": 9.713945172824791e-08, | |
| "logits/chosen": -3.09375, | |
| "logits/rejected": -3.40625, | |
| "logps/chosen": -124.5, | |
| "logps/rejected": -330.0, | |
| "loss": 0.0052, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.034912109375, | |
| "rewards/margins": 13.3125, | |
| "rewards/rejected": -13.375, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 2.508038585209003, | |
| "grad_norm": 0.02411362444063136, | |
| "learning_rate": 9.117997616209773e-08, | |
| "logits/chosen": -3.140625, | |
| "logits/rejected": -3.390625, | |
| "logps/chosen": -146.0, | |
| "logps/rejected": -356.0, | |
| "loss": 0.0025, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 0.3046875, | |
| "rewards/margins": 13.75, | |
| "rewards/rejected": -13.5, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 2.540192926045016, | |
| "grad_norm": 0.14361577541299492, | |
| "learning_rate": 8.522050059594756e-08, | |
| "logits/chosen": -3.15625, | |
| "logits/rejected": -3.40625, | |
| "logps/chosen": -148.0, | |
| "logps/rejected": -340.0, | |
| "loss": 0.0008, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 0.2734375, | |
| "rewards/margins": 13.5, | |
| "rewards/rejected": -13.25, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 2.572347266881029, | |
| "grad_norm": 3.7709142939626923, | |
| "learning_rate": 7.926102502979737e-08, | |
| "logits/chosen": -3.171875, | |
| "logits/rejected": -3.40625, | |
| "logps/chosen": -126.0, | |
| "logps/rejected": -344.0, | |
| "loss": 0.0013, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 0.62890625, | |
| "rewards/margins": 13.6875, | |
| "rewards/rejected": -13.0625, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 2.604501607717042, | |
| "grad_norm": 0.6815929170420622, | |
| "learning_rate": 7.33015494636472e-08, | |
| "logits/chosen": -3.171875, | |
| "logits/rejected": -3.421875, | |
| "logps/chosen": -122.0, | |
| "logps/rejected": -306.0, | |
| "loss": 0.0094, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 0.006866455078125, | |
| "rewards/margins": 12.75, | |
| "rewards/rejected": -12.75, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 2.6366559485530545, | |
| "grad_norm": 1.1968332787733935, | |
| "learning_rate": 6.734207389749702e-08, | |
| "logits/chosen": -3.09375, | |
| "logits/rejected": -3.4375, | |
| "logps/chosen": -110.0, | |
| "logps/rejected": -320.0, | |
| "loss": 0.0075, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": 0.94140625, | |
| "rewards/margins": 13.75, | |
| "rewards/rejected": -12.75, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 2.6688102893890675, | |
| "grad_norm": 0.2654328339493102, | |
| "learning_rate": 6.138259833134683e-08, | |
| "logits/chosen": -3.15625, | |
| "logits/rejected": -3.484375, | |
| "logps/chosen": -127.5, | |
| "logps/rejected": -328.0, | |
| "loss": 0.0081, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 0.453125, | |
| "rewards/margins": 13.9375, | |
| "rewards/rejected": -13.5, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 2.7009646302250805, | |
| "grad_norm": 0.2744508216663156, | |
| "learning_rate": 5.542312276519666e-08, | |
| "logits/chosen": -3.140625, | |
| "logits/rejected": -3.375, | |
| "logps/chosen": -119.5, | |
| "logps/rejected": -346.0, | |
| "loss": 0.005, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": 0.10107421875, | |
| "rewards/margins": 13.125, | |
| "rewards/rejected": -13.0625, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 2.733118971061093, | |
| "grad_norm": 0.03947959588228437, | |
| "learning_rate": 4.9463647199046485e-08, | |
| "logits/chosen": -3.203125, | |
| "logits/rejected": -3.34375, | |
| "logps/chosen": -120.5, | |
| "logps/rejected": -316.0, | |
| "loss": 0.0073, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": 0.224609375, | |
| "rewards/margins": 12.875, | |
| "rewards/rejected": -12.6875, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 2.765273311897106, | |
| "grad_norm": 0.6863210482116098, | |
| "learning_rate": 4.3504171632896303e-08, | |
| "logits/chosen": -3.25, | |
| "logits/rejected": -3.5, | |
| "logps/chosen": -136.0, | |
| "logps/rejected": -354.0, | |
| "loss": 0.0105, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 0.59375, | |
| "rewards/margins": 13.75, | |
| "rewards/rejected": -13.125, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 2.797427652733119, | |
| "grad_norm": 2.8280463279363106, | |
| "learning_rate": 3.754469606674612e-08, | |
| "logits/chosen": -3.0625, | |
| "logits/rejected": -3.390625, | |
| "logps/chosen": -116.5, | |
| "logps/rejected": -318.0, | |
| "loss": 0.0078, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": 0.255859375, | |
| "rewards/margins": 13.25, | |
| "rewards/rejected": -13.0, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 2.829581993569132, | |
| "grad_norm": 4.677533654072525, | |
| "learning_rate": 3.158522050059595e-08, | |
| "logits/chosen": -3.21875, | |
| "logits/rejected": -3.421875, | |
| "logps/chosen": -147.0, | |
| "logps/rejected": -312.0, | |
| "loss": 0.0227, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": 0.1044921875, | |
| "rewards/margins": 12.375, | |
| "rewards/rejected": -12.25, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 2.861736334405145, | |
| "grad_norm": 4.075516801733505, | |
| "learning_rate": 2.562574493444577e-08, | |
| "logits/chosen": -3.140625, | |
| "logits/rejected": -3.390625, | |
| "logps/chosen": -130.0, | |
| "logps/rejected": -328.0, | |
| "loss": 0.0112, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": 0.1201171875, | |
| "rewards/margins": 13.125, | |
| "rewards/rejected": -13.0, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 2.8938906752411575, | |
| "grad_norm": 0.31091685277937586, | |
| "learning_rate": 1.966626936829559e-08, | |
| "logits/chosen": -3.171875, | |
| "logits/rejected": -3.453125, | |
| "logps/chosen": -114.5, | |
| "logps/rejected": -316.0, | |
| "loss": 0.0116, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": 0.498046875, | |
| "rewards/margins": 13.3125, | |
| "rewards/rejected": -12.8125, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 2.9260450160771705, | |
| "grad_norm": 0.008132094709855322, | |
| "learning_rate": 1.370679380214541e-08, | |
| "logits/chosen": -3.125, | |
| "logits/rejected": -3.5, | |
| "logps/chosen": -124.0, | |
| "logps/rejected": -356.0, | |
| "loss": 0.0056, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 0.0111083984375, | |
| "rewards/margins": 13.625, | |
| "rewards/rejected": -13.625, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 2.958199356913183, | |
| "grad_norm": 10.508744046977588, | |
| "learning_rate": 7.747318235995233e-09, | |
| "logits/chosen": -3.28125, | |
| "logits/rejected": -3.5625, | |
| "logps/chosen": -115.0, | |
| "logps/rejected": -316.0, | |
| "loss": 0.0031, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 0.1298828125, | |
| "rewards/margins": 13.125, | |
| "rewards/rejected": -13.0, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 2.990353697749196, | |
| "grad_norm": 0.053140421396559286, | |
| "learning_rate": 1.7878426698450536e-09, | |
| "logits/chosen": -3.1875, | |
| "logits/rejected": -3.40625, | |
| "logps/chosen": -117.5, | |
| "logps/rejected": -308.0, | |
| "loss": 0.0051, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 0.203125, | |
| "rewards/margins": 12.9375, | |
| "rewards/rejected": -12.75, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_logits/chosen": -3.515625, | |
| "eval_logits/rejected": -3.59375, | |
| "eval_logps/chosen": -245.0, | |
| "eval_logps/rejected": -292.0, | |
| "eval_loss": 1.0541672706604004, | |
| "eval_rewards/accuracies": 0.6964285969734192, | |
| "eval_rewards/chosen": -5.96875, | |
| "eval_rewards/margins": 4.125, | |
| "eval_rewards/rejected": -10.0625, | |
| "eval_runtime": 15.892, | |
| "eval_samples_per_second": 12.522, | |
| "eval_steps_per_second": 0.44, | |
| "step": 933 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "step": 933, | |
| "total_flos": 0.0, | |
| "train_loss": 0.09964664405788767, | |
| "train_runtime": 5808.5282, | |
| "train_samples_per_second": 5.14, | |
| "train_steps_per_second": 0.161 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 933, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 0.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |