{ "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": 1.1991031976026794, "learning_rate": 1.7421602787456448e-07, "logits/chosen": -11.75, "logits/rejected": -11.8125, "logps/chosen": -488.0, "logps/rejected": -436.0, "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": 1.1712300437505658, "learning_rate": 1.7421602787456445e-06, "logits/chosen": -10.875, "logits/rejected": -11.25, "logps/chosen": -510.0, "logps/rejected": -440.0, "loss": 0.6919, "rewards/accuracies": 0.1666666716337204, "rewards/chosen": 0.00083160400390625, "rewards/margins": -0.00011920928955078125, "rewards/rejected": 0.000949859619140625, "step": 10 }, { "epoch": 0.020942408376963352, "grad_norm": 1.1208003399144875, "learning_rate": 3.484320557491289e-06, "logits/chosen": -10.75, "logits/rejected": -10.875, "logps/chosen": -446.0, "logps/rejected": -426.0, "loss": 0.6902, "rewards/accuracies": 0.3187499940395355, "rewards/chosen": 0.004119873046875, "rewards/margins": 0.0023956298828125, "rewards/rejected": 0.00171661376953125, "step": 20 }, { "epoch": 0.031413612565445025, "grad_norm": 1.0317246362077097, "learning_rate": 5.226480836236934e-06, "logits/chosen": -10.8125, "logits/rejected": -11.0, "logps/chosen": -458.0, "logps/rejected": -424.0, "loss": 0.6806, "rewards/accuracies": 0.6625000238418579, "rewards/chosen": 0.009521484375, "rewards/margins": 0.022705078125, "rewards/rejected": -0.01318359375, "step": 30 }, { "epoch": 0.041884816753926704, "grad_norm": 1.1689300517618306, "learning_rate": 6.968641114982578e-06, "logits/chosen": -10.6875, "logits/rejected": -10.875, "logps/chosen": -472.0, "logps/rejected": -446.0, "loss": 0.6546, "rewards/accuracies": 0.800000011920929, "rewards/chosen": -0.03515625, "rewards/margins": 0.0830078125, "rewards/rejected": -0.1181640625, "step": 40 }, { "epoch": 0.05235602094240838, "grad_norm": 1.7331389557317003, "learning_rate": 8.710801393728225e-06, "logits/chosen": -10.1875, "logits/rejected": -10.3125, "logps/chosen": -524.0, "logps/rejected": -500.0, "loss": 0.5881, "rewards/accuracies": 0.8374999761581421, "rewards/chosen": -0.5, "rewards/margins": 0.236328125, "rewards/rejected": -0.73828125, "step": 50 }, { "epoch": 0.06282722513089005, "grad_norm": 3.056138756064425, "learning_rate": 1.0452961672473868e-05, "logits/chosen": -9.0, "logits/rejected": -9.25, "logps/chosen": -716.0, "logps/rejected": -732.0, "loss": 0.462, "rewards/accuracies": 0.831250011920929, "rewards/chosen": -3.078125, "rewards/margins": 0.71875, "rewards/rejected": -3.796875, "step": 60 }, { "epoch": 0.07329842931937172, "grad_norm": 6.486331824464859, "learning_rate": 1.2195121951219513e-05, "logits/chosen": -8.8125, "logits/rejected": -9.0, "logps/chosen": -968.0, "logps/rejected": -1104.0, "loss": 0.3627, "rewards/accuracies": 0.8812500238418579, "rewards/chosen": -4.59375, "rewards/margins": 1.9921875, "rewards/rejected": -6.59375, "step": 70 }, { "epoch": 0.08376963350785341, "grad_norm": 1.7532818812068962, "learning_rate": 1.3937282229965156e-05, "logits/chosen": -7.1875, "logits/rejected": -7.21875, "logps/chosen": -1088.0, "logps/rejected": -1264.0, "loss": 0.261, "rewards/accuracies": 0.90625, "rewards/chosen": -6.15625, "rewards/margins": 2.546875, "rewards/rejected": -8.6875, "step": 80 }, { "epoch": 0.09424083769633508, "grad_norm": 3.7777465097739644, "learning_rate": 1.56794425087108e-05, "logits/chosen": -7.78125, "logits/rejected": -7.65625, "logps/chosen": -1304.0, "logps/rejected": -1784.0, "loss": 0.2354, "rewards/accuracies": 0.8812500238418579, "rewards/chosen": -8.9375, "rewards/margins": 5.375, "rewards/rejected": -14.3125, "step": 90 }, { "epoch": 0.10471204188481675, "grad_norm": 13.660286084226847, "learning_rate": 1.742160278745645e-05, "logits/chosen": -5.96875, "logits/rejected": -5.46875, "logps/chosen": -1728.0, "logps/rejected": -2640.0, "loss": 0.2459, "rewards/accuracies": 0.8999999761581421, "rewards/chosen": -12.25, "rewards/margins": 9.3125, "rewards/rejected": -21.625, "step": 100 }, { "epoch": 0.10471204188481675, "eval_logits/chosen": -5.96875, "eval_logits/rejected": -5.5625, "eval_logps/chosen": -1528.0, "eval_logps/rejected": -1344.0, "eval_loss": 3.8620312213897705, "eval_rewards/accuracies": 0.453125, "eval_rewards/chosen": -12.0625, "eval_rewards/margins": -1.5390625, "eval_rewards/rejected": -10.5, "eval_runtime": 48.4569, "eval_samples_per_second": 41.274, "eval_steps_per_second": 0.66, "step": 100 }, { "epoch": 0.11518324607329843, "grad_norm": 1.1692656731568394, "learning_rate": 1.9163763066202093e-05, "logits/chosen": -6.4375, "logits/rejected": -6.03125, "logps/chosen": -1384.0, "logps/rejected": -1920.0, "loss": 0.1962, "rewards/accuracies": 0.925000011920929, "rewards/chosen": -9.125, "rewards/margins": 5.8125, "rewards/rejected": -14.9375, "step": 110 }, { "epoch": 0.1256544502617801, "grad_norm": 1.4373017570995457, "learning_rate": 2.0905923344947736e-05, "logits/chosen": -5.25, "logits/rejected": -4.59375, "logps/chosen": -1464.0, "logps/rejected": -2176.0, "loss": 0.1554, "rewards/accuracies": 0.956250011920929, "rewards/chosen": -10.25, "rewards/margins": 7.65625, "rewards/rejected": -17.875, "step": 120 }, { "epoch": 0.13612565445026178, "grad_norm": 2.050292449823905, "learning_rate": 2.264808362369338e-05, "logits/chosen": -4.09375, "logits/rejected": -3.265625, "logps/chosen": -1384.0, "logps/rejected": -2256.0, "loss": 0.146, "rewards/accuracies": 0.9624999761581421, "rewards/chosen": -9.3125, "rewards/margins": 9.0625, "rewards/rejected": -18.375, "step": 130 }, { "epoch": 0.14659685863874344, "grad_norm": 1.4595989283853212, "learning_rate": 2.4390243902439026e-05, "logits/chosen": -4.375, "logits/rejected": -3.421875, "logps/chosen": -1784.0, "logps/rejected": -2576.0, "loss": 0.1543, "rewards/accuracies": 0.96875, "rewards/chosen": -12.5625, "rewards/margins": 8.6875, "rewards/rejected": -21.25, "step": 140 }, { "epoch": 0.15706806282722513, "grad_norm": 3.216390453334537, "learning_rate": 2.6132404181184672e-05, "logits/chosen": -5.5, "logits/rejected": -4.8125, "logps/chosen": -1800.0, "logps/rejected": -2544.0, "loss": 0.1739, "rewards/accuracies": 0.9375, "rewards/chosen": -13.1875, "rewards/margins": 8.0625, "rewards/rejected": -21.25, "step": 150 }, { "epoch": 0.16753926701570682, "grad_norm": 6.056034568064105, "learning_rate": 2.7874564459930312e-05, "logits/chosen": -5.625, "logits/rejected": -5.0625, "logps/chosen": -1520.0, "logps/rejected": -2320.0, "loss": 0.1543, "rewards/accuracies": 0.949999988079071, "rewards/chosen": -10.625, "rewards/margins": 8.625, "rewards/rejected": -19.25, "step": 160 }, { "epoch": 0.17801047120418848, "grad_norm": 1.8298757840769393, "learning_rate": 2.9616724738675962e-05, "logits/chosen": -5.375, "logits/rejected": -4.34375, "logps/chosen": -1496.0, "logps/rejected": -2496.0, "loss": 0.1397, "rewards/accuracies": 0.956250011920929, "rewards/chosen": -9.8125, "rewards/margins": 10.625, "rewards/rejected": -20.5, "step": 170 }, { "epoch": 0.18848167539267016, "grad_norm": 0.9937646525369369, "learning_rate": 3.13588850174216e-05, "logits/chosen": -3.609375, "logits/rejected": -2.5625, "logps/chosen": -1344.0, "logps/rejected": -2048.0, "loss": 0.1368, "rewards/accuracies": 0.96875, "rewards/chosen": -8.5625, "rewards/margins": 7.59375, "rewards/rejected": -16.125, "step": 180 }, { "epoch": 0.19895287958115182, "grad_norm": 1.840092244661011, "learning_rate": 3.310104529616725e-05, "logits/chosen": -6.03125, "logits/rejected": -5.0625, "logps/chosen": -1672.0, "logps/rejected": -2560.0, "loss": 0.1793, "rewards/accuracies": 0.956250011920929, "rewards/chosen": -11.75, "rewards/margins": 9.8125, "rewards/rejected": -21.5, "step": 190 }, { "epoch": 0.2094240837696335, "grad_norm": 4.50714817512215, "learning_rate": 3.48432055749129e-05, "logits/chosen": -3.40625, "logits/rejected": -2.484375, "logps/chosen": -2256.0, "logps/rejected": -3456.0, "loss": 0.1787, "rewards/accuracies": 0.96875, "rewards/chosen": -17.875, "rewards/margins": 12.125, "rewards/rejected": -30.0, "step": 200 }, { "epoch": 0.2094240837696335, "eval_logits/chosen": -1.8828125, "eval_logits/rejected": -1.0546875, "eval_logps/chosen": -1624.0, "eval_logps/rejected": -1408.0, "eval_loss": 4.223593711853027, "eval_rewards/accuracies": 0.443359375, "eval_rewards/chosen": -13.0625, "eval_rewards/margins": -1.9140625, "eval_rewards/rejected": -11.1875, "eval_runtime": 46.7116, "eval_samples_per_second": 42.816, "eval_steps_per_second": 0.685, "step": 200 }, { "epoch": 0.2198952879581152, "grad_norm": 4.250925824647202, "learning_rate": 3.6585365853658535e-05, "logits/chosen": -3.484375, "logits/rejected": -2.3125, "logps/chosen": -1432.0, "logps/rejected": -2240.0, "loss": 0.1471, "rewards/accuracies": 0.956250011920929, "rewards/chosen": -10.3125, "rewards/margins": 8.375, "rewards/rejected": -18.625, "step": 210 }, { "epoch": 0.23036649214659685, "grad_norm": 1.6860619113388613, "learning_rate": 3.8327526132404185e-05, "logits/chosen": -5.125, "logits/rejected": -4.28125, "logps/chosen": -1320.0, "logps/rejected": -2024.0, "loss": 0.1281, "rewards/accuracies": 0.9624999761581421, "rewards/chosen": -8.6875, "rewards/margins": 7.34375, "rewards/rejected": -16.0, "step": 220 }, { "epoch": 0.24083769633507854, "grad_norm": 0.8928819803501318, "learning_rate": 4.006968641114983e-05, "logits/chosen": -5.25, "logits/rejected": -4.0625, "logps/chosen": -1792.0, "logps/rejected": -2704.0, "loss": 0.1358, "rewards/accuracies": 0.9437500238418579, "rewards/chosen": -13.5, "rewards/margins": 9.5625, "rewards/rejected": -23.125, "step": 230 }, { "epoch": 0.2513089005235602, "grad_norm": 1.693104590992371, "learning_rate": 4.181184668989547e-05, "logits/chosen": -3.1875, "logits/rejected": -2.203125, "logps/chosen": -1712.0, "logps/rejected": -2624.0, "loss": 0.1677, "rewards/accuracies": 0.956250011920929, "rewards/chosen": -12.75, "rewards/margins": 9.625, "rewards/rejected": -22.375, "step": 240 }, { "epoch": 0.2617801047120419, "grad_norm": 2.6808184998193045, "learning_rate": 4.3554006968641115e-05, "logits/chosen": -3.671875, "logits/rejected": -2.515625, "logps/chosen": -2096.0, "logps/rejected": -3120.0, "loss": 0.1074, "rewards/accuracies": 0.96875, "rewards/chosen": -15.8125, "rewards/margins": 10.8125, "rewards/rejected": -26.625, "step": 250 }, { "epoch": 0.27225130890052357, "grad_norm": 3.169846505508325, "learning_rate": 4.529616724738676e-05, "logits/chosen": -3.390625, "logits/rejected": -1.59375, "logps/chosen": -2128.0, "logps/rejected": -3408.0, "loss": 0.1035, "rewards/accuracies": 0.9125000238418579, "rewards/chosen": -17.0, "rewards/margins": 13.25, "rewards/rejected": -30.375, "step": 260 }, { "epoch": 0.28272251308900526, "grad_norm": 1.1046683408082556, "learning_rate": 4.703832752613241e-05, "logits/chosen": -3.171875, "logits/rejected": -1.3515625, "logps/chosen": -2144.0, "logps/rejected": -3376.0, "loss": 0.1898, "rewards/accuracies": 0.949999988079071, "rewards/chosen": -17.375, "rewards/margins": 12.5625, "rewards/rejected": -29.875, "step": 270 }, { "epoch": 0.2931937172774869, "grad_norm": 1.9127547814933203, "learning_rate": 4.878048780487805e-05, "logits/chosen": -2.328125, "logits/rejected": -0.95703125, "logps/chosen": -1432.0, "logps/rejected": -2128.0, "loss": 0.1215, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -10.0625, "rewards/margins": 7.5625, "rewards/rejected": -17.625, "step": 280 }, { "epoch": 0.3036649214659686, "grad_norm": 1.554329529332899, "learning_rate": 4.999983293481418e-05, "logits/chosen": -2.78125, "logits/rejected": -0.87109375, "logps/chosen": -2528.0, "logps/rejected": -4048.0, "loss": 0.103, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -20.625, "rewards/margins": 15.9375, "rewards/rejected": -36.5, "step": 290 }, { "epoch": 0.31413612565445026, "grad_norm": 2.9794411700812495, "learning_rate": 4.9996862949181595e-05, "logits/chosen": -2.359375, "logits/rejected": 0.2099609375, "logps/chosen": -1840.0, "logps/rejected": -2944.0, "loss": 0.1064, "rewards/accuracies": 0.956250011920929, "rewards/chosen": -14.125, "rewards/margins": 11.25, "rewards/rejected": -25.375, "step": 300 }, { "epoch": 0.31413612565445026, "eval_logits/chosen": 1.828125, "eval_logits/rejected": 2.640625, "eval_logps/chosen": -2256.0, "eval_logps/rejected": -1984.0, "eval_loss": 5.558437347412109, "eval_rewards/accuracies": 0.43359375, "eval_rewards/chosen": -19.5, "eval_rewards/margins": -2.515625, "eval_rewards/rejected": -16.875, "eval_runtime": 46.6947, "eval_samples_per_second": 42.831, "eval_steps_per_second": 0.685, "step": 300 }, { "epoch": 0.32460732984293195, "grad_norm": 3.8794407926661045, "learning_rate": 4.9990180911526896e-05, "logits/chosen": -1.5625, "logits/rejected": 1.125, "logps/chosen": -1400.0, "logps/rejected": -2352.0, "loss": 0.1257, "rewards/accuracies": 0.96875, "rewards/chosen": -9.6875, "rewards/margins": 10.0625, "rewards/rejected": -19.75, "step": 310 }, { "epoch": 0.33507853403141363, "grad_norm": 0.7300720242228556, "learning_rate": 4.997978781413745e-05, "logits/chosen": 0.3125, "logits/rejected": 2.09375, "logps/chosen": -2096.0, "logps/rejected": -3360.0, "loss": 0.1282, "rewards/accuracies": 0.956250011920929, "rewards/chosen": -16.125, "rewards/margins": 13.0625, "rewards/rejected": -29.25, "step": 320 }, { "epoch": 0.34554973821989526, "grad_norm": 1.221416701526911, "learning_rate": 4.996568520039565e-05, "logits/chosen": -1.9140625, "logits/rejected": -0.322265625, "logps/chosen": -1696.0, "logps/rejected": -2704.0, "loss": 0.1256, "rewards/accuracies": 0.949999988079071, "rewards/chosen": -12.3125, "rewards/margins": 10.5, "rewards/rejected": -22.875, "step": 330 }, { "epoch": 0.35602094240837695, "grad_norm": 0.6292920124847436, "learning_rate": 4.994787516454973e-05, "logits/chosen": -2.984375, "logits/rejected": -0.96484375, "logps/chosen": -1856.0, "logps/rejected": -3040.0, "loss": 0.1016, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -13.8125, "rewards/margins": 12.5, "rewards/rejected": -26.375, "step": 340 }, { "epoch": 0.36649214659685864, "grad_norm": 2.0347194193473297, "learning_rate": 4.992636035140273e-05, "logits/chosen": -4.28125, "logits/rejected": -2.03125, "logps/chosen": -1456.0, "logps/rejected": -2688.0, "loss": 0.0804, "rewards/accuracies": 0.96875, "rewards/chosen": -10.125, "rewards/margins": 12.875, "rewards/rejected": -23.0, "step": 350 }, { "epoch": 0.3769633507853403, "grad_norm": 1.0823333369998072, "learning_rate": 4.99011439559198e-05, "logits/chosen": -6.46875, "logits/rejected": -3.3125, "logps/chosen": -1728.0, "logps/rejected": -3184.0, "loss": 0.1143, "rewards/accuracies": 0.9624999761581421, "rewards/chosen": -12.625, "rewards/margins": 15.125, "rewards/rejected": -27.75, "step": 360 }, { "epoch": 0.387434554973822, "grad_norm": 0.8905037931783505, "learning_rate": 4.987222972275369e-05, "logits/chosen": -3.65625, "logits/rejected": -0.380859375, "logps/chosen": -1904.0, "logps/rejected": -2992.0, "loss": 0.1549, "rewards/accuracies": 0.956250011920929, "rewards/chosen": -14.375, "rewards/margins": 11.375, "rewards/rejected": -25.75, "step": 370 }, { "epoch": 0.39790575916230364, "grad_norm": 0.5328758687523436, "learning_rate": 4.983962194568872e-05, "logits/chosen": -6.375, "logits/rejected": -3.625, "logps/chosen": -1512.0, "logps/rejected": -2288.0, "loss": 0.1113, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -10.375, "rewards/margins": 8.25, "rewards/rejected": -18.625, "step": 380 }, { "epoch": 0.4083769633507853, "grad_norm": 1.100275679715039, "learning_rate": 4.9803325467003084e-05, "logits/chosen": -6.3125, "logits/rejected": -2.96875, "logps/chosen": -2528.0, "logps/rejected": -3984.0, "loss": 0.1279, "rewards/accuracies": 0.9375, "rewards/chosen": -20.375, "rewards/margins": 15.25, "rewards/rejected": -35.5, "step": 390 }, { "epoch": 0.418848167539267, "grad_norm": 1.1753830566327925, "learning_rate": 4.9763345676749816e-05, "logits/chosen": -8.9375, "logits/rejected": -6.0, "logps/chosen": -1776.0, "logps/rejected": -2928.0, "loss": 0.1114, "rewards/accuracies": 0.956250011920929, "rewards/chosen": -13.0625, "rewards/margins": 12.125, "rewards/rejected": -25.25, "step": 400 }, { "epoch": 0.418848167539267, "eval_logits/chosen": -3.296875, "eval_logits/rejected": -2.390625, "eval_logps/chosen": -2480.0, "eval_logps/rejected": -2240.0, "eval_loss": 5.962562561035156, "eval_rewards/accuracies": 0.447265625, "eval_rewards/chosen": -21.625, "eval_rewards/margins": -2.109375, "eval_rewards/rejected": -19.5, "eval_runtime": 46.679, "eval_samples_per_second": 42.846, "eval_steps_per_second": 0.686, "step": 400 }, { "epoch": 0.4293193717277487, "grad_norm": 6.232625206460211, "learning_rate": 4.9719688511956374e-05, "logits/chosen": -5.8125, "logits/rejected": -2.875, "logps/chosen": -2448.0, "logps/rejected": -4192.0, "loss": 0.146, "rewards/accuracies": 0.949999988079071, "rewards/chosen": -19.75, "rewards/margins": 17.75, "rewards/rejected": -37.5, "step": 410 }, { "epoch": 0.4397905759162304, "grad_norm": 1.0285136093587235, "learning_rate": 4.967236045574294e-05, "logits/chosen": -8.625, "logits/rejected": -6.125, "logps/chosen": -1408.0, "logps/rejected": -2544.0, "loss": 0.1382, "rewards/accuracies": 0.949999988079071, "rewards/chosen": -9.4375, "rewards/margins": 11.9375, "rewards/rejected": -21.375, "step": 420 }, { "epoch": 0.450261780104712, "grad_norm": 0.36705904069465384, "learning_rate": 4.96213685363597e-05, "logits/chosen": -6.375, "logits/rejected": -3.703125, "logps/chosen": -1496.0, "logps/rejected": -2320.0, "loss": 0.0874, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -10.3125, "rewards/margins": 8.8125, "rewards/rejected": -19.125, "step": 430 }, { "epoch": 0.4607329842931937, "grad_norm": 0.8508885519731393, "learning_rate": 4.956672032614315e-05, "logits/chosen": -5.125, "logits/rejected": -2.359375, "logps/chosen": -1808.0, "logps/rejected": -2800.0, "loss": 0.0894, "rewards/accuracies": 0.956250011920929, "rewards/chosen": -13.3125, "rewards/margins": 10.5, "rewards/rejected": -23.75, "step": 440 }, { "epoch": 0.4712041884816754, "grad_norm": 1.2053403643652096, "learning_rate": 4.950842394039157e-05, "logits/chosen": -4.84375, "logits/rejected": -1.9375, "logps/chosen": -2144.0, "logps/rejected": -3344.0, "loss": 0.0934, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -17.125, "rewards/margins": 12.4375, "rewards/rejected": -29.5, "step": 450 }, { "epoch": 0.4816753926701571, "grad_norm": 0.8916688345956255, "learning_rate": 4.944648803615994e-05, "logits/chosen": -3.15625, "logits/rejected": -0.12158203125, "logps/chosen": -1776.0, "logps/rejected": -2944.0, "loss": 0.1127, "rewards/accuracies": 0.949999988079071, "rewards/chosen": -13.125, "rewards/margins": 12.1875, "rewards/rejected": -25.375, "step": 460 }, { "epoch": 0.49214659685863876, "grad_norm": 0.6268103117145335, "learning_rate": 4.938092181097433e-05, "logits/chosen": -3.125, "logits/rejected": -1.3515625, "logps/chosen": -1760.0, "logps/rejected": -2928.0, "loss": 0.0896, "rewards/accuracies": 0.956250011920929, "rewards/chosen": -13.0, "rewards/margins": 12.1875, "rewards/rejected": -25.25, "step": 470 }, { "epoch": 0.5026178010471204, "grad_norm": 6.048886392395671, "learning_rate": 4.931173500146607e-05, "logits/chosen": -2.265625, "logits/rejected": 0.361328125, "logps/chosen": -2352.0, "logps/rejected": -4096.0, "loss": 0.0971, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -19.125, "rewards/margins": 18.125, "rewards/rejected": -37.25, "step": 480 }, { "epoch": 0.5130890052356021, "grad_norm": 1.2810235152580198, "learning_rate": 4.923893788192582e-05, "logits/chosen": -4.34375, "logits/rejected": -1.5078125, "logps/chosen": -2752.0, "logps/rejected": -4352.0, "loss": 0.1132, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -22.625, "rewards/margins": 16.5, "rewards/rejected": -39.25, "step": 490 }, { "epoch": 0.5235602094240838, "grad_norm": 0.6059260552557658, "learning_rate": 4.9162541262777906e-05, "logits/chosen": -5.0, "logits/rejected": -2.015625, "logps/chosen": -1960.0, "logps/rejected": -3088.0, "loss": 0.0803, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -14.8125, "rewards/margins": 11.8125, "rewards/rejected": -26.625, "step": 500 }, { "epoch": 0.5235602094240838, "eval_logits/chosen": 2.484375, "eval_logits/rejected": 3.65625, "eval_logps/chosen": -2800.0, "eval_logps/rejected": -2624.0, "eval_loss": 6.103968620300293, "eval_rewards/accuracies": 0.4921875, "eval_rewards/chosen": -24.75, "eval_rewards/margins": -1.4140625, "eval_rewards/rejected": -23.375, "eval_runtime": 46.6698, "eval_samples_per_second": 42.854, "eval_steps_per_second": 0.686, "step": 500 }, { "epoch": 0.5340314136125655, "grad_norm": 0.6598365951748475, "learning_rate": 4.90825564889749e-05, "logits/chosen": -3.40625, "logits/rejected": 0.95703125, "logps/chosen": -1888.0, "logps/rejected": -3200.0, "loss": 0.1102, "rewards/accuracies": 0.949999988079071, "rewards/chosen": -14.1875, "rewards/margins": 13.4375, "rewards/rejected": -27.625, "step": 510 }, { "epoch": 0.5445026178010471, "grad_norm": 1.8295330296014316, "learning_rate": 4.899899543831288e-05, "logits/chosen": -2.34375, "logits/rejected": 0.87890625, "logps/chosen": -2176.0, "logps/rejected": -3280.0, "loss": 0.0967, "rewards/accuracies": 0.925000011920929, "rewards/chosen": -16.75, "rewards/margins": 11.6875, "rewards/rejected": -28.375, "step": 520 }, { "epoch": 0.5549738219895288, "grad_norm": 0.8730362144631139, "learning_rate": 4.891187051966765e-05, "logits/chosen": -3.671875, "logits/rejected": -0.87890625, "logps/chosen": -1480.0, "logps/rejected": -2432.0, "loss": 0.086, "rewards/accuracies": 0.949999988079071, "rewards/chosen": -10.375, "rewards/margins": 9.6875, "rewards/rejected": -20.0, "step": 530 }, { "epoch": 0.5654450261780105, "grad_norm": 1.2038937188803909, "learning_rate": 4.882119467115191e-05, "logits/chosen": -4.21875, "logits/rejected": -0.98046875, "logps/chosen": -1744.0, "logps/rejected": -2688.0, "loss": 0.0813, "rewards/accuracies": 0.956250011920929, "rewards/chosen": -12.75, "rewards/margins": 10.0, "rewards/rejected": -22.75, "step": 540 }, { "epoch": 0.5759162303664922, "grad_norm": 2.4431030688271904, "learning_rate": 4.872698135819402e-05, "logits/chosen": -2.34375, "logits/rejected": 0.240234375, "logps/chosen": -2448.0, "logps/rejected": -3872.0, "loss": 0.0883, "rewards/accuracies": 0.9375, "rewards/chosen": -19.875, "rewards/margins": 14.8125, "rewards/rejected": -34.75, "step": 550 }, { "epoch": 0.5863874345549738, "grad_norm": 1.733904404687087, "learning_rate": 4.862924457153831e-05, "logits/chosen": -3.265625, "logits/rejected": 0.470703125, "logps/chosen": -2048.0, "logps/rejected": -3344.0, "loss": 0.0848, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -16.5, "rewards/margins": 13.5, "rewards/rejected": -30.0, "step": 560 }, { "epoch": 0.5968586387434555, "grad_norm": 2.3556103965686095, "learning_rate": 4.8527998825167484e-05, "logits/chosen": -3.3125, "logits/rejected": 0.404296875, "logps/chosen": -1920.0, "logps/rejected": -3072.0, "loss": 0.0826, "rewards/accuracies": 0.9624999761581421, "rewards/chosen": -15.0, "rewards/margins": 12.0, "rewards/rejected": -27.0, "step": 570 }, { "epoch": 0.6073298429319371, "grad_norm": 0.5472877512246217, "learning_rate": 4.842325915414728e-05, "logits/chosen": -1.7734375, "logits/rejected": 0.6484375, "logps/chosen": -2040.0, "logps/rejected": -3200.0, "loss": 0.0852, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -16.0, "rewards/margins": 12.125, "rewards/rejected": -28.25, "step": 580 }, { "epoch": 0.6178010471204188, "grad_norm": 4.384971462464364, "learning_rate": 4.831504111239373e-05, "logits/chosen": -3.578125, "logits/rejected": -0.3671875, "logps/chosen": -2224.0, "logps/rejected": -3552.0, "loss": 0.0776, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -17.375, "rewards/margins": 13.75, "rewards/rejected": -31.125, "step": 590 }, { "epoch": 0.6282722513089005, "grad_norm": 1.4143813556457387, "learning_rate": 4.8203360770363425e-05, "logits/chosen": -3.359375, "logits/rejected": 0.69921875, "logps/chosen": -2352.0, "logps/rejected": -3984.0, "loss": 0.0999, "rewards/accuracies": 0.949999988079071, "rewards/chosen": -18.875, "rewards/margins": 16.75, "rewards/rejected": -35.75, "step": 600 }, { "epoch": 0.6282722513089005, "eval_logits/chosen": 1.296875, "eval_logits/rejected": 2.671875, "eval_logps/chosen": -2576.0, "eval_logps/rejected": -2336.0, "eval_loss": 5.522375106811523, "eval_rewards/accuracies": 0.439453125, "eval_rewards/chosen": -22.5, "eval_rewards/margins": -2.125, "eval_rewards/rejected": -20.375, "eval_runtime": 46.691, "eval_samples_per_second": 42.835, "eval_steps_per_second": 0.685, "step": 600 }, { "epoch": 0.6387434554973822, "grad_norm": 1.7478821045832178, "learning_rate": 4.808823471266702e-05, "logits/chosen": -3.78125, "logits/rejected": -0.396484375, "logps/chosen": -1760.0, "logps/rejected": -2768.0, "loss": 0.1195, "rewards/accuracies": 0.956250011920929, "rewards/chosen": -13.1875, "rewards/margins": 10.4375, "rewards/rejected": -23.625, "step": 610 }, { "epoch": 0.6492146596858639, "grad_norm": 0.9870986139282529, "learning_rate": 4.796968003560639e-05, "logits/chosen": -4.75, "logits/rejected": -2.28125, "logps/chosen": -1888.0, "logps/rejected": -2832.0, "loss": 0.1063, "rewards/accuracies": 0.96875, "rewards/chosen": -14.1875, "rewards/margins": 10.0625, "rewards/rejected": -24.25, "step": 620 }, { "epoch": 0.6596858638743456, "grad_norm": 2.2394524896853003, "learning_rate": 4.7847714344635864e-05, "logits/chosen": -3.5, "logits/rejected": -1.25, "logps/chosen": -1832.0, "logps/rejected": -2704.0, "loss": 0.1051, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -14.125, "rewards/margins": 8.9375, "rewards/rejected": -23.0, "step": 630 }, { "epoch": 0.6701570680628273, "grad_norm": 2.118057301959731, "learning_rate": 4.772235575174776e-05, "logits/chosen": -4.65625, "logits/rejected": -1.984375, "logps/chosen": -2240.0, "logps/rejected": -3232.0, "loss": 0.1032, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -17.375, "rewards/margins": 10.5, "rewards/rejected": -27.875, "step": 640 }, { "epoch": 0.680628272251309, "grad_norm": 4.193375939487143, "learning_rate": 4.7593622872782754e-05, "logits/chosen": -6.15625, "logits/rejected": -2.140625, "logps/chosen": -2192.0, "logps/rejected": -3504.0, "loss": 0.0858, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -17.625, "rewards/margins": 13.5, "rewards/rejected": -31.125, "step": 650 }, { "epoch": 0.6910994764397905, "grad_norm": 0.8596137623025543, "learning_rate": 4.746153482466542e-05, "logits/chosen": -6.40625, "logits/rejected": -2.703125, "logps/chosen": -2016.0, "logps/rejected": -3264.0, "loss": 0.1291, "rewards/accuracies": 0.9312499761581421, "rewards/chosen": -15.625, "rewards/margins": 13.125, "rewards/rejected": -28.75, "step": 660 }, { "epoch": 0.7015706806282722, "grad_norm": 1.729863367652938, "learning_rate": 4.732611122256537e-05, "logits/chosen": -6.34375, "logits/rejected": -3.5, "logps/chosen": -2128.0, "logps/rejected": -3136.0, "loss": 0.0725, "rewards/accuracies": 0.9624999761581421, "rewards/chosen": -16.5, "rewards/margins": 10.4375, "rewards/rejected": -26.875, "step": 670 }, { "epoch": 0.7120418848167539, "grad_norm": 4.509604278551911, "learning_rate": 4.718737217698435e-05, "logits/chosen": -5.03125, "logits/rejected": -1.640625, "logps/chosen": -2096.0, "logps/rejected": -3168.0, "loss": 0.0833, "rewards/accuracies": 0.956250011920929, "rewards/chosen": -17.0, "rewards/margins": 10.875, "rewards/rejected": -27.875, "step": 680 }, { "epoch": 0.7225130890052356, "grad_norm": 1.0122920441912262, "learning_rate": 4.704533829076982e-05, "logits/chosen": -5.09375, "logits/rejected": -0.80859375, "logps/chosen": -2560.0, "logps/rejected": -4448.0, "loss": 0.0654, "rewards/accuracies": 0.96875, "rewards/chosen": -21.125, "rewards/margins": 19.5, "rewards/rejected": -40.5, "step": 690 }, { "epoch": 0.7329842931937173, "grad_norm": 0.8084831048634067, "learning_rate": 4.6900030656055455e-05, "logits/chosen": -7.1875, "logits/rejected": -2.921875, "logps/chosen": -1904.0, "logps/rejected": -3200.0, "loss": 0.0767, "rewards/accuracies": 0.9437500238418579, "rewards/chosen": -14.5, "rewards/margins": 13.4375, "rewards/rejected": -28.0, "step": 700 }, { "epoch": 0.7329842931937173, "eval_logits/chosen": -0.44140625, "eval_logits/rejected": 0.578125, "eval_logps/chosen": -2736.0, "eval_logps/rejected": -2544.0, "eval_loss": 5.996812343597412, "eval_rewards/accuracies": 0.46484375, "eval_rewards/chosen": -24.25, "eval_rewards/margins": -1.6953125, "eval_rewards/rejected": -22.5, "eval_runtime": 46.6936, "eval_samples_per_second": 42.832, "eval_steps_per_second": 0.685, "step": 700 }, { "epoch": 0.743455497382199, "grad_norm": 2.060217484330229, "learning_rate": 4.6751470851128906e-05, "logits/chosen": -3.984375, "logits/rejected": -0.640625, "logps/chosen": -2208.0, "logps/rejected": -3552.0, "loss": 0.0845, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -17.125, "rewards/margins": 14.0, "rewards/rejected": -31.125, "step": 710 }, { "epoch": 0.7539267015706806, "grad_norm": 1.8103653722673227, "learning_rate": 4.6599680937227454e-05, "logits/chosen": -4.03125, "logits/rejected": -0.072265625, "logps/chosen": -2224.0, "logps/rejected": -3552.0, "loss": 0.0972, "rewards/accuracies": 0.96875, "rewards/chosen": -17.25, "rewards/margins": 13.8125, "rewards/rejected": -31.125, "step": 720 }, { "epoch": 0.7643979057591623, "grad_norm": 1.5149912406723094, "learning_rate": 4.644468345526183e-05, "logits/chosen": -5.53125, "logits/rejected": -1.0546875, "logps/chosen": -1792.0, "logps/rejected": -2864.0, "loss": 0.0816, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -13.125, "rewards/margins": 11.3125, "rewards/rejected": -24.5, "step": 730 }, { "epoch": 0.774869109947644, "grad_norm": 1.2174068919585266, "learning_rate": 4.628650142246894e-05, "logits/chosen": -8.375, "logits/rejected": -3.75, "logps/chosen": -1552.0, "logps/rejected": -2656.0, "loss": 0.0843, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -11.0625, "rewards/margins": 11.5625, "rewards/rejected": -22.625, "step": 740 }, { "epoch": 0.7853403141361257, "grad_norm": 1.1574179017830328, "learning_rate": 4.6125158328993765e-05, "logits/chosen": -7.0625, "logits/rejected": -3.015625, "logps/chosen": -1872.0, "logps/rejected": -3136.0, "loss": 0.0847, "rewards/accuracies": 0.949999988079071, "rewards/chosen": -14.1875, "rewards/margins": 13.25, "rewards/rejected": -27.375, "step": 750 }, { "epoch": 0.7958115183246073, "grad_norm": 1.8256106070467697, "learning_rate": 4.5960678134401064e-05, "logits/chosen": -8.4375, "logits/rejected": -4.71875, "logps/chosen": -1704.0, "logps/rejected": -2960.0, "loss": 0.0808, "rewards/accuracies": 0.9624999761581421, "rewards/chosen": -12.1875, "rewards/margins": 13.25, "rewards/rejected": -25.375, "step": 760 }, { "epoch": 0.806282722513089, "grad_norm": 5.723317159961658, "learning_rate": 4.579308526411734e-05, "logits/chosen": -7.96875, "logits/rejected": -4.40625, "logps/chosen": -1888.0, "logps/rejected": -3152.0, "loss": 0.109, "rewards/accuracies": 0.9624999761581421, "rewards/chosen": -14.0625, "rewards/margins": 13.5, "rewards/rejected": -27.5, "step": 770 }, { "epoch": 0.8167539267015707, "grad_norm": 1.1292029592474329, "learning_rate": 4.562240460580367e-05, "logits/chosen": -6.03125, "logits/rejected": -3.046875, "logps/chosen": -2080.0, "logps/rejected": -3520.0, "loss": 0.1293, "rewards/accuracies": 0.9624999761581421, "rewards/chosen": -16.25, "rewards/margins": 14.8125, "rewards/rejected": -31.0, "step": 780 }, { "epoch": 0.8272251308900523, "grad_norm": 2.1505177772567157, "learning_rate": 4.544866150565985e-05, "logits/chosen": -4.84375, "logits/rejected": -2.3125, "logps/chosen": -2016.0, "logps/rejected": -3040.0, "loss": 0.0768, "rewards/accuracies": 0.956250011920929, "rewards/chosen": -15.6875, "rewards/margins": 10.5625, "rewards/rejected": -26.25, "step": 790 }, { "epoch": 0.837696335078534, "grad_norm": 0.5958488194248702, "learning_rate": 4.527188176466051e-05, "logits/chosen": -5.125, "logits/rejected": -1.828125, "logps/chosen": -2128.0, "logps/rejected": -3232.0, "loss": 0.0891, "rewards/accuracies": 0.96875, "rewards/chosen": -16.5, "rewards/margins": 11.5, "rewards/rejected": -28.0, "step": 800 }, { "epoch": 0.837696335078534, "eval_logits/chosen": -1.375, "eval_logits/rejected": -0.365234375, "eval_logps/chosen": -2400.0, "eval_logps/rejected": -2208.0, "eval_loss": 4.992062568664551, "eval_rewards/accuracies": 0.45703125, "eval_rewards/chosen": -20.875, "eval_rewards/margins": -1.71875, "eval_rewards/rejected": -19.125, "eval_runtime": 46.6885, "eval_samples_per_second": 42.837, "eval_steps_per_second": 0.685, "step": 800 }, { "epoch": 0.8481675392670157, "grad_norm": 0.9607752643046815, "learning_rate": 4.509209163472357e-05, "logits/chosen": -5.5625, "logits/rejected": -2.125, "logps/chosen": -2032.0, "logps/rejected": -3168.0, "loss": 0.0711, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -15.5, "rewards/margins": 12.0, "rewards/rejected": -27.375, "step": 810 }, { "epoch": 0.8586387434554974, "grad_norm": 0.8124604937467872, "learning_rate": 4.4909317814811896e-05, "logits/chosen": -5.03125, "logits/rejected": -2.328125, "logps/chosen": -2080.0, "logps/rejected": -3424.0, "loss": 0.058, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -16.125, "rewards/margins": 13.8125, "rewards/rejected": -29.875, "step": 820 }, { "epoch": 0.8691099476439791, "grad_norm": 1.5779040941861349, "learning_rate": 4.472358744696845e-05, "logits/chosen": -5.5, "logits/rejected": -3.109375, "logps/chosen": -2432.0, "logps/rejected": -3888.0, "loss": 0.0807, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -19.5, "rewards/margins": 15.3125, "rewards/rejected": -35.0, "step": 830 }, { "epoch": 0.8795811518324608, "grad_norm": 1.5585580358827753, "learning_rate": 4.4534928112285645e-05, "logits/chosen": -7.0, "logits/rejected": -4.375, "logps/chosen": -2096.0, "logps/rejected": -3744.0, "loss": 0.0818, "rewards/accuracies": 0.9624999761581421, "rewards/chosen": -16.375, "rewards/margins": 16.875, "rewards/rejected": -33.25, "step": 840 }, { "epoch": 0.8900523560209425, "grad_norm": 0.3922108780267044, "learning_rate": 4.43433678268096e-05, "logits/chosen": -7.59375, "logits/rejected": -4.46875, "logps/chosen": -1952.0, "logps/rejected": -3456.0, "loss": 0.0527, "rewards/accuracies": 0.96875, "rewards/chosen": -15.0625, "rewards/margins": 15.5, "rewards/rejected": -30.625, "step": 850 }, { "epoch": 0.900523560209424, "grad_norm": 0.9590868208490266, "learning_rate": 4.414893503737968e-05, "logits/chosen": -5.0625, "logits/rejected": -2.640625, "logps/chosen": -2576.0, "logps/rejected": -4192.0, "loss": 0.0759, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -21.625, "rewards/margins": 16.5, "rewards/rejected": -38.25, "step": 860 }, { "epoch": 0.9109947643979057, "grad_norm": 0.964378765740042, "learning_rate": 4.395165861740414e-05, "logits/chosen": -6.71875, "logits/rejected": -4.3125, "logps/chosen": -2112.0, "logps/rejected": -3456.0, "loss": 0.0627, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -16.5, "rewards/margins": 14.125, "rewards/rejected": -30.625, "step": 870 }, { "epoch": 0.9214659685863874, "grad_norm": 1.7721301087427972, "learning_rate": 4.375156786257241e-05, "logits/chosen": -8.75, "logits/rejected": -6.78125, "logps/chosen": -1920.0, "logps/rejected": -3168.0, "loss": 0.0897, "rewards/accuracies": 0.9624999761581421, "rewards/chosen": -14.3125, "rewards/margins": 13.0625, "rewards/rejected": -27.375, "step": 880 }, { "epoch": 0.9319371727748691, "grad_norm": 0.23410090359235883, "learning_rate": 4.354869248650466e-05, "logits/chosen": -7.4375, "logits/rejected": -4.5, "logps/chosen": -2272.0, "logps/rejected": -3760.0, "loss": 0.0783, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -17.875, "rewards/margins": 15.375, "rewards/rejected": -33.25, "step": 890 }, { "epoch": 0.9424083769633508, "grad_norm": 3.5731454620127474, "learning_rate": 4.334306261633931e-05, "logits/chosen": -6.75, "logits/rejected": -4.0, "logps/chosen": -1800.0, "logps/rejected": -3008.0, "loss": 0.0907, "rewards/accuracies": 0.9624999761581421, "rewards/chosen": -12.9375, "rewards/margins": 12.8125, "rewards/rejected": -25.75, "step": 900 }, { "epoch": 0.9424083769633508, "eval_logits/chosen": -3.046875, "eval_logits/rejected": -2.078125, "eval_logps/chosen": -2040.0, "eval_logps/rejected": -1896.0, "eval_loss": 3.986875057220459, "eval_rewards/accuracies": 0.478515625, "eval_rewards/chosen": -17.25, "eval_rewards/margins": -1.1328125, "eval_rewards/rejected": -16.125, "eval_runtime": 46.6641, "eval_samples_per_second": 42.86, "eval_steps_per_second": 0.686, "step": 900 }, { "epoch": 0.9528795811518325, "grad_norm": 0.6076535472092232, "learning_rate": 4.313470878825911e-05, "logits/chosen": -5.375, "logits/rejected": -2.609375, "logps/chosen": -1720.0, "logps/rejected": -2624.0, "loss": 0.0699, "rewards/accuracies": 0.956250011920929, "rewards/chosen": -12.5625, "rewards/margins": 9.6875, "rewards/rejected": -22.25, "step": 910 }, { "epoch": 0.9633507853403142, "grad_norm": 0.27748934559957394, "learning_rate": 4.292366194295652e-05, "logits/chosen": -4.5, "logits/rejected": -0.8203125, "logps/chosen": -1960.0, "logps/rejected": -3376.0, "loss": 0.121, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -14.9375, "rewards/margins": 14.6875, "rewards/rejected": -29.625, "step": 920 }, { "epoch": 0.9738219895287958, "grad_norm": 3.783087311906835, "learning_rate": 4.2709953421038964e-05, "logits/chosen": -5.6875, "logits/rejected": -1.9140625, "logps/chosen": -2024.0, "logps/rejected": -3376.0, "loss": 0.0945, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -15.75, "rewards/margins": 14.1875, "rewards/rejected": -29.875, "step": 930 }, { "epoch": 0.9842931937172775, "grad_norm": 1.259847464257079, "learning_rate": 4.249361495837478e-05, "logits/chosen": -5.625, "logits/rejected": -2.171875, "logps/chosen": -1784.0, "logps/rejected": -2912.0, "loss": 0.0567, "rewards/accuracies": 0.96875, "rewards/chosen": -13.4375, "rewards/margins": 11.9375, "rewards/rejected": -25.375, "step": 940 }, { "epoch": 0.9947643979057592, "grad_norm": 0.8257586320126123, "learning_rate": 4.227467868138035e-05, "logits/chosen": -3.328125, "logits/rejected": 0.125, "logps/chosen": -2592.0, "logps/rejected": -3920.0, "loss": 0.0682, "rewards/accuracies": 0.96875, "rewards/chosen": -21.0, "rewards/margins": 13.9375, "rewards/rejected": -35.0, "step": 950 }, { "epoch": 1.0052356020942408, "grad_norm": 0.9903787507809574, "learning_rate": 4.205317710224938e-05, "logits/chosen": -6.0625, "logits/rejected": -1.9453125, "logps/chosen": -2096.0, "logps/rejected": -3632.0, "loss": 0.0492, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -16.25, "rewards/margins": 15.875, "rewards/rejected": -32.0, "step": 960 }, { "epoch": 1.0157068062827226, "grad_norm": 0.9999519847351056, "learning_rate": 4.1829143114124733e-05, "logits/chosen": -5.65625, "logits/rejected": -1.484375, "logps/chosen": -2240.0, "logps/rejected": -3856.0, "loss": 0.0391, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -17.5, "rewards/margins": 16.75, "rewards/rejected": -34.25, "step": 970 }, { "epoch": 1.0261780104712042, "grad_norm": 0.31515563654809836, "learning_rate": 4.160260998621387e-05, "logits/chosen": -5.21875, "logits/rejected": -0.953125, "logps/chosen": -2576.0, "logps/rejected": -4832.0, "loss": 0.0335, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -20.75, "rewards/margins": 23.0, "rewards/rejected": -43.75, "step": 980 }, { "epoch": 1.036649214659686, "grad_norm": 0.23064159228128733, "learning_rate": 4.137361135884824e-05, "logits/chosen": -5.03125, "logits/rejected": -0.7109375, "logps/chosen": -2560.0, "logps/rejected": -4768.0, "loss": 0.0242, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -20.75, "rewards/margins": 22.5, "rewards/rejected": -43.25, "step": 990 }, { "epoch": 1.0471204188481675, "grad_norm": 0.43632340692639004, "learning_rate": 4.114218123848777e-05, "logits/chosen": -5.90625, "logits/rejected": -1.4140625, "logps/chosen": -2160.0, "logps/rejected": -4192.0, "loss": 0.028, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -17.125, "rewards/margins": 20.5, "rewards/rejected": -37.5, "step": 1000 }, { "epoch": 1.0471204188481675, "eval_logits/chosen": -2.609375, "eval_logits/rejected": -1.6328125, "eval_logps/chosen": -3104.0, "eval_logps/rejected": -2896.0, "eval_loss": 7.599374771118164, "eval_rewards/accuracies": 0.482421875, "eval_rewards/chosen": -27.75, "eval_rewards/margins": -1.7421875, "eval_rewards/rejected": -26.0, "eval_runtime": 46.6798, "eval_samples_per_second": 42.845, "eval_steps_per_second": 0.686, "step": 1000 }, { "epoch": 1.057591623036649, "grad_norm": 0.3601532383027415, "learning_rate": 4.0908353992670824e-05, "logits/chosen": -5.625, "logits/rejected": -2.140625, "logps/chosen": -2560.0, "logps/rejected": -4352.0, "loss": 0.0485, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -21.125, "rewards/margins": 18.5, "rewards/rejected": -39.75, "step": 1010 }, { "epoch": 1.068062827225131, "grad_norm": 0.30758009064143593, "learning_rate": 4.067216434491059e-05, "logits/chosen": -4.8125, "logits/rejected": -1.1953125, "logps/chosen": -2976.0, "logps/rejected": -5280.0, "loss": 0.0282, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -24.875, "rewards/margins": 23.75, "rewards/rejected": -48.5, "step": 1020 }, { "epoch": 1.0785340314136125, "grad_norm": 0.1427324340343653, "learning_rate": 4.043364736953866e-05, "logits/chosen": -3.140625, "logits/rejected": 0.48046875, "logps/chosen": -3376.0, "logps/rejected": -6304.0, "loss": 0.0572, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -29.0, "rewards/margins": 29.25, "rewards/rejected": -58.25, "step": 1030 }, { "epoch": 1.0890052356020943, "grad_norm": 0.20651262192959718, "learning_rate": 4.019283848649643e-05, "logits/chosen": -4.09375, "logits/rejected": -0.60546875, "logps/chosen": -3088.0, "logps/rejected": -6048.0, "loss": 0.0206, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -26.0, "rewards/margins": 30.25, "rewards/rejected": -56.25, "step": 1040 }, { "epoch": 1.0994764397905759, "grad_norm": 1.486049159077319, "learning_rate": 3.9949773456075215e-05, "logits/chosen": -4.125, "logits/rejected": -0.36328125, "logps/chosen": -2960.0, "logps/rejected": -5504.0, "loss": 0.0343, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -25.5, "rewards/margins": 25.625, "rewards/rejected": -51.0, "step": 1050 }, { "epoch": 1.1099476439790577, "grad_norm": 1.9891084820894762, "learning_rate": 3.970448837360585e-05, "logits/chosen": -5.84375, "logits/rejected": -1.234375, "logps/chosen": -2832.0, "logps/rejected": -5664.0, "loss": 0.0326, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -23.0, "rewards/margins": 28.875, "rewards/rejected": -52.0, "step": 1060 }, { "epoch": 1.1204188481675392, "grad_norm": 0.5257254167121985, "learning_rate": 3.945701966409846e-05, "logits/chosen": -4.5, "logits/rejected": 0.1611328125, "logps/chosen": -3568.0, "logps/rejected": -6880.0, "loss": 0.0351, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -30.625, "rewards/margins": 34.0, "rewards/rejected": -64.5, "step": 1070 }, { "epoch": 1.130890052356021, "grad_norm": 0.9787452931157342, "learning_rate": 3.9207404076833374e-05, "logits/chosen": -2.046875, "logits/rejected": 1.46875, "logps/chosen": -4000.0, "logps/rejected": -6880.0, "loss": 0.0363, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -35.0, "rewards/margins": 29.375, "rewards/rejected": -64.5, "step": 1080 }, { "epoch": 1.1413612565445026, "grad_norm": 0.5535978493008736, "learning_rate": 3.895567867990379e-05, "logits/chosen": -5.25, "logits/rejected": -1.5625, "logps/chosen": -2992.0, "logps/rejected": -5664.0, "loss": 0.0586, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -24.625, "rewards/margins": 27.375, "rewards/rejected": -52.0, "step": 1090 }, { "epoch": 1.1518324607329844, "grad_norm": 0.28543569180603323, "learning_rate": 3.870188085471114e-05, "logits/chosen": -6.65625, "logits/rejected": -2.6875, "logps/chosen": -2112.0, "logps/rejected": -4160.0, "loss": 0.0329, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -16.625, "rewards/margins": 20.75, "rewards/rejected": -37.5, "step": 1100 }, { "epoch": 1.1518324607329844, "eval_logits/chosen": -0.25390625, "eval_logits/rejected": 0.80859375, "eval_logps/chosen": -3776.0, "eval_logps/rejected": -3584.0, "eval_loss": 8.876625061035156, "eval_rewards/accuracies": 0.470703125, "eval_rewards/chosen": -34.5, "eval_rewards/margins": -1.734375, "eval_rewards/rejected": -33.0, "eval_runtime": 46.68, "eval_samples_per_second": 42.845, "eval_steps_per_second": 0.686, "step": 1100 }, { "epoch": 1.162303664921466, "grad_norm": 3.1277231877904543, "learning_rate": 3.844604829041395e-05, "logits/chosen": -3.5, "logits/rejected": -0.0732421875, "logps/chosen": -2624.0, "logps/rejected": -4736.0, "loss": 0.055, "rewards/accuracies": 1.0, "rewards/chosen": -22.125, "rewards/margins": 21.5, "rewards/rejected": -43.5, "step": 1110 }, { "epoch": 1.1727748691099475, "grad_norm": 0.2171874449711856, "learning_rate": 3.818821897833095e-05, "logits/chosen": -3.015625, "logits/rejected": -0.1943359375, "logps/chosen": -2624.0, "logps/rejected": -4320.0, "loss": 0.0345, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -21.5, "rewards/margins": 17.625, "rewards/rejected": -39.0, "step": 1120 }, { "epoch": 1.1832460732984293, "grad_norm": 0.43319452287073673, "learning_rate": 3.792843120629935e-05, "logits/chosen": -5.40625, "logits/rejected": -1.7109375, "logps/chosen": -2192.0, "logps/rejected": -3632.0, "loss": 0.0356, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -17.375, "rewards/margins": 14.8125, "rewards/rejected": -32.25, "step": 1130 }, { "epoch": 1.193717277486911, "grad_norm": 0.45474339491025195, "learning_rate": 3.766672355298907e-05, "logits/chosen": -4.90625, "logits/rejected": -0.83203125, "logps/chosen": -2040.0, "logps/rejected": -3520.0, "loss": 0.0377, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -16.375, "rewards/margins": 15.1875, "rewards/rejected": -31.625, "step": 1140 }, { "epoch": 1.2041884816753927, "grad_norm": 0.31401921450586406, "learning_rate": 3.740313488217373e-05, "logits/chosen": -5.15625, "logits/rejected": -0.51171875, "logps/chosen": -2448.0, "logps/rejected": -4224.0, "loss": 0.0749, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -20.0, "rewards/margins": 18.5, "rewards/rejected": -38.5, "step": 1150 }, { "epoch": 1.2146596858638743, "grad_norm": 0.36406749802613714, "learning_rate": 3.7137704336959464e-05, "logits/chosen": -3.171875, "logits/rejected": 1.84375, "logps/chosen": -2512.0, "logps/rejected": -4352.0, "loss": 0.043, "rewards/accuracies": 0.96875, "rewards/chosen": -20.75, "rewards/margins": 19.0, "rewards/rejected": -39.75, "step": 1160 }, { "epoch": 1.225130890052356, "grad_norm": 1.0264965556776318, "learning_rate": 3.687047133397201e-05, "logits/chosen": -4.34375, "logits/rejected": 1.203125, "logps/chosen": -2384.0, "logps/rejected": -4224.0, "loss": 0.0413, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -19.125, "rewards/margins": 19.0, "rewards/rejected": -38.25, "step": 1170 }, { "epoch": 1.2356020942408377, "grad_norm": 0.2293952174369065, "learning_rate": 3.6601475557503415e-05, "logits/chosen": -3.265625, "logits/rejected": 0.94140625, "logps/chosen": -2624.0, "logps/rejected": -4512.0, "loss": 0.0285, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -21.5, "rewards/margins": 19.625, "rewards/rejected": -41.0, "step": 1180 }, { "epoch": 1.2460732984293195, "grad_norm": 1.5314714492236332, "learning_rate": 3.633075695361882e-05, "logits/chosen": -4.4375, "logits/rejected": 0.388671875, "logps/chosen": -2128.0, "logps/rejected": -3824.0, "loss": 0.0558, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -16.875, "rewards/margins": 17.5, "rewards/rejected": -34.25, "step": 1190 }, { "epoch": 1.256544502617801, "grad_norm": 0.1944064146993958, "learning_rate": 3.6058355724224477e-05, "logits/chosen": -4.84375, "logits/rejected": -0.734375, "logps/chosen": -2320.0, "logps/rejected": -3968.0, "loss": 0.0288, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -18.625, "rewards/margins": 17.0, "rewards/rejected": -35.5, "step": 1200 }, { "epoch": 1.256544502617801, "eval_logits/chosen": -0.1513671875, "eval_logits/rejected": 0.796875, "eval_logps/chosen": -3344.0, "eval_logps/rejected": -3072.0, "eval_loss": 7.4045000076293945, "eval_rewards/accuracies": 0.453125, "eval_rewards/chosen": -30.25, "eval_rewards/margins": -2.34375, "eval_rewards/rejected": -27.875, "eval_runtime": 46.7054, "eval_samples_per_second": 42.822, "eval_steps_per_second": 0.685, "step": 1200 }, { "epoch": 1.2670157068062826, "grad_norm": 0.5815552405801894, "learning_rate": 3.5784312321097744e-05, "logits/chosen": -4.84375, "logits/rejected": -0.8828125, "logps/chosen": -2496.0, "logps/rejected": -4320.0, "loss": 0.0466, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -20.25, "rewards/margins": 18.625, "rewards/rejected": -39.0, "step": 1210 }, { "epoch": 1.2774869109947644, "grad_norm": 1.4417929895525565, "learning_rate": 3.550866743987994e-05, "logits/chosen": -5.28125, "logits/rejected": -1.15625, "logps/chosen": -2512.0, "logps/rejected": -4064.0, "loss": 0.0432, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -20.0, "rewards/margins": 16.25, "rewards/rejected": -36.25, "step": 1220 }, { "epoch": 1.2879581151832462, "grad_norm": 0.46394763021073915, "learning_rate": 3.5231462014033024e-05, "logits/chosen": -6.78125, "logits/rejected": -1.890625, "logps/chosen": -2064.0, "logps/rejected": -3488.0, "loss": 0.0234, "rewards/accuracies": 1.0, "rewards/chosen": -16.125, "rewards/margins": 14.875, "rewards/rejected": -31.0, "step": 1230 }, { "epoch": 1.2984293193717278, "grad_norm": 0.5317050289774921, "learning_rate": 3.495273720876095e-05, "logits/chosen": -5.53125, "logits/rejected": -0.4375, "logps/chosen": -2736.0, "logps/rejected": -4448.0, "loss": 0.0557, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -22.75, "rewards/margins": 17.75, "rewards/rejected": -40.5, "step": 1240 }, { "epoch": 1.3089005235602094, "grad_norm": 0.561051774042172, "learning_rate": 3.4672534414896616e-05, "logits/chosen": -3.796875, "logits/rejected": -0.05908203125, "logps/chosen": -3456.0, "logps/rejected": -5760.0, "loss": 0.0348, "rewards/accuracies": 0.9624999761581421, "rewards/chosen": -29.75, "rewards/margins": 23.625, "rewards/rejected": -53.25, "step": 1250 }, { "epoch": 1.3193717277486912, "grad_norm": 0.0811805532009588, "learning_rate": 3.439089524275526e-05, "logits/chosen": -4.03125, "logits/rejected": -0.1904296875, "logps/chosen": -3184.0, "logps/rejected": -5152.0, "loss": 0.052, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -27.625, "rewards/margins": 20.125, "rewards/rejected": -47.75, "step": 1260 }, { "epoch": 1.3298429319371727, "grad_norm": 0.5824623765263743, "learning_rate": 3.410786151595533e-05, "logits/chosen": -5.375, "logits/rejected": -1.4609375, "logps/chosen": -2224.0, "logps/rejected": -3552.0, "loss": 0.0292, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -17.875, "rewards/margins": 13.8125, "rewards/rejected": -31.75, "step": 1270 }, { "epoch": 1.3403141361256545, "grad_norm": 0.5400828064601035, "learning_rate": 3.382347526520765e-05, "logits/chosen": -4.875, "logits/rejected": -1.3828125, "logps/chosen": -2320.0, "logps/rejected": -3712.0, "loss": 0.0361, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -18.75, "rewards/margins": 14.375, "rewards/rejected": -33.25, "step": 1280 }, { "epoch": 1.350785340314136, "grad_norm": 0.03860557825495785, "learning_rate": 3.353777872207381e-05, "logits/chosen": -5.8125, "logits/rejected": -2.1875, "logps/chosen": -2512.0, "logps/rejected": -4048.0, "loss": 0.0311, "rewards/accuracies": 1.0, "rewards/chosen": -20.5, "rewards/margins": 15.8125, "rewards/rejected": -36.25, "step": 1290 }, { "epoch": 1.3612565445026177, "grad_norm": 0.8286851596181335, "learning_rate": 3.325081431269475e-05, "logits/chosen": -6.4375, "logits/rejected": -2.46875, "logps/chosen": -2352.0, "logps/rejected": -4016.0, "loss": 0.0403, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -19.125, "rewards/margins": 17.25, "rewards/rejected": -36.5, "step": 1300 }, { "epoch": 1.3612565445026177, "eval_logits/chosen": -3.8125, "eval_logits/rejected": -2.96875, "eval_logps/chosen": -3088.0, "eval_logps/rejected": -2864.0, "eval_loss": 6.609875202178955, "eval_rewards/accuracies": 0.453125, "eval_rewards/chosen": -27.75, "eval_rewards/margins": -1.984375, "eval_rewards/rejected": -25.75, "eval_runtime": 46.6635, "eval_samples_per_second": 42.86, "eval_steps_per_second": 0.686, "step": 1300 }, { "epoch": 1.3717277486910995, "grad_norm": 0.22377441361000777, "learning_rate": 3.296262465149046e-05, "logits/chosen": -8.125, "logits/rejected": -4.46875, "logps/chosen": -2080.0, "logps/rejected": -3760.0, "loss": 0.0238, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -15.9375, "rewards/margins": 17.5, "rewards/rejected": -33.5, "step": 1310 }, { "epoch": 1.3821989528795813, "grad_norm": 0.34920850966331374, "learning_rate": 3.2673252534831687e-05, "logits/chosen": -7.84375, "logits/rejected": -4.03125, "logps/chosen": -2480.0, "logps/rejected": -4288.0, "loss": 0.0382, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -20.375, "rewards/margins": 18.625, "rewards/rejected": -38.75, "step": 1320 }, { "epoch": 1.3926701570680629, "grad_norm": 0.9764003281459026, "learning_rate": 3.2382740934684696e-05, "logits/chosen": -7.1875, "logits/rejected": -2.765625, "logps/chosen": -2576.0, "logps/rejected": -4576.0, "loss": 0.0287, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -21.125, "rewards/margins": 20.625, "rewards/rejected": -41.75, "step": 1330 }, { "epoch": 1.4031413612565444, "grad_norm": 0.27933570133215424, "learning_rate": 3.209113299222982e-05, "logits/chosen": -6.6875, "logits/rejected": -1.53125, "logps/chosen": -2432.0, "logps/rejected": -4288.0, "loss": 0.0374, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -20.25, "rewards/margins": 19.25, "rewards/rejected": -39.5, "step": 1340 }, { "epoch": 1.4136125654450262, "grad_norm": 0.5211324152589242, "learning_rate": 3.179847201145505e-05, "logits/chosen": -5.75, "logits/rejected": -1.3125, "logps/chosen": -2368.0, "logps/rejected": -3984.0, "loss": 0.0345, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -18.875, "rewards/margins": 16.75, "rewards/rejected": -35.5, "step": 1350 }, { "epoch": 1.4240837696335078, "grad_norm": 0.23463942523916043, "learning_rate": 3.150480145272528e-05, "logits/chosen": -6.15625, "logits/rejected": -1.296875, "logps/chosen": -2432.0, "logps/rejected": -4288.0, "loss": 0.0242, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -19.625, "rewards/margins": 19.125, "rewards/rejected": -38.75, "step": 1360 }, { "epoch": 1.4345549738219896, "grad_norm": 0.34711605116362065, "learning_rate": 3.121016492632848e-05, "logits/chosen": -6.25, "logits/rejected": -1.8359375, "logps/chosen": -2496.0, "logps/rejected": -4544.0, "loss": 0.0361, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -20.25, "rewards/margins": 21.0, "rewards/rejected": -41.25, "step": 1370 }, { "epoch": 1.4450261780104712, "grad_norm": 1.682075012745004, "learning_rate": 3.091460618599951e-05, "logits/chosen": -4.03125, "logits/rejected": 0.359375, "logps/chosen": -3072.0, "logps/rejected": -5344.0, "loss": 0.0244, "rewards/accuracies": 1.0, "rewards/chosen": -26.125, "rewards/margins": 23.25, "rewards/rejected": -49.5, "step": 1380 }, { "epoch": 1.4554973821989527, "grad_norm": 0.20646706160668513, "learning_rate": 3.061816912242265e-05, "logits/chosen": -4.5625, "logits/rejected": 0.515625, "logps/chosen": -2944.0, "logps/rejected": -5408.0, "loss": 0.0237, "rewards/accuracies": 1.0, "rewards/chosen": -24.75, "rewards/margins": 25.125, "rewards/rejected": -49.75, "step": 1390 }, { "epoch": 1.4659685863874345, "grad_norm": 1.002892186666077, "learning_rate": 3.0320897756713785e-05, "logits/chosen": -4.8125, "logits/rejected": -0.208984375, "logps/chosen": -3184.0, "logps/rejected": -6368.0, "loss": 0.0286, "rewards/accuracies": 0.96875, "rewards/chosen": -26.875, "rewards/margins": 32.5, "rewards/rejected": -59.25, "step": 1400 }, { "epoch": 1.4659685863874345, "eval_logits/chosen": -0.0228271484375, "eval_logits/rejected": 0.94921875, "eval_logps/chosen": -4672.0, "eval_logps/rejected": -4288.0, "eval_loss": 12.43274974822998, "eval_rewards/accuracies": 0.46875, "eval_rewards/chosen": -43.75, "eval_rewards/margins": -3.875, "eval_rewards/rejected": -39.75, "eval_runtime": 46.6866, "eval_samples_per_second": 42.839, "eval_steps_per_second": 0.685, "step": 1400 }, { "epoch": 1.4764397905759163, "grad_norm": 0.07978425534496958, "learning_rate": 3.002283623388332e-05, "logits/chosen": -5.125, "logits/rejected": -0.63671875, "logps/chosen": -2960.0, "logps/rejected": -5984.0, "loss": 0.0309, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -25.25, "rewards/margins": 30.875, "rewards/rejected": -56.25, "step": 1410 }, { "epoch": 1.486910994764398, "grad_norm": 1.1900857096571082, "learning_rate": 2.9724028816280507e-05, "logits/chosen": -4.34375, "logits/rejected": -0.1279296875, "logps/chosen": -3376.0, "logps/rejected": -6432.0, "loss": 0.0306, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -29.5, "rewards/margins": 30.875, "rewards/rejected": -60.25, "step": 1420 }, { "epoch": 1.4973821989528795, "grad_norm": 0.3487398209894109, "learning_rate": 2.9424519877020524e-05, "logits/chosen": -4.28125, "logits/rejected": -0.3046875, "logps/chosen": -4192.0, "logps/rejected": -8064.0, "loss": 0.0274, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -37.0, "rewards/margins": 39.25, "rewards/rejected": -76.0, "step": 1430 }, { "epoch": 1.5078534031413613, "grad_norm": 0.22708442653598893, "learning_rate": 2.912435389339504e-05, "logits/chosen": -3.578125, "logits/rejected": -0.01611328125, "logps/chosen": -3408.0, "logps/rejected": -6496.0, "loss": 0.0248, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -29.5, "rewards/margins": 31.375, "rewards/rejected": -60.75, "step": 1440 }, { "epoch": 1.518324607329843, "grad_norm": 1.571784535660224, "learning_rate": 2.8823575440267226e-05, "logits/chosen": -3.859375, "logits/rejected": 0.064453125, "logps/chosen": -3744.0, "logps/rejected": -6784.0, "loss": 0.0333, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -32.5, "rewards/margins": 31.375, "rewards/rejected": -64.0, "step": 1450 }, { "epoch": 1.5287958115183247, "grad_norm": 7.404401252470361, "learning_rate": 2.8522229183452465e-05, "logits/chosen": -4.28125, "logits/rejected": 0.1650390625, "logps/chosen": -3616.0, "logps/rejected": -7040.0, "loss": 0.0646, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -31.625, "rewards/margins": 34.75, "rewards/rejected": -66.5, "step": 1460 }, { "epoch": 1.5392670157068062, "grad_norm": 0.39921320664245197, "learning_rate": 2.8220359873085373e-05, "logits/chosen": -4.875, "logits/rejected": -1.28125, "logps/chosen": -3008.0, "logps/rejected": -5184.0, "loss": 0.0232, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -25.5, "rewards/margins": 22.125, "rewards/rejected": -47.75, "step": 1470 }, { "epoch": 1.5497382198952878, "grad_norm": 0.1702420101790886, "learning_rate": 2.7918012336974388e-05, "logits/chosen": -5.03125, "logits/rejected": -1.2109375, "logps/chosen": -2640.0, "logps/rejected": -4384.0, "loss": 0.0253, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -21.875, "rewards/margins": 17.875, "rewards/rejected": -39.75, "step": 1480 }, { "epoch": 1.5602094240837696, "grad_norm": 2.279427436005394, "learning_rate": 2.7615231473944815e-05, "logits/chosen": -3.8125, "logits/rejected": 0.11865234375, "logps/chosen": -3040.0, "logps/rejected": -4992.0, "loss": 0.0419, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -26.0, "rewards/margins": 20.125, "rewards/rejected": -46.0, "step": 1490 }, { "epoch": 1.5706806282722514, "grad_norm": 0.3205025677294034, "learning_rate": 2.7312062247171327e-05, "logits/chosen": -4.875, "logits/rejected": -0.76953125, "logps/chosen": -2768.0, "logps/rejected": -4832.0, "loss": 0.0237, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -23.375, "rewards/margins": 21.25, "rewards/rejected": -44.5, "step": 1500 }, { "epoch": 1.5706806282722514, "eval_logits/chosen": 0.37890625, "eval_logits/rejected": 1.4140625, "eval_logps/chosen": -4016.0, "eval_logps/rejected": -3664.0, "eval_loss": 9.634187698364258, "eval_rewards/accuracies": 0.44140625, "eval_rewards/chosen": -37.0, "eval_rewards/margins": -3.25, "eval_rewards/rejected": -33.75, "eval_runtime": 46.6862, "eval_samples_per_second": 42.839, "eval_steps_per_second": 0.685, "step": 1500 }, { "epoch": 1.581151832460733, "grad_norm": 0.36916418040296595, "learning_rate": 2.7008549677500876e-05, "logits/chosen": -5.1875, "logits/rejected": -1.1328125, "logps/chosen": -2864.0, "logps/rejected": -5248.0, "loss": 0.0327, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -23.625, "rewards/margins": 24.5, "rewards/rejected": -48.0, "step": 1510 }, { "epoch": 1.5916230366492146, "grad_norm": 1.3927718855245257, "learning_rate": 2.670473883676709e-05, "logits/chosen": -5.125, "logits/rejected": -1.1328125, "logps/chosen": -2496.0, "logps/rejected": -4672.0, "loss": 0.0345, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -20.375, "rewards/margins": 22.125, "rewards/rejected": -42.5, "step": 1520 }, { "epoch": 1.6020942408376964, "grad_norm": 10.027917154946675, "learning_rate": 2.6400674841097074e-05, "logits/chosen": -3.734375, "logits/rejected": -0.291015625, "logps/chosen": -3664.0, "logps/rejected": -6464.0, "loss": 0.0437, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -32.0, "rewards/margins": 28.625, "rewards/rejected": -60.5, "step": 1530 }, { "epoch": 1.6125654450261782, "grad_norm": 0.426370412366889, "learning_rate": 2.609640284421158e-05, "logits/chosen": -3.578125, "logits/rejected": 0.0849609375, "logps/chosen": -3232.0, "logps/rejected": -5920.0, "loss": 0.0315, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -27.75, "rewards/margins": 27.5, "rewards/rejected": -55.25, "step": 1540 }, { "epoch": 1.6230366492146597, "grad_norm": 0.2833158216827099, "learning_rate": 2.5791968030719726e-05, "logits/chosen": -4.53125, "logits/rejected": -0.625, "logps/chosen": -3104.0, "logps/rejected": -5632.0, "loss": 0.0353, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -26.125, "rewards/margins": 25.625, "rewards/rejected": -51.75, "step": 1550 }, { "epoch": 1.6335078534031413, "grad_norm": 0.40953108589831894, "learning_rate": 2.548741560940902e-05, "logits/chosen": -4.03125, "logits/rejected": -0.291015625, "logps/chosen": -3040.0, "logps/rejected": -5536.0, "loss": 0.0491, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -25.75, "rewards/margins": 25.5, "rewards/rejected": -51.25, "step": 1560 }, { "epoch": 1.6439790575916229, "grad_norm": 0.6366721759012215, "learning_rate": 2.5182790806531787e-05, "logits/chosen": -4.0625, "logits/rejected": -0.12158203125, "logps/chosen": -2672.0, "logps/rejected": -4768.0, "loss": 0.0453, "rewards/accuracies": 0.96875, "rewards/chosen": -22.125, "rewards/margins": 21.5, "rewards/rejected": -43.5, "step": 1570 }, { "epoch": 1.6544502617801047, "grad_norm": 0.44031786955716545, "learning_rate": 2.4878138859089074e-05, "logits/chosen": -4.625, "logits/rejected": -0.92578125, "logps/chosen": -2544.0, "logps/rejected": -4352.0, "loss": 0.0218, "rewards/accuracies": 1.0, "rewards/chosen": -21.125, "rewards/margins": 18.25, "rewards/rejected": -39.25, "step": 1580 }, { "epoch": 1.6649214659685865, "grad_norm": 0.05896262622666351, "learning_rate": 2.4573505008112924e-05, "logits/chosen": -5.0625, "logits/rejected": -1.453125, "logps/chosen": -2800.0, "logps/rejected": -4960.0, "loss": 0.0351, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -23.5, "rewards/margins": 22.125, "rewards/rejected": -45.5, "step": 1590 }, { "epoch": 1.675392670157068, "grad_norm": 0.4125669335218442, "learning_rate": 2.4268934491948032e-05, "logits/chosen": -4.59375, "logits/rejected": -1.3203125, "logps/chosen": -2752.0, "logps/rejected": -4800.0, "loss": 0.0231, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -22.875, "rewards/margins": 20.875, "rewards/rejected": -43.75, "step": 1600 }, { "epoch": 1.675392670157068, "eval_logits/chosen": 0.16796875, "eval_logits/rejected": 1.1015625, "eval_logps/chosen": -4160.0, "eval_logps/rejected": -3776.0, "eval_loss": 9.662375450134277, "eval_rewards/accuracies": 0.453125, "eval_rewards/chosen": -38.25, "eval_rewards/margins": -3.515625, "eval_rewards/rejected": -34.75, "eval_runtime": 46.672, "eval_samples_per_second": 42.852, "eval_steps_per_second": 0.686, "step": 1600 }, { "epoch": 1.6858638743455496, "grad_norm": 0.6035961782630841, "learning_rate": 2.396447253953385e-05, "logits/chosen": -5.21875, "logits/rejected": -1.65625, "logps/chosen": -3376.0, "logps/rejected": -5856.0, "loss": 0.0303, "rewards/accuracies": 0.9624999761581421, "rewards/chosen": -28.75, "rewards/margins": 25.25, "rewards/rejected": -54.0, "step": 1610 }, { "epoch": 1.6963350785340314, "grad_norm": 0.9317904949574454, "learning_rate": 2.3660164363687998e-05, "logits/chosen": -4.5, "logits/rejected": -0.83203125, "logps/chosen": -3392.0, "logps/rejected": -6304.0, "loss": 0.0271, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -28.75, "rewards/margins": 29.75, "rewards/rejected": -58.5, "step": 1620 }, { "epoch": 1.7068062827225132, "grad_norm": 0.19726900736489286, "learning_rate": 2.33560551543922e-05, "logits/chosen": -4.375, "logits/rejected": -0.421875, "logps/chosen": -3120.0, "logps/rejected": -5920.0, "loss": 0.0265, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -26.25, "rewards/margins": 28.625, "rewards/rejected": -55.0, "step": 1630 }, { "epoch": 1.7172774869109948, "grad_norm": 0.5366057896162931, "learning_rate": 2.3052190072081493e-05, "logits/chosen": -5.125, "logits/rejected": -1.359375, "logps/chosen": -2816.0, "logps/rejected": -5120.0, "loss": 0.0205, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -23.5, "rewards/margins": 23.375, "rewards/rejected": -47.0, "step": 1640 }, { "epoch": 1.7277486910994764, "grad_norm": 0.288713837684222, "learning_rate": 2.2748614240937864e-05, "logits/chosen": -4.59375, "logits/rejected": -0.7890625, "logps/chosen": -2592.0, "logps/rejected": -4992.0, "loss": 0.0203, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -21.375, "rewards/margins": 24.625, "rewards/rejected": -46.0, "step": 1650 }, { "epoch": 1.738219895287958, "grad_norm": 0.8099367937443963, "learning_rate": 2.2445372742189334e-05, "logits/chosen": -4.3125, "logits/rejected": -0.6640625, "logps/chosen": -2816.0, "logps/rejected": -5216.0, "loss": 0.0299, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -23.5, "rewards/margins": 24.375, "rewards/rejected": -48.0, "step": 1660 }, { "epoch": 1.7486910994764397, "grad_norm": 0.08287616443823331, "learning_rate": 2.214251060741528e-05, "logits/chosen": -4.8125, "logits/rejected": -0.90234375, "logps/chosen": -2672.0, "logps/rejected": -5408.0, "loss": 0.0261, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -22.0, "rewards/margins": 27.625, "rewards/rejected": -49.5, "step": 1670 }, { "epoch": 1.7591623036649215, "grad_norm": 0.2244909870748531, "learning_rate": 2.1840072811859302e-05, "logits/chosen": -4.75, "logits/rejected": -0.70703125, "logps/chosen": -3040.0, "logps/rejected": -5664.0, "loss": 0.0326, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -25.5, "rewards/margins": 26.875, "rewards/rejected": -52.25, "step": 1680 }, { "epoch": 1.7696335078534031, "grad_norm": 0.03818992665954291, "learning_rate": 2.1538104267750287e-05, "logits/chosen": -4.21875, "logits/rejected": 0.2353515625, "logps/chosen": -2912.0, "logps/rejected": -5600.0, "loss": 0.0224, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -24.5, "rewards/margins": 27.5, "rewards/rejected": -52.0, "step": 1690 }, { "epoch": 1.7801047120418847, "grad_norm": 0.661947709768003, "learning_rate": 2.1236649817632952e-05, "logits/chosen": -4.34375, "logits/rejected": -0.267578125, "logps/chosen": -3168.0, "logps/rejected": -6208.0, "loss": 0.0199, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -27.5, "rewards/margins": 31.0, "rewards/rejected": -58.5, "step": 1700 }, { "epoch": 1.7801047120418847, "eval_logits/chosen": 0.90625, "eval_logits/rejected": 1.84375, "eval_logps/chosen": -5152.0, "eval_logps/rejected": -4672.0, "eval_loss": 13.210624694824219, "eval_rewards/accuracies": 0.451171875, "eval_rewards/chosen": -48.5, "eval_rewards/margins": -4.75, "eval_rewards/rejected": -43.75, "eval_runtime": 46.6746, "eval_samples_per_second": 42.85, "eval_steps_per_second": 0.686, "step": 1700 }, { "epoch": 1.7905759162303665, "grad_norm": 3.63127429065245, "learning_rate": 2.093575422770872e-05, "logits/chosen": -3.28125, "logits/rejected": 0.6796875, "logps/chosen": -3056.0, "logps/rejected": -5824.0, "loss": 0.0551, "rewards/accuracies": 0.96875, "rewards/chosen": -26.125, "rewards/margins": 28.375, "rewards/rejected": -54.5, "step": 1710 }, { "epoch": 1.8010471204188483, "grad_norm": 0.433710887587725, "learning_rate": 2.0635462181187827e-05, "logits/chosen": -4.8125, "logits/rejected": -1.1171875, "logps/chosen": -3040.0, "logps/rejected": -5536.0, "loss": 0.0265, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -25.75, "rewards/margins": 25.625, "rewards/rejected": -51.5, "step": 1720 }, { "epoch": 1.8115183246073299, "grad_norm": 0.40532659167832996, "learning_rate": 2.0335818271653917e-05, "logits/chosen": -4.0, "logits/rejected": -0.146484375, "logps/chosen": -3328.0, "logps/rejected": -6080.0, "loss": 0.0369, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -28.625, "rewards/margins": 28.0, "rewards/rejected": -56.5, "step": 1730 }, { "epoch": 1.8219895287958114, "grad_norm": 0.7622564530024274, "learning_rate": 2.0036866996441815e-05, "logits/chosen": -3.328125, "logits/rejected": 0.5, "logps/chosen": -3184.0, "logps/rejected": -5760.0, "loss": 0.0254, "rewards/accuracies": 1.0, "rewards/chosen": -27.375, "rewards/margins": 26.0, "rewards/rejected": -53.5, "step": 1740 }, { "epoch": 1.8324607329842932, "grad_norm": 1.3697788994875846, "learning_rate": 1.9738652750029677e-05, "logits/chosen": -3.40625, "logits/rejected": 0.470703125, "logps/chosen": -3376.0, "logps/rejected": -5984.0, "loss": 0.0267, "rewards/accuracies": 0.96875, "rewards/chosen": -29.25, "rewards/margins": 26.5, "rewards/rejected": -55.75, "step": 1750 }, { "epoch": 1.8429319371727748, "grad_norm": 2.6650218009278746, "learning_rate": 1.944121981744631e-05, "logits/chosen": -5.25, "logits/rejected": -0.61328125, "logps/chosen": -2912.0, "logps/rejected": -5792.0, "loss": 0.0266, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -24.375, "rewards/margins": 28.875, "rewards/rejected": -53.25, "step": 1760 }, { "epoch": 1.8534031413612566, "grad_norm": 0.05386844135170857, "learning_rate": 1.9144612367694887e-05, "logits/chosen": -5.21875, "logits/rejected": -0.64453125, "logps/chosen": -2848.0, "logps/rejected": -5760.0, "loss": 0.0223, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -23.875, "rewards/margins": 29.625, "rewards/rejected": -53.5, "step": 1770 }, { "epoch": 1.8638743455497382, "grad_norm": 0.1709010645553243, "learning_rate": 1.8848874447193803e-05, "logits/chosen": -5.375, "logits/rejected": -1.359375, "logps/chosen": -2928.0, "logps/rejected": -5248.0, "loss": 0.0282, "rewards/accuracies": 0.9624999761581421, "rewards/chosen": -24.875, "rewards/margins": 23.75, "rewards/rejected": -48.75, "step": 1780 }, { "epoch": 1.8743455497382198, "grad_norm": 0.4264726318514752, "learning_rate": 1.855404997323571e-05, "logits/chosen": -5.3125, "logits/rejected": -0.55078125, "logps/chosen": -3376.0, "logps/rejected": -6080.0, "loss": 0.0329, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -29.0, "rewards/margins": 28.0, "rewards/rejected": -56.75, "step": 1790 }, { "epoch": 1.8848167539267016, "grad_norm": 1.0255932743747602, "learning_rate": 1.82601827274658e-05, "logits/chosen": -5.28125, "logits/rejected": -0.58203125, "logps/chosen": -2992.0, "logps/rejected": -5536.0, "loss": 0.0202, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -25.5, "rewards/margins": 26.0, "rewards/rejected": -51.5, "step": 1800 }, { "epoch": 1.8848167539267016, "eval_logits/chosen": -0.29296875, "eval_logits/rejected": 0.6640625, "eval_logps/chosen": -4416.0, "eval_logps/rejected": -4080.0, "eval_loss": 10.321125030517578, "eval_rewards/accuracies": 0.44921875, "eval_rewards/chosen": -41.0, "eval_rewards/margins": -3.234375, "eval_rewards/rejected": -37.75, "eval_runtime": 46.6805, "eval_samples_per_second": 42.844, "eval_steps_per_second": 0.686, "step": 1800 }, { "epoch": 1.8952879581151834, "grad_norm": 3.3576466838068404, "learning_rate": 1.7967316349380213e-05, "logits/chosen": -5.75, "logits/rejected": -1.296875, "logps/chosen": -2672.0, "logps/rejected": -5216.0, "loss": 0.0385, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -22.375, "rewards/margins": 26.125, "rewards/rejected": -48.5, "step": 1810 }, { "epoch": 1.905759162303665, "grad_norm": 0.23409946188149564, "learning_rate": 1.7675494329845516e-05, "logits/chosen": -5.375, "logits/rejected": -1.1640625, "logps/chosen": -2640.0, "logps/rejected": -4736.0, "loss": 0.0149, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -21.875, "rewards/margins": 21.5, "rewards/rejected": -43.5, "step": 1820 }, { "epoch": 1.9162303664921465, "grad_norm": 0.5284118246632048, "learning_rate": 1.738476000464026e-05, "logits/chosen": -4.78125, "logits/rejected": -0.9140625, "logps/chosen": -2656.0, "logps/rejected": -4864.0, "loss": 0.0246, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -22.125, "rewards/margins": 22.375, "rewards/rejected": -44.5, "step": 1830 }, { "epoch": 1.9267015706806283, "grad_norm": 0.42844287782798285, "learning_rate": 1.7095156548019646e-05, "logits/chosen": -5.1875, "logits/rejected": -1.171875, "logps/chosen": -2560.0, "logps/rejected": -4640.0, "loss": 0.0264, "rewards/accuracies": 0.96875, "rewards/chosen": -21.375, "rewards/margins": 21.375, "rewards/rejected": -42.75, "step": 1840 }, { "epoch": 1.93717277486911, "grad_norm": 0.14049245129154458, "learning_rate": 1.6806726966304064e-05, "logits/chosen": -5.40625, "logits/rejected": -0.8125, "logps/chosen": -2736.0, "logps/rejected": -5376.0, "loss": 0.0304, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -22.5, "rewards/margins": 26.75, "rewards/rejected": -49.25, "step": 1850 }, { "epoch": 1.9476439790575917, "grad_norm": 0.15437894704887348, "learning_rate": 1.6519514091492625e-05, "logits/chosen": -5.0, "logits/rejected": -0.75, "logps/chosen": -2496.0, "logps/rejected": -4544.0, "loss": 0.0271, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -20.625, "rewards/margins": 21.0, "rewards/rejected": -41.5, "step": 1860 }, { "epoch": 1.9581151832460733, "grad_norm": 0.046542503877578335, "learning_rate": 1.6233560574902627e-05, "logits/chosen": -4.96875, "logits/rejected": -0.67578125, "logps/chosen": -2928.0, "logps/rejected": -5248.0, "loss": 0.0225, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -25.125, "rewards/margins": 23.75, "rewards/rejected": -49.0, "step": 1870 }, { "epoch": 1.9685863874345548, "grad_norm": 0.3007524485605509, "learning_rate": 1.5948908880835753e-05, "logits/chosen": -4.9375, "logits/rejected": -0.25, "logps/chosen": -2736.0, "logps/rejected": -5152.0, "loss": 0.0476, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -22.875, "rewards/margins": 24.625, "rewards/rejected": -47.5, "step": 1880 }, { "epoch": 1.9790575916230366, "grad_norm": 0.5954410234408122, "learning_rate": 1.5665601280272124e-05, "logits/chosen": -6.875, "logits/rejected": -2.375, "logps/chosen": -2384.0, "logps/rejected": -4640.0, "loss": 0.0284, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -19.375, "rewards/margins": 22.875, "rewards/rejected": -42.25, "step": 1890 }, { "epoch": 1.9895287958115184, "grad_norm": 0.13010785028420152, "learning_rate": 1.538367984459298e-05, "logits/chosen": -5.71875, "logits/rejected": -1.4609375, "logps/chosen": -2448.0, "logps/rejected": -4768.0, "loss": 0.0305, "rewards/accuracies": 0.949999988079071, "rewards/chosen": -20.25, "rewards/margins": 23.625, "rewards/rejected": -44.0, "step": 1900 }, { "epoch": 1.9895287958115184, "eval_logits/chosen": -1.5, "eval_logits/rejected": -0.5703125, "eval_logps/chosen": -3856.0, "eval_logps/rejected": -3616.0, "eval_loss": 9.091375350952148, "eval_rewards/accuracies": 0.4609375, "eval_rewards/chosen": -35.5, "eval_rewards/margins": -2.0625, "eval_rewards/rejected": -33.25, "eval_runtime": 46.6856, "eval_samples_per_second": 42.84, "eval_steps_per_second": 0.685, "step": 1900 }, { "epoch": 2.0, "grad_norm": 0.27989090766679425, "learning_rate": 1.5103186439333133e-05, "logits/chosen": -6.0625, "logits/rejected": -1.40625, "logps/chosen": -2640.0, "logps/rejected": -4800.0, "loss": 0.0187, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -22.0, "rewards/margins": 22.0, "rewards/rejected": -44.0, "step": 1910 }, { "epoch": 2.0104712041884816, "grad_norm": 0.04547420644741809, "learning_rate": 1.4824162717963829e-05, "logits/chosen": -6.6875, "logits/rejected": -1.8828125, "logps/chosen": -2352.0, "logps/rejected": -4864.0, "loss": 0.0121, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -19.0, "rewards/margins": 26.0, "rewards/rejected": -45.0, "step": 1920 }, { "epoch": 2.020942408376963, "grad_norm": 0.22273320495640803, "learning_rate": 1.4546650115707192e-05, "logits/chosen": -6.78125, "logits/rejected": -2.40625, "logps/chosen": -2480.0, "logps/rejected": -5152.0, "loss": 0.0085, "rewards/accuracies": 1.0, "rewards/chosen": -20.5, "rewards/margins": 27.375, "rewards/rejected": -47.75, "step": 1930 }, { "epoch": 2.031413612565445, "grad_norm": 0.029080064380412527, "learning_rate": 1.4270689843383111e-05, "logits/chosen": -5.21875, "logits/rejected": -1.0078125, "logps/chosen": -2656.0, "logps/rejected": -5120.0, "loss": 0.0104, "rewards/accuracies": 1.0, "rewards/chosen": -22.0, "rewards/margins": 25.0, "rewards/rejected": -47.0, "step": 1940 }, { "epoch": 2.0418848167539267, "grad_norm": 0.11941407790950143, "learning_rate": 1.3996322881289347e-05, "logits/chosen": -6.1875, "logits/rejected": -1.28125, "logps/chosen": -2880.0, "logps/rejected": -5792.0, "loss": 0.005, "rewards/accuracies": 1.0, "rewards/chosen": -24.5, "rewards/margins": 29.75, "rewards/rejected": -54.25, "step": 1950 }, { "epoch": 2.0523560209424083, "grad_norm": 0.07464545884138041, "learning_rate": 1.3723589973115963e-05, "logits/chosen": -6.375, "logits/rejected": -1.1015625, "logps/chosen": -2592.0, "logps/rejected": -5376.0, "loss": 0.0077, "rewards/accuracies": 1.0, "rewards/chosen": -21.5, "rewards/margins": 28.375, "rewards/rejected": -49.75, "step": 1960 }, { "epoch": 2.06282722513089, "grad_norm": 0.22073344253214078, "learning_rate": 1.3452531619894837e-05, "logits/chosen": -5.0625, "logits/rejected": -0.181640625, "logps/chosen": -3232.0, "logps/rejected": -5984.0, "loss": 0.0048, "rewards/accuracies": 1.0, "rewards/chosen": -27.875, "rewards/margins": 28.0, "rewards/rejected": -55.75, "step": 1970 }, { "epoch": 2.073298429319372, "grad_norm": 0.043518671775384925, "learning_rate": 1.3183188073985245e-05, "logits/chosen": -5.8125, "logits/rejected": -1.0078125, "logps/chosen": -3168.0, "logps/rejected": -5984.0, "loss": 0.0079, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -27.125, "rewards/margins": 28.875, "rewards/rejected": -56.0, "step": 1980 }, { "epoch": 2.0837696335078535, "grad_norm": 0.22402092801520945, "learning_rate": 1.2915599333096351e-05, "logits/chosen": -6.09375, "logits/rejected": -1.3984375, "logps/chosen": -3392.0, "logps/rejected": -6368.0, "loss": 0.0054, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -29.125, "rewards/margins": 30.5, "rewards/rejected": -59.5, "step": 1990 }, { "epoch": 2.094240837696335, "grad_norm": 0.4339045702528522, "learning_rate": 1.264980513434752e-05, "logits/chosen": -4.875, "logits/rejected": -0.87109375, "logps/chosen": -3312.0, "logps/rejected": -6464.0, "loss": 0.0093, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -28.375, "rewards/margins": 31.875, "rewards/rejected": -60.25, "step": 2000 }, { "epoch": 2.094240837696335, "eval_logits/chosen": -0.1796875, "eval_logits/rejected": 0.796875, "eval_logps/chosen": -4864.0, "eval_logps/rejected": -4480.0, "eval_loss": 12.383999824523926, "eval_rewards/accuracies": 0.451171875, "eval_rewards/chosen": -45.75, "eval_rewards/margins": -3.59375, "eval_rewards/rejected": -42.0, "eval_runtime": 46.6756, "eval_samples_per_second": 42.849, "eval_steps_per_second": 0.686, "step": 2000 }, { "epoch": 2.1047120418848166, "grad_norm": 0.04372675809102335, "learning_rate": 1.2385844948367324e-05, "logits/chosen": -5.5625, "logits/rejected": -1.0703125, "logps/chosen": -3152.0, "logps/rejected": -6720.0, "loss": 0.0081, "rewards/accuracies": 1.0, "rewards/chosen": -27.25, "rewards/margins": 36.25, "rewards/rejected": -63.25, "step": 2010 }, { "epoch": 2.115183246073298, "grad_norm": 0.021928665141441248, "learning_rate": 1.2123757973432114e-05, "logits/chosen": -5.25, "logits/rejected": -1.0390625, "logps/chosen": -3952.0, "logps/rejected": -7232.0, "loss": 0.0062, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -34.75, "rewards/margins": 33.5, "rewards/rejected": -68.0, "step": 2020 }, { "epoch": 2.1256544502617802, "grad_norm": 0.27310370527484834, "learning_rate": 1.1863583129645106e-05, "logits/chosen": -5.21875, "logits/rejected": -0.1728515625, "logps/chosen": -3440.0, "logps/rejected": -6944.0, "loss": 0.0102, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -29.875, "rewards/margins": 35.5, "rewards/rejected": -65.5, "step": 2030 }, { "epoch": 2.136125654450262, "grad_norm": 0.07436827446740518, "learning_rate": 1.1605359053156607e-05, "logits/chosen": -5.34375, "logits/rejected": -0.6796875, "logps/chosen": -3904.0, "logps/rejected": -7904.0, "loss": 0.0066, "rewards/accuracies": 1.0, "rewards/chosen": -34.25, "rewards/margins": 40.5, "rewards/rejected": -74.5, "step": 2040 }, { "epoch": 2.1465968586387434, "grad_norm": 0.08735651557545136, "learning_rate": 1.1349124090426644e-05, "logits/chosen": -4.78125, "logits/rejected": 0.06005859375, "logps/chosen": -4000.0, "logps/rejected": -7968.0, "loss": 0.0033, "rewards/accuracies": 1.0, "rewards/chosen": -35.0, "rewards/margins": 40.5, "rewards/rejected": -75.5, "step": 2050 }, { "epoch": 2.157068062827225, "grad_norm": 0.14831855782284922, "learning_rate": 1.1094916292530406e-05, "logits/chosen": -4.53125, "logits/rejected": 0.33984375, "logps/chosen": -3808.0, "logps/rejected": -7488.0, "loss": 0.0047, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -33.5, "rewards/margins": 37.25, "rewards/rejected": -70.5, "step": 2060 }, { "epoch": 2.167539267015707, "grad_norm": 0.13954845437425645, "learning_rate": 1.0842773409507623e-05, "logits/chosen": -4.5625, "logits/rejected": 0.2578125, "logps/chosen": -3424.0, "logps/rejected": -6656.0, "loss": 0.0103, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -29.875, "rewards/margins": 33.0, "rewards/rejected": -63.0, "step": 2070 }, { "epoch": 2.1780104712041886, "grad_norm": 0.15150270978611693, "learning_rate": 1.0592732884756754e-05, "logits/chosen": -5.125, "logits/rejected": -0.0283203125, "logps/chosen": -3984.0, "logps/rejected": -7936.0, "loss": 0.0059, "rewards/accuracies": 1.0, "rewards/chosen": -34.75, "rewards/margins": 40.25, "rewards/rejected": -75.0, "step": 2080 }, { "epoch": 2.18848167539267, "grad_norm": 0.31245323297028604, "learning_rate": 1.0344831849474507e-05, "logits/chosen": -4.8125, "logits/rejected": 0.2451171875, "logps/chosen": -3472.0, "logps/rejected": -6880.0, "loss": 0.0052, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -30.125, "rewards/margins": 34.75, "rewards/rejected": -65.0, "step": 2090 }, { "epoch": 2.1989528795811517, "grad_norm": 0.03531403187325075, "learning_rate": 1.0099107117141879e-05, "logits/chosen": -4.3125, "logits/rejected": 0.373046875, "logps/chosen": -3248.0, "logps/rejected": -6432.0, "loss": 0.006, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -28.125, "rewards/margins": 32.5, "rewards/rejected": -60.5, "step": 2100 }, { "epoch": 2.1989528795811517, "eval_logits/chosen": 0.427734375, "eval_logits/rejected": 1.40625, "eval_logps/chosen": -5280.0, "eval_logps/rejected": -4832.0, "eval_loss": 13.616874694824219, "eval_rewards/accuracies": 0.453125, "eval_rewards/chosen": -49.5, "eval_rewards/margins": -4.21875, "eval_rewards/rejected": -45.25, "eval_runtime": 46.6739, "eval_samples_per_second": 42.851, "eval_steps_per_second": 0.686, "step": 2100 }, { "epoch": 2.2094240837696333, "grad_norm": 0.11064477380251503, "learning_rate": 9.855595178057334e-06, "logits/chosen": -4.625, "logits/rejected": 0.322265625, "logps/chosen": -3696.0, "logps/rejected": -7072.0, "loss": 0.0069, "rewards/accuracies": 1.0, "rewards/chosen": -32.25, "rewards/margins": 34.25, "rewards/rejected": -66.5, "step": 2110 }, { "epoch": 2.2198952879581153, "grad_norm": 0.02515522722732333, "learning_rate": 9.614332193917949e-06, "logits/chosen": -4.4375, "logits/rejected": 0.71484375, "logps/chosen": -3744.0, "logps/rejected": -7136.0, "loss": 0.0073, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -32.5, "rewards/margins": 34.25, "rewards/rejected": -67.0, "step": 2120 }, { "epoch": 2.230366492146597, "grad_norm": 0.16592609430180663, "learning_rate": 9.375353992449385e-06, "logits/chosen": -4.75, "logits/rejected": 0.408203125, "logps/chosen": -3616.0, "logps/rejected": -7296.0, "loss": 0.0107, "rewards/accuracies": 1.0, "rewards/chosen": -31.375, "rewards/margins": 37.5, "rewards/rejected": -69.0, "step": 2130 }, { "epoch": 2.2408376963350785, "grad_norm": 0.41363967697912485, "learning_rate": 9.138696062085442e-06, "logits/chosen": -4.90625, "logits/rejected": 0.380859375, "logps/chosen": -3616.0, "logps/rejected": -7264.0, "loss": 0.0108, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -31.625, "rewards/margins": 37.0, "rewards/rejected": -68.5, "step": 2140 }, { "epoch": 2.25130890052356, "grad_norm": 0.062817914691297, "learning_rate": 8.904393546698007e-06, "logits/chosen": -4.5625, "logits/rejected": 0.5546875, "logps/chosen": -3776.0, "logps/rejected": -7424.0, "loss": 0.0051, "rewards/accuracies": 1.0, "rewards/chosen": -33.25, "rewards/margins": 37.0, "rewards/rejected": -70.5, "step": 2150 }, { "epoch": 2.261780104712042, "grad_norm": 0.032751016825559566, "learning_rate": 8.672481240378142e-06, "logits/chosen": -4.78125, "logits/rejected": 0.05126953125, "logps/chosen": -3168.0, "logps/rejected": -6144.0, "loss": 0.0058, "rewards/accuracies": 1.0, "rewards/chosen": -27.5, "rewards/margins": 30.5, "rewards/rejected": -58.0, "step": 2160 }, { "epoch": 2.2722513089005236, "grad_norm": 3.384619711301097, "learning_rate": 8.44299358226919e-06, "logits/chosen": -4.875, "logits/rejected": 0.043701171875, "logps/chosen": -3344.0, "logps/rejected": -6400.0, "loss": 0.0105, "rewards/accuracies": 1.0, "rewards/chosen": -28.75, "rewards/margins": 31.25, "rewards/rejected": -60.0, "step": 2170 }, { "epoch": 2.282722513089005, "grad_norm": 0.052384210423619786, "learning_rate": 8.215964651452457e-06, "logits/chosen": -4.9375, "logits/rejected": 0.107421875, "logps/chosen": -3152.0, "logps/rejected": -6112.0, "loss": 0.0063, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -26.875, "rewards/margins": 30.375, "rewards/rejected": -57.25, "step": 2180 }, { "epoch": 2.2931937172774868, "grad_norm": 0.3999452897331087, "learning_rate": 7.991428161886502e-06, "logits/chosen": -5.25, "logits/rejected": -0.703125, "logps/chosen": -3216.0, "logps/rejected": -6016.0, "loss": 0.0085, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -27.375, "rewards/margins": 28.5, "rewards/rejected": -56.0, "step": 2190 }, { "epoch": 2.303664921465969, "grad_norm": 0.11635565541841181, "learning_rate": 7.76941745740061e-06, "logits/chosen": -5.0, "logits/rejected": -0.453125, "logps/chosen": -3584.0, "logps/rejected": -6624.0, "loss": 0.0119, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -31.0, "rewards/margins": 30.875, "rewards/rejected": -61.75, "step": 2200 }, { "epoch": 2.303664921465969, "eval_logits/chosen": 0.478515625, "eval_logits/rejected": 1.4453125, "eval_logps/chosen": -4896.0, "eval_logps/rejected": -4480.0, "eval_loss": 12.226374626159668, "eval_rewards/accuracies": 0.453125, "eval_rewards/chosen": -45.75, "eval_rewards/margins": -3.984375, "eval_rewards/rejected": -41.75, "eval_runtime": 46.5522, "eval_samples_per_second": 42.963, "eval_steps_per_second": 0.687, "step": 2200 }, { "epoch": 2.3141361256544504, "grad_norm": 0.056604612548133815, "learning_rate": 7.549965506743123e-06, "logits/chosen": -4.25, "logits/rejected": 0.318359375, "logps/chosen": -3344.0, "logps/rejected": -6336.0, "loss": 0.008, "rewards/accuracies": 1.0, "rewards/chosen": -29.0, "rewards/margins": 30.75, "rewards/rejected": -59.5, "step": 2210 }, { "epoch": 2.324607329842932, "grad_norm": 0.09288994863474215, "learning_rate": 7.333104898685631e-06, "logits/chosen": -5.0, "logits/rejected": -0.65625, "logps/chosen": -3104.0, "logps/rejected": -6016.0, "loss": 0.0076, "rewards/accuracies": 1.0, "rewards/chosen": -26.75, "rewards/margins": 29.25, "rewards/rejected": -56.0, "step": 2220 }, { "epoch": 2.3350785340314135, "grad_norm": 0.029507753056812136, "learning_rate": 7.118867837183518e-06, "logits/chosen": -4.3125, "logits/rejected": -0.412109375, "logps/chosen": -3824.0, "logps/rejected": -7104.0, "loss": 0.0087, "rewards/accuracies": 1.0, "rewards/chosen": -33.25, "rewards/margins": 33.25, "rewards/rejected": -66.5, "step": 2230 }, { "epoch": 2.345549738219895, "grad_norm": 0.18895225331830107, "learning_rate": 6.9072861365936045e-06, "logits/chosen": -4.84375, "logits/rejected": -0.40234375, "logps/chosen": -3920.0, "logps/rejected": -7616.0, "loss": 0.0098, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -34.0, "rewards/margins": 37.5, "rewards/rejected": -71.5, "step": 2240 }, { "epoch": 2.356020942408377, "grad_norm": 0.18650349884451983, "learning_rate": 6.6983912169497005e-06, "logits/chosen": -4.4375, "logits/rejected": -0.1298828125, "logps/chosen": -3600.0, "logps/rejected": -7232.0, "loss": 0.0069, "rewards/accuracies": 1.0, "rewards/chosen": -31.25, "rewards/margins": 36.5, "rewards/rejected": -68.0, "step": 2250 }, { "epoch": 2.3664921465968587, "grad_norm": 0.0671185872396581, "learning_rate": 6.49221409929677e-06, "logits/chosen": -4.125, "logits/rejected": 0.412109375, "logps/chosen": -3152.0, "logps/rejected": -6176.0, "loss": 0.0319, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -27.5, "rewards/margins": 30.75, "rewards/rejected": -58.25, "step": 2260 }, { "epoch": 2.3769633507853403, "grad_norm": 0.1754009258071977, "learning_rate": 6.2887854010842075e-06, "logits/chosen": -5.28125, "logits/rejected": -0.3515625, "logps/chosen": -3232.0, "logps/rejected": -6304.0, "loss": 0.0085, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -28.0, "rewards/margins": 31.125, "rewards/rejected": -59.0, "step": 2270 }, { "epoch": 2.387434554973822, "grad_norm": 0.12213889145542803, "learning_rate": 6.0881353316191385e-06, "logits/chosen": -4.9375, "logits/rejected": -0.609375, "logps/chosen": -3344.0, "logps/rejected": -6496.0, "loss": 0.0073, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -28.625, "rewards/margins": 32.0, "rewards/rejected": -60.5, "step": 2280 }, { "epoch": 2.3979057591623034, "grad_norm": 0.024553664996777146, "learning_rate": 5.890293687580381e-06, "logits/chosen": -4.59375, "logits/rejected": -0.310546875, "logps/chosen": -3792.0, "logps/rejected": -6720.0, "loss": 0.0068, "rewards/accuracies": 1.0, "rewards/chosen": -33.5, "rewards/margins": 29.625, "rewards/rejected": -63.0, "step": 2290 }, { "epoch": 2.4083769633507854, "grad_norm": 0.04029870906251963, "learning_rate": 5.695289848593532e-06, "logits/chosen": -5.25, "logits/rejected": -0.458984375, "logps/chosen": -3600.0, "logps/rejected": -6976.0, "loss": 0.0105, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -31.25, "rewards/margins": 34.5, "rewards/rejected": -66.0, "step": 2300 }, { "epoch": 2.4083769633507854, "eval_logits/chosen": 0.45703125, "eval_logits/rejected": 1.40625, "eval_logps/chosen": -5056.0, "eval_logps/rejected": -4608.0, "eval_loss": 12.744000434875488, "eval_rewards/accuracies": 0.453125, "eval_rewards/chosen": -47.5, "eval_rewards/margins": -4.125, "eval_rewards/rejected": -43.25, "eval_runtime": 46.6046, "eval_samples_per_second": 42.914, "eval_steps_per_second": 0.687, "step": 2300 }, { "epoch": 2.418848167539267, "grad_norm": 0.042509990755605846, "learning_rate": 5.50315277286812e-06, "logits/chosen": -5.625, "logits/rejected": -0.416015625, "logps/chosen": -3232.0, "logps/rejected": -6528.0, "loss": 0.0059, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -27.875, "rewards/margins": 33.5, "rewards/rejected": -61.5, "step": 2310 }, { "epoch": 2.4293193717277486, "grad_norm": 0.09728023515762632, "learning_rate": 5.313910992897281e-06, "logits/chosen": -4.1875, "logits/rejected": 0.1513671875, "logps/chosen": -3312.0, "logps/rejected": -6368.0, "loss": 0.0048, "rewards/accuracies": 1.0, "rewards/chosen": -28.375, "rewards/margins": 31.0, "rewards/rejected": -59.5, "step": 2320 }, { "epoch": 2.4397905759162306, "grad_norm": 0.06507825993198593, "learning_rate": 5.127592611220658e-06, "logits/chosen": -4.5, "logits/rejected": -0.11328125, "logps/chosen": -3968.0, "logps/rejected": -7264.0, "loss": 0.0146, "rewards/accuracies": 0.9624999761581421, "rewards/chosen": -34.75, "rewards/margins": 33.5, "rewards/rejected": -68.5, "step": 2330 }, { "epoch": 2.450261780104712, "grad_norm": 0.024802189784689375, "learning_rate": 4.944225296251159e-06, "logits/chosen": -4.875, "logits/rejected": -0.044189453125, "logps/chosen": -3376.0, "logps/rejected": -6656.0, "loss": 0.0047, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -29.375, "rewards/margins": 33.25, "rewards/rejected": -62.75, "step": 2340 }, { "epoch": 2.4607329842931938, "grad_norm": 0.1467476054116497, "learning_rate": 4.763836278166172e-06, "logits/chosen": -4.28125, "logits/rejected": 0.447265625, "logps/chosen": -3408.0, "logps/rejected": -6720.0, "loss": 0.0086, "rewards/accuracies": 1.0, "rewards/chosen": -29.5, "rewards/margins": 33.75, "rewards/rejected": -63.25, "step": 2350 }, { "epoch": 2.4712041884816753, "grad_norm": 0.01406081652511781, "learning_rate": 4.586452344863876e-06, "logits/chosen": -4.28125, "logits/rejected": 0.34375, "logps/chosen": -3664.0, "logps/rejected": -7232.0, "loss": 0.0063, "rewards/accuracies": 1.0, "rewards/chosen": -31.875, "rewards/margins": 36.25, "rewards/rejected": -68.0, "step": 2360 }, { "epoch": 2.481675392670157, "grad_norm": 0.25677748957659263, "learning_rate": 4.412099837985192e-06, "logits/chosen": -4.625, "logits/rejected": 0.431640625, "logps/chosen": -3488.0, "logps/rejected": -6976.0, "loss": 0.0097, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -30.125, "rewards/margins": 35.25, "rewards/rejected": -65.5, "step": 2370 }, { "epoch": 2.492146596858639, "grad_norm": 0.07542759531269545, "learning_rate": 4.240804649002089e-06, "logits/chosen": -4.5625, "logits/rejected": 0.5234375, "logps/chosen": -3552.0, "logps/rejected": -7104.0, "loss": 0.0071, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -31.375, "rewards/margins": 36.0, "rewards/rejected": -67.5, "step": 2380 }, { "epoch": 2.5026178010471205, "grad_norm": 0.07232832893871259, "learning_rate": 4.072592215372598e-06, "logits/chosen": -5.125, "logits/rejected": -0.56640625, "logps/chosen": -3696.0, "logps/rejected": -7136.0, "loss": 0.0056, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -32.0, "rewards/margins": 35.25, "rewards/rejected": -67.0, "step": 2390 }, { "epoch": 2.513089005235602, "grad_norm": 0.05292714902356548, "learning_rate": 3.907487516763389e-06, "logits/chosen": -4.9375, "logits/rejected": 0.0673828125, "logps/chosen": -3360.0, "logps/rejected": -6944.0, "loss": 0.0077, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -29.25, "rewards/margins": 36.25, "rewards/rejected": -65.5, "step": 2400 }, { "epoch": 2.513089005235602, "eval_logits/chosen": 0.8125, "eval_logits/rejected": 1.765625, "eval_logps/chosen": -5344.0, "eval_logps/rejected": -4864.0, "eval_loss": 13.484375, "eval_rewards/accuracies": 0.451171875, "eval_rewards/chosen": -50.25, "eval_rewards/margins": -4.3125, "eval_rewards/rejected": -45.75, "eval_runtime": 46.5606, "eval_samples_per_second": 42.955, "eval_steps_per_second": 0.687, "step": 2400 }, { "epoch": 2.5235602094240837, "grad_norm": 0.08588599958598957, "learning_rate": 3.745515071340286e-06, "logits/chosen": -4.5, "logits/rejected": -0.00714111328125, "logps/chosen": -3632.0, "logps/rejected": -7008.0, "loss": 0.0078, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -31.625, "rewards/margins": 34.25, "rewards/rejected": -66.0, "step": 2410 }, { "epoch": 2.5340314136125652, "grad_norm": 0.0462499396019616, "learning_rate": 3.586698932127236e-06, "logits/chosen": -4.34375, "logits/rejected": -0.0230712890625, "logps/chosen": -3744.0, "logps/rejected": -7328.0, "loss": 0.0061, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -32.75, "rewards/margins": 36.5, "rewards/rejected": -69.0, "step": 2420 }, { "epoch": 2.5445026178010473, "grad_norm": 0.06028596671517465, "learning_rate": 3.4310626834344744e-06, "logits/chosen": -4.65625, "logits/rejected": 0.2470703125, "logps/chosen": -4000.0, "logps/rejected": -8192.0, "loss": 0.0038, "rewards/accuracies": 1.0, "rewards/chosen": -35.25, "rewards/margins": 42.5, "rewards/rejected": -77.5, "step": 2430 }, { "epoch": 2.554973821989529, "grad_norm": 0.44321119124761416, "learning_rate": 3.2786294373562343e-06, "logits/chosen": -4.65625, "logits/rejected": 0.1513671875, "logps/chosen": -3632.0, "logps/rejected": -7200.0, "loss": 0.0057, "rewards/accuracies": 1.0, "rewards/chosen": -31.625, "rewards/margins": 36.25, "rewards/rejected": -68.0, "step": 2440 }, { "epoch": 2.5654450261780104, "grad_norm": 0.0890498870087265, "learning_rate": 3.1294218303385526e-06, "logits/chosen": -4.53125, "logits/rejected": 0.5390625, "logps/chosen": -3728.0, "logps/rejected": -7552.0, "loss": 0.0048, "rewards/accuracies": 1.0, "rewards/chosen": -32.5, "rewards/margins": 38.75, "rewards/rejected": -71.0, "step": 2450 }, { "epoch": 2.5759162303664924, "grad_norm": 0.15344716610491824, "learning_rate": 2.9834620198177483e-06, "logits/chosen": -4.5625, "logits/rejected": 0.1142578125, "logps/chosen": -3504.0, "logps/rejected": -6720.0, "loss": 0.0086, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -30.375, "rewards/margins": 32.75, "rewards/rejected": -63.0, "step": 2460 }, { "epoch": 2.5863874345549736, "grad_norm": 0.07560438224025691, "learning_rate": 2.840771680930068e-06, "logits/chosen": -4.6875, "logits/rejected": 0.328125, "logps/chosen": -4192.0, "logps/rejected": -7968.0, "loss": 0.0016, "rewards/accuracies": 1.0, "rewards/chosen": -36.75, "rewards/margins": 38.5, "rewards/rejected": -75.5, "step": 2470 }, { "epoch": 2.5968586387434556, "grad_norm": 0.12234132796627159, "learning_rate": 2.701372003292868e-06, "logits/chosen": -3.96875, "logits/rejected": 0.87890625, "logps/chosen": -3616.0, "logps/rejected": -6912.0, "loss": 0.0073, "rewards/accuracies": 1.0, "rewards/chosen": -31.75, "rewards/margins": 33.5, "rewards/rejected": -65.0, "step": 2480 }, { "epoch": 2.607329842931937, "grad_norm": 0.002028804484259438, "learning_rate": 2.56528368785795e-06, "logits/chosen": -4.78125, "logits/rejected": -0.041015625, "logps/chosen": -3520.0, "logps/rejected": -7136.0, "loss": 0.0052, "rewards/accuracies": 1.0, "rewards/chosen": -30.5, "rewards/margins": 36.5, "rewards/rejected": -67.0, "step": 2490 }, { "epoch": 2.6178010471204187, "grad_norm": 0.1051261111139172, "learning_rate": 2.432526943837471e-06, "logits/chosen": -4.46875, "logits/rejected": 0.71875, "logps/chosen": -3424.0, "logps/rejected": -7072.0, "loss": 0.0149, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -29.875, "rewards/margins": 37.0, "rewards/rejected": -67.0, "step": 2500 }, { "epoch": 2.6178010471204187, "eval_logits/chosen": 0.703125, "eval_logits/rejected": 1.65625, "eval_logps/chosen": -5408.0, "eval_logps/rejected": -4960.0, "eval_loss": 13.776000022888184, "eval_rewards/accuracies": 0.455078125, "eval_rewards/chosen": -51.0, "eval_rewards/margins": -4.34375, "eval_rewards/rejected": -46.75, "eval_runtime": 46.6844, "eval_samples_per_second": 42.841, "eval_steps_per_second": 0.685, "step": 2500 }, { "epoch": 2.6282722513089007, "grad_norm": 0.36708285379249034, "learning_rate": 2.3031214857028446e-06, "logits/chosen": -4.25, "logits/rejected": 0.640625, "logps/chosen": -3600.0, "logps/rejected": -7008.0, "loss": 0.0083, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -31.5, "rewards/margins": 34.25, "rewards/rejected": -65.5, "step": 2510 }, { "epoch": 2.6387434554973823, "grad_norm": 0.6262748191189448, "learning_rate": 2.177086530257139e-06, "logits/chosen": -4.53125, "logits/rejected": 0.056396484375, "logps/chosen": -3728.0, "logps/rejected": -7200.0, "loss": 0.0096, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -32.75, "rewards/margins": 35.25, "rewards/rejected": -68.0, "step": 2520 }, { "epoch": 2.649214659685864, "grad_norm": 0.0217333391939512, "learning_rate": 2.054440793781354e-06, "logits/chosen": -4.40625, "logits/rejected": 0.26171875, "logps/chosen": -3616.0, "logps/rejected": -7200.0, "loss": 0.0074, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -31.5, "rewards/margins": 36.5, "rewards/rejected": -68.0, "step": 2530 }, { "epoch": 2.6596858638743455, "grad_norm": 0.02563773247160278, "learning_rate": 1.9352024892550587e-06, "logits/chosen": -4.84375, "logits/rejected": -0.1025390625, "logps/chosen": -3632.0, "logps/rejected": -7296.0, "loss": 0.0039, "rewards/accuracies": 1.0, "rewards/chosen": -31.75, "rewards/margins": 37.0, "rewards/rejected": -68.5, "step": 2540 }, { "epoch": 2.670157068062827, "grad_norm": 0.010736953300293634, "learning_rate": 1.8193893236517379e-06, "logits/chosen": -4.875, "logits/rejected": -0.255859375, "logps/chosen": -3856.0, "logps/rejected": -7232.0, "loss": 0.0057, "rewards/accuracies": 1.0, "rewards/chosen": -33.75, "rewards/margins": 34.25, "rewards/rejected": -68.0, "step": 2550 }, { "epoch": 2.680628272251309, "grad_norm": 0.00958637601339211, "learning_rate": 1.7070184953092956e-06, "logits/chosen": -4.53125, "logits/rejected": 0.244140625, "logps/chosen": -3456.0, "logps/rejected": -7392.0, "loss": 0.0044, "rewards/accuracies": 1.0, "rewards/chosen": -29.75, "rewards/margins": 40.0, "rewards/rejected": -69.5, "step": 2560 }, { "epoch": 2.6910994764397906, "grad_norm": 0.18254622234870518, "learning_rate": 1.5981066913760918e-06, "logits/chosen": -4.3125, "logits/rejected": 0.61328125, "logps/chosen": -3200.0, "logps/rejected": -6176.0, "loss": 0.0064, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -27.5, "rewards/margins": 30.625, "rewards/rejected": -58.0, "step": 2570 }, { "epoch": 2.701570680628272, "grad_norm": 0.10684248348789567, "learning_rate": 1.492670085332884e-06, "logits/chosen": -4.40625, "logits/rejected": 0.177734375, "logps/chosen": -3696.0, "logps/rejected": -7488.0, "loss": 0.0101, "rewards/accuracies": 1.0, "rewards/chosen": -32.25, "rewards/margins": 38.25, "rewards/rejected": -70.5, "step": 2580 }, { "epoch": 2.712041884816754, "grad_norm": 0.07659535342093675, "learning_rate": 1.3907243345910787e-06, "logits/chosen": -3.640625, "logits/rejected": 0.7890625, "logps/chosen": -3824.0, "logps/rejected": -7264.0, "loss": 0.0059, "rewards/accuracies": 1.0, "rewards/chosen": -33.5, "rewards/margins": 34.75, "rewards/rejected": -68.0, "step": 2590 }, { "epoch": 2.7225130890052354, "grad_norm": 0.0534004130829911, "learning_rate": 1.292284578167538e-06, "logits/chosen": -4.5, "logits/rejected": 0.10498046875, "logps/chosen": -3472.0, "logps/rejected": -6848.0, "loss": 0.0045, "rewards/accuracies": 1.0, "rewards/chosen": -30.25, "rewards/margins": 34.25, "rewards/rejected": -64.5, "step": 2600 }, { "epoch": 2.7225130890052354, "eval_logits/chosen": 1.0078125, "eval_logits/rejected": 1.9765625, "eval_logps/chosen": -5600.0, "eval_logps/rejected": -5120.0, "eval_loss": 14.25837516784668, "eval_rewards/accuracies": 0.455078125, "eval_rewards/chosen": -52.75, "eval_rewards/margins": -4.5, "eval_rewards/rejected": -48.25, "eval_runtime": 46.6896, "eval_samples_per_second": 42.836, "eval_steps_per_second": 0.685, "step": 2600 }, { "epoch": 2.7329842931937174, "grad_norm": 0.2484074056879596, "learning_rate": 1.1973654344364925e-06, "logits/chosen": -4.3125, "logits/rejected": 0.5390625, "logps/chosen": -4048.0, "logps/rejected": -7904.0, "loss": 0.01, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -35.75, "rewards/margins": 39.0, "rewards/rejected": -75.0, "step": 2610 }, { "epoch": 2.743455497382199, "grad_norm": 1.7017548803889362, "learning_rate": 1.1059809989586583e-06, "logits/chosen": -3.96875, "logits/rejected": 1.0625, "logps/chosen": -3616.0, "logps/rejected": -7232.0, "loss": 0.0062, "rewards/accuracies": 1.0, "rewards/chosen": -31.25, "rewards/margins": 37.0, "rewards/rejected": -68.0, "step": 2620 }, { "epoch": 2.7539267015706805, "grad_norm": 0.0815000755620004, "learning_rate": 1.018144842388044e-06, "logits/chosen": -3.6875, "logits/rejected": 0.68359375, "logps/chosen": -3696.0, "logps/rejected": -7072.0, "loss": 0.0108, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -32.25, "rewards/margins": 34.0, "rewards/rejected": -66.5, "step": 2630 }, { "epoch": 2.7643979057591626, "grad_norm": 0.3569594594546677, "learning_rate": 9.338700084567109e-07, "logits/chosen": -4.46875, "logits/rejected": 0.1435546875, "logps/chosen": -3632.0, "logps/rejected": -6912.0, "loss": 0.0056, "rewards/accuracies": 1.0, "rewards/chosen": -31.75, "rewards/margins": 33.5, "rewards/rejected": -65.5, "step": 2640 }, { "epoch": 2.774869109947644, "grad_norm": 0.02121476138473284, "learning_rate": 8.531690120377606e-07, "logits/chosen": -3.796875, "logits/rejected": 0.59375, "logps/chosen": -4288.0, "logps/rejected": -8000.0, "loss": 0.0131, "rewards/accuracies": 1.0, "rewards/chosen": -37.75, "rewards/margins": 37.75, "rewards/rejected": -75.5, "step": 2650 }, { "epoch": 2.7853403141361257, "grad_norm": 0.04290212582120115, "learning_rate": 7.760538372868636e-07, "logits/chosen": -4.4375, "logits/rejected": 0.11181640625, "logps/chosen": -3680.0, "logps/rejected": -7616.0, "loss": 0.0081, "rewards/accuracies": 1.0, "rewards/chosen": -32.25, "rewards/margins": 40.0, "rewards/rejected": -72.0, "step": 2660 }, { "epoch": 2.7958115183246073, "grad_norm": 0.04763046415405345, "learning_rate": 7.025359358626166e-07, "logits/chosen": -3.828125, "logits/rejected": 0.7265625, "logps/chosen": -4128.0, "logps/rejected": -7936.0, "loss": 0.0046, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -36.5, "rewards/margins": 38.75, "rewards/rejected": -75.0, "step": 2670 }, { "epoch": 2.806282722513089, "grad_norm": 0.34310917424273324, "learning_rate": 6.326262252259463e-07, "logits/chosen": -4.53125, "logits/rejected": -0.119140625, "logps/chosen": -3600.0, "logps/rejected": -6624.0, "loss": 0.0076, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -31.5, "rewards/margins": 30.625, "rewards/rejected": -62.0, "step": 2680 }, { "epoch": 2.816753926701571, "grad_norm": 0.25278642430909287, "learning_rate": 5.66335087018871e-07, "logits/chosen": -4.625, "logits/rejected": 0.373046875, "logps/chosen": -3904.0, "logps/rejected": -7520.0, "loss": 0.0065, "rewards/accuracies": 1.0, "rewards/chosen": -34.25, "rewards/margins": 36.5, "rewards/rejected": -71.0, "step": 2690 }, { "epoch": 2.8272251308900525, "grad_norm": 0.06974410770068638, "learning_rate": 5.036723655228226e-07, "logits/chosen": -4.25, "logits/rejected": 0.271484375, "logps/chosen": -3584.0, "logps/rejected": -7136.0, "loss": 0.0105, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -31.625, "rewards/margins": 35.75, "rewards/rejected": -67.5, "step": 2700 }, { "epoch": 2.8272251308900525, "eval_logits/chosen": 0.8515625, "eval_logits/rejected": 1.8203125, "eval_logps/chosen": -5472.0, "eval_logps/rejected": -4992.0, "eval_loss": 13.871999740600586, "eval_rewards/accuracies": 0.455078125, "eval_rewards/chosen": -51.5, "eval_rewards/margins": -4.375, "eval_rewards/rejected": -47.0, "eval_runtime": 46.6798, "eval_samples_per_second": 42.845, "eval_steps_per_second": 0.686, "step": 2700 }, { "epoch": 2.837696335078534, "grad_norm": 0.039987558113219336, "learning_rate": 4.446473661967432e-07, "logits/chosen": -4.1875, "logits/rejected": 0.5390625, "logps/chosen": -3808.0, "logps/rejected": -7456.0, "loss": 0.0108, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -33.5, "rewards/margins": 37.0, "rewards/rejected": -70.5, "step": 2710 }, { "epoch": 2.8481675392670156, "grad_norm": 0.12978954737477066, "learning_rate": 3.892688542952411e-07, "logits/chosen": -4.09375, "logits/rejected": 0.55078125, "logps/chosen": -3728.0, "logps/rejected": -7136.0, "loss": 0.0102, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -33.0, "rewards/margins": 34.5, "rewards/rejected": -67.5, "step": 2720 }, { "epoch": 2.858638743455497, "grad_norm": 0.038811231130778424, "learning_rate": 3.3754505356694e-07, "logits/chosen": -5.375, "logits/rejected": -0.3125, "logps/chosen": -3584.0, "logps/rejected": -7488.0, "loss": 0.0076, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -30.875, "rewards/margins": 39.5, "rewards/rejected": -70.5, "step": 2730 }, { "epoch": 2.869109947643979, "grad_norm": 0.11521370259821771, "learning_rate": 2.8948364503322276e-07, "logits/chosen": -4.84375, "logits/rejected": 0.291015625, "logps/chosen": -3824.0, "logps/rejected": -7616.0, "loss": 0.0099, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -33.75, "rewards/margins": 38.5, "rewards/rejected": -72.0, "step": 2740 }, { "epoch": 2.8795811518324608, "grad_norm": 0.047203707815717054, "learning_rate": 2.4509176584762706e-07, "logits/chosen": -4.5, "logits/rejected": 0.134765625, "logps/chosen": -3696.0, "logps/rejected": -7232.0, "loss": 0.0079, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -32.25, "rewards/margins": 36.0, "rewards/rejected": -68.5, "step": 2750 }, { "epoch": 2.8900523560209423, "grad_norm": 0.42433661818647306, "learning_rate": 2.0437600823595694e-07, "logits/chosen": -4.9375, "logits/rejected": -0.474609375, "logps/chosen": -3552.0, "logps/rejected": -7136.0, "loss": 0.0044, "rewards/accuracies": 1.0, "rewards/chosen": -30.625, "rewards/margins": 36.25, "rewards/rejected": -67.0, "step": 2760 }, { "epoch": 2.900523560209424, "grad_norm": 0.09063818632092174, "learning_rate": 1.6734241851733e-07, "logits/chosen": -4.40625, "logits/rejected": 0.388671875, "logps/chosen": -4160.0, "logps/rejected": -7680.0, "loss": 0.0067, "rewards/accuracies": 1.0, "rewards/chosen": -36.75, "rewards/margins": 35.75, "rewards/rejected": -72.5, "step": 2770 }, { "epoch": 2.9109947643979055, "grad_norm": 0.2664527034765327, "learning_rate": 1.339964962062984e-07, "logits/chosen": -4.625, "logits/rejected": 0.3515625, "logps/chosen": -4128.0, "logps/rejected": -7584.0, "loss": 0.0044, "rewards/accuracies": 1.0, "rewards/chosen": -36.5, "rewards/margins": 35.5, "rewards/rejected": -72.0, "step": 2780 }, { "epoch": 2.9214659685863875, "grad_norm": 0.09844660199032843, "learning_rate": 1.0434319319617136e-07, "logits/chosen": -3.796875, "logits/rejected": 0.65234375, "logps/chosen": -3600.0, "logps/rejected": -7232.0, "loss": 0.0055, "rewards/accuracies": 1.0, "rewards/chosen": -31.5, "rewards/margins": 36.5, "rewards/rejected": -68.0, "step": 2790 }, { "epoch": 2.931937172774869, "grad_norm": 0.041672234115281474, "learning_rate": 7.838691302365363e-08, "logits/chosen": -4.0625, "logits/rejected": 0.03076171875, "logps/chosen": -3488.0, "logps/rejected": -7232.0, "loss": 0.0065, "rewards/accuracies": 1.0, "rewards/chosen": -30.5, "rewards/margins": 37.75, "rewards/rejected": -68.0, "step": 2800 }, { "epoch": 2.931937172774869, "eval_logits/chosen": 0.84765625, "eval_logits/rejected": 1.8125, "eval_logps/chosen": -5440.0, "eval_logps/rejected": -4960.0, "eval_loss": 13.76099967956543, "eval_rewards/accuracies": 0.45703125, "eval_rewards/chosen": -51.0, "eval_rewards/margins": -4.34375, "eval_rewards/rejected": -46.75, "eval_runtime": 46.6762, "eval_samples_per_second": 42.848, "eval_steps_per_second": 0.686, "step": 2800 }, { "epoch": 2.9424083769633507, "grad_norm": 0.019161353008805088, "learning_rate": 5.613151021490459e-08, "logits/chosen": -4.3125, "logits/rejected": 0.423828125, "logps/chosen": -3168.0, "logps/rejected": -6048.0, "loss": 0.0057, "rewards/accuracies": 1.0, "rewards/chosen": -27.75, "rewards/margins": 29.25, "rewards/rejected": -57.0, "step": 2810 }, { "epoch": 2.9528795811518327, "grad_norm": 0.03575388722075906, "learning_rate": 3.7580289713159965e-08, "logits/chosen": -4.15625, "logits/rejected": 0.291015625, "logps/chosen": -3632.0, "logps/rejected": -7392.0, "loss": 0.0091, "rewards/accuracies": 1.0, "rewards/chosen": -31.5, "rewards/margins": 38.0, "rewards/rejected": -69.5, "step": 2820 }, { "epoch": 2.9633507853403143, "grad_norm": 0.227898927708421, "learning_rate": 2.2736006387927277e-08, "logits/chosen": -4.21875, "logits/rejected": 0.5546875, "logps/chosen": -4096.0, "logps/rejected": -7808.0, "loss": 0.0062, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -36.25, "rewards/margins": 37.5, "rewards/rejected": -74.0, "step": 2830 }, { "epoch": 2.973821989528796, "grad_norm": 0.32810234899558244, "learning_rate": 1.1600864625893604e-08, "logits/chosen": -4.40625, "logits/rejected": 0.0208740234375, "logps/chosen": -3792.0, "logps/rejected": -7584.0, "loss": 0.0078, "rewards/accuracies": 1.0, "rewards/chosen": -33.25, "rewards/margins": 38.25, "rewards/rejected": -71.5, "step": 2840 }, { "epoch": 2.9842931937172774, "grad_norm": 0.009283058224114698, "learning_rate": 4.176518003573549e-09, "logits/chosen": -4.125, "logits/rejected": -0.099609375, "logps/chosen": -4064.0, "logps/rejected": -7296.0, "loss": 0.0111, "rewards/accuracies": 1.0, "rewards/chosen": -36.0, "rewards/margins": 32.75, "rewards/rejected": -68.5, "step": 2850 }, { "epoch": 2.994764397905759, "grad_norm": 0.08156074334653951, "learning_rate": 4.640690417528903e-10, "logits/chosen": -4.6875, "logits/rejected": 0.34765625, "logps/chosen": -3520.0, "logps/rejected": -7360.0, "loss": 0.0141, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -30.875, "rewards/margins": 38.75, "rewards/rejected": -69.5, "step": 2860 }, { "epoch": 3.0, "step": 2865, "total_flos": 0.0, "train_loss": 0.06252696459388025, "train_runtime": 12308.2002, "train_samples_per_second": 14.897, "train_steps_per_second": 0.233 } ], "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 }