| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.0, | |
| "eval_steps": 100, | |
| "global_step": 2865, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0010471204188481676, | |
| "grad_norm": 5.294534803424339, | |
| "learning_rate": 1.7421602787456448e-07, | |
| "logits/chosen": -11.9375, | |
| "logits/rejected": -9.5, | |
| "logps/chosen": -420.0, | |
| "logps/rejected": -1792.0, | |
| "loss": 0.6914, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.010471204188481676, | |
| "grad_norm": 5.367171584066136, | |
| "learning_rate": 1.7421602787456445e-06, | |
| "logits/chosen": -10.9375, | |
| "logits/rejected": -8.3125, | |
| "logps/chosen": -396.0, | |
| "logps/rejected": -2040.0, | |
| "loss": 0.6886, | |
| "rewards/accuracies": 0.2569444477558136, | |
| "rewards/chosen": 0.00133514404296875, | |
| "rewards/margins": 0.009765625, | |
| "rewards/rejected": -0.0084228515625, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.020942408376963352, | |
| "grad_norm": 4.59977383551253, | |
| "learning_rate": 3.484320557491289e-06, | |
| "logits/chosen": -10.6875, | |
| "logits/rejected": -8.375, | |
| "logps/chosen": -380.0, | |
| "logps/rejected": -1936.0, | |
| "loss": 0.6314, | |
| "rewards/accuracies": 0.78125, | |
| "rewards/chosen": 0.00238037109375, | |
| "rewards/margins": 0.1201171875, | |
| "rewards/rejected": -0.11767578125, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.031413612565445025, | |
| "grad_norm": 1.7708360223618325, | |
| "learning_rate": 5.226480836236934e-06, | |
| "logits/chosen": -11.25, | |
| "logits/rejected": -8.9375, | |
| "logps/chosen": -392.0, | |
| "logps/rejected": -2096.0, | |
| "loss": 0.3399, | |
| "rewards/accuracies": 0.9437500238418579, | |
| "rewards/chosen": -0.09033203125, | |
| "rewards/margins": 1.125, | |
| "rewards/rejected": -1.21875, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.041884816753926704, | |
| "grad_norm": 1.1703071033321846, | |
| "learning_rate": 6.968641114982578e-06, | |
| "logits/chosen": -11.6875, | |
| "logits/rejected": -9.6875, | |
| "logps/chosen": -462.0, | |
| "logps/rejected": -2448.0, | |
| "loss": 0.1442, | |
| "rewards/accuracies": 0.9125000238418579, | |
| "rewards/chosen": -0.6328125, | |
| "rewards/margins": 4.9375, | |
| "rewards/rejected": -5.59375, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.05235602094240838, | |
| "grad_norm": 0.8257109366827098, | |
| "learning_rate": 8.710801393728225e-06, | |
| "logits/chosen": -12.9375, | |
| "logits/rejected": -10.6875, | |
| "logps/chosen": -492.0, | |
| "logps/rejected": -3296.0, | |
| "loss": 0.084, | |
| "rewards/accuracies": 0.9437500238418579, | |
| "rewards/chosen": -1.078125, | |
| "rewards/margins": 10.9375, | |
| "rewards/rejected": -12.0, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.06282722513089005, | |
| "grad_norm": 0.7656107565278699, | |
| "learning_rate": 1.0452961672473868e-05, | |
| "logits/chosen": -14.875, | |
| "logits/rejected": -12.625, | |
| "logps/chosen": -428.0, | |
| "logps/rejected": -3264.0, | |
| "loss": 0.1035, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -1.0390625, | |
| "rewards/margins": 12.25, | |
| "rewards/rejected": -13.25, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.07329842931937172, | |
| "grad_norm": 0.7007862852468707, | |
| "learning_rate": 1.2195121951219513e-05, | |
| "logits/chosen": -15.625, | |
| "logits/rejected": -12.9375, | |
| "logps/chosen": -496.0, | |
| "logps/rejected": -3184.0, | |
| "loss": 0.0917, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -0.9765625, | |
| "rewards/margins": 10.5625, | |
| "rewards/rejected": -11.5625, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.08376963350785341, | |
| "grad_norm": 0.7117529742981135, | |
| "learning_rate": 1.3937282229965156e-05, | |
| "logits/chosen": -15.0625, | |
| "logits/rejected": -13.0, | |
| "logps/chosen": -474.0, | |
| "logps/rejected": -3024.0, | |
| "loss": 0.0785, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -1.0703125, | |
| "rewards/margins": 10.5625, | |
| "rewards/rejected": -11.625, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.09424083769633508, | |
| "grad_norm": 1.3956151807921877, | |
| "learning_rate": 1.56794425087108e-05, | |
| "logits/chosen": -13.5625, | |
| "logits/rejected": -11.875, | |
| "logps/chosen": -482.0, | |
| "logps/rejected": -3712.0, | |
| "loss": 0.0657, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -1.4765625, | |
| "rewards/margins": 15.6875, | |
| "rewards/rejected": -17.125, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.10471204188481675, | |
| "grad_norm": 0.30084573847599866, | |
| "learning_rate": 1.742160278745645e-05, | |
| "logits/chosen": -13.8125, | |
| "logits/rejected": -12.1875, | |
| "logps/chosen": -632.0, | |
| "logps/rejected": -3808.0, | |
| "loss": 0.0972, | |
| "rewards/accuracies": 0.96875, | |
| "rewards/chosen": -1.9375, | |
| "rewards/margins": 16.875, | |
| "rewards/rejected": -18.75, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.10471204188481675, | |
| "eval_logits/chosen": -13.75, | |
| "eval_logits/rejected": -13.625, | |
| "eval_logps/chosen": -454.0, | |
| "eval_logps/rejected": -446.0, | |
| "eval_loss": 0.7388749718666077, | |
| "eval_rewards/accuracies": 0.552734375, | |
| "eval_rewards/chosen": -1.3515625, | |
| "eval_rewards/margins": 0.220703125, | |
| "eval_rewards/rejected": -1.578125, | |
| "eval_runtime": 49.1993, | |
| "eval_samples_per_second": 40.651, | |
| "eval_steps_per_second": 0.65, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.11518324607329843, | |
| "grad_norm": 1.8377225141738138, | |
| "learning_rate": 1.9163763066202093e-05, | |
| "logits/chosen": -14.1875, | |
| "logits/rejected": -12.375, | |
| "logps/chosen": -528.0, | |
| "logps/rejected": -3968.0, | |
| "loss": 0.0811, | |
| "rewards/accuracies": 0.96875, | |
| "rewards/chosen": -1.4921875, | |
| "rewards/margins": 17.875, | |
| "rewards/rejected": -19.375, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.1256544502617801, | |
| "grad_norm": 1.644495224043385, | |
| "learning_rate": 2.0905923344947736e-05, | |
| "logits/chosen": -12.5, | |
| "logits/rejected": -11.1875, | |
| "logps/chosen": -454.0, | |
| "logps/rejected": -3872.0, | |
| "loss": 0.0645, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -1.171875, | |
| "rewards/margins": 17.5, | |
| "rewards/rejected": -18.75, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.13612565445026178, | |
| "grad_norm": 1.8727105167383522, | |
| "learning_rate": 2.264808362369338e-05, | |
| "logits/chosen": -13.0625, | |
| "logits/rejected": -11.625, | |
| "logps/chosen": -552.0, | |
| "logps/rejected": -4256.0, | |
| "loss": 0.0718, | |
| "rewards/accuracies": 0.956250011920929, | |
| "rewards/chosen": -1.7578125, | |
| "rewards/margins": 20.5, | |
| "rewards/rejected": -22.25, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.14659685863874344, | |
| "grad_norm": 1.3826585814060648, | |
| "learning_rate": 2.4390243902439026e-05, | |
| "logits/chosen": -14.4375, | |
| "logits/rejected": -13.125, | |
| "logps/chosen": -600.0, | |
| "logps/rejected": -3680.0, | |
| "loss": 0.0573, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": -1.859375, | |
| "rewards/margins": 16.875, | |
| "rewards/rejected": -18.75, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.15706806282722513, | |
| "grad_norm": 0.9509856503717125, | |
| "learning_rate": 2.6132404181184672e-05, | |
| "logits/chosen": -12.625, | |
| "logits/rejected": -11.125, | |
| "logps/chosen": -544.0, | |
| "logps/rejected": -3584.0, | |
| "loss": 0.0725, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -1.5390625, | |
| "rewards/margins": 14.125, | |
| "rewards/rejected": -15.6875, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.16753926701570682, | |
| "grad_norm": 0.6492092710859492, | |
| "learning_rate": 2.7874564459930312e-05, | |
| "logits/chosen": -11.25, | |
| "logits/rejected": -9.5625, | |
| "logps/chosen": -544.0, | |
| "logps/rejected": -3488.0, | |
| "loss": 0.0648, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": -1.78125, | |
| "rewards/margins": 13.75, | |
| "rewards/rejected": -15.5, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.17801047120418848, | |
| "grad_norm": 1.3510046947357488, | |
| "learning_rate": 2.9616724738675962e-05, | |
| "logits/chosen": -11.8125, | |
| "logits/rejected": -10.3125, | |
| "logps/chosen": -536.0, | |
| "logps/rejected": -3616.0, | |
| "loss": 0.0538, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -1.4296875, | |
| "rewards/margins": 14.6875, | |
| "rewards/rejected": -16.125, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.18848167539267016, | |
| "grad_norm": 0.21771027438036225, | |
| "learning_rate": 3.13588850174216e-05, | |
| "logits/chosen": -12.875, | |
| "logits/rejected": -11.4375, | |
| "logps/chosen": -568.0, | |
| "logps/rejected": -3584.0, | |
| "loss": 0.0822, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -1.703125, | |
| "rewards/margins": 15.5, | |
| "rewards/rejected": -17.25, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.19895287958115182, | |
| "grad_norm": 1.3545568551652185, | |
| "learning_rate": 3.310104529616725e-05, | |
| "logits/chosen": -11.9375, | |
| "logits/rejected": -10.25, | |
| "logps/chosen": -506.0, | |
| "logps/rejected": -3648.0, | |
| "loss": 0.056, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -1.375, | |
| "rewards/margins": 14.75, | |
| "rewards/rejected": -16.125, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.2094240837696335, | |
| "grad_norm": 0.881724805173648, | |
| "learning_rate": 3.48432055749129e-05, | |
| "logits/chosen": -12.0, | |
| "logits/rejected": -10.1875, | |
| "logps/chosen": -728.0, | |
| "logps/rejected": -4224.0, | |
| "loss": 0.0661, | |
| "rewards/accuracies": 0.949999988079071, | |
| "rewards/chosen": -3.109375, | |
| "rewards/margins": 19.125, | |
| "rewards/rejected": -22.25, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.2094240837696335, | |
| "eval_logits/chosen": -12.9375, | |
| "eval_logits/rejected": -12.6875, | |
| "eval_logps/chosen": -560.0, | |
| "eval_logps/rejected": -568.0, | |
| "eval_loss": 0.9070937633514404, | |
| "eval_rewards/accuracies": 0.552734375, | |
| "eval_rewards/chosen": -2.421875, | |
| "eval_rewards/margins": 0.37109375, | |
| "eval_rewards/rejected": -2.796875, | |
| "eval_runtime": 46.7056, | |
| "eval_samples_per_second": 42.821, | |
| "eval_steps_per_second": 0.685, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.2198952879581152, | |
| "grad_norm": 0.8063827549996978, | |
| "learning_rate": 3.6585365853658535e-05, | |
| "logits/chosen": -14.125, | |
| "logits/rejected": -11.0625, | |
| "logps/chosen": -490.0, | |
| "logps/rejected": -3232.0, | |
| "loss": 0.0594, | |
| "rewards/accuracies": 0.949999988079071, | |
| "rewards/chosen": -1.515625, | |
| "rewards/margins": 10.9375, | |
| "rewards/rejected": -12.5, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.23036649214659685, | |
| "grad_norm": 1.9446003589320062, | |
| "learning_rate": 3.8327526132404185e-05, | |
| "logits/chosen": -13.375, | |
| "logits/rejected": -11.125, | |
| "logps/chosen": -600.0, | |
| "logps/rejected": -3344.0, | |
| "loss": 0.0516, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -2.03125, | |
| "rewards/margins": 12.5, | |
| "rewards/rejected": -14.5, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.24083769633507854, | |
| "grad_norm": 0.5815073948131051, | |
| "learning_rate": 4.006968641114983e-05, | |
| "logits/chosen": -13.0, | |
| "logits/rejected": -10.8125, | |
| "logps/chosen": -604.0, | |
| "logps/rejected": -4016.0, | |
| "loss": 0.0666, | |
| "rewards/accuracies": 0.956250011920929, | |
| "rewards/chosen": -2.46875, | |
| "rewards/margins": 18.75, | |
| "rewards/rejected": -21.25, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.2513089005235602, | |
| "grad_norm": 0.4604815018249142, | |
| "learning_rate": 4.181184668989547e-05, | |
| "logits/chosen": -13.375, | |
| "logits/rejected": -10.9375, | |
| "logps/chosen": -572.0, | |
| "logps/rejected": -4000.0, | |
| "loss": 0.0619, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -2.28125, | |
| "rewards/margins": 18.375, | |
| "rewards/rejected": -20.625, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.2617801047120419, | |
| "grad_norm": 0.6578003098209557, | |
| "learning_rate": 4.3554006968641115e-05, | |
| "logits/chosen": -11.9375, | |
| "logits/rejected": -10.0625, | |
| "logps/chosen": -612.0, | |
| "logps/rejected": -3504.0, | |
| "loss": 0.0639, | |
| "rewards/accuracies": 0.96875, | |
| "rewards/chosen": -2.015625, | |
| "rewards/margins": 14.625, | |
| "rewards/rejected": -16.625, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.27225130890052357, | |
| "grad_norm": 1.2975457720860102, | |
| "learning_rate": 4.529616724738676e-05, | |
| "logits/chosen": -9.25, | |
| "logits/rejected": -7.9375, | |
| "logps/chosen": -604.0, | |
| "logps/rejected": -4000.0, | |
| "loss": 0.0601, | |
| "rewards/accuracies": 0.949999988079071, | |
| "rewards/chosen": -2.546875, | |
| "rewards/margins": 18.125, | |
| "rewards/rejected": -20.625, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.28272251308900526, | |
| "grad_norm": 0.8550507515013447, | |
| "learning_rate": 4.703832752613241e-05, | |
| "logits/chosen": -12.5, | |
| "logits/rejected": -10.5, | |
| "logps/chosen": -676.0, | |
| "logps/rejected": -3920.0, | |
| "loss": 0.1163, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -3.28125, | |
| "rewards/margins": 18.125, | |
| "rewards/rejected": -21.375, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.2931937172774869, | |
| "grad_norm": 0.7293051061360788, | |
| "learning_rate": 4.878048780487805e-05, | |
| "logits/chosen": -15.5625, | |
| "logits/rejected": -12.625, | |
| "logps/chosen": -592.0, | |
| "logps/rejected": -3056.0, | |
| "loss": 0.0625, | |
| "rewards/accuracies": 0.96875, | |
| "rewards/chosen": -2.578125, | |
| "rewards/margins": 9.9375, | |
| "rewards/rejected": -12.5, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.3036649214659686, | |
| "grad_norm": 0.5328107937258684, | |
| "learning_rate": 4.999983293481418e-05, | |
| "logits/chosen": -15.75, | |
| "logits/rejected": -12.0625, | |
| "logps/chosen": -600.0, | |
| "logps/rejected": -3584.0, | |
| "loss": 0.0617, | |
| "rewards/accuracies": 0.96875, | |
| "rewards/chosen": -2.28125, | |
| "rewards/margins": 13.9375, | |
| "rewards/rejected": -16.25, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.31413612565445026, | |
| "grad_norm": 1.1466823320197184, | |
| "learning_rate": 4.9996862949181595e-05, | |
| "logits/chosen": -14.5625, | |
| "logits/rejected": -10.9375, | |
| "logps/chosen": -596.0, | |
| "logps/rejected": -3648.0, | |
| "loss": 0.0937, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": -2.203125, | |
| "rewards/margins": 14.75, | |
| "rewards/rejected": -16.875, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.31413612565445026, | |
| "eval_logits/chosen": -15.0, | |
| "eval_logits/rejected": -14.6875, | |
| "eval_logps/chosen": -624.0, | |
| "eval_logps/rejected": -624.0, | |
| "eval_loss": 0.9645781517028809, | |
| "eval_rewards/accuracies": 0.515625, | |
| "eval_rewards/chosen": -3.046875, | |
| "eval_rewards/margins": 0.318359375, | |
| "eval_rewards/rejected": -3.375, | |
| "eval_runtime": 46.7091, | |
| "eval_samples_per_second": 42.818, | |
| "eval_steps_per_second": 0.685, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.32460732984293195, | |
| "grad_norm": 0.9240446323921724, | |
| "learning_rate": 4.9990180911526896e-05, | |
| "logits/chosen": -14.9375, | |
| "logits/rejected": -10.75, | |
| "logps/chosen": -588.0, | |
| "logps/rejected": -4320.0, | |
| "loss": 0.0703, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -2.390625, | |
| "rewards/margins": 19.25, | |
| "rewards/rejected": -21.625, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.33507853403141363, | |
| "grad_norm": 3.1706203769916326, | |
| "learning_rate": 4.997978781413745e-05, | |
| "logits/chosen": -14.875, | |
| "logits/rejected": -11.875, | |
| "logps/chosen": -700.0, | |
| "logps/rejected": -3904.0, | |
| "loss": 0.1103, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": -2.84375, | |
| "rewards/margins": 17.5, | |
| "rewards/rejected": -20.25, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.34554973821989526, | |
| "grad_norm": 6.512271998410246, | |
| "learning_rate": 4.996568520039565e-05, | |
| "logits/chosen": -15.6875, | |
| "logits/rejected": -12.1875, | |
| "logps/chosen": -668.0, | |
| "logps/rejected": -4512.0, | |
| "loss": 0.0958, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -2.921875, | |
| "rewards/margins": 22.75, | |
| "rewards/rejected": -25.75, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.35602094240837695, | |
| "grad_norm": 1.6902543940279116, | |
| "learning_rate": 4.994787516454973e-05, | |
| "logits/chosen": -15.25, | |
| "logits/rejected": -11.0625, | |
| "logps/chosen": -720.0, | |
| "logps/rejected": -5760.0, | |
| "loss": 0.0711, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -3.5625, | |
| "rewards/margins": 34.5, | |
| "rewards/rejected": -38.0, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.36649214659685864, | |
| "grad_norm": 0.9686314628160185, | |
| "learning_rate": 4.992636035140273e-05, | |
| "logits/chosen": -14.4375, | |
| "logits/rejected": -10.5, | |
| "logps/chosen": -628.0, | |
| "logps/rejected": -4640.0, | |
| "loss": 0.0825, | |
| "rewards/accuracies": 0.96875, | |
| "rewards/chosen": -2.828125, | |
| "rewards/margins": 23.625, | |
| "rewards/rejected": -26.5, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.3769633507853403, | |
| "grad_norm": 0.1884420574446141, | |
| "learning_rate": 4.99011439559198e-05, | |
| "logits/chosen": -13.25, | |
| "logits/rejected": -9.3125, | |
| "logps/chosen": -640.0, | |
| "logps/rejected": -3904.0, | |
| "loss": 0.0509, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -2.625, | |
| "rewards/margins": 17.5, | |
| "rewards/rejected": -20.125, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.387434554973822, | |
| "grad_norm": 0.8819465762268481, | |
| "learning_rate": 4.987222972275369e-05, | |
| "logits/chosen": -11.9375, | |
| "logits/rejected": -9.4375, | |
| "logps/chosen": -616.0, | |
| "logps/rejected": -3632.0, | |
| "loss": 0.0513, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -2.5, | |
| "rewards/margins": 14.125, | |
| "rewards/rejected": -16.625, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.39790575916230364, | |
| "grad_norm": 1.3374420877268833, | |
| "learning_rate": 4.983962194568872e-05, | |
| "logits/chosen": -12.125, | |
| "logits/rejected": -9.1875, | |
| "logps/chosen": -700.0, | |
| "logps/rejected": -4032.0, | |
| "loss": 0.0828, | |
| "rewards/accuracies": 0.96875, | |
| "rewards/chosen": -3.140625, | |
| "rewards/margins": 16.125, | |
| "rewards/rejected": -19.375, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.4083769633507853, | |
| "grad_norm": 2.670190812947824, | |
| "learning_rate": 4.9803325467003084e-05, | |
| "logits/chosen": -13.0625, | |
| "logits/rejected": -9.4375, | |
| "logps/chosen": -732.0, | |
| "logps/rejected": -4224.0, | |
| "loss": 0.0645, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -3.484375, | |
| "rewards/margins": 19.75, | |
| "rewards/rejected": -23.25, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.418848167539267, | |
| "grad_norm": 1.300302144580882, | |
| "learning_rate": 4.9763345676749816e-05, | |
| "logits/chosen": -13.25, | |
| "logits/rejected": -9.4375, | |
| "logps/chosen": -716.0, | |
| "logps/rejected": -5248.0, | |
| "loss": 0.183, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -3.515625, | |
| "rewards/margins": 29.0, | |
| "rewards/rejected": -32.5, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.418848167539267, | |
| "eval_logits/chosen": -14.125, | |
| "eval_logits/rejected": -13.875, | |
| "eval_logps/chosen": -616.0, | |
| "eval_logps/rejected": -596.0, | |
| "eval_loss": 0.9572968482971191, | |
| "eval_rewards/accuracies": 0.5, | |
| "eval_rewards/chosen": -2.96875, | |
| "eval_rewards/margins": 0.08984375, | |
| "eval_rewards/rejected": -3.0625, | |
| "eval_runtime": 46.6892, | |
| "eval_samples_per_second": 42.836, | |
| "eval_steps_per_second": 0.685, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.4293193717277487, | |
| "grad_norm": 1.5044102990853008, | |
| "learning_rate": 4.9719688511956374e-05, | |
| "logits/chosen": -14.4375, | |
| "logits/rejected": -9.5625, | |
| "logps/chosen": -680.0, | |
| "logps/rejected": -4096.0, | |
| "loss": 0.1, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": -2.84375, | |
| "rewards/margins": 18.25, | |
| "rewards/rejected": -21.0, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.4397905759162304, | |
| "grad_norm": 0.5100007845935307, | |
| "learning_rate": 4.967236045574294e-05, | |
| "logits/chosen": -14.9375, | |
| "logits/rejected": -9.5, | |
| "logps/chosen": -620.0, | |
| "logps/rejected": -3488.0, | |
| "loss": 0.0828, | |
| "rewards/accuracies": 0.949999988079071, | |
| "rewards/chosen": -2.4375, | |
| "rewards/margins": 12.125, | |
| "rewards/rejected": -14.5625, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.450261780104712, | |
| "grad_norm": 1.02294811267431, | |
| "learning_rate": 4.96213685363597e-05, | |
| "logits/chosen": -15.875, | |
| "logits/rejected": -10.9375, | |
| "logps/chosen": -644.0, | |
| "logps/rejected": -3776.0, | |
| "loss": 0.0823, | |
| "rewards/accuracies": 0.9437500238418579, | |
| "rewards/chosen": -2.828125, | |
| "rewards/margins": 16.75, | |
| "rewards/rejected": -19.5, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.4607329842931937, | |
| "grad_norm": 1.7346999214052725, | |
| "learning_rate": 4.956672032614315e-05, | |
| "logits/chosen": -15.0, | |
| "logits/rejected": -10.5, | |
| "logps/chosen": -724.0, | |
| "logps/rejected": -3840.0, | |
| "loss": 0.0442, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -3.328125, | |
| "rewards/margins": 15.6875, | |
| "rewards/rejected": -19.0, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.4712041884816754, | |
| "grad_norm": 0.48995260381026173, | |
| "learning_rate": 4.950842394039157e-05, | |
| "logits/chosen": -16.375, | |
| "logits/rejected": -10.75, | |
| "logps/chosen": -704.0, | |
| "logps/rejected": -4192.0, | |
| "loss": 0.0591, | |
| "rewards/accuracies": 0.96875, | |
| "rewards/chosen": -3.5, | |
| "rewards/margins": 18.125, | |
| "rewards/rejected": -21.625, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.4816753926701571, | |
| "grad_norm": 0.2055426155067474, | |
| "learning_rate": 4.944648803615994e-05, | |
| "logits/chosen": -17.125, | |
| "logits/rejected": -12.3125, | |
| "logps/chosen": -736.0, | |
| "logps/rejected": -4192.0, | |
| "loss": 0.0522, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -3.453125, | |
| "rewards/margins": 18.875, | |
| "rewards/rejected": -22.25, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.49214659685863876, | |
| "grad_norm": 0.32077309628293404, | |
| "learning_rate": 4.938092181097433e-05, | |
| "logits/chosen": -15.5625, | |
| "logits/rejected": -11.25, | |
| "logps/chosen": -680.0, | |
| "logps/rejected": -4160.0, | |
| "loss": 0.0652, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -3.03125, | |
| "rewards/margins": 19.0, | |
| "rewards/rejected": -22.0, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.5026178010471204, | |
| "grad_norm": 4.474490964734582, | |
| "learning_rate": 4.931173500146607e-05, | |
| "logits/chosen": -15.0, | |
| "logits/rejected": -10.125, | |
| "logps/chosen": -668.0, | |
| "logps/rejected": -4448.0, | |
| "loss": 0.087, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -3.21875, | |
| "rewards/margins": 20.375, | |
| "rewards/rejected": -23.5, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.5130890052356021, | |
| "grad_norm": 0.5610180125895814, | |
| "learning_rate": 4.923893788192582e-05, | |
| "logits/chosen": -13.125, | |
| "logits/rejected": -5.9375, | |
| "logps/chosen": -780.0, | |
| "logps/rejected": -4128.0, | |
| "loss": 0.06, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -3.984375, | |
| "rewards/margins": 18.75, | |
| "rewards/rejected": -22.75, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.5235602094240838, | |
| "grad_norm": 1.3486779373172177, | |
| "learning_rate": 4.9162541262777906e-05, | |
| "logits/chosen": -12.9375, | |
| "logits/rejected": -4.03125, | |
| "logps/chosen": -816.0, | |
| "logps/rejected": -4896.0, | |
| "loss": 0.0827, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": -4.40625, | |
| "rewards/margins": 24.25, | |
| "rewards/rejected": -28.625, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.5235602094240838, | |
| "eval_logits/chosen": -12.875, | |
| "eval_logits/rejected": -12.5625, | |
| "eval_logps/chosen": -692.0, | |
| "eval_logps/rejected": -700.0, | |
| "eval_loss": 1.1729687452316284, | |
| "eval_rewards/accuracies": 0.490234375, | |
| "eval_rewards/chosen": -3.734375, | |
| "eval_rewards/margins": 0.388671875, | |
| "eval_rewards/rejected": -4.125, | |
| "eval_runtime": 46.6967, | |
| "eval_samples_per_second": 42.83, | |
| "eval_steps_per_second": 0.685, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.5340314136125655, | |
| "grad_norm": 0.8745124436798694, | |
| "learning_rate": 4.90825564889749e-05, | |
| "logits/chosen": -14.0, | |
| "logits/rejected": -6.53125, | |
| "logps/chosen": -796.0, | |
| "logps/rejected": -4048.0, | |
| "loss": 0.0923, | |
| "rewards/accuracies": 0.956250011920929, | |
| "rewards/chosen": -3.859375, | |
| "rewards/margins": 17.375, | |
| "rewards/rejected": -21.25, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.5445026178010471, | |
| "grad_norm": 4.483578603706967, | |
| "learning_rate": 4.899899543831288e-05, | |
| "logits/chosen": -12.625, | |
| "logits/rejected": -8.0625, | |
| "logps/chosen": -732.0, | |
| "logps/rejected": -4096.0, | |
| "loss": 0.069, | |
| "rewards/accuracies": 0.96875, | |
| "rewards/chosen": -3.1875, | |
| "rewards/margins": 17.25, | |
| "rewards/rejected": -20.5, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.5549738219895288, | |
| "grad_norm": 0.8721470857743174, | |
| "learning_rate": 4.891187051966765e-05, | |
| "logits/chosen": -13.875, | |
| "logits/rejected": -9.8125, | |
| "logps/chosen": -688.0, | |
| "logps/rejected": -3856.0, | |
| "loss": 0.0578, | |
| "rewards/accuracies": 0.96875, | |
| "rewards/chosen": -2.984375, | |
| "rewards/margins": 16.5, | |
| "rewards/rejected": -19.375, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.5654450261780105, | |
| "grad_norm": 2.761766525086082, | |
| "learning_rate": 4.882119467115191e-05, | |
| "logits/chosen": -16.5, | |
| "logits/rejected": -10.25, | |
| "logps/chosen": -680.0, | |
| "logps/rejected": -4416.0, | |
| "loss": 0.1286, | |
| "rewards/accuracies": 0.96875, | |
| "rewards/chosen": -3.046875, | |
| "rewards/margins": 21.5, | |
| "rewards/rejected": -24.625, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.5759162303664922, | |
| "grad_norm": 0.41499842423599675, | |
| "learning_rate": 4.872698135819402e-05, | |
| "logits/chosen": -16.375, | |
| "logits/rejected": -8.4375, | |
| "logps/chosen": -704.0, | |
| "logps/rejected": -4960.0, | |
| "loss": 0.0615, | |
| "rewards/accuracies": 0.96875, | |
| "rewards/chosen": -3.40625, | |
| "rewards/margins": 25.375, | |
| "rewards/rejected": -28.75, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.5863874345549738, | |
| "grad_norm": 0.1200145160347875, | |
| "learning_rate": 4.862924457153831e-05, | |
| "logits/chosen": -15.625, | |
| "logits/rejected": -7.75, | |
| "logps/chosen": -632.0, | |
| "logps/rejected": -4480.0, | |
| "loss": 0.051, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -3.140625, | |
| "rewards/margins": 20.375, | |
| "rewards/rejected": -23.5, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.5968586387434555, | |
| "grad_norm": 0.36302533563513756, | |
| "learning_rate": 4.8527998825167484e-05, | |
| "logits/chosen": -16.125, | |
| "logits/rejected": -9.1875, | |
| "logps/chosen": -660.0, | |
| "logps/rejected": -3616.0, | |
| "loss": 0.0425, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -3.0, | |
| "rewards/margins": 15.0625, | |
| "rewards/rejected": -18.125, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.6073298429319371, | |
| "grad_norm": 1.3546319923494652, | |
| "learning_rate": 4.842325915414728e-05, | |
| "logits/chosen": -17.0, | |
| "logits/rejected": -10.125, | |
| "logps/chosen": -708.0, | |
| "logps/rejected": -4608.0, | |
| "loss": 0.0429, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -3.640625, | |
| "rewards/margins": 23.125, | |
| "rewards/rejected": -26.75, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.6178010471204188, | |
| "grad_norm": 0.6599663206698803, | |
| "learning_rate": 4.831504111239373e-05, | |
| "logits/chosen": -18.25, | |
| "logits/rejected": -12.0, | |
| "logps/chosen": -800.0, | |
| "logps/rejected": -4864.0, | |
| "loss": 0.056, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -3.96875, | |
| "rewards/margins": 25.0, | |
| "rewards/rejected": -29.0, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.6282722513089005, | |
| "grad_norm": 0.40480679105115547, | |
| "learning_rate": 4.8203360770363425e-05, | |
| "logits/chosen": -19.375, | |
| "logits/rejected": -14.0625, | |
| "logps/chosen": -708.0, | |
| "logps/rejected": -4320.0, | |
| "loss": 0.07, | |
| "rewards/accuracies": 0.96875, | |
| "rewards/chosen": -3.21875, | |
| "rewards/margins": 22.125, | |
| "rewards/rejected": -25.375, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.6282722513089005, | |
| "eval_logits/chosen": -19.0, | |
| "eval_logits/rejected": -19.125, | |
| "eval_logps/chosen": -620.0, | |
| "eval_logps/rejected": -616.0, | |
| "eval_loss": 1.064843773841858, | |
| "eval_rewards/accuracies": 0.517578125, | |
| "eval_rewards/chosen": -3.0, | |
| "eval_rewards/margins": 0.26953125, | |
| "eval_rewards/rejected": -3.265625, | |
| "eval_runtime": 46.6672, | |
| "eval_samples_per_second": 42.857, | |
| "eval_steps_per_second": 0.686, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.6387434554973822, | |
| "grad_norm": 0.9536577561440287, | |
| "learning_rate": 4.808823471266702e-05, | |
| "logits/chosen": -19.125, | |
| "logits/rejected": -12.125, | |
| "logps/chosen": -640.0, | |
| "logps/rejected": -4064.0, | |
| "loss": 0.057, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -2.75, | |
| "rewards/margins": 19.375, | |
| "rewards/rejected": -22.125, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.6492146596858639, | |
| "grad_norm": 1.9626062783500162, | |
| "learning_rate": 4.796968003560639e-05, | |
| "logits/chosen": -18.625, | |
| "logits/rejected": -11.6875, | |
| "logps/chosen": -696.0, | |
| "logps/rejected": -5536.0, | |
| "loss": 0.0798, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -3.234375, | |
| "rewards/margins": 31.0, | |
| "rewards/rejected": -34.25, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.6596858638743456, | |
| "grad_norm": 0.46846523468891155, | |
| "learning_rate": 4.7847714344635864e-05, | |
| "logits/chosen": -18.125, | |
| "logits/rejected": -11.75, | |
| "logps/chosen": -720.0, | |
| "logps/rejected": -5216.0, | |
| "loss": 0.061, | |
| "rewards/accuracies": 0.956250011920929, | |
| "rewards/chosen": -3.59375, | |
| "rewards/margins": 29.875, | |
| "rewards/rejected": -33.5, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.6701570680628273, | |
| "grad_norm": 3.7729051122717703, | |
| "learning_rate": 4.772235575174776e-05, | |
| "logits/chosen": -16.5, | |
| "logits/rejected": -10.6875, | |
| "logps/chosen": -792.0, | |
| "logps/rejected": -5184.0, | |
| "loss": 0.0747, | |
| "rewards/accuracies": 0.949999988079071, | |
| "rewards/chosen": -4.03125, | |
| "rewards/margins": 27.75, | |
| "rewards/rejected": -31.75, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.680628272251309, | |
| "grad_norm": 2.3916018370407093, | |
| "learning_rate": 4.7593622872782754e-05, | |
| "logits/chosen": -18.125, | |
| "logits/rejected": -12.375, | |
| "logps/chosen": -624.0, | |
| "logps/rejected": -4704.0, | |
| "loss": 0.0521, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -2.6875, | |
| "rewards/margins": 25.5, | |
| "rewards/rejected": -28.25, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.6910994764397905, | |
| "grad_norm": 4.308132901198597, | |
| "learning_rate": 4.746153482466542e-05, | |
| "logits/chosen": -18.625, | |
| "logits/rejected": -11.0, | |
| "logps/chosen": -696.0, | |
| "logps/rejected": -6976.0, | |
| "loss": 0.0777, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -3.40625, | |
| "rewards/margins": 46.75, | |
| "rewards/rejected": -50.0, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.7015706806282722, | |
| "grad_norm": 3.473901409726854, | |
| "learning_rate": 4.732611122256537e-05, | |
| "logits/chosen": -17.5, | |
| "logits/rejected": -10.125, | |
| "logps/chosen": -796.0, | |
| "logps/rejected": -7040.0, | |
| "loss": 0.071, | |
| "rewards/accuracies": 0.9437500238418579, | |
| "rewards/chosen": -3.875, | |
| "rewards/margins": 47.25, | |
| "rewards/rejected": -51.25, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.7120418848167539, | |
| "grad_norm": 1.0361518162052883, | |
| "learning_rate": 4.718737217698435e-05, | |
| "logits/chosen": -16.625, | |
| "logits/rejected": -10.8125, | |
| "logps/chosen": -676.0, | |
| "logps/rejected": -4576.0, | |
| "loss": 0.1198, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -3.15625, | |
| "rewards/margins": 22.625, | |
| "rewards/rejected": -25.75, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.7225130890052356, | |
| "grad_norm": 1.2611218236257373, | |
| "learning_rate": 4.704533829076982e-05, | |
| "logits/chosen": -14.9375, | |
| "logits/rejected": -11.6875, | |
| "logps/chosen": -664.0, | |
| "logps/rejected": -3200.0, | |
| "loss": 0.1049, | |
| "rewards/accuracies": 0.956250011920929, | |
| "rewards/chosen": -2.875, | |
| "rewards/margins": 9.25, | |
| "rewards/rejected": -12.125, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.7329842931937173, | |
| "grad_norm": 5.014124395550331, | |
| "learning_rate": 4.6900030656055455e-05, | |
| "logits/chosen": -11.6875, | |
| "logits/rejected": -9.6875, | |
| "logps/chosen": -668.0, | |
| "logps/rejected": -3280.0, | |
| "loss": 0.0789, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -2.875, | |
| "rewards/margins": 10.125, | |
| "rewards/rejected": -13.0, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.7329842931937173, | |
| "eval_logits/chosen": -12.5, | |
| "eval_logits/rejected": -12.0, | |
| "eval_logps/chosen": -604.0, | |
| "eval_logps/rejected": -584.0, | |
| "eval_loss": 0.8502343893051147, | |
| "eval_rewards/accuracies": 0.5078125, | |
| "eval_rewards/chosen": -2.84375, | |
| "eval_rewards/margins": 0.099609375, | |
| "eval_rewards/rejected": -2.953125, | |
| "eval_runtime": 46.6971, | |
| "eval_samples_per_second": 42.829, | |
| "eval_steps_per_second": 0.685, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.743455497382199, | |
| "grad_norm": 0.877658360964006, | |
| "learning_rate": 4.6751470851128906e-05, | |
| "logits/chosen": -14.1875, | |
| "logits/rejected": -11.6875, | |
| "logps/chosen": -692.0, | |
| "logps/rejected": -3424.0, | |
| "loss": 0.0485, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": -2.890625, | |
| "rewards/margins": 12.6875, | |
| "rewards/rejected": -15.5625, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.7539267015706806, | |
| "grad_norm": 0.8933101260289281, | |
| "learning_rate": 4.6599680937227454e-05, | |
| "logits/chosen": -15.125, | |
| "logits/rejected": -12.4375, | |
| "logps/chosen": -688.0, | |
| "logps/rejected": -3936.0, | |
| "loss": 0.0603, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -2.984375, | |
| "rewards/margins": 16.5, | |
| "rewards/rejected": -19.5, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.7643979057591623, | |
| "grad_norm": 0.2745144413043669, | |
| "learning_rate": 4.644468345526183e-05, | |
| "logits/chosen": -17.375, | |
| "logits/rejected": -13.75, | |
| "logps/chosen": -664.0, | |
| "logps/rejected": -3776.0, | |
| "loss": 0.0566, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -2.8125, | |
| "rewards/margins": 14.625, | |
| "rewards/rejected": -17.375, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.774869109947644, | |
| "grad_norm": 2.8617740847265614, | |
| "learning_rate": 4.628650142246894e-05, | |
| "logits/chosen": -18.25, | |
| "logits/rejected": -14.0625, | |
| "logps/chosen": -612.0, | |
| "logps/rejected": -3312.0, | |
| "loss": 0.0732, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -2.578125, | |
| "rewards/margins": 11.25, | |
| "rewards/rejected": -13.8125, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.7853403141361257, | |
| "grad_norm": 1.5773222299838061, | |
| "learning_rate": 4.6125158328993765e-05, | |
| "logits/chosen": -18.5, | |
| "logits/rejected": -13.8125, | |
| "logps/chosen": -664.0, | |
| "logps/rejected": -3360.0, | |
| "loss": 0.0662, | |
| "rewards/accuracies": 0.96875, | |
| "rewards/chosen": -3.015625, | |
| "rewards/margins": 10.5625, | |
| "rewards/rejected": -13.5625, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.7958115183246073, | |
| "grad_norm": 1.3483888057453328, | |
| "learning_rate": 4.5960678134401064e-05, | |
| "logits/chosen": -18.5, | |
| "logits/rejected": -10.8125, | |
| "logps/chosen": -716.0, | |
| "logps/rejected": -3632.0, | |
| "loss": 0.0875, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -3.390625, | |
| "rewards/margins": 13.5625, | |
| "rewards/rejected": -16.875, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.806282722513089, | |
| "grad_norm": 0.7522302649894089, | |
| "learning_rate": 4.579308526411734e-05, | |
| "logits/chosen": -18.5, | |
| "logits/rejected": -10.375, | |
| "logps/chosen": -648.0, | |
| "logps/rejected": -3760.0, | |
| "loss": 0.0513, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -2.921875, | |
| "rewards/margins": 15.0, | |
| "rewards/rejected": -17.875, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.8167539267015707, | |
| "grad_norm": 1.8247878739965429, | |
| "learning_rate": 4.562240460580367e-05, | |
| "logits/chosen": -18.375, | |
| "logits/rejected": -10.5, | |
| "logps/chosen": -680.0, | |
| "logps/rejected": -3856.0, | |
| "loss": 0.0483, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -3.078125, | |
| "rewards/margins": 15.8125, | |
| "rewards/rejected": -18.875, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.8272251308900523, | |
| "grad_norm": 0.5943538720238456, | |
| "learning_rate": 4.544866150565985e-05, | |
| "logits/chosen": -17.75, | |
| "logits/rejected": -10.125, | |
| "logps/chosen": -704.0, | |
| "logps/rejected": -3952.0, | |
| "loss": 0.0476, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -3.359375, | |
| "rewards/margins": 16.5, | |
| "rewards/rejected": -19.875, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.837696335078534, | |
| "grad_norm": 1.0306252352234817, | |
| "learning_rate": 4.527188176466051e-05, | |
| "logits/chosen": -17.375, | |
| "logits/rejected": -10.1875, | |
| "logps/chosen": -800.0, | |
| "logps/rejected": -4192.0, | |
| "loss": 0.061, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -4.0625, | |
| "rewards/margins": 18.125, | |
| "rewards/rejected": -22.125, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.837696335078534, | |
| "eval_logits/chosen": -17.625, | |
| "eval_logits/rejected": -17.375, | |
| "eval_logps/chosen": -712.0, | |
| "eval_logps/rejected": -708.0, | |
| "eval_loss": 1.201812505722046, | |
| "eval_rewards/accuracies": 0.515625, | |
| "eval_rewards/chosen": -3.953125, | |
| "eval_rewards/margins": 0.2578125, | |
| "eval_rewards/rejected": -4.21875, | |
| "eval_runtime": 46.68, | |
| "eval_samples_per_second": 42.845, | |
| "eval_steps_per_second": 0.686, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.8481675392670157, | |
| "grad_norm": 1.1616767671449386, | |
| "learning_rate": 4.509209163472357e-05, | |
| "logits/chosen": -17.875, | |
| "logits/rejected": -11.1875, | |
| "logps/chosen": -728.0, | |
| "logps/rejected": -3520.0, | |
| "loss": 0.0763, | |
| "rewards/accuracies": 0.96875, | |
| "rewards/chosen": -3.4375, | |
| "rewards/margins": 13.0625, | |
| "rewards/rejected": -16.5, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.8586387434554974, | |
| "grad_norm": 1.223837632577751, | |
| "learning_rate": 4.4909317814811896e-05, | |
| "logits/chosen": -17.625, | |
| "logits/rejected": -11.375, | |
| "logps/chosen": -664.0, | |
| "logps/rejected": -3424.0, | |
| "loss": 0.0583, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -2.8125, | |
| "rewards/margins": 10.875, | |
| "rewards/rejected": -13.6875, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.8691099476439791, | |
| "grad_norm": 0.4174719476328254, | |
| "learning_rate": 4.472358744696845e-05, | |
| "logits/chosen": -18.0, | |
| "logits/rejected": -12.75, | |
| "logps/chosen": -664.0, | |
| "logps/rejected": -3216.0, | |
| "loss": 0.0496, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -2.921875, | |
| "rewards/margins": 11.3125, | |
| "rewards/rejected": -14.25, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.8795811518324608, | |
| "grad_norm": 0.8375351428363348, | |
| "learning_rate": 4.4534928112285645e-05, | |
| "logits/chosen": -17.5, | |
| "logits/rejected": -12.625, | |
| "logps/chosen": -688.0, | |
| "logps/rejected": -3536.0, | |
| "loss": 0.0619, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -3.109375, | |
| "rewards/margins": 13.4375, | |
| "rewards/rejected": -16.625, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.8900523560209425, | |
| "grad_norm": 0.39612664539622144, | |
| "learning_rate": 4.43433678268096e-05, | |
| "logits/chosen": -17.625, | |
| "logits/rejected": -12.1875, | |
| "logps/chosen": -716.0, | |
| "logps/rejected": -4064.0, | |
| "loss": 0.0496, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -3.546875, | |
| "rewards/margins": 17.0, | |
| "rewards/rejected": -20.5, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.900523560209424, | |
| "grad_norm": 0.8987318228701537, | |
| "learning_rate": 4.414893503737968e-05, | |
| "logits/chosen": -16.875, | |
| "logits/rejected": -11.0, | |
| "logps/chosen": -748.0, | |
| "logps/rejected": -4320.0, | |
| "loss": 0.0284, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -4.0625, | |
| "rewards/margins": 20.125, | |
| "rewards/rejected": -24.125, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.9109947643979057, | |
| "grad_norm": 1.2399119249150083, | |
| "learning_rate": 4.395165861740414e-05, | |
| "logits/chosen": -17.125, | |
| "logits/rejected": -11.75, | |
| "logps/chosen": -736.0, | |
| "logps/rejected": -4192.0, | |
| "loss": 0.0758, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -3.90625, | |
| "rewards/margins": 19.75, | |
| "rewards/rejected": -23.625, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.9214659685863874, | |
| "grad_norm": 0.6085176240008444, | |
| "learning_rate": 4.375156786257241e-05, | |
| "logits/chosen": -16.25, | |
| "logits/rejected": -11.25, | |
| "logps/chosen": -752.0, | |
| "logps/rejected": -3744.0, | |
| "loss": 0.0552, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -3.53125, | |
| "rewards/margins": 13.875, | |
| "rewards/rejected": -17.5, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.9319371727748691, | |
| "grad_norm": 0.36777610819115614, | |
| "learning_rate": 4.354869248650466e-05, | |
| "logits/chosen": -17.625, | |
| "logits/rejected": -12.625, | |
| "logps/chosen": -732.0, | |
| "logps/rejected": -3456.0, | |
| "loss": 0.0586, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -3.390625, | |
| "rewards/margins": 10.625, | |
| "rewards/rejected": -14.0, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.9424083769633508, | |
| "grad_norm": 1.0756587675637965, | |
| "learning_rate": 4.334306261633931e-05, | |
| "logits/chosen": -17.5, | |
| "logits/rejected": -13.3125, | |
| "logps/chosen": -736.0, | |
| "logps/rejected": -3376.0, | |
| "loss": 0.0536, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": -3.328125, | |
| "rewards/margins": 10.8125, | |
| "rewards/rejected": -14.1875, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.9424083769633508, | |
| "eval_logits/chosen": -18.125, | |
| "eval_logits/rejected": -18.0, | |
| "eval_logps/chosen": -688.0, | |
| "eval_logps/rejected": -672.0, | |
| "eval_loss": 1.0159218311309814, | |
| "eval_rewards/accuracies": 0.53125, | |
| "eval_rewards/chosen": -3.6875, | |
| "eval_rewards/margins": 0.1298828125, | |
| "eval_rewards/rejected": -3.8125, | |
| "eval_runtime": 46.6872, | |
| "eval_samples_per_second": 42.838, | |
| "eval_steps_per_second": 0.685, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.9528795811518325, | |
| "grad_norm": 0.454905245775056, | |
| "learning_rate": 4.313470878825911e-05, | |
| "logits/chosen": -17.75, | |
| "logits/rejected": -13.0, | |
| "logps/chosen": -732.0, | |
| "logps/rejected": -3792.0, | |
| "loss": 0.0496, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -3.640625, | |
| "rewards/margins": 14.125, | |
| "rewards/rejected": -17.75, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.9633507853403142, | |
| "grad_norm": 0.9185132981686277, | |
| "learning_rate": 4.292366194295652e-05, | |
| "logits/chosen": -17.25, | |
| "logits/rejected": -12.1875, | |
| "logps/chosen": -792.0, | |
| "logps/rejected": -3616.0, | |
| "loss": 0.0689, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -4.15625, | |
| "rewards/margins": 13.25, | |
| "rewards/rejected": -17.375, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.9738219895287958, | |
| "grad_norm": 0.5182584904868646, | |
| "learning_rate": 4.2709953421038964e-05, | |
| "logits/chosen": -17.625, | |
| "logits/rejected": -12.375, | |
| "logps/chosen": -772.0, | |
| "logps/rejected": -3840.0, | |
| "loss": 0.0498, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -4.125, | |
| "rewards/margins": 14.375, | |
| "rewards/rejected": -18.5, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.9842931937172775, | |
| "grad_norm": 2.0466597503674406, | |
| "learning_rate": 4.249361495837478e-05, | |
| "logits/chosen": -17.875, | |
| "logits/rejected": -12.375, | |
| "logps/chosen": -792.0, | |
| "logps/rejected": -4128.0, | |
| "loss": 0.0723, | |
| "rewards/accuracies": 0.956250011920929, | |
| "rewards/chosen": -4.28125, | |
| "rewards/margins": 16.25, | |
| "rewards/rejected": -20.5, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.9947643979057592, | |
| "grad_norm": 0.6751205692867492, | |
| "learning_rate": 4.227467868138035e-05, | |
| "logits/chosen": -17.75, | |
| "logits/rejected": -13.1875, | |
| "logps/chosen": -744.0, | |
| "logps/rejected": -3648.0, | |
| "loss": 0.0422, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -3.609375, | |
| "rewards/margins": 13.625, | |
| "rewards/rejected": -17.25, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.0052356020942408, | |
| "grad_norm": 0.03492379347725474, | |
| "learning_rate": 4.205317710224938e-05, | |
| "logits/chosen": -17.125, | |
| "logits/rejected": -13.5625, | |
| "logps/chosen": -736.0, | |
| "logps/rejected": -3680.0, | |
| "loss": 0.0268, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -3.484375, | |
| "rewards/margins": 14.1875, | |
| "rewards/rejected": -17.625, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 1.0157068062827226, | |
| "grad_norm": 0.7964234048124855, | |
| "learning_rate": 4.1829143114124733e-05, | |
| "logits/chosen": -17.375, | |
| "logits/rejected": -13.5, | |
| "logps/chosen": -684.0, | |
| "logps/rejected": -3840.0, | |
| "loss": 0.0187, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -3.15625, | |
| "rewards/margins": 15.75, | |
| "rewards/rejected": -19.0, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 1.0261780104712042, | |
| "grad_norm": 0.23928691391421453, | |
| "learning_rate": 4.160260998621387e-05, | |
| "logits/chosen": -17.75, | |
| "logits/rejected": -13.0625, | |
| "logps/chosen": -736.0, | |
| "logps/rejected": -4000.0, | |
| "loss": 0.0122, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -3.46875, | |
| "rewards/margins": 17.25, | |
| "rewards/rejected": -20.75, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 1.036649214659686, | |
| "grad_norm": 0.049049647996422883, | |
| "learning_rate": 4.137361135884824e-05, | |
| "logits/chosen": -16.375, | |
| "logits/rejected": -11.6875, | |
| "logps/chosen": -772.0, | |
| "logps/rejected": -4480.0, | |
| "loss": 0.0105, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -3.890625, | |
| "rewards/margins": 20.25, | |
| "rewards/rejected": -24.125, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 1.0471204188481675, | |
| "grad_norm": 0.38257437987248777, | |
| "learning_rate": 4.114218123848777e-05, | |
| "logits/chosen": -16.125, | |
| "logits/rejected": -11.8125, | |
| "logps/chosen": -736.0, | |
| "logps/rejected": -4384.0, | |
| "loss": 0.0051, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -3.734375, | |
| "rewards/margins": 20.125, | |
| "rewards/rejected": -23.875, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.0471204188481675, | |
| "eval_logits/chosen": -16.125, | |
| "eval_logits/rejected": -15.6875, | |
| "eval_logps/chosen": -752.0, | |
| "eval_logps/rejected": -756.0, | |
| "eval_loss": 1.0945625305175781, | |
| "eval_rewards/accuracies": 0.515625, | |
| "eval_rewards/chosen": -4.34375, | |
| "eval_rewards/margins": 0.337890625, | |
| "eval_rewards/rejected": -4.6875, | |
| "eval_runtime": 46.6914, | |
| "eval_samples_per_second": 42.834, | |
| "eval_steps_per_second": 0.685, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.057591623036649, | |
| "grad_norm": 0.16612348156992668, | |
| "learning_rate": 4.0908353992670824e-05, | |
| "logits/chosen": -17.0, | |
| "logits/rejected": -12.1875, | |
| "logps/chosen": -696.0, | |
| "logps/rejected": -4352.0, | |
| "loss": 0.0201, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -3.390625, | |
| "rewards/margins": 20.5, | |
| "rewards/rejected": -23.875, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 1.068062827225131, | |
| "grad_norm": 0.13216288380440386, | |
| "learning_rate": 4.067216434491059e-05, | |
| "logits/chosen": -16.75, | |
| "logits/rejected": -11.625, | |
| "logps/chosen": -776.0, | |
| "logps/rejected": -4480.0, | |
| "loss": 0.0095, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -4.15625, | |
| "rewards/margins": 21.25, | |
| "rewards/rejected": -25.375, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 1.0785340314136125, | |
| "grad_norm": 3.742189598510517, | |
| "learning_rate": 4.043364736953866e-05, | |
| "logits/chosen": -16.375, | |
| "logits/rejected": -11.1875, | |
| "logps/chosen": -888.0, | |
| "logps/rejected": -4512.0, | |
| "loss": 0.0242, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -4.6875, | |
| "rewards/margins": 20.75, | |
| "rewards/rejected": -25.375, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 1.0890052356020943, | |
| "grad_norm": 0.3027917540903447, | |
| "learning_rate": 4.019283848649643e-05, | |
| "logits/chosen": -16.125, | |
| "logits/rejected": -10.5, | |
| "logps/chosen": -828.0, | |
| "logps/rejected": -4544.0, | |
| "loss": 0.0173, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.46875, | |
| "rewards/margins": 20.875, | |
| "rewards/rejected": -25.375, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 1.0994764397905759, | |
| "grad_norm": 0.03655942341237458, | |
| "learning_rate": 3.9949773456075215e-05, | |
| "logits/chosen": -16.5, | |
| "logits/rejected": -10.75, | |
| "logps/chosen": -768.0, | |
| "logps/rejected": -4544.0, | |
| "loss": 0.013, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.15625, | |
| "rewards/margins": 21.75, | |
| "rewards/rejected": -25.875, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.1099476439790577, | |
| "grad_norm": 0.4476479359530186, | |
| "learning_rate": 3.970448837360585e-05, | |
| "logits/chosen": -16.75, | |
| "logits/rejected": -10.875, | |
| "logps/chosen": -864.0, | |
| "logps/rejected": -4384.0, | |
| "loss": 0.014, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.46875, | |
| "rewards/margins": 21.0, | |
| "rewards/rejected": -25.5, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 1.1204188481675392, | |
| "grad_norm": 0.13002112543173103, | |
| "learning_rate": 3.945701966409846e-05, | |
| "logits/chosen": -16.875, | |
| "logits/rejected": -10.9375, | |
| "logps/chosen": -788.0, | |
| "logps/rejected": -4672.0, | |
| "loss": 0.0092, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.21875, | |
| "rewards/margins": 21.75, | |
| "rewards/rejected": -25.875, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 1.130890052356021, | |
| "grad_norm": 0.09233177496904184, | |
| "learning_rate": 3.9207404076833374e-05, | |
| "logits/chosen": -16.25, | |
| "logits/rejected": -10.875, | |
| "logps/chosen": -848.0, | |
| "logps/rejected": -4640.0, | |
| "loss": 0.0152, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.40625, | |
| "rewards/margins": 22.0, | |
| "rewards/rejected": -26.5, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 1.1413612565445026, | |
| "grad_norm": 0.08038949876060705, | |
| "learning_rate": 3.895567867990379e-05, | |
| "logits/chosen": -17.25, | |
| "logits/rejected": -11.125, | |
| "logps/chosen": -852.0, | |
| "logps/rejected": -4608.0, | |
| "loss": 0.0077, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -4.40625, | |
| "rewards/margins": 21.625, | |
| "rewards/rejected": -26.0, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 1.1518324607329844, | |
| "grad_norm": 0.30816879790295787, | |
| "learning_rate": 3.870188085471114e-05, | |
| "logits/chosen": -17.625, | |
| "logits/rejected": -11.4375, | |
| "logps/chosen": -844.0, | |
| "logps/rejected": -4704.0, | |
| "loss": 0.0049, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.5625, | |
| "rewards/margins": 23.125, | |
| "rewards/rejected": -27.75, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 1.1518324607329844, | |
| "eval_logits/chosen": -17.875, | |
| "eval_logits/rejected": -17.75, | |
| "eval_logps/chosen": -788.0, | |
| "eval_logps/rejected": -808.0, | |
| "eval_loss": 1.232968807220459, | |
| "eval_rewards/accuracies": 0.509765625, | |
| "eval_rewards/chosen": -4.71875, | |
| "eval_rewards/margins": 0.46484375, | |
| "eval_rewards/rejected": -5.1875, | |
| "eval_runtime": 46.692, | |
| "eval_samples_per_second": 42.834, | |
| "eval_steps_per_second": 0.685, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 1.162303664921466, | |
| "grad_norm": 0.030165669216804965, | |
| "learning_rate": 3.844604829041395e-05, | |
| "logits/chosen": -18.0, | |
| "logits/rejected": -11.625, | |
| "logps/chosen": -784.0, | |
| "logps/rejected": -4608.0, | |
| "loss": 0.0116, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -4.375, | |
| "rewards/margins": 23.5, | |
| "rewards/rejected": -27.875, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 1.1727748691099475, | |
| "grad_norm": 0.21260453736458534, | |
| "learning_rate": 3.818821897833095e-05, | |
| "logits/chosen": -17.375, | |
| "logits/rejected": -10.5625, | |
| "logps/chosen": -868.0, | |
| "logps/rejected": -4896.0, | |
| "loss": 0.0135, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.8125, | |
| "rewards/margins": 24.875, | |
| "rewards/rejected": -29.75, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 1.1832460732984293, | |
| "grad_norm": 0.08240359251592858, | |
| "learning_rate": 3.792843120629935e-05, | |
| "logits/chosen": -17.25, | |
| "logits/rejected": -11.5625, | |
| "logps/chosen": -784.0, | |
| "logps/rejected": -4640.0, | |
| "loss": 0.004, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.21875, | |
| "rewards/margins": 22.75, | |
| "rewards/rejected": -27.0, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 1.193717277486911, | |
| "grad_norm": 0.08918772045474595, | |
| "learning_rate": 3.766672355298907e-05, | |
| "logits/chosen": -17.5, | |
| "logits/rejected": -11.625, | |
| "logps/chosen": -740.0, | |
| "logps/rejected": -4576.0, | |
| "loss": 0.0126, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -3.890625, | |
| "rewards/margins": 22.625, | |
| "rewards/rejected": -26.5, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 1.2041884816753927, | |
| "grad_norm": 0.13814580953176903, | |
| "learning_rate": 3.740313488217373e-05, | |
| "logits/chosen": -17.25, | |
| "logits/rejected": -11.8125, | |
| "logps/chosen": -732.0, | |
| "logps/rejected": -4384.0, | |
| "loss": 0.0166, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -3.84375, | |
| "rewards/margins": 21.25, | |
| "rewards/rejected": -25.125, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 1.2146596858638743, | |
| "grad_norm": 0.0670734525298685, | |
| "learning_rate": 3.7137704336959464e-05, | |
| "logits/chosen": -16.75, | |
| "logits/rejected": -11.125, | |
| "logps/chosen": -740.0, | |
| "logps/rejected": -4576.0, | |
| "loss": 0.0153, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -3.9375, | |
| "rewards/margins": 21.5, | |
| "rewards/rejected": -25.375, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 1.225130890052356, | |
| "grad_norm": 0.2052376945112177, | |
| "learning_rate": 3.687047133397201e-05, | |
| "logits/chosen": -17.375, | |
| "logits/rejected": -11.1875, | |
| "logps/chosen": -720.0, | |
| "logps/rejected": -4928.0, | |
| "loss": 0.0064, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -3.578125, | |
| "rewards/margins": 25.25, | |
| "rewards/rejected": -28.875, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 1.2356020942408377, | |
| "grad_norm": 0.06177922766814115, | |
| "learning_rate": 3.6601475557503415e-05, | |
| "logits/chosen": -17.5, | |
| "logits/rejected": -11.3125, | |
| "logps/chosen": -788.0, | |
| "logps/rejected": -4800.0, | |
| "loss": 0.0067, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.09375, | |
| "rewards/margins": 25.125, | |
| "rewards/rejected": -29.25, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 1.2460732984293195, | |
| "grad_norm": 0.041520450753771254, | |
| "learning_rate": 3.633075695361882e-05, | |
| "logits/chosen": -17.75, | |
| "logits/rejected": -12.0, | |
| "logps/chosen": -736.0, | |
| "logps/rejected": -5088.0, | |
| "loss": 0.0159, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -3.84375, | |
| "rewards/margins": 26.875, | |
| "rewards/rejected": -30.75, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 1.256544502617801, | |
| "grad_norm": 0.03756717909900222, | |
| "learning_rate": 3.6058355724224477e-05, | |
| "logits/chosen": -18.375, | |
| "logits/rejected": -12.125, | |
| "logps/chosen": -792.0, | |
| "logps/rejected": -5184.0, | |
| "loss": 0.0037, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.25, | |
| "rewards/margins": 27.25, | |
| "rewards/rejected": -31.5, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 1.256544502617801, | |
| "eval_logits/chosen": -18.375, | |
| "eval_logits/rejected": -18.5, | |
| "eval_logps/chosen": -796.0, | |
| "eval_logps/rejected": -812.0, | |
| "eval_loss": 1.2517812252044678, | |
| "eval_rewards/accuracies": 0.515625, | |
| "eval_rewards/chosen": -4.75, | |
| "eval_rewards/margins": 0.470703125, | |
| "eval_rewards/rejected": -5.25, | |
| "eval_runtime": 46.7138, | |
| "eval_samples_per_second": 42.814, | |
| "eval_steps_per_second": 0.685, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 1.2670157068062826, | |
| "grad_norm": 0.09607507041317226, | |
| "learning_rate": 3.5784312321097744e-05, | |
| "logits/chosen": -18.125, | |
| "logits/rejected": -11.875, | |
| "logps/chosen": -784.0, | |
| "logps/rejected": -4960.0, | |
| "loss": 0.0059, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.125, | |
| "rewards/margins": 27.0, | |
| "rewards/rejected": -31.125, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 1.2774869109947644, | |
| "grad_norm": 0.8646262708803255, | |
| "learning_rate": 3.550866743987994e-05, | |
| "logits/chosen": -18.0, | |
| "logits/rejected": -12.125, | |
| "logps/chosen": -840.0, | |
| "logps/rejected": -4896.0, | |
| "loss": 0.0271, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -4.5, | |
| "rewards/margins": 26.125, | |
| "rewards/rejected": -30.625, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 1.2879581151832462, | |
| "grad_norm": 0.033705549961800946, | |
| "learning_rate": 3.5231462014033024e-05, | |
| "logits/chosen": -18.375, | |
| "logits/rejected": -12.25, | |
| "logps/chosen": -716.0, | |
| "logps/rejected": -5216.0, | |
| "loss": 0.0042, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -3.6875, | |
| "rewards/margins": 28.0, | |
| "rewards/rejected": -31.75, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 1.2984293193717278, | |
| "grad_norm": 0.01788823047874542, | |
| "learning_rate": 3.495273720876095e-05, | |
| "logits/chosen": -18.0, | |
| "logits/rejected": -11.875, | |
| "logps/chosen": -780.0, | |
| "logps/rejected": -5088.0, | |
| "loss": 0.0137, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -3.96875, | |
| "rewards/margins": 27.25, | |
| "rewards/rejected": -31.125, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 1.3089005235602094, | |
| "grad_norm": 0.13093003022830835, | |
| "learning_rate": 3.4672534414896616e-05, | |
| "logits/chosen": -18.0, | |
| "logits/rejected": -12.4375, | |
| "logps/chosen": -764.0, | |
| "logps/rejected": -4864.0, | |
| "loss": 0.0108, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -3.78125, | |
| "rewards/margins": 25.375, | |
| "rewards/rejected": -29.125, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 1.3193717277486912, | |
| "grad_norm": 0.04201927686460754, | |
| "learning_rate": 3.439089524275526e-05, | |
| "logits/chosen": -17.375, | |
| "logits/rejected": -11.625, | |
| "logps/chosen": -724.0, | |
| "logps/rejected": -4832.0, | |
| "loss": 0.0044, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -3.640625, | |
| "rewards/margins": 24.875, | |
| "rewards/rejected": -28.5, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 1.3298429319371727, | |
| "grad_norm": 0.11736392726396039, | |
| "learning_rate": 3.410786151595533e-05, | |
| "logits/chosen": -17.625, | |
| "logits/rejected": -11.375, | |
| "logps/chosen": -716.0, | |
| "logps/rejected": -4832.0, | |
| "loss": 0.0044, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -3.484375, | |
| "rewards/margins": 25.375, | |
| "rewards/rejected": -28.875, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 1.3403141361256545, | |
| "grad_norm": 0.1366074458803577, | |
| "learning_rate": 3.382347526520765e-05, | |
| "logits/chosen": -18.375, | |
| "logits/rejected": -10.9375, | |
| "logps/chosen": -724.0, | |
| "logps/rejected": -4960.0, | |
| "loss": 0.0076, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -3.546875, | |
| "rewards/margins": 26.375, | |
| "rewards/rejected": -29.875, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 1.350785340314136, | |
| "grad_norm": 0.18048268397507528, | |
| "learning_rate": 3.353777872207381e-05, | |
| "logits/chosen": -18.25, | |
| "logits/rejected": -11.125, | |
| "logps/chosen": -724.0, | |
| "logps/rejected": -4672.0, | |
| "loss": 0.0101, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -3.578125, | |
| "rewards/margins": 24.75, | |
| "rewards/rejected": -28.25, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 1.3612565445026177, | |
| "grad_norm": 0.08119318142298289, | |
| "learning_rate": 3.325081431269475e-05, | |
| "logits/chosen": -18.0, | |
| "logits/rejected": -10.9375, | |
| "logps/chosen": -680.0, | |
| "logps/rejected": -4992.0, | |
| "loss": 0.0122, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -3.4375, | |
| "rewards/margins": 26.5, | |
| "rewards/rejected": -30.0, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 1.3612565445026177, | |
| "eval_logits/chosen": -18.375, | |
| "eval_logits/rejected": -18.25, | |
| "eval_logps/chosen": -716.0, | |
| "eval_logps/rejected": -720.0, | |
| "eval_loss": 1.0438125133514404, | |
| "eval_rewards/accuracies": 0.53125, | |
| "eval_rewards/chosen": -3.96875, | |
| "eval_rewards/margins": 0.34765625, | |
| "eval_rewards/rejected": -4.3125, | |
| "eval_runtime": 46.6958, | |
| "eval_samples_per_second": 42.83, | |
| "eval_steps_per_second": 0.685, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 1.3717277486910995, | |
| "grad_norm": 0.16815981785951112, | |
| "learning_rate": 3.296262465149046e-05, | |
| "logits/chosen": -17.875, | |
| "logits/rejected": -10.125, | |
| "logps/chosen": -756.0, | |
| "logps/rejected": -5088.0, | |
| "loss": 0.0072, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -3.765625, | |
| "rewards/margins": 26.5, | |
| "rewards/rejected": -30.25, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 1.3821989528795813, | |
| "grad_norm": 0.1877599590344403, | |
| "learning_rate": 3.2673252534831687e-05, | |
| "logits/chosen": -17.875, | |
| "logits/rejected": -11.3125, | |
| "logps/chosen": -748.0, | |
| "logps/rejected": -4768.0, | |
| "loss": 0.0067, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -3.765625, | |
| "rewards/margins": 24.625, | |
| "rewards/rejected": -28.375, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 1.3926701570680629, | |
| "grad_norm": 0.06285595104499013, | |
| "learning_rate": 3.2382740934684696e-05, | |
| "logits/chosen": -18.125, | |
| "logits/rejected": -10.875, | |
| "logps/chosen": -728.0, | |
| "logps/rejected": -4736.0, | |
| "loss": 0.0324, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -3.671875, | |
| "rewards/margins": 24.25, | |
| "rewards/rejected": -28.0, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 1.4031413612565444, | |
| "grad_norm": 1.3597835366420297, | |
| "learning_rate": 3.209113299222982e-05, | |
| "logits/chosen": -18.25, | |
| "logits/rejected": -11.5625, | |
| "logps/chosen": -672.0, | |
| "logps/rejected": -4704.0, | |
| "loss": 0.0053, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -3.46875, | |
| "rewards/margins": 23.5, | |
| "rewards/rejected": -27.0, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 1.4136125654450262, | |
| "grad_norm": 0.027805443688728032, | |
| "learning_rate": 3.179847201145505e-05, | |
| "logits/chosen": -17.75, | |
| "logits/rejected": -9.125, | |
| "logps/chosen": -780.0, | |
| "logps/rejected": -5120.0, | |
| "loss": 0.0086, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -3.921875, | |
| "rewards/margins": 27.0, | |
| "rewards/rejected": -30.875, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 1.4240837696335078, | |
| "grad_norm": 4.493926699111642, | |
| "learning_rate": 3.150480145272528e-05, | |
| "logits/chosen": -16.375, | |
| "logits/rejected": -7.6875, | |
| "logps/chosen": -828.0, | |
| "logps/rejected": -5408.0, | |
| "loss": 0.0476, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -4.59375, | |
| "rewards/margins": 29.625, | |
| "rewards/rejected": -34.25, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 1.4345549738219896, | |
| "grad_norm": 0.03125383872435537, | |
| "learning_rate": 3.121016492632848e-05, | |
| "logits/chosen": -16.375, | |
| "logits/rejected": -8.25, | |
| "logps/chosen": -784.0, | |
| "logps/rejected": -4256.0, | |
| "loss": 0.0109, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.0625, | |
| "rewards/margins": 20.125, | |
| "rewards/rejected": -24.25, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 1.4450261780104712, | |
| "grad_norm": 0.02754813584439254, | |
| "learning_rate": 3.091460618599951e-05, | |
| "logits/chosen": -16.75, | |
| "logits/rejected": -8.25, | |
| "logps/chosen": -712.0, | |
| "logps/rejected": -3984.0, | |
| "loss": 0.0049, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -3.53125, | |
| "rewards/margins": 17.75, | |
| "rewards/rejected": -21.25, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 1.4554973821989527, | |
| "grad_norm": 0.197031650549633, | |
| "learning_rate": 3.061816912242265e-05, | |
| "logits/chosen": -16.5, | |
| "logits/rejected": -7.8125, | |
| "logps/chosen": -768.0, | |
| "logps/rejected": -4224.0, | |
| "loss": 0.007, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -3.84375, | |
| "rewards/margins": 18.375, | |
| "rewards/rejected": -22.25, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 1.4659685863874345, | |
| "grad_norm": 0.03135084746404854, | |
| "learning_rate": 3.0320897756713785e-05, | |
| "logits/chosen": -16.375, | |
| "logits/rejected": -8.125, | |
| "logps/chosen": -780.0, | |
| "logps/rejected": -4544.0, | |
| "loss": 0.0082, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -3.953125, | |
| "rewards/margins": 20.375, | |
| "rewards/rejected": -24.25, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 1.4659685863874345, | |
| "eval_logits/chosen": -17.125, | |
| "eval_logits/rejected": -16.875, | |
| "eval_logps/chosen": -792.0, | |
| "eval_logps/rejected": -796.0, | |
| "eval_loss": 1.2435468435287476, | |
| "eval_rewards/accuracies": 0.5078125, | |
| "eval_rewards/chosen": -4.75, | |
| "eval_rewards/margins": 0.3203125, | |
| "eval_rewards/rejected": -5.0625, | |
| "eval_runtime": 46.6956, | |
| "eval_samples_per_second": 42.831, | |
| "eval_steps_per_second": 0.685, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 1.4764397905759163, | |
| "grad_norm": 0.06152955846616178, | |
| "learning_rate": 3.002283623388332e-05, | |
| "logits/chosen": -16.75, | |
| "logits/rejected": -9.5, | |
| "logps/chosen": -784.0, | |
| "logps/rejected": -4416.0, | |
| "loss": 0.0332, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.375, | |
| "rewards/margins": 20.5, | |
| "rewards/rejected": -24.875, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 1.486910994764398, | |
| "grad_norm": 1.3897309315800281, | |
| "learning_rate": 2.9724028816280507e-05, | |
| "logits/chosen": -16.875, | |
| "logits/rejected": -8.875, | |
| "logps/chosen": -816.0, | |
| "logps/rejected": -4192.0, | |
| "loss": 0.0145, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -4.59375, | |
| "rewards/margins": 18.0, | |
| "rewards/rejected": -22.5, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 1.4973821989528795, | |
| "grad_norm": 0.46962933463094425, | |
| "learning_rate": 2.9424519877020524e-05, | |
| "logits/chosen": -16.625, | |
| "logits/rejected": -8.25, | |
| "logps/chosen": -840.0, | |
| "logps/rejected": -3968.0, | |
| "loss": 0.0073, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.4375, | |
| "rewards/margins": 15.6875, | |
| "rewards/rejected": -20.125, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 1.5078534031413613, | |
| "grad_norm": 0.2069393854602299, | |
| "learning_rate": 2.912435389339504e-05, | |
| "logits/chosen": -16.75, | |
| "logits/rejected": -8.875, | |
| "logps/chosen": -744.0, | |
| "logps/rejected": -3760.0, | |
| "loss": 0.005, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -3.8125, | |
| "rewards/margins": 14.5625, | |
| "rewards/rejected": -18.375, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 1.518324607329843, | |
| "grad_norm": 0.03870032498975125, | |
| "learning_rate": 2.8823575440267226e-05, | |
| "logits/chosen": -16.25, | |
| "logits/rejected": -8.375, | |
| "logps/chosen": -760.0, | |
| "logps/rejected": -4160.0, | |
| "loss": 0.0042, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.0625, | |
| "rewards/margins": 16.5, | |
| "rewards/rejected": -20.625, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 1.5287958115183247, | |
| "grad_norm": 0.7250241653094004, | |
| "learning_rate": 2.8522229183452465e-05, | |
| "logits/chosen": -16.5, | |
| "logits/rejected": -8.75, | |
| "logps/chosen": -796.0, | |
| "logps/rejected": -3968.0, | |
| "loss": 0.0217, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.125, | |
| "rewards/margins": 15.4375, | |
| "rewards/rejected": -19.625, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 1.5392670157068062, | |
| "grad_norm": 0.14607344865595595, | |
| "learning_rate": 2.8220359873085373e-05, | |
| "logits/chosen": -16.25, | |
| "logits/rejected": -9.5625, | |
| "logps/chosen": -780.0, | |
| "logps/rejected": -3856.0, | |
| "loss": 0.0055, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.03125, | |
| "rewards/margins": 15.125, | |
| "rewards/rejected": -19.125, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 1.5497382198952878, | |
| "grad_norm": 0.5605283207094957, | |
| "learning_rate": 2.7918012336974388e-05, | |
| "logits/chosen": -16.125, | |
| "logits/rejected": -8.4375, | |
| "logps/chosen": -740.0, | |
| "logps/rejected": -3808.0, | |
| "loss": 0.081, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -3.71875, | |
| "rewards/margins": 15.9375, | |
| "rewards/rejected": -19.625, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 1.5602094240837696, | |
| "grad_norm": 0.11209505617163172, | |
| "learning_rate": 2.7615231473944815e-05, | |
| "logits/chosen": -16.0, | |
| "logits/rejected": -8.6875, | |
| "logps/chosen": -744.0, | |
| "logps/rejected": -3984.0, | |
| "loss": 0.0092, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -3.921875, | |
| "rewards/margins": 16.625, | |
| "rewards/rejected": -20.5, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 1.5706806282722514, | |
| "grad_norm": 0.541906647366998, | |
| "learning_rate": 2.7312062247171327e-05, | |
| "logits/chosen": -15.75, | |
| "logits/rejected": -8.5, | |
| "logps/chosen": -772.0, | |
| "logps/rejected": -4288.0, | |
| "loss": 0.0039, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.3125, | |
| "rewards/margins": 18.375, | |
| "rewards/rejected": -22.625, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.5706806282722514, | |
| "eval_logits/chosen": -15.75, | |
| "eval_logits/rejected": -15.3125, | |
| "eval_logps/chosen": -808.0, | |
| "eval_logps/rejected": -816.0, | |
| "eval_loss": 1.2731250524520874, | |
| "eval_rewards/accuracies": 0.50390625, | |
| "eval_rewards/chosen": -4.875, | |
| "eval_rewards/margins": 0.40234375, | |
| "eval_rewards/rejected": -5.28125, | |
| "eval_runtime": 46.7087, | |
| "eval_samples_per_second": 42.819, | |
| "eval_steps_per_second": 0.685, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.581151832460733, | |
| "grad_norm": 0.08874575726622319, | |
| "learning_rate": 2.7008549677500876e-05, | |
| "logits/chosen": -15.875, | |
| "logits/rejected": -8.0, | |
| "logps/chosen": -820.0, | |
| "logps/rejected": -4384.0, | |
| "loss": 0.0048, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.34375, | |
| "rewards/margins": 20.375, | |
| "rewards/rejected": -24.75, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 1.5916230366492146, | |
| "grad_norm": 0.2592247322122908, | |
| "learning_rate": 2.670473883676709e-05, | |
| "logits/chosen": -15.3125, | |
| "logits/rejected": -6.53125, | |
| "logps/chosen": -864.0, | |
| "logps/rejected": -4736.0, | |
| "loss": 0.0026, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.8125, | |
| "rewards/margins": 22.5, | |
| "rewards/rejected": -27.25, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 1.6020942408376964, | |
| "grad_norm": 0.7216376052414317, | |
| "learning_rate": 2.6400674841097074e-05, | |
| "logits/chosen": -14.5625, | |
| "logits/rejected": -6.09375, | |
| "logps/chosen": -896.0, | |
| "logps/rejected": -4960.0, | |
| "loss": 0.0122, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -5.21875, | |
| "rewards/margins": 23.875, | |
| "rewards/rejected": -29.125, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 1.6125654450261782, | |
| "grad_norm": 0.5608343776940902, | |
| "learning_rate": 2.609640284421158e-05, | |
| "logits/chosen": -15.5, | |
| "logits/rejected": -7.375, | |
| "logps/chosen": -824.0, | |
| "logps/rejected": -4704.0, | |
| "loss": 0.0181, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -4.53125, | |
| "rewards/margins": 22.5, | |
| "rewards/rejected": -27.0, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 1.6230366492146597, | |
| "grad_norm": 0.009659152298821585, | |
| "learning_rate": 2.5791968030719726e-05, | |
| "logits/chosen": -15.6875, | |
| "logits/rejected": -6.8125, | |
| "logps/chosen": -840.0, | |
| "logps/rejected": -4672.0, | |
| "loss": 0.0171, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.34375, | |
| "rewards/margins": 22.25, | |
| "rewards/rejected": -26.625, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 1.6335078534031413, | |
| "grad_norm": 0.1084865800422241, | |
| "learning_rate": 2.548741560940902e-05, | |
| "logits/chosen": -15.6875, | |
| "logits/rejected": -7.65625, | |
| "logps/chosen": -760.0, | |
| "logps/rejected": -4352.0, | |
| "loss": 0.0049, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -3.984375, | |
| "rewards/margins": 20.375, | |
| "rewards/rejected": -24.25, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 1.6439790575916229, | |
| "grad_norm": 0.38608005944963747, | |
| "learning_rate": 2.5182790806531787e-05, | |
| "logits/chosen": -15.75, | |
| "logits/rejected": -8.0, | |
| "logps/chosen": -812.0, | |
| "logps/rejected": -4448.0, | |
| "loss": 0.0026, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.1875, | |
| "rewards/margins": 22.0, | |
| "rewards/rejected": -26.125, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 1.6544502617801047, | |
| "grad_norm": 0.374640121892769, | |
| "learning_rate": 2.4878138859089074e-05, | |
| "logits/chosen": -15.875, | |
| "logits/rejected": -7.34375, | |
| "logps/chosen": -800.0, | |
| "logps/rejected": -4672.0, | |
| "loss": 0.0097, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.1875, | |
| "rewards/margins": 22.375, | |
| "rewards/rejected": -26.625, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 1.6649214659685865, | |
| "grad_norm": 0.02537557716181889, | |
| "learning_rate": 2.4573505008112924e-05, | |
| "logits/chosen": -15.875, | |
| "logits/rejected": -8.0625, | |
| "logps/chosen": -788.0, | |
| "logps/rejected": -4352.0, | |
| "loss": 0.0036, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.125, | |
| "rewards/margins": 20.5, | |
| "rewards/rejected": -24.75, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 1.675392670157068, | |
| "grad_norm": 0.19661403477317288, | |
| "learning_rate": 2.4268934491948032e-05, | |
| "logits/chosen": -15.5625, | |
| "logits/rejected": -7.6875, | |
| "logps/chosen": -828.0, | |
| "logps/rejected": -4576.0, | |
| "loss": 0.0021, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.40625, | |
| "rewards/margins": 21.875, | |
| "rewards/rejected": -26.25, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 1.675392670157068, | |
| "eval_logits/chosen": -16.125, | |
| "eval_logits/rejected": -15.8125, | |
| "eval_logps/chosen": -808.0, | |
| "eval_logps/rejected": -820.0, | |
| "eval_loss": 1.3171093463897705, | |
| "eval_rewards/accuracies": 0.509765625, | |
| "eval_rewards/chosen": -4.875, | |
| "eval_rewards/margins": 0.416015625, | |
| "eval_rewards/rejected": -5.28125, | |
| "eval_runtime": 46.6839, | |
| "eval_samples_per_second": 42.841, | |
| "eval_steps_per_second": 0.685, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 1.6858638743455496, | |
| "grad_norm": 0.05230137510476896, | |
| "learning_rate": 2.396447253953385e-05, | |
| "logits/chosen": -15.5, | |
| "logits/rejected": -6.875, | |
| "logps/chosen": -848.0, | |
| "logps/rejected": -4832.0, | |
| "loss": 0.0083, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.46875, | |
| "rewards/margins": 22.25, | |
| "rewards/rejected": -26.75, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 1.6963350785340314, | |
| "grad_norm": 0.20893248885159454, | |
| "learning_rate": 2.3660164363687998e-05, | |
| "logits/chosen": -15.6875, | |
| "logits/rejected": -7.46875, | |
| "logps/chosen": -860.0, | |
| "logps/rejected": -4704.0, | |
| "loss": 0.0093, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -4.5, | |
| "rewards/margins": 21.625, | |
| "rewards/rejected": -26.125, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 1.7068062827225132, | |
| "grad_norm": 0.36232213103664296, | |
| "learning_rate": 2.33560551543922e-05, | |
| "logits/chosen": -15.9375, | |
| "logits/rejected": -9.1875, | |
| "logps/chosen": -804.0, | |
| "logps/rejected": -4288.0, | |
| "loss": 0.0073, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.0, | |
| "rewards/margins": 20.25, | |
| "rewards/rejected": -24.25, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 1.7172774869109948, | |
| "grad_norm": 0.04917124166701055, | |
| "learning_rate": 2.3052190072081493e-05, | |
| "logits/chosen": -16.25, | |
| "logits/rejected": -9.75, | |
| "logps/chosen": -808.0, | |
| "logps/rejected": -4448.0, | |
| "loss": 0.0052, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.1875, | |
| "rewards/margins": 21.0, | |
| "rewards/rejected": -25.125, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 1.7277486910994764, | |
| "grad_norm": 0.3380241189344471, | |
| "learning_rate": 2.2748614240937864e-05, | |
| "logits/chosen": -15.75, | |
| "logits/rejected": -8.125, | |
| "logps/chosen": -752.0, | |
| "logps/rejected": -4736.0, | |
| "loss": 0.0031, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -3.984375, | |
| "rewards/margins": 23.75, | |
| "rewards/rejected": -27.75, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 1.738219895287958, | |
| "grad_norm": 0.02734848445098051, | |
| "learning_rate": 2.2445372742189334e-05, | |
| "logits/chosen": -15.5625, | |
| "logits/rejected": -8.25, | |
| "logps/chosen": -828.0, | |
| "logps/rejected": -4832.0, | |
| "loss": 0.0036, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.375, | |
| "rewards/margins": 24.625, | |
| "rewards/rejected": -29.0, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 1.7486910994764397, | |
| "grad_norm": 1.97563389885794, | |
| "learning_rate": 2.214251060741528e-05, | |
| "logits/chosen": -16.125, | |
| "logits/rejected": -8.375, | |
| "logps/chosen": -860.0, | |
| "logps/rejected": -4928.0, | |
| "loss": 0.0147, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.5625, | |
| "rewards/margins": 24.875, | |
| "rewards/rejected": -29.375, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 1.7591623036649215, | |
| "grad_norm": 0.22516857871355894, | |
| "learning_rate": 2.1840072811859302e-05, | |
| "logits/chosen": -16.0, | |
| "logits/rejected": -8.0, | |
| "logps/chosen": -844.0, | |
| "logps/rejected": -4992.0, | |
| "loss": 0.0032, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.375, | |
| "rewards/margins": 25.5, | |
| "rewards/rejected": -29.875, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 1.7696335078534031, | |
| "grad_norm": 0.04285341699844151, | |
| "learning_rate": 2.1538104267750287e-05, | |
| "logits/chosen": -16.375, | |
| "logits/rejected": -8.0625, | |
| "logps/chosen": -772.0, | |
| "logps/rejected": -5120.0, | |
| "loss": 0.005, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -4.09375, | |
| "rewards/margins": 26.5, | |
| "rewards/rejected": -30.5, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 1.7801047120418847, | |
| "grad_norm": 0.06848616832684133, | |
| "learning_rate": 2.1236649817632952e-05, | |
| "logits/chosen": -15.9375, | |
| "logits/rejected": -8.5625, | |
| "logps/chosen": -736.0, | |
| "logps/rejected": -4640.0, | |
| "loss": 0.0146, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -3.953125, | |
| "rewards/margins": 23.125, | |
| "rewards/rejected": -27.0, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 1.7801047120418847, | |
| "eval_logits/chosen": -16.125, | |
| "eval_logits/rejected": -16.0, | |
| "eval_logps/chosen": -780.0, | |
| "eval_logps/rejected": -792.0, | |
| "eval_loss": 1.2652031183242798, | |
| "eval_rewards/accuracies": 0.501953125, | |
| "eval_rewards/chosen": -4.625, | |
| "eval_rewards/margins": 0.4140625, | |
| "eval_rewards/rejected": -5.03125, | |
| "eval_runtime": 46.7097, | |
| "eval_samples_per_second": 42.818, | |
| "eval_steps_per_second": 0.685, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 1.7905759162303665, | |
| "grad_norm": 0.03474738390846604, | |
| "learning_rate": 2.093575422770872e-05, | |
| "logits/chosen": -15.6875, | |
| "logits/rejected": -8.75, | |
| "logps/chosen": -736.0, | |
| "logps/rejected": -4544.0, | |
| "loss": 0.0086, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -3.84375, | |
| "rewards/margins": 22.375, | |
| "rewards/rejected": -26.25, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 1.8010471204188483, | |
| "grad_norm": 0.034033443949527425, | |
| "learning_rate": 2.0635462181187827e-05, | |
| "logits/chosen": -15.875, | |
| "logits/rejected": -9.125, | |
| "logps/chosen": -776.0, | |
| "logps/rejected": -4512.0, | |
| "loss": 0.0055, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -3.953125, | |
| "rewards/margins": 21.5, | |
| "rewards/rejected": -25.375, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 1.8115183246073299, | |
| "grad_norm": 0.2534692001412973, | |
| "learning_rate": 2.0335818271653917e-05, | |
| "logits/chosen": -16.125, | |
| "logits/rejected": -9.5, | |
| "logps/chosen": -776.0, | |
| "logps/rejected": -4480.0, | |
| "loss": 0.0052, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -3.984375, | |
| "rewards/margins": 21.125, | |
| "rewards/rejected": -25.125, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 1.8219895287958114, | |
| "grad_norm": 0.013329997626451026, | |
| "learning_rate": 2.0036866996441815e-05, | |
| "logits/chosen": -16.625, | |
| "logits/rejected": -9.5, | |
| "logps/chosen": -808.0, | |
| "logps/rejected": -4704.0, | |
| "loss": 0.0142, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.28125, | |
| "rewards/margins": 21.875, | |
| "rewards/rejected": -26.25, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 1.8324607329842932, | |
| "grad_norm": 0.01070034590297931, | |
| "learning_rate": 1.9738652750029677e-05, | |
| "logits/chosen": -15.9375, | |
| "logits/rejected": -9.25, | |
| "logps/chosen": -816.0, | |
| "logps/rejected": -4448.0, | |
| "loss": 0.005, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.15625, | |
| "rewards/margins": 21.25, | |
| "rewards/rejected": -25.5, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 1.8429319371727748, | |
| "grad_norm": 0.05415359250883316, | |
| "learning_rate": 1.944121981744631e-05, | |
| "logits/chosen": -15.8125, | |
| "logits/rejected": -9.25, | |
| "logps/chosen": -864.0, | |
| "logps/rejected": -4288.0, | |
| "loss": 0.0309, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -4.46875, | |
| "rewards/margins": 20.875, | |
| "rewards/rejected": -25.25, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 1.8534031413612566, | |
| "grad_norm": 0.03171886566354555, | |
| "learning_rate": 1.9144612367694887e-05, | |
| "logits/chosen": -16.375, | |
| "logits/rejected": -9.5, | |
| "logps/chosen": -748.0, | |
| "logps/rejected": -4352.0, | |
| "loss": 0.0071, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -3.84375, | |
| "rewards/margins": 21.0, | |
| "rewards/rejected": -24.875, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 1.8638743455497382, | |
| "grad_norm": 0.06399177625906781, | |
| "learning_rate": 1.8848874447193803e-05, | |
| "logits/chosen": -15.8125, | |
| "logits/rejected": -9.125, | |
| "logps/chosen": -752.0, | |
| "logps/rejected": -4480.0, | |
| "loss": 0.0028, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -3.953125, | |
| "rewards/margins": 20.75, | |
| "rewards/rejected": -24.625, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 1.8743455497382198, | |
| "grad_norm": 0.1734560991093733, | |
| "learning_rate": 1.855404997323571e-05, | |
| "logits/chosen": -16.125, | |
| "logits/rejected": -9.625, | |
| "logps/chosen": -800.0, | |
| "logps/rejected": -4352.0, | |
| "loss": 0.0049, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.1875, | |
| "rewards/margins": 19.875, | |
| "rewards/rejected": -24.0, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 1.8848167539267016, | |
| "grad_norm": 0.15731942695308787, | |
| "learning_rate": 1.82601827274658e-05, | |
| "logits/chosen": -16.5, | |
| "logits/rejected": -9.5, | |
| "logps/chosen": -764.0, | |
| "logps/rejected": -4384.0, | |
| "loss": 0.0034, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.125, | |
| "rewards/margins": 21.625, | |
| "rewards/rejected": -25.625, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 1.8848167539267016, | |
| "eval_logits/chosen": -16.25, | |
| "eval_logits/rejected": -16.0, | |
| "eval_logps/chosen": -788.0, | |
| "eval_logps/rejected": -788.0, | |
| "eval_loss": 1.2840312719345093, | |
| "eval_rewards/accuracies": 0.5234375, | |
| "eval_rewards/chosen": -4.6875, | |
| "eval_rewards/margins": 0.302734375, | |
| "eval_rewards/rejected": -4.96875, | |
| "eval_runtime": 46.6928, | |
| "eval_samples_per_second": 42.833, | |
| "eval_steps_per_second": 0.685, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 1.8952879581151834, | |
| "grad_norm": 0.1092778208442411, | |
| "learning_rate": 1.7967316349380213e-05, | |
| "logits/chosen": -16.5, | |
| "logits/rejected": -9.0625, | |
| "logps/chosen": -756.0, | |
| "logps/rejected": -4384.0, | |
| "loss": 0.0041, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.0625, | |
| "rewards/margins": 21.75, | |
| "rewards/rejected": -25.75, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 1.905759162303665, | |
| "grad_norm": 0.6126740987610202, | |
| "learning_rate": 1.7675494329845516e-05, | |
| "logits/chosen": -16.125, | |
| "logits/rejected": -8.875, | |
| "logps/chosen": -812.0, | |
| "logps/rejected": -4896.0, | |
| "loss": 0.0049, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.4375, | |
| "rewards/margins": 24.125, | |
| "rewards/rejected": -28.5, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 1.9162303664921465, | |
| "grad_norm": 0.07427146968932585, | |
| "learning_rate": 1.738476000464026e-05, | |
| "logits/chosen": -16.125, | |
| "logits/rejected": -8.8125, | |
| "logps/chosen": -788.0, | |
| "logps/rejected": -4704.0, | |
| "loss": 0.0019, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.1875, | |
| "rewards/margins": 23.0, | |
| "rewards/rejected": -27.25, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 1.9267015706806283, | |
| "grad_norm": 0.43381946942806676, | |
| "learning_rate": 1.7095156548019646e-05, | |
| "logits/chosen": -16.5, | |
| "logits/rejected": -8.6875, | |
| "logps/chosen": -796.0, | |
| "logps/rejected": -4896.0, | |
| "loss": 0.0044, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.34375, | |
| "rewards/margins": 24.75, | |
| "rewards/rejected": -29.125, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 1.93717277486911, | |
| "grad_norm": 0.06946023463286834, | |
| "learning_rate": 1.6806726966304064e-05, | |
| "logits/chosen": -16.125, | |
| "logits/rejected": -8.5, | |
| "logps/chosen": -820.0, | |
| "logps/rejected": -4832.0, | |
| "loss": 0.0024, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.28125, | |
| "rewards/margins": 24.375, | |
| "rewards/rejected": -28.625, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 1.9476439790575917, | |
| "grad_norm": 0.3035423599466412, | |
| "learning_rate": 1.6519514091492625e-05, | |
| "logits/chosen": -15.8125, | |
| "logits/rejected": -8.5, | |
| "logps/chosen": -772.0, | |
| "logps/rejected": -4864.0, | |
| "loss": 0.0103, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -4.3125, | |
| "rewards/margins": 25.125, | |
| "rewards/rejected": -29.5, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 1.9581151832460733, | |
| "grad_norm": 0.02390442972192678, | |
| "learning_rate": 1.6233560574902627e-05, | |
| "logits/chosen": -16.5, | |
| "logits/rejected": -8.4375, | |
| "logps/chosen": -744.0, | |
| "logps/rejected": -4928.0, | |
| "loss": 0.017, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.09375, | |
| "rewards/margins": 25.25, | |
| "rewards/rejected": -29.375, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 1.9685863874345548, | |
| "grad_norm": 2.0042759567999524, | |
| "learning_rate": 1.5948908880835753e-05, | |
| "logits/chosen": -16.375, | |
| "logits/rejected": -9.0, | |
| "logps/chosen": -772.0, | |
| "logps/rejected": -4800.0, | |
| "loss": 0.0152, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.125, | |
| "rewards/margins": 24.125, | |
| "rewards/rejected": -28.25, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 1.9790575916230366, | |
| "grad_norm": 0.5281626383524519, | |
| "learning_rate": 1.5665601280272124e-05, | |
| "logits/chosen": -16.125, | |
| "logits/rejected": -9.6875, | |
| "logps/chosen": -764.0, | |
| "logps/rejected": -4608.0, | |
| "loss": 0.0056, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -4.125, | |
| "rewards/margins": 23.0, | |
| "rewards/rejected": -27.125, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 1.9895287958115184, | |
| "grad_norm": 0.5068404981201945, | |
| "learning_rate": 1.538367984459298e-05, | |
| "logits/chosen": -15.8125, | |
| "logits/rejected": -9.375, | |
| "logps/chosen": -752.0, | |
| "logps/rejected": -4544.0, | |
| "loss": 0.0031, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.03125, | |
| "rewards/margins": 21.875, | |
| "rewards/rejected": -25.875, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 1.9895287958115184, | |
| "eval_logits/chosen": -16.25, | |
| "eval_logits/rejected": -16.125, | |
| "eval_logps/chosen": -772.0, | |
| "eval_logps/rejected": -772.0, | |
| "eval_loss": 1.265531301498413, | |
| "eval_rewards/accuracies": 0.51171875, | |
| "eval_rewards/chosen": -4.53125, | |
| "eval_rewards/margins": 0.30078125, | |
| "eval_rewards/rejected": -4.84375, | |
| "eval_runtime": 46.7072, | |
| "eval_samples_per_second": 42.82, | |
| "eval_steps_per_second": 0.685, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "grad_norm": 0.02318310230155886, | |
| "learning_rate": 1.5103186439333133e-05, | |
| "logits/chosen": -16.125, | |
| "logits/rejected": -9.9375, | |
| "logps/chosen": -748.0, | |
| "logps/rejected": -4448.0, | |
| "loss": 0.0284, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -3.890625, | |
| "rewards/margins": 21.25, | |
| "rewards/rejected": -25.25, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 2.0104712041884816, | |
| "grad_norm": 0.015520177085299257, | |
| "learning_rate": 1.4824162717963829e-05, | |
| "logits/chosen": -16.125, | |
| "logits/rejected": -9.875, | |
| "logps/chosen": -712.0, | |
| "logps/rejected": -4320.0, | |
| "loss": 0.0022, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -3.578125, | |
| "rewards/margins": 20.125, | |
| "rewards/rejected": -23.75, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 2.020942408376963, | |
| "grad_norm": 0.016572349470184163, | |
| "learning_rate": 1.4546650115707192e-05, | |
| "logits/chosen": -16.125, | |
| "logits/rejected": -10.0, | |
| "logps/chosen": -736.0, | |
| "logps/rejected": -4416.0, | |
| "loss": 0.0003, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -3.890625, | |
| "rewards/margins": 20.875, | |
| "rewards/rejected": -24.75, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 2.031413612565445, | |
| "grad_norm": 0.03569456580194663, | |
| "learning_rate": 1.4270689843383111e-05, | |
| "logits/chosen": -16.5, | |
| "logits/rejected": -9.9375, | |
| "logps/chosen": -744.0, | |
| "logps/rejected": -4480.0, | |
| "loss": 0.0018, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -3.671875, | |
| "rewards/margins": 20.25, | |
| "rewards/rejected": -24.0, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 2.0418848167539267, | |
| "grad_norm": 0.032305953859107756, | |
| "learning_rate": 1.3996322881289347e-05, | |
| "logits/chosen": -16.375, | |
| "logits/rejected": -10.0625, | |
| "logps/chosen": -692.0, | |
| "logps/rejected": -4416.0, | |
| "loss": 0.0008, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -3.5625, | |
| "rewards/margins": 20.75, | |
| "rewards/rejected": -24.25, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 2.0523560209424083, | |
| "grad_norm": 0.01326643794664419, | |
| "learning_rate": 1.3723589973115963e-05, | |
| "logits/chosen": -16.125, | |
| "logits/rejected": -10.3125, | |
| "logps/chosen": -724.0, | |
| "logps/rejected": -4512.0, | |
| "loss": 0.0013, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -3.75, | |
| "rewards/margins": 21.25, | |
| "rewards/rejected": -25.0, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 2.06282722513089, | |
| "grad_norm": 0.01451401449012035, | |
| "learning_rate": 1.3452531619894837e-05, | |
| "logits/chosen": -16.25, | |
| "logits/rejected": -10.0625, | |
| "logps/chosen": -748.0, | |
| "logps/rejected": -4448.0, | |
| "loss": 0.0008, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -3.875, | |
| "rewards/margins": 21.625, | |
| "rewards/rejected": -25.5, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 2.073298429319372, | |
| "grad_norm": 0.05922001767819958, | |
| "learning_rate": 1.3183188073985245e-05, | |
| "logits/chosen": -16.125, | |
| "logits/rejected": -10.0625, | |
| "logps/chosen": -744.0, | |
| "logps/rejected": -4736.0, | |
| "loss": 0.0009, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.0, | |
| "rewards/margins": 22.375, | |
| "rewards/rejected": -26.375, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 2.0837696335078535, | |
| "grad_norm": 0.007567883544913513, | |
| "learning_rate": 1.2915599333096351e-05, | |
| "logits/chosen": -16.125, | |
| "logits/rejected": -9.875, | |
| "logps/chosen": -804.0, | |
| "logps/rejected": -4704.0, | |
| "loss": 0.0009, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.15625, | |
| "rewards/margins": 22.75, | |
| "rewards/rejected": -27.0, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 2.094240837696335, | |
| "grad_norm": 0.020553640994480184, | |
| "learning_rate": 1.264980513434752e-05, | |
| "logits/chosen": -16.375, | |
| "logits/rejected": -9.875, | |
| "logps/chosen": -816.0, | |
| "logps/rejected": -4672.0, | |
| "loss": 0.0007, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.28125, | |
| "rewards/margins": 22.125, | |
| "rewards/rejected": -26.375, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 2.094240837696335, | |
| "eval_logits/chosen": -16.625, | |
| "eval_logits/rejected": -16.5, | |
| "eval_logps/chosen": -804.0, | |
| "eval_logps/rejected": -812.0, | |
| "eval_loss": 1.3137656450271606, | |
| "eval_rewards/accuracies": 0.5078125, | |
| "eval_rewards/chosen": -4.875, | |
| "eval_rewards/margins": 0.369140625, | |
| "eval_rewards/rejected": -5.25, | |
| "eval_runtime": 46.703, | |
| "eval_samples_per_second": 42.824, | |
| "eval_steps_per_second": 0.685, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 2.1047120418848166, | |
| "grad_norm": 0.019862221644006747, | |
| "learning_rate": 1.2385844948367324e-05, | |
| "logits/chosen": -16.375, | |
| "logits/rejected": -9.875, | |
| "logps/chosen": -756.0, | |
| "logps/rejected": -4800.0, | |
| "loss": 0.0007, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.21875, | |
| "rewards/margins": 23.625, | |
| "rewards/rejected": -27.875, | |
| "step": 2010 | |
| }, | |
| { | |
| "epoch": 2.115183246073298, | |
| "grad_norm": 0.014630496583183702, | |
| "learning_rate": 1.2123757973432114e-05, | |
| "logits/chosen": -15.9375, | |
| "logits/rejected": -9.8125, | |
| "logps/chosen": -844.0, | |
| "logps/rejected": -4736.0, | |
| "loss": 0.0012, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.53125, | |
| "rewards/margins": 22.75, | |
| "rewards/rejected": -27.25, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 2.1256544502617802, | |
| "grad_norm": 0.07347149098989865, | |
| "learning_rate": 1.1863583129645106e-05, | |
| "logits/chosen": -16.375, | |
| "logits/rejected": -9.375, | |
| "logps/chosen": -788.0, | |
| "logps/rejected": -4864.0, | |
| "loss": 0.0008, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.25, | |
| "rewards/margins": 23.75, | |
| "rewards/rejected": -28.0, | |
| "step": 2030 | |
| }, | |
| { | |
| "epoch": 2.136125654450262, | |
| "grad_norm": 0.008265748457053025, | |
| "learning_rate": 1.1605359053156607e-05, | |
| "logits/chosen": -16.875, | |
| "logits/rejected": -9.75, | |
| "logps/chosen": -804.0, | |
| "logps/rejected": -4864.0, | |
| "loss": 0.0005, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.28125, | |
| "rewards/margins": 24.125, | |
| "rewards/rejected": -28.375, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 2.1465968586387434, | |
| "grad_norm": 0.027592268148150217, | |
| "learning_rate": 1.1349124090426644e-05, | |
| "logits/chosen": -16.25, | |
| "logits/rejected": -9.4375, | |
| "logps/chosen": -796.0, | |
| "logps/rejected": -4768.0, | |
| "loss": 0.0008, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.25, | |
| "rewards/margins": 23.5, | |
| "rewards/rejected": -27.75, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 2.157068062827225, | |
| "grad_norm": 0.06940828078217133, | |
| "learning_rate": 1.1094916292530406e-05, | |
| "logits/chosen": -16.375, | |
| "logits/rejected": -9.8125, | |
| "logps/chosen": -816.0, | |
| "logps/rejected": -4672.0, | |
| "loss": 0.0008, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.375, | |
| "rewards/margins": 22.875, | |
| "rewards/rejected": -27.25, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 2.167539267015707, | |
| "grad_norm": 0.024289375299553218, | |
| "learning_rate": 1.0842773409507623e-05, | |
| "logits/chosen": -16.375, | |
| "logits/rejected": -9.1875, | |
| "logps/chosen": -784.0, | |
| "logps/rejected": -4896.0, | |
| "loss": 0.0015, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.40625, | |
| "rewards/margins": 24.25, | |
| "rewards/rejected": -28.75, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 2.1780104712041886, | |
| "grad_norm": 0.03832291761523684, | |
| "learning_rate": 1.0592732884756754e-05, | |
| "logits/chosen": -16.125, | |
| "logits/rejected": -9.75, | |
| "logps/chosen": -796.0, | |
| "logps/rejected": -4544.0, | |
| "loss": 0.0011, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.25, | |
| "rewards/margins": 22.75, | |
| "rewards/rejected": -27.125, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 2.18848167539267, | |
| "grad_norm": 0.007557808844138426, | |
| "learning_rate": 1.0344831849474507e-05, | |
| "logits/chosen": -16.5, | |
| "logits/rejected": -9.375, | |
| "logps/chosen": -860.0, | |
| "logps/rejected": -4992.0, | |
| "loss": 0.0017, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.78125, | |
| "rewards/margins": 24.75, | |
| "rewards/rejected": -29.625, | |
| "step": 2090 | |
| }, | |
| { | |
| "epoch": 2.1989528795811517, | |
| "grad_norm": 0.04652530216526909, | |
| "learning_rate": 1.0099107117141879e-05, | |
| "logits/chosen": -16.875, | |
| "logits/rejected": -9.75, | |
| "logps/chosen": -772.0, | |
| "logps/rejected": -4864.0, | |
| "loss": 0.0209, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.25, | |
| "rewards/margins": 25.0, | |
| "rewards/rejected": -29.25, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 2.1989528795811517, | |
| "eval_logits/chosen": -16.875, | |
| "eval_logits/rejected": -16.75, | |
| "eval_logps/chosen": -844.0, | |
| "eval_logps/rejected": -856.0, | |
| "eval_loss": 1.385015606880188, | |
| "eval_rewards/accuracies": 0.51171875, | |
| "eval_rewards/chosen": -5.25, | |
| "eval_rewards/margins": 0.42578125, | |
| "eval_rewards/rejected": -5.65625, | |
| "eval_runtime": 46.7093, | |
| "eval_samples_per_second": 42.818, | |
| "eval_steps_per_second": 0.685, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 2.2094240837696333, | |
| "grad_norm": 0.004585861613642097, | |
| "learning_rate": 9.855595178057334e-06, | |
| "logits/chosen": -16.25, | |
| "logits/rejected": -9.5625, | |
| "logps/chosen": -864.0, | |
| "logps/rejected": -4736.0, | |
| "loss": 0.0005, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.78125, | |
| "rewards/margins": 23.25, | |
| "rewards/rejected": -28.0, | |
| "step": 2110 | |
| }, | |
| { | |
| "epoch": 2.2198952879581153, | |
| "grad_norm": 0.6269416440115462, | |
| "learning_rate": 9.614332193917949e-06, | |
| "logits/chosen": -16.75, | |
| "logits/rejected": -9.0, | |
| "logps/chosen": -920.0, | |
| "logps/rejected": -4896.0, | |
| "loss": 0.0012, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -5.125, | |
| "rewards/margins": 24.625, | |
| "rewards/rejected": -29.75, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 2.230366492146597, | |
| "grad_norm": 0.006956964044466042, | |
| "learning_rate": 9.375353992449385e-06, | |
| "logits/chosen": -16.75, | |
| "logits/rejected": -10.0, | |
| "logps/chosen": -816.0, | |
| "logps/rejected": -4640.0, | |
| "loss": 0.0003, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.46875, | |
| "rewards/margins": 24.0, | |
| "rewards/rejected": -28.5, | |
| "step": 2130 | |
| }, | |
| { | |
| "epoch": 2.2408376963350785, | |
| "grad_norm": 0.02777718193894219, | |
| "learning_rate": 9.138696062085442e-06, | |
| "logits/chosen": -16.625, | |
| "logits/rejected": -9.625, | |
| "logps/chosen": -808.0, | |
| "logps/rejected": -5088.0, | |
| "loss": 0.0012, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.5625, | |
| "rewards/margins": 25.75, | |
| "rewards/rejected": -30.375, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 2.25130890052356, | |
| "grad_norm": 0.017558980618208167, | |
| "learning_rate": 8.904393546698007e-06, | |
| "logits/chosen": -17.0, | |
| "logits/rejected": -9.5, | |
| "logps/chosen": -852.0, | |
| "logps/rejected": -5024.0, | |
| "loss": 0.0012, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.625, | |
| "rewards/margins": 25.75, | |
| "rewards/rejected": -30.375, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 2.261780104712042, | |
| "grad_norm": 0.005707933297978994, | |
| "learning_rate": 8.672481240378142e-06, | |
| "logits/chosen": -16.75, | |
| "logits/rejected": -9.6875, | |
| "logps/chosen": -784.0, | |
| "logps/rejected": -4800.0, | |
| "loss": 0.0004, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.5625, | |
| "rewards/margins": 24.75, | |
| "rewards/rejected": -29.25, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 2.2722513089005236, | |
| "grad_norm": 0.03350831301454641, | |
| "learning_rate": 8.44299358226919e-06, | |
| "logits/chosen": -16.5, | |
| "logits/rejected": -9.25, | |
| "logps/chosen": -820.0, | |
| "logps/rejected": -5056.0, | |
| "loss": 0.0007, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.625, | |
| "rewards/margins": 25.375, | |
| "rewards/rejected": -30.0, | |
| "step": 2170 | |
| }, | |
| { | |
| "epoch": 2.282722513089005, | |
| "grad_norm": 0.02445817375151862, | |
| "learning_rate": 8.215964651452457e-06, | |
| "logits/chosen": -16.5, | |
| "logits/rejected": -8.6875, | |
| "logps/chosen": -840.0, | |
| "logps/rejected": -5024.0, | |
| "loss": 0.0009, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.8125, | |
| "rewards/margins": 25.75, | |
| "rewards/rejected": -30.625, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 2.2931937172774868, | |
| "grad_norm": 0.00915496832852379, | |
| "learning_rate": 7.991428161886502e-06, | |
| "logits/chosen": -16.5, | |
| "logits/rejected": -9.625, | |
| "logps/chosen": -856.0, | |
| "logps/rejected": -4736.0, | |
| "loss": 0.0004, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.78125, | |
| "rewards/margins": 24.625, | |
| "rewards/rejected": -29.375, | |
| "step": 2190 | |
| }, | |
| { | |
| "epoch": 2.303664921465969, | |
| "grad_norm": 0.014155661257121331, | |
| "learning_rate": 7.76941745740061e-06, | |
| "logits/chosen": -16.375, | |
| "logits/rejected": -9.25, | |
| "logps/chosen": -884.0, | |
| "logps/rejected": -4992.0, | |
| "loss": 0.0007, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.96875, | |
| "rewards/margins": 25.0, | |
| "rewards/rejected": -30.0, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 2.303664921465969, | |
| "eval_logits/chosen": -16.875, | |
| "eval_logits/rejected": -16.625, | |
| "eval_logps/chosen": -876.0, | |
| "eval_logps/rejected": -896.0, | |
| "eval_loss": 1.4691718816757202, | |
| "eval_rewards/accuracies": 0.5234375, | |
| "eval_rewards/chosen": -5.5625, | |
| "eval_rewards/margins": 0.498046875, | |
| "eval_rewards/rejected": -6.0625, | |
| "eval_runtime": 46.7159, | |
| "eval_samples_per_second": 42.812, | |
| "eval_steps_per_second": 0.685, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 2.3141361256544504, | |
| "grad_norm": 0.0024207942166172944, | |
| "learning_rate": 7.549965506743123e-06, | |
| "logits/chosen": -17.0, | |
| "logits/rejected": -9.375, | |
| "logps/chosen": -828.0, | |
| "logps/rejected": -5088.0, | |
| "loss": 0.0023, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.65625, | |
| "rewards/margins": 26.25, | |
| "rewards/rejected": -31.0, | |
| "step": 2210 | |
| }, | |
| { | |
| "epoch": 2.324607329842932, | |
| "grad_norm": 0.011500163017294868, | |
| "learning_rate": 7.333104898685631e-06, | |
| "logits/chosen": -16.5, | |
| "logits/rejected": -9.5, | |
| "logps/chosen": -840.0, | |
| "logps/rejected": -4896.0, | |
| "loss": 0.0006, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.75, | |
| "rewards/margins": 25.625, | |
| "rewards/rejected": -30.375, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 2.3350785340314135, | |
| "grad_norm": 0.008543884614552025, | |
| "learning_rate": 7.118867837183518e-06, | |
| "logits/chosen": -16.5, | |
| "logits/rejected": -9.3125, | |
| "logps/chosen": -928.0, | |
| "logps/rejected": -4992.0, | |
| "loss": 0.0004, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -5.0, | |
| "rewards/margins": 24.875, | |
| "rewards/rejected": -29.875, | |
| "step": 2230 | |
| }, | |
| { | |
| "epoch": 2.345549738219895, | |
| "grad_norm": 0.009319619405825642, | |
| "learning_rate": 6.9072861365936045e-06, | |
| "logits/chosen": -16.75, | |
| "logits/rejected": -9.1875, | |
| "logps/chosen": -888.0, | |
| "logps/rejected": -5088.0, | |
| "loss": 0.0003, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.78125, | |
| "rewards/margins": 26.25, | |
| "rewards/rejected": -31.0, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 2.356020942408377, | |
| "grad_norm": 0.0035591525557576815, | |
| "learning_rate": 6.6983912169497005e-06, | |
| "logits/chosen": -16.625, | |
| "logits/rejected": -9.75, | |
| "logps/chosen": -888.0, | |
| "logps/rejected": -4928.0, | |
| "loss": 0.0007, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.75, | |
| "rewards/margins": 25.375, | |
| "rewards/rejected": -30.0, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 2.3664921465968587, | |
| "grad_norm": 0.022955207098826827, | |
| "learning_rate": 6.49221409929677e-06, | |
| "logits/chosen": -16.875, | |
| "logits/rejected": -8.875, | |
| "logps/chosen": -816.0, | |
| "logps/rejected": -5216.0, | |
| "loss": 0.0008, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.75, | |
| "rewards/margins": 27.75, | |
| "rewards/rejected": -32.5, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 2.3769633507853403, | |
| "grad_norm": 0.014106227936770445, | |
| "learning_rate": 6.2887854010842075e-06, | |
| "logits/chosen": -16.5, | |
| "logits/rejected": -9.0, | |
| "logps/chosen": -816.0, | |
| "logps/rejected": -5248.0, | |
| "loss": 0.0005, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.65625, | |
| "rewards/margins": 27.5, | |
| "rewards/rejected": -32.25, | |
| "step": 2270 | |
| }, | |
| { | |
| "epoch": 2.387434554973822, | |
| "grad_norm": 0.021666402580970915, | |
| "learning_rate": 6.0881353316191385e-06, | |
| "logits/chosen": -16.25, | |
| "logits/rejected": -9.0625, | |
| "logps/chosen": -908.0, | |
| "logps/rejected": -4800.0, | |
| "loss": 0.0018, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -5.15625, | |
| "rewards/margins": 24.875, | |
| "rewards/rejected": -30.125, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 2.3979057591623034, | |
| "grad_norm": 0.29662024118875624, | |
| "learning_rate": 5.890293687580381e-06, | |
| "logits/chosen": -16.625, | |
| "logits/rejected": -9.9375, | |
| "logps/chosen": -884.0, | |
| "logps/rejected": -4864.0, | |
| "loss": 0.001, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -5.0625, | |
| "rewards/margins": 25.0, | |
| "rewards/rejected": -30.0, | |
| "step": 2290 | |
| }, | |
| { | |
| "epoch": 2.4083769633507854, | |
| "grad_norm": 0.06978047451232525, | |
| "learning_rate": 5.695289848593532e-06, | |
| "logits/chosen": -16.625, | |
| "logits/rejected": -9.3125, | |
| "logps/chosen": -820.0, | |
| "logps/rejected": -4928.0, | |
| "loss": 0.0008, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.65625, | |
| "rewards/margins": 26.0, | |
| "rewards/rejected": -30.625, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 2.4083769633507854, | |
| "eval_logits/chosen": -16.75, | |
| "eval_logits/rejected": -16.5, | |
| "eval_logps/chosen": -900.0, | |
| "eval_logps/rejected": -924.0, | |
| "eval_loss": 1.5070468187332153, | |
| "eval_rewards/accuracies": 0.517578125, | |
| "eval_rewards/chosen": -5.8125, | |
| "eval_rewards/margins": 0.53125, | |
| "eval_rewards/rejected": -6.34375, | |
| "eval_runtime": 46.7009, | |
| "eval_samples_per_second": 42.826, | |
| "eval_steps_per_second": 0.685, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 2.418848167539267, | |
| "grad_norm": 0.010034633693248394, | |
| "learning_rate": 5.50315277286812e-06, | |
| "logits/chosen": -17.0, | |
| "logits/rejected": -9.4375, | |
| "logps/chosen": -832.0, | |
| "logps/rejected": -5088.0, | |
| "loss": 0.0004, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.9375, | |
| "rewards/margins": 27.0, | |
| "rewards/rejected": -31.875, | |
| "step": 2310 | |
| }, | |
| { | |
| "epoch": 2.4293193717277486, | |
| "grad_norm": 0.012388994578689047, | |
| "learning_rate": 5.313910992897281e-06, | |
| "logits/chosen": -16.375, | |
| "logits/rejected": -8.625, | |
| "logps/chosen": -828.0, | |
| "logps/rejected": -5120.0, | |
| "loss": 0.0007, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.6875, | |
| "rewards/margins": 27.25, | |
| "rewards/rejected": -31.875, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 2.4397905759162306, | |
| "grad_norm": 0.006945961061839208, | |
| "learning_rate": 5.127592611220658e-06, | |
| "logits/chosen": -16.375, | |
| "logits/rejected": -9.0, | |
| "logps/chosen": -896.0, | |
| "logps/rejected": -5152.0, | |
| "loss": 0.0006, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -5.0625, | |
| "rewards/margins": 26.875, | |
| "rewards/rejected": -32.0, | |
| "step": 2330 | |
| }, | |
| { | |
| "epoch": 2.450261780104712, | |
| "grad_norm": 0.06571927999165755, | |
| "learning_rate": 4.944225296251159e-06, | |
| "logits/chosen": -16.5, | |
| "logits/rejected": -8.4375, | |
| "logps/chosen": -836.0, | |
| "logps/rejected": -5376.0, | |
| "loss": 0.0013, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.96875, | |
| "rewards/margins": 29.125, | |
| "rewards/rejected": -34.25, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 2.4607329842931938, | |
| "grad_norm": 0.007348485389758178, | |
| "learning_rate": 4.763836278166172e-06, | |
| "logits/chosen": -16.75, | |
| "logits/rejected": -9.375, | |
| "logps/chosen": -848.0, | |
| "logps/rejected": -5024.0, | |
| "loss": 0.0007, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.9375, | |
| "rewards/margins": 26.125, | |
| "rewards/rejected": -31.0, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 2.4712041884816753, | |
| "grad_norm": 0.00577919335252458, | |
| "learning_rate": 4.586452344863876e-06, | |
| "logits/chosen": -16.75, | |
| "logits/rejected": -9.0, | |
| "logps/chosen": -856.0, | |
| "logps/rejected": -4896.0, | |
| "loss": 0.0009, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.875, | |
| "rewards/margins": 25.75, | |
| "rewards/rejected": -30.625, | |
| "step": 2360 | |
| }, | |
| { | |
| "epoch": 2.481675392670157, | |
| "grad_norm": 0.0023976973502127987, | |
| "learning_rate": 4.412099837985192e-06, | |
| "logits/chosen": -16.5, | |
| "logits/rejected": -8.75, | |
| "logps/chosen": -876.0, | |
| "logps/rejected": -5120.0, | |
| "loss": 0.0002, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -5.0, | |
| "rewards/margins": 26.125, | |
| "rewards/rejected": -31.125, | |
| "step": 2370 | |
| }, | |
| { | |
| "epoch": 2.492146596858639, | |
| "grad_norm": 0.016089648327252022, | |
| "learning_rate": 4.240804649002089e-06, | |
| "logits/chosen": -16.5, | |
| "logits/rejected": -8.75, | |
| "logps/chosen": -816.0, | |
| "logps/rejected": -5184.0, | |
| "loss": 0.0027, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.84375, | |
| "rewards/margins": 27.125, | |
| "rewards/rejected": -32.0, | |
| "step": 2380 | |
| }, | |
| { | |
| "epoch": 2.5026178010471205, | |
| "grad_norm": 0.0028493740824482704, | |
| "learning_rate": 4.072592215372598e-06, | |
| "logits/chosen": -16.375, | |
| "logits/rejected": -8.875, | |
| "logps/chosen": -940.0, | |
| "logps/rejected": -4928.0, | |
| "loss": 0.0027, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -5.375, | |
| "rewards/margins": 24.25, | |
| "rewards/rejected": -29.625, | |
| "step": 2390 | |
| }, | |
| { | |
| "epoch": 2.513089005235602, | |
| "grad_norm": 0.004115929739447787, | |
| "learning_rate": 3.907487516763389e-06, | |
| "logits/chosen": -16.75, | |
| "logits/rejected": -9.0, | |
| "logps/chosen": -884.0, | |
| "logps/rejected": -5120.0, | |
| "loss": 0.0003, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -5.1875, | |
| "rewards/margins": 27.375, | |
| "rewards/rejected": -32.5, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 2.513089005235602, | |
| "eval_logits/chosen": -16.75, | |
| "eval_logits/rejected": -16.625, | |
| "eval_logps/chosen": -880.0, | |
| "eval_logps/rejected": -900.0, | |
| "eval_loss": 1.464937448501587, | |
| "eval_rewards/accuracies": 0.5234375, | |
| "eval_rewards/chosen": -5.625, | |
| "eval_rewards/margins": 0.50390625, | |
| "eval_rewards/rejected": -6.125, | |
| "eval_runtime": 46.711, | |
| "eval_samples_per_second": 42.816, | |
| "eval_steps_per_second": 0.685, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 2.5235602094240837, | |
| "grad_norm": 0.044668647879899444, | |
| "learning_rate": 3.745515071340286e-06, | |
| "logits/chosen": -16.5, | |
| "logits/rejected": -8.625, | |
| "logps/chosen": -936.0, | |
| "logps/rejected": -5120.0, | |
| "loss": 0.002, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -5.46875, | |
| "rewards/margins": 26.625, | |
| "rewards/rejected": -32.25, | |
| "step": 2410 | |
| }, | |
| { | |
| "epoch": 2.5340314136125652, | |
| "grad_norm": 0.024244164584912388, | |
| "learning_rate": 3.586698932127236e-06, | |
| "logits/chosen": -16.5, | |
| "logits/rejected": -9.125, | |
| "logps/chosen": -848.0, | |
| "logps/rejected": -5152.0, | |
| "loss": 0.0025, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.875, | |
| "rewards/margins": 26.875, | |
| "rewards/rejected": -31.75, | |
| "step": 2420 | |
| }, | |
| { | |
| "epoch": 2.5445026178010473, | |
| "grad_norm": 0.03693103543925516, | |
| "learning_rate": 3.4310626834344744e-06, | |
| "logits/chosen": -16.75, | |
| "logits/rejected": -9.4375, | |
| "logps/chosen": -872.0, | |
| "logps/rejected": -4864.0, | |
| "loss": 0.0015, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.84375, | |
| "rewards/margins": 25.375, | |
| "rewards/rejected": -30.125, | |
| "step": 2430 | |
| }, | |
| { | |
| "epoch": 2.554973821989529, | |
| "grad_norm": 0.045566799580578, | |
| "learning_rate": 3.2786294373562343e-06, | |
| "logits/chosen": -16.125, | |
| "logits/rejected": -8.5, | |
| "logps/chosen": -884.0, | |
| "logps/rejected": -5344.0, | |
| "loss": 0.0011, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -5.0, | |
| "rewards/margins": 27.375, | |
| "rewards/rejected": -32.5, | |
| "step": 2440 | |
| }, | |
| { | |
| "epoch": 2.5654450261780104, | |
| "grad_norm": 0.020864858503065246, | |
| "learning_rate": 3.1294218303385526e-06, | |
| "logits/chosen": -15.9375, | |
| "logits/rejected": -8.625, | |
| "logps/chosen": -900.0, | |
| "logps/rejected": -5024.0, | |
| "loss": 0.0004, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.96875, | |
| "rewards/margins": 25.375, | |
| "rewards/rejected": -30.375, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 2.5759162303664924, | |
| "grad_norm": 0.012382389140262351, | |
| "learning_rate": 2.9834620198177483e-06, | |
| "logits/chosen": -16.125, | |
| "logits/rejected": -9.625, | |
| "logps/chosen": -884.0, | |
| "logps/rejected": -4832.0, | |
| "loss": 0.0005, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -5.0625, | |
| "rewards/margins": 25.5, | |
| "rewards/rejected": -30.5, | |
| "step": 2460 | |
| }, | |
| { | |
| "epoch": 2.5863874345549736, | |
| "grad_norm": 0.0027925494540995484, | |
| "learning_rate": 2.840771680930068e-06, | |
| "logits/chosen": -16.625, | |
| "logits/rejected": -8.625, | |
| "logps/chosen": -928.0, | |
| "logps/rejected": -5312.0, | |
| "loss": 0.0003, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -5.3125, | |
| "rewards/margins": 27.5, | |
| "rewards/rejected": -32.75, | |
| "step": 2470 | |
| }, | |
| { | |
| "epoch": 2.5968586387434556, | |
| "grad_norm": 0.0016360092495325761, | |
| "learning_rate": 2.701372003292868e-06, | |
| "logits/chosen": -16.5, | |
| "logits/rejected": -8.5625, | |
| "logps/chosen": -832.0, | |
| "logps/rejected": -5184.0, | |
| "loss": 0.0003, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.6875, | |
| "rewards/margins": 27.125, | |
| "rewards/rejected": -31.875, | |
| "step": 2480 | |
| }, | |
| { | |
| "epoch": 2.607329842931937, | |
| "grad_norm": 0.017243919932531254, | |
| "learning_rate": 2.56528368785795e-06, | |
| "logits/chosen": -16.75, | |
| "logits/rejected": -8.4375, | |
| "logps/chosen": -888.0, | |
| "logps/rejected": -5248.0, | |
| "loss": 0.0003, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.90625, | |
| "rewards/margins": 27.0, | |
| "rewards/rejected": -31.875, | |
| "step": 2490 | |
| }, | |
| { | |
| "epoch": 2.6178010471204187, | |
| "grad_norm": 0.01371157797145213, | |
| "learning_rate": 2.432526943837471e-06, | |
| "logits/chosen": -16.625, | |
| "logits/rejected": -8.875, | |
| "logps/chosen": -836.0, | |
| "logps/rejected": -5088.0, | |
| "loss": 0.0003, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.75, | |
| "rewards/margins": 26.875, | |
| "rewards/rejected": -31.625, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 2.6178010471204187, | |
| "eval_logits/chosen": -16.75, | |
| "eval_logits/rejected": -16.5, | |
| "eval_logps/chosen": -872.0, | |
| "eval_logps/rejected": -892.0, | |
| "eval_loss": 1.436843752861023, | |
| "eval_rewards/accuracies": 0.517578125, | |
| "eval_rewards/chosen": -5.53125, | |
| "eval_rewards/margins": 0.498046875, | |
| "eval_rewards/rejected": -6.03125, | |
| "eval_runtime": 46.6739, | |
| "eval_samples_per_second": 42.851, | |
| "eval_steps_per_second": 0.686, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 2.6282722513089007, | |
| "grad_norm": 0.020663148763788317, | |
| "learning_rate": 2.3031214857028446e-06, | |
| "logits/chosen": -16.375, | |
| "logits/rejected": -8.9375, | |
| "logps/chosen": -880.0, | |
| "logps/rejected": -5056.0, | |
| "loss": 0.0004, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.875, | |
| "rewards/margins": 26.375, | |
| "rewards/rejected": -31.25, | |
| "step": 2510 | |
| }, | |
| { | |
| "epoch": 2.6387434554973823, | |
| "grad_norm": 0.024700514579579457, | |
| "learning_rate": 2.177086530257139e-06, | |
| "logits/chosen": -16.75, | |
| "logits/rejected": -9.3125, | |
| "logps/chosen": -820.0, | |
| "logps/rejected": -5312.0, | |
| "loss": 0.0003, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.71875, | |
| "rewards/margins": 27.875, | |
| "rewards/rejected": -32.5, | |
| "step": 2520 | |
| }, | |
| { | |
| "epoch": 2.649214659685864, | |
| "grad_norm": 0.0016085920818353391, | |
| "learning_rate": 2.054440793781354e-06, | |
| "logits/chosen": -16.5, | |
| "logits/rejected": -8.9375, | |
| "logps/chosen": -816.0, | |
| "logps/rejected": -5152.0, | |
| "loss": 0.0004, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.53125, | |
| "rewards/margins": 27.125, | |
| "rewards/rejected": -31.625, | |
| "step": 2530 | |
| }, | |
| { | |
| "epoch": 2.6596858638743455, | |
| "grad_norm": 0.007406101304927939, | |
| "learning_rate": 1.9352024892550587e-06, | |
| "logits/chosen": -16.625, | |
| "logits/rejected": -9.125, | |
| "logps/chosen": -840.0, | |
| "logps/rejected": -5152.0, | |
| "loss": 0.0006, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.6875, | |
| "rewards/margins": 27.25, | |
| "rewards/rejected": -31.875, | |
| "step": 2540 | |
| }, | |
| { | |
| "epoch": 2.670157068062827, | |
| "grad_norm": 0.03527119941965091, | |
| "learning_rate": 1.8193893236517379e-06, | |
| "logits/chosen": -16.375, | |
| "logits/rejected": -9.125, | |
| "logps/chosen": -880.0, | |
| "logps/rejected": -5248.0, | |
| "loss": 0.0007, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -5.0, | |
| "rewards/margins": 26.75, | |
| "rewards/rejected": -31.75, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 2.680628272251309, | |
| "grad_norm": 0.0028971611021014073, | |
| "learning_rate": 1.7070184953092956e-06, | |
| "logits/chosen": -16.5, | |
| "logits/rejected": -9.1875, | |
| "logps/chosen": -848.0, | |
| "logps/rejected": -5024.0, | |
| "loss": 0.0004, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.71875, | |
| "rewards/margins": 26.875, | |
| "rewards/rejected": -31.625, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 2.6910994764397906, | |
| "grad_norm": 0.01640388047082497, | |
| "learning_rate": 1.5981066913760918e-06, | |
| "logits/chosen": -16.5, | |
| "logits/rejected": -9.4375, | |
| "logps/chosen": -800.0, | |
| "logps/rejected": -4896.0, | |
| "loss": 0.0005, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.46875, | |
| "rewards/margins": 25.875, | |
| "rewards/rejected": -30.375, | |
| "step": 2570 | |
| }, | |
| { | |
| "epoch": 2.701570680628272, | |
| "grad_norm": 0.024791878539349648, | |
| "learning_rate": 1.492670085332884e-06, | |
| "logits/chosen": -16.75, | |
| "logits/rejected": -9.0, | |
| "logps/chosen": -888.0, | |
| "logps/rejected": -5376.0, | |
| "loss": 0.0005, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -5.0, | |
| "rewards/margins": 26.75, | |
| "rewards/rejected": -31.75, | |
| "step": 2580 | |
| }, | |
| { | |
| "epoch": 2.712041884816754, | |
| "grad_norm": 0.023998208640493034, | |
| "learning_rate": 1.3907243345910787e-06, | |
| "logits/chosen": -16.625, | |
| "logits/rejected": -8.5, | |
| "logps/chosen": -896.0, | |
| "logps/rejected": -5120.0, | |
| "loss": 0.001, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.9375, | |
| "rewards/margins": 25.75, | |
| "rewards/rejected": -30.75, | |
| "step": 2590 | |
| }, | |
| { | |
| "epoch": 2.7225130890052354, | |
| "grad_norm": 0.011511351536525903, | |
| "learning_rate": 1.292284578167538e-06, | |
| "logits/chosen": -16.375, | |
| "logits/rejected": -8.8125, | |
| "logps/chosen": -832.0, | |
| "logps/rejected": -5088.0, | |
| "loss": 0.0007, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.625, | |
| "rewards/margins": 27.125, | |
| "rewards/rejected": -31.75, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 2.7225130890052354, | |
| "eval_logits/chosen": -16.75, | |
| "eval_logits/rejected": -16.5, | |
| "eval_logps/chosen": -876.0, | |
| "eval_logps/rejected": -896.0, | |
| "eval_loss": 1.4452344179153442, | |
| "eval_rewards/accuracies": 0.521484375, | |
| "eval_rewards/chosen": -5.5625, | |
| "eval_rewards/margins": 0.50390625, | |
| "eval_rewards/rejected": -6.09375, | |
| "eval_runtime": 46.7157, | |
| "eval_samples_per_second": 42.812, | |
| "eval_steps_per_second": 0.685, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 2.7329842931937174, | |
| "grad_norm": 5.424214158979469, | |
| "learning_rate": 1.1973654344364925e-06, | |
| "logits/chosen": -16.5, | |
| "logits/rejected": -9.25, | |
| "logps/chosen": -848.0, | |
| "logps/rejected": -4928.0, | |
| "loss": 0.0029, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.625, | |
| "rewards/margins": 26.125, | |
| "rewards/rejected": -30.75, | |
| "step": 2610 | |
| }, | |
| { | |
| "epoch": 2.743455497382199, | |
| "grad_norm": 0.044904576481916446, | |
| "learning_rate": 1.1059809989586583e-06, | |
| "logits/chosen": -17.0, | |
| "logits/rejected": -8.5625, | |
| "logps/chosen": -816.0, | |
| "logps/rejected": -5184.0, | |
| "loss": 0.0017, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.46875, | |
| "rewards/margins": 28.125, | |
| "rewards/rejected": -32.5, | |
| "step": 2620 | |
| }, | |
| { | |
| "epoch": 2.7539267015706805, | |
| "grad_norm": 0.011089274695951617, | |
| "learning_rate": 1.018144842388044e-06, | |
| "logits/chosen": -16.625, | |
| "logits/rejected": -8.75, | |
| "logps/chosen": -880.0, | |
| "logps/rejected": -5344.0, | |
| "loss": 0.0003, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.8125, | |
| "rewards/margins": 27.75, | |
| "rewards/rejected": -32.5, | |
| "step": 2630 | |
| }, | |
| { | |
| "epoch": 2.7643979057591626, | |
| "grad_norm": 0.00600884911900604, | |
| "learning_rate": 9.338700084567109e-07, | |
| "logits/chosen": -16.25, | |
| "logits/rejected": -8.875, | |
| "logps/chosen": -848.0, | |
| "logps/rejected": -5344.0, | |
| "loss": 0.0003, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.8125, | |
| "rewards/margins": 28.25, | |
| "rewards/rejected": -33.0, | |
| "step": 2640 | |
| }, | |
| { | |
| "epoch": 2.774869109947644, | |
| "grad_norm": 0.0077011113450599, | |
| "learning_rate": 8.531690120377606e-07, | |
| "logits/chosen": -16.5, | |
| "logits/rejected": -9.1875, | |
| "logps/chosen": -884.0, | |
| "logps/rejected": -4960.0, | |
| "loss": 0.0008, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.84375, | |
| "rewards/margins": 25.5, | |
| "rewards/rejected": -30.375, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 2.7853403141361257, | |
| "grad_norm": 0.002069386970454753, | |
| "learning_rate": 7.760538372868636e-07, | |
| "logits/chosen": -16.625, | |
| "logits/rejected": -9.0625, | |
| "logps/chosen": -848.0, | |
| "logps/rejected": -5184.0, | |
| "loss": 0.0004, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.84375, | |
| "rewards/margins": 26.875, | |
| "rewards/rejected": -31.75, | |
| "step": 2660 | |
| }, | |
| { | |
| "epoch": 2.7958115183246073, | |
| "grad_norm": 0.01979807938743199, | |
| "learning_rate": 7.025359358626166e-07, | |
| "logits/chosen": -16.5, | |
| "logits/rejected": -8.625, | |
| "logps/chosen": -912.0, | |
| "logps/rejected": -5248.0, | |
| "loss": 0.0004, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -5.125, | |
| "rewards/margins": 27.75, | |
| "rewards/rejected": -33.0, | |
| "step": 2670 | |
| }, | |
| { | |
| "epoch": 2.806282722513089, | |
| "grad_norm": 0.017642667236110365, | |
| "learning_rate": 6.326262252259463e-07, | |
| "logits/chosen": -16.25, | |
| "logits/rejected": -8.375, | |
| "logps/chosen": -900.0, | |
| "logps/rejected": -5312.0, | |
| "loss": 0.0004, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -5.0, | |
| "rewards/margins": 28.0, | |
| "rewards/rejected": -33.0, | |
| "step": 2680 | |
| }, | |
| { | |
| "epoch": 2.816753926701571, | |
| "grad_norm": 0.017205575540545966, | |
| "learning_rate": 5.66335087018871e-07, | |
| "logits/chosen": -16.375, | |
| "logits/rejected": -8.5625, | |
| "logps/chosen": -896.0, | |
| "logps/rejected": -5184.0, | |
| "loss": 0.0052, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -5.0, | |
| "rewards/margins": 27.375, | |
| "rewards/rejected": -32.5, | |
| "step": 2690 | |
| }, | |
| { | |
| "epoch": 2.8272251308900525, | |
| "grad_norm": 0.024534963171825245, | |
| "learning_rate": 5.036723655228226e-07, | |
| "logits/chosen": -16.5, | |
| "logits/rejected": -8.875, | |
| "logps/chosen": -844.0, | |
| "logps/rejected": -5120.0, | |
| "loss": 0.0009, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.875, | |
| "rewards/margins": 27.125, | |
| "rewards/rejected": -32.0, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 2.8272251308900525, | |
| "eval_logits/chosen": -16.75, | |
| "eval_logits/rejected": -16.5, | |
| "eval_logps/chosen": -880.0, | |
| "eval_logps/rejected": -900.0, | |
| "eval_loss": 1.4518905878067017, | |
| "eval_rewards/accuracies": 0.5234375, | |
| "eval_rewards/chosen": -5.59375, | |
| "eval_rewards/margins": 0.5078125, | |
| "eval_rewards/rejected": -6.125, | |
| "eval_runtime": 46.6906, | |
| "eval_samples_per_second": 42.835, | |
| "eval_steps_per_second": 0.685, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 2.837696335078534, | |
| "grad_norm": 0.018661925383804283, | |
| "learning_rate": 4.446473661967432e-07, | |
| "logits/chosen": -16.75, | |
| "logits/rejected": -9.25, | |
| "logps/chosen": -848.0, | |
| "logps/rejected": -5024.0, | |
| "loss": 0.0013, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.84375, | |
| "rewards/margins": 26.625, | |
| "rewards/rejected": -31.5, | |
| "step": 2710 | |
| }, | |
| { | |
| "epoch": 2.8481675392670156, | |
| "grad_norm": 0.021211200996309684, | |
| "learning_rate": 3.892688542952411e-07, | |
| "logits/chosen": -16.875, | |
| "logits/rejected": -8.875, | |
| "logps/chosen": -796.0, | |
| "logps/rejected": -4832.0, | |
| "loss": 0.0004, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.40625, | |
| "rewards/margins": 25.75, | |
| "rewards/rejected": -30.125, | |
| "step": 2720 | |
| }, | |
| { | |
| "epoch": 2.858638743455497, | |
| "grad_norm": 0.017013765166187345, | |
| "learning_rate": 3.3754505356694e-07, | |
| "logits/chosen": -16.625, | |
| "logits/rejected": -8.3125, | |
| "logps/chosen": -848.0, | |
| "logps/rejected": -5408.0, | |
| "loss": 0.0005, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.71875, | |
| "rewards/margins": 28.625, | |
| "rewards/rejected": -33.25, | |
| "step": 2730 | |
| }, | |
| { | |
| "epoch": 2.869109947643979, | |
| "grad_norm": 0.011474155460371553, | |
| "learning_rate": 2.8948364503322276e-07, | |
| "logits/chosen": -16.5, | |
| "logits/rejected": -9.125, | |
| "logps/chosen": -796.0, | |
| "logps/rejected": -4928.0, | |
| "loss": 0.0026, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.5, | |
| "rewards/margins": 26.375, | |
| "rewards/rejected": -30.875, | |
| "step": 2740 | |
| }, | |
| { | |
| "epoch": 2.8795811518324608, | |
| "grad_norm": 0.007837253277418682, | |
| "learning_rate": 2.4509176584762706e-07, | |
| "logits/chosen": -16.625, | |
| "logits/rejected": -8.75, | |
| "logps/chosen": -824.0, | |
| "logps/rejected": -4864.0, | |
| "loss": 0.0007, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.625, | |
| "rewards/margins": 25.875, | |
| "rewards/rejected": -30.5, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 2.8900523560209423, | |
| "grad_norm": 0.009081754545139684, | |
| "learning_rate": 2.0437600823595694e-07, | |
| "logits/chosen": -16.25, | |
| "logits/rejected": -9.0, | |
| "logps/chosen": -888.0, | |
| "logps/rejected": -5152.0, | |
| "loss": 0.0006, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.875, | |
| "rewards/margins": 27.0, | |
| "rewards/rejected": -31.875, | |
| "step": 2760 | |
| }, | |
| { | |
| "epoch": 2.900523560209424, | |
| "grad_norm": 0.007291213301121672, | |
| "learning_rate": 1.6734241851733e-07, | |
| "logits/chosen": -16.625, | |
| "logits/rejected": -8.6875, | |
| "logps/chosen": -924.0, | |
| "logps/rejected": -5120.0, | |
| "loss": 0.0002, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -5.09375, | |
| "rewards/margins": 26.625, | |
| "rewards/rejected": -31.625, | |
| "step": 2770 | |
| }, | |
| { | |
| "epoch": 2.9109947643979055, | |
| "grad_norm": 0.03006213904020165, | |
| "learning_rate": 1.339964962062984e-07, | |
| "logits/chosen": -16.5, | |
| "logits/rejected": -9.0625, | |
| "logps/chosen": -832.0, | |
| "logps/rejected": -5312.0, | |
| "loss": 0.0025, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.90625, | |
| "rewards/margins": 28.875, | |
| "rewards/rejected": -33.75, | |
| "step": 2780 | |
| }, | |
| { | |
| "epoch": 2.9214659685863875, | |
| "grad_norm": 0.015452254820070541, | |
| "learning_rate": 1.0434319319617136e-07, | |
| "logits/chosen": -16.75, | |
| "logits/rejected": -8.875, | |
| "logps/chosen": -832.0, | |
| "logps/rejected": -5344.0, | |
| "loss": 0.0011, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.8125, | |
| "rewards/margins": 28.625, | |
| "rewards/rejected": -33.5, | |
| "step": 2790 | |
| }, | |
| { | |
| "epoch": 2.931937172774869, | |
| "grad_norm": 0.037567299428383, | |
| "learning_rate": 7.838691302365363e-08, | |
| "logits/chosen": -16.625, | |
| "logits/rejected": -9.25, | |
| "logps/chosen": -836.0, | |
| "logps/rejected": -4928.0, | |
| "loss": 0.0005, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.625, | |
| "rewards/margins": 26.25, | |
| "rewards/rejected": -30.875, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 2.931937172774869, | |
| "eval_logits/chosen": -16.75, | |
| "eval_logits/rejected": -16.5, | |
| "eval_logps/chosen": -880.0, | |
| "eval_logps/rejected": -900.0, | |
| "eval_loss": 1.4524531364440918, | |
| "eval_rewards/accuracies": 0.5234375, | |
| "eval_rewards/chosen": -5.59375, | |
| "eval_rewards/margins": 0.5078125, | |
| "eval_rewards/rejected": -6.125, | |
| "eval_runtime": 46.7036, | |
| "eval_samples_per_second": 42.823, | |
| "eval_steps_per_second": 0.685, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 2.9424083769633507, | |
| "grad_norm": 0.005736478908413097, | |
| "learning_rate": 5.613151021490459e-08, | |
| "logits/chosen": -16.75, | |
| "logits/rejected": -8.6875, | |
| "logps/chosen": -784.0, | |
| "logps/rejected": -5120.0, | |
| "loss": 0.0004, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.625, | |
| "rewards/margins": 27.75, | |
| "rewards/rejected": -32.25, | |
| "step": 2810 | |
| }, | |
| { | |
| "epoch": 2.9528795811518327, | |
| "grad_norm": 0.021326505606858653, | |
| "learning_rate": 3.7580289713159965e-08, | |
| "logits/chosen": -16.625, | |
| "logits/rejected": -9.0, | |
| "logps/chosen": -880.0, | |
| "logps/rejected": -4864.0, | |
| "loss": 0.0004, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.84375, | |
| "rewards/margins": 25.625, | |
| "rewards/rejected": -30.5, | |
| "step": 2820 | |
| }, | |
| { | |
| "epoch": 2.9633507853403143, | |
| "grad_norm": 0.016046250675575274, | |
| "learning_rate": 2.2736006387927277e-08, | |
| "logits/chosen": -16.625, | |
| "logits/rejected": -8.6875, | |
| "logps/chosen": -876.0, | |
| "logps/rejected": -5184.0, | |
| "loss": 0.0006, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.9375, | |
| "rewards/margins": 27.375, | |
| "rewards/rejected": -32.25, | |
| "step": 2830 | |
| }, | |
| { | |
| "epoch": 2.973821989528796, | |
| "grad_norm": 0.0016206262708230927, | |
| "learning_rate": 1.1600864625893604e-08, | |
| "logits/chosen": -16.625, | |
| "logits/rejected": -9.0, | |
| "logps/chosen": -824.0, | |
| "logps/rejected": -5248.0, | |
| "loss": 0.0003, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.53125, | |
| "rewards/margins": 27.75, | |
| "rewards/rejected": -32.25, | |
| "step": 2840 | |
| }, | |
| { | |
| "epoch": 2.9842931937172774, | |
| "grad_norm": 0.005086526160486453, | |
| "learning_rate": 4.176518003573549e-09, | |
| "logits/chosen": -16.5, | |
| "logits/rejected": -9.0, | |
| "logps/chosen": -860.0, | |
| "logps/rejected": -5120.0, | |
| "loss": 0.0004, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.90625, | |
| "rewards/margins": 26.5, | |
| "rewards/rejected": -31.5, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 2.994764397905759, | |
| "grad_norm": 0.007654692754901971, | |
| "learning_rate": 4.640690417528903e-10, | |
| "logits/chosen": -16.5, | |
| "logits/rejected": -9.0, | |
| "logps/chosen": -828.0, | |
| "logps/rejected": -5216.0, | |
| "loss": 0.0005, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -4.78125, | |
| "rewards/margins": 28.375, | |
| "rewards/rejected": -33.25, | |
| "step": 2860 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "step": 2865, | |
| "total_flos": 0.0, | |
| "train_loss": 0.03265144428803893, | |
| "train_runtime": 12446.6394, | |
| "train_samples_per_second": 14.731, | |
| "train_steps_per_second": 0.23 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 2865, | |
| "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": false, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 0.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |