{ "best_metric": null, "best_model_checkpoint": null, "epoch": 4.99738356881214, "eval_steps": 100, "global_step": 4775, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0010465724751439038, "grad_norm": 1.5610575235354534, "learning_rate": 1.0460251046025104e-07, "logits/chosen": -9.375, "logits/rejected": -9.5, "logps/chosen": -448.0, "logps/rejected": -398.0, "loss": 0.6914, "rewards/accuracies": 0.0, "rewards/chosen": 0.0, "rewards/margins": 0.0, "rewards/rejected": 0.0, "step": 1 }, { "epoch": 0.010465724751439037, "grad_norm": 1.349006948243697, "learning_rate": 1.0460251046025104e-06, "logits/chosen": -9.5, "logits/rejected": -9.625, "logps/chosen": -296.0, "logps/rejected": -264.0, "loss": 0.6925, "rewards/accuracies": 0.2083333283662796, "rewards/chosen": 0.0012054443359375, "rewards/margins": 0.000579833984375, "rewards/rejected": 0.0006256103515625, "step": 10 }, { "epoch": 0.020931449502878074, "grad_norm": 1.5264399374921611, "learning_rate": 2.092050209205021e-06, "logits/chosen": -9.5, "logits/rejected": -9.4375, "logps/chosen": -324.0, "logps/rejected": -310.0, "loss": 0.6918, "rewards/accuracies": 0.3187499940395355, "rewards/chosen": 0.00299072265625, "rewards/margins": 0.003204345703125, "rewards/rejected": -0.00020885467529296875, "step": 20 }, { "epoch": 0.03139717425431711, "grad_norm": 1.583736957016309, "learning_rate": 3.1380753138075313e-06, "logits/chosen": -9.6875, "logits/rejected": -9.5, "logps/chosen": -314.0, "logps/rejected": -310.0, "loss": 0.6908, "rewards/accuracies": 0.39375001192092896, "rewards/chosen": 0.00579833984375, "rewards/margins": 0.001678466796875, "rewards/rejected": 0.00408935546875, "step": 30 }, { "epoch": 0.04186289900575615, "grad_norm": 1.4892011169858126, "learning_rate": 4.184100418410042e-06, "logits/chosen": -9.5625, "logits/rejected": -9.8125, "logps/chosen": -334.0, "logps/rejected": -308.0, "loss": 0.6868, "rewards/accuracies": 0.5062500238418579, "rewards/chosen": 0.00885009765625, "rewards/margins": 0.0108642578125, "rewards/rejected": -0.0019683837890625, "step": 40 }, { "epoch": 0.052328623757195186, "grad_norm": 1.3871844092880496, "learning_rate": 5.230125523012553e-06, "logits/chosen": -9.5625, "logits/rejected": -9.5, "logps/chosen": -316.0, "logps/rejected": -270.0, "loss": 0.6818, "rewards/accuracies": 0.5562499761581421, "rewards/chosen": -0.040771484375, "rewards/margins": 0.0247802734375, "rewards/rejected": -0.0654296875, "step": 50 }, { "epoch": 0.06279434850863422, "grad_norm": 1.5257182239483598, "learning_rate": 6.276150627615063e-06, "logits/chosen": -9.0625, "logits/rejected": -9.0, "logps/chosen": -364.0, "logps/rejected": -324.0, "loss": 0.6721, "rewards/accuracies": 0.643750011920929, "rewards/chosen": -0.11474609375, "rewards/margins": 0.049072265625, "rewards/rejected": -0.1640625, "step": 60 }, { "epoch": 0.07326007326007326, "grad_norm": 1.5659951271442103, "learning_rate": 7.3221757322175736e-06, "logits/chosen": -9.875, "logits/rejected": -9.5, "logps/chosen": -326.0, "logps/rejected": -320.0, "loss": 0.6597, "rewards/accuracies": 0.6187499761581421, "rewards/chosen": -0.2734375, "rewards/margins": 0.0859375, "rewards/rejected": -0.359375, "step": 70 }, { "epoch": 0.0837257980115123, "grad_norm": 1.8241831527626855, "learning_rate": 8.368200836820084e-06, "logits/chosen": -9.6875, "logits/rejected": -10.125, "logps/chosen": -360.0, "logps/rejected": -328.0, "loss": 0.6419, "rewards/accuracies": 0.699999988079071, "rewards/chosen": -0.32421875, "rewards/margins": 0.166015625, "rewards/rejected": -0.490234375, "step": 80 }, { "epoch": 0.09419152276295134, "grad_norm": 1.6581722991160308, "learning_rate": 9.414225941422594e-06, "logits/chosen": -9.9375, "logits/rejected": -9.9375, "logps/chosen": -348.0, "logps/rejected": -364.0, "loss": 0.6405, "rewards/accuracies": 0.59375, "rewards/chosen": -0.6015625, "rewards/margins": 0.0888671875, "rewards/rejected": -0.69140625, "step": 90 }, { "epoch": 0.10465724751439037, "grad_norm": 1.9270010320276483, "learning_rate": 1.0460251046025105e-05, "logits/chosen": -8.75, "logits/rejected": -8.625, "logps/chosen": -416.0, "logps/rejected": -402.0, "loss": 0.6232, "rewards/accuracies": 0.6812499761581421, "rewards/chosen": -0.6875, "rewards/margins": 0.25, "rewards/rejected": -0.9375, "step": 100 }, { "epoch": 0.10465724751439037, "eval_logits/chosen": -8.5, "eval_logits/rejected": -8.25, "eval_logps/chosen": -374.0, "eval_logps/rejected": -368.0, "eval_loss": 0.6285156011581421, "eval_rewards/accuracies": 0.666015625, "eval_rewards/chosen": -0.60546875, "eval_rewards/margins": 0.2177734375, "eval_rewards/rejected": -0.82421875, "eval_runtime": 53.3726, "eval_samples_per_second": 37.472, "eval_steps_per_second": 0.6, "step": 100 }, { "epoch": 0.1151229722658294, "grad_norm": 1.8649818778865919, "learning_rate": 1.1506276150627615e-05, "logits/chosen": -8.4375, "logits/rejected": -8.3125, "logps/chosen": -352.0, "logps/rejected": -360.0, "loss": 0.6325, "rewards/accuracies": 0.6312500238418579, "rewards/chosen": -0.703125, "rewards/margins": 0.21484375, "rewards/rejected": -0.91796875, "step": 110 }, { "epoch": 0.12558869701726844, "grad_norm": 3.1874202308933772, "learning_rate": 1.2552301255230125e-05, "logits/chosen": -9.0, "logits/rejected": -8.6875, "logps/chosen": -386.0, "logps/rejected": -426.0, "loss": 0.6012, "rewards/accuracies": 0.6875, "rewards/chosen": -0.78125, "rewards/margins": 0.2734375, "rewards/rejected": -1.0546875, "step": 120 }, { "epoch": 0.1360544217687075, "grad_norm": 2.8031812157693743, "learning_rate": 1.3598326359832636e-05, "logits/chosen": -8.75, "logits/rejected": -8.5, "logps/chosen": -452.0, "logps/rejected": -452.0, "loss": 0.6069, "rewards/accuracies": 0.6812499761581421, "rewards/chosen": -1.21875, "rewards/margins": 0.435546875, "rewards/rejected": -1.65625, "step": 130 }, { "epoch": 0.14652014652014653, "grad_norm": 2.197923048367805, "learning_rate": 1.4644351464435147e-05, "logits/chosen": -10.1875, "logits/rejected": -10.0625, "logps/chosen": -400.0, "logps/rejected": -412.0, "loss": 0.5996, "rewards/accuracies": 0.6812499761581421, "rewards/chosen": -0.65625, "rewards/margins": 0.357421875, "rewards/rejected": -1.015625, "step": 140 }, { "epoch": 0.15698587127158556, "grad_norm": 1.9425267890042368, "learning_rate": 1.5690376569037658e-05, "logits/chosen": -11.4375, "logits/rejected": -11.0625, "logps/chosen": -418.0, "logps/rejected": -402.0, "loss": 0.6035, "rewards/accuracies": 0.675000011920929, "rewards/chosen": -0.8359375, "rewards/margins": 0.373046875, "rewards/rejected": -1.2109375, "step": 150 }, { "epoch": 0.1674515960230246, "grad_norm": 2.6369764941319116, "learning_rate": 1.6736401673640167e-05, "logits/chosen": -11.625, "logits/rejected": -11.375, "logps/chosen": -440.0, "logps/rejected": -418.0, "loss": 0.5842, "rewards/accuracies": 0.699999988079071, "rewards/chosen": -0.92578125, "rewards/margins": 0.43359375, "rewards/rejected": -1.359375, "step": 160 }, { "epoch": 0.17791732077446362, "grad_norm": 2.206623859417659, "learning_rate": 1.7782426778242676e-05, "logits/chosen": -10.0625, "logits/rejected": -9.5625, "logps/chosen": -470.0, "logps/rejected": -480.0, "loss": 0.5908, "rewards/accuracies": 0.6812499761581421, "rewards/chosen": -1.5546875, "rewards/margins": 0.55078125, "rewards/rejected": -2.109375, "step": 170 }, { "epoch": 0.18838304552590268, "grad_norm": 2.3093801072587725, "learning_rate": 1.882845188284519e-05, "logits/chosen": -6.21875, "logits/rejected": -5.53125, "logps/chosen": -426.0, "logps/rejected": -470.0, "loss": 0.5717, "rewards/accuracies": 0.675000011920929, "rewards/chosen": -1.4296875, "rewards/margins": 0.5546875, "rewards/rejected": -1.984375, "step": 180 }, { "epoch": 0.1988487702773417, "grad_norm": 2.4317513580230186, "learning_rate": 1.9874476987447698e-05, "logits/chosen": -4.8125, "logits/rejected": -4.4375, "logps/chosen": -434.0, "logps/rejected": -480.0, "loss": 0.5755, "rewards/accuracies": 0.6937500238418579, "rewards/chosen": -1.5859375, "rewards/margins": 0.4453125, "rewards/rejected": -2.03125, "step": 190 }, { "epoch": 0.20931449502878074, "grad_norm": 2.5804173107312147, "learning_rate": 2.092050209205021e-05, "logits/chosen": -6.9375, "logits/rejected": -6.4375, "logps/chosen": -516.0, "logps/rejected": -490.0, "loss": 0.5729, "rewards/accuracies": 0.675000011920929, "rewards/chosen": -1.5625, "rewards/margins": 0.51171875, "rewards/rejected": -2.078125, "step": 200 }, { "epoch": 0.20931449502878074, "eval_logits/chosen": -8.4375, "eval_logits/rejected": -7.96875, "eval_logps/chosen": -478.0, "eval_logps/rejected": -498.0, "eval_loss": 0.595703125, "eval_rewards/accuracies": 0.69921875, "eval_rewards/chosen": -1.6328125, "eval_rewards/margins": 0.4765625, "eval_rewards/rejected": -2.109375, "eval_runtime": 52.7631, "eval_samples_per_second": 37.905, "eval_steps_per_second": 0.606, "step": 200 }, { "epoch": 0.21978021978021978, "grad_norm": 2.028055345355448, "learning_rate": 2.196652719665272e-05, "logits/chosen": -9.25, "logits/rejected": -8.8125, "logps/chosen": -504.0, "logps/rejected": -512.0, "loss": 0.616, "rewards/accuracies": 0.6499999761581421, "rewards/chosen": -1.765625, "rewards/margins": 0.283203125, "rewards/rejected": -2.046875, "step": 210 }, { "epoch": 0.2302459445316588, "grad_norm": 2.4926314967465726, "learning_rate": 2.301255230125523e-05, "logits/chosen": -8.1875, "logits/rejected": -7.90625, "logps/chosen": -516.0, "logps/rejected": -488.0, "loss": 0.5839, "rewards/accuracies": 0.699999988079071, "rewards/chosen": -1.8515625, "rewards/margins": 0.470703125, "rewards/rejected": -2.328125, "step": 220 }, { "epoch": 0.24071166928309787, "grad_norm": 2.2773378614378026, "learning_rate": 2.405857740585774e-05, "logits/chosen": -7.4375, "logits/rejected": -7.0, "logps/chosen": -502.0, "logps/rejected": -512.0, "loss": 0.5985, "rewards/accuracies": 0.6937500238418579, "rewards/chosen": -1.7109375, "rewards/margins": 0.400390625, "rewards/rejected": -2.109375, "step": 230 }, { "epoch": 0.25117739403453687, "grad_norm": 2.1409273439247754, "learning_rate": 2.510460251046025e-05, "logits/chosen": -9.8125, "logits/rejected": -9.5, "logps/chosen": -472.0, "logps/rejected": -492.0, "loss": 0.5804, "rewards/accuracies": 0.6875, "rewards/chosen": -1.6328125, "rewards/margins": 0.439453125, "rewards/rejected": -2.0625, "step": 240 }, { "epoch": 0.2616431187859759, "grad_norm": 2.8417736797719093, "learning_rate": 2.615062761506276e-05, "logits/chosen": -7.5, "logits/rejected": -7.0, "logps/chosen": -552.0, "logps/rejected": -572.0, "loss": 0.5424, "rewards/accuracies": 0.6875, "rewards/chosen": -2.296875, "rewards/margins": 0.5546875, "rewards/rejected": -2.84375, "step": 250 }, { "epoch": 0.272108843537415, "grad_norm": 2.1201241469645438, "learning_rate": 2.7196652719665273e-05, "logits/chosen": -9.25, "logits/rejected": -8.8125, "logps/chosen": -572.0, "logps/rejected": -596.0, "loss": 0.5807, "rewards/accuracies": 0.65625, "rewards/chosen": -2.40625, "rewards/margins": 0.42578125, "rewards/rejected": -2.828125, "step": 260 }, { "epoch": 0.282574568288854, "grad_norm": 2.809629729626212, "learning_rate": 2.824267782426778e-05, "logits/chosen": -6.875, "logits/rejected": -6.3125, "logps/chosen": -544.0, "logps/rejected": -528.0, "loss": 0.5719, "rewards/accuracies": 0.6625000238418579, "rewards/chosen": -1.9453125, "rewards/margins": 0.4765625, "rewards/rejected": -2.421875, "step": 270 }, { "epoch": 0.29304029304029305, "grad_norm": 2.0280888428773807, "learning_rate": 2.9288702928870294e-05, "logits/chosen": -5.6875, "logits/rejected": -5.125, "logps/chosen": -494.0, "logps/rejected": -510.0, "loss": 0.5969, "rewards/accuracies": 0.6937500238418579, "rewards/chosen": -1.5, "rewards/margins": 0.443359375, "rewards/rejected": -1.9453125, "step": 280 }, { "epoch": 0.3035060177917321, "grad_norm": 2.5524631342772692, "learning_rate": 3.0334728033472803e-05, "logits/chosen": -6.65625, "logits/rejected": -6.25, "logps/chosen": -486.0, "logps/rejected": -468.0, "loss": 0.6021, "rewards/accuracies": 0.6875, "rewards/chosen": -1.6875, "rewards/margins": 0.3515625, "rewards/rejected": -2.03125, "step": 290 }, { "epoch": 0.3139717425431711, "grad_norm": 2.353526727859833, "learning_rate": 3.1380753138075316e-05, "logits/chosen": -6.8125, "logits/rejected": -6.65625, "logps/chosen": -544.0, "logps/rejected": -564.0, "loss": 0.6122, "rewards/accuracies": 0.6312500238418579, "rewards/chosen": -2.328125, "rewards/margins": 0.375, "rewards/rejected": -2.703125, "step": 300 }, { "epoch": 0.3139717425431711, "eval_logits/chosen": -6.15625, "eval_logits/rejected": -5.59375, "eval_logps/chosen": -474.0, "eval_logps/rejected": -496.0, "eval_loss": 0.5750937461853027, "eval_rewards/accuracies": 0.712890625, "eval_rewards/chosen": -1.6015625, "eval_rewards/margins": 0.5, "eval_rewards/rejected": -2.109375, "eval_runtime": 52.7701, "eval_samples_per_second": 37.9, "eval_steps_per_second": 0.606, "step": 300 }, { "epoch": 0.32443746729461015, "grad_norm": 2.534963819166362, "learning_rate": 3.2426778242677825e-05, "logits/chosen": -7.0, "logits/rejected": -5.84375, "logps/chosen": -492.0, "logps/rejected": -456.0, "loss": 0.588, "rewards/accuracies": 0.731249988079071, "rewards/chosen": -1.5703125, "rewards/margins": 0.494140625, "rewards/rejected": -2.0625, "step": 310 }, { "epoch": 0.3349031920460492, "grad_norm": 1.9350591762386424, "learning_rate": 3.3472803347280334e-05, "logits/chosen": -6.53125, "logits/rejected": -6.0625, "logps/chosen": -476.0, "logps/rejected": -516.0, "loss": 0.6021, "rewards/accuracies": 0.643750011920929, "rewards/chosen": -1.8828125, "rewards/margins": 0.50390625, "rewards/rejected": -2.390625, "step": 320 }, { "epoch": 0.3453689167974882, "grad_norm": 2.255741471296445, "learning_rate": 3.4518828451882844e-05, "logits/chosen": -7.5, "logits/rejected": -7.09375, "logps/chosen": -528.0, "logps/rejected": -536.0, "loss": 0.582, "rewards/accuracies": 0.6875, "rewards/chosen": -2.296875, "rewards/margins": 0.455078125, "rewards/rejected": -2.75, "step": 330 }, { "epoch": 0.35583464154892724, "grad_norm": 2.829695974746096, "learning_rate": 3.556485355648535e-05, "logits/chosen": -7.65625, "logits/rejected": -7.0625, "logps/chosen": -524.0, "logps/rejected": -520.0, "loss": 0.5962, "rewards/accuracies": 0.6312500238418579, "rewards/chosen": -2.15625, "rewards/margins": 0.384765625, "rewards/rejected": -2.53125, "step": 340 }, { "epoch": 0.3663003663003663, "grad_norm": 1.9387450136987523, "learning_rate": 3.661087866108787e-05, "logits/chosen": -5.3125, "logits/rejected": -4.59375, "logps/chosen": -612.0, "logps/rejected": -608.0, "loss": 0.5795, "rewards/accuracies": 0.6937500238418579, "rewards/chosen": -2.703125, "rewards/margins": 0.54296875, "rewards/rejected": -3.25, "step": 350 }, { "epoch": 0.37676609105180536, "grad_norm": 2.4596243424669755, "learning_rate": 3.765690376569038e-05, "logits/chosen": -5.3125, "logits/rejected": -4.59375, "logps/chosen": -540.0, "logps/rejected": -560.0, "loss": 0.5594, "rewards/accuracies": 0.7437499761581421, "rewards/chosen": -2.21875, "rewards/margins": 0.53515625, "rewards/rejected": -2.765625, "step": 360 }, { "epoch": 0.3872318158032444, "grad_norm": 2.38807244727826, "learning_rate": 3.870292887029289e-05, "logits/chosen": -6.46875, "logits/rejected": -6.09375, "logps/chosen": -588.0, "logps/rejected": -596.0, "loss": 0.5757, "rewards/accuracies": 0.668749988079071, "rewards/chosen": -2.796875, "rewards/margins": 0.466796875, "rewards/rejected": -3.265625, "step": 370 }, { "epoch": 0.3976975405546834, "grad_norm": 2.4861103037966816, "learning_rate": 3.9748953974895396e-05, "logits/chosen": -6.125, "logits/rejected": -5.65625, "logps/chosen": -568.0, "logps/rejected": -600.0, "loss": 0.5829, "rewards/accuracies": 0.643750011920929, "rewards/chosen": -2.71875, "rewards/margins": 0.4609375, "rewards/rejected": -3.1875, "step": 380 }, { "epoch": 0.40816326530612246, "grad_norm": 2.1200670617456407, "learning_rate": 4.0794979079497905e-05, "logits/chosen": -6.28125, "logits/rejected": -5.6875, "logps/chosen": -516.0, "logps/rejected": -528.0, "loss": 0.6082, "rewards/accuracies": 0.637499988079071, "rewards/chosen": -1.90625, "rewards/margins": 0.3671875, "rewards/rejected": -2.28125, "step": 390 }, { "epoch": 0.4186289900575615, "grad_norm": 3.353244167214107, "learning_rate": 4.184100418410042e-05, "logits/chosen": -6.21875, "logits/rejected": -5.4375, "logps/chosen": -510.0, "logps/rejected": -536.0, "loss": 0.5905, "rewards/accuracies": 0.6875, "rewards/chosen": -1.953125, "rewards/margins": 0.455078125, "rewards/rejected": -2.40625, "step": 400 }, { "epoch": 0.4186289900575615, "eval_logits/chosen": -6.34375, "eval_logits/rejected": -5.5625, "eval_logps/chosen": -478.0, "eval_logps/rejected": -504.0, "eval_loss": 0.599359393119812, "eval_rewards/accuracies": 0.66796875, "eval_rewards/chosen": -1.6328125, "eval_rewards/margins": 0.55078125, "eval_rewards/rejected": -2.1875, "eval_runtime": 52.7697, "eval_samples_per_second": 37.901, "eval_steps_per_second": 0.606, "step": 400 }, { "epoch": 0.4290947148090005, "grad_norm": 1.9400560513665543, "learning_rate": 4.288702928870293e-05, "logits/chosen": -5.875, "logits/rejected": -5.0625, "logps/chosen": -532.0, "logps/rejected": -540.0, "loss": 0.604, "rewards/accuracies": 0.65625, "rewards/chosen": -2.109375, "rewards/margins": 0.515625, "rewards/rejected": -2.625, "step": 410 }, { "epoch": 0.43956043956043955, "grad_norm": 1.801312683220209, "learning_rate": 4.393305439330544e-05, "logits/chosen": -5.0, "logits/rejected": -4.125, "logps/chosen": -564.0, "logps/rejected": -560.0, "loss": 0.6109, "rewards/accuracies": 0.6937500238418579, "rewards/chosen": -2.328125, "rewards/margins": 0.42578125, "rewards/rejected": -2.75, "step": 420 }, { "epoch": 0.4500261643118786, "grad_norm": 1.7147949043618933, "learning_rate": 4.497907949790795e-05, "logits/chosen": -6.125, "logits/rejected": -5.5625, "logps/chosen": -478.0, "logps/rejected": -502.0, "loss": 0.6254, "rewards/accuracies": 0.699999988079071, "rewards/chosen": -1.7890625, "rewards/margins": 0.369140625, "rewards/rejected": -2.15625, "step": 430 }, { "epoch": 0.4604918890633176, "grad_norm": 2.796041254164824, "learning_rate": 4.602510460251046e-05, "logits/chosen": -7.71875, "logits/rejected": -7.3125, "logps/chosen": -528.0, "logps/rejected": -504.0, "loss": 0.611, "rewards/accuracies": 0.731249988079071, "rewards/chosen": -1.90625, "rewards/margins": 0.392578125, "rewards/rejected": -2.296875, "step": 440 }, { "epoch": 0.47095761381475665, "grad_norm": 2.1381334456656744, "learning_rate": 4.7071129707112974e-05, "logits/chosen": -8.3125, "logits/rejected": -7.96875, "logps/chosen": -520.0, "logps/rejected": -556.0, "loss": 0.5771, "rewards/accuracies": 0.7437499761581421, "rewards/chosen": -1.875, "rewards/margins": 0.58984375, "rewards/rejected": -2.46875, "step": 450 }, { "epoch": 0.48142333856619574, "grad_norm": 1.8842575828374497, "learning_rate": 4.811715481171548e-05, "logits/chosen": -8.5, "logits/rejected": -8.375, "logps/chosen": -532.0, "logps/rejected": -552.0, "loss": 0.5877, "rewards/accuracies": 0.668749988079071, "rewards/chosen": -2.3125, "rewards/margins": 0.5078125, "rewards/rejected": -2.828125, "step": 460 }, { "epoch": 0.49188906331763477, "grad_norm": 2.295050225174663, "learning_rate": 4.916317991631799e-05, "logits/chosen": -9.3125, "logits/rejected": -8.8125, "logps/chosen": -500.0, "logps/rejected": -508.0, "loss": 0.5301, "rewards/accuracies": 0.699999988079071, "rewards/chosen": -1.7421875, "rewards/margins": 0.62109375, "rewards/rejected": -2.359375, "step": 470 }, { "epoch": 0.5023547880690737, "grad_norm": 2.159411861590043, "learning_rate": 4.9999973273692244e-05, "logits/chosen": -9.625, "logits/rejected": -9.25, "logps/chosen": -532.0, "logps/rejected": -576.0, "loss": 0.5703, "rewards/accuracies": 0.675000011920929, "rewards/chosen": -2.203125, "rewards/margins": 0.734375, "rewards/rejected": -2.9375, "step": 480 }, { "epoch": 0.5128205128205128, "grad_norm": 1.8524703002374596, "learning_rate": 4.9999037858921e-05, "logits/chosen": -7.875, "logits/rejected": -7.125, "logps/chosen": -620.0, "logps/rejected": -620.0, "loss": 0.5705, "rewards/accuracies": 0.7437499761581421, "rewards/chosen": -2.546875, "rewards/margins": 0.7890625, "rewards/rejected": -3.34375, "step": 490 }, { "epoch": 0.5232862375719518, "grad_norm": 2.3441132459203655, "learning_rate": 4.9996766185905256e-05, "logits/chosen": -6.75, "logits/rejected": -6.125, "logps/chosen": -490.0, "logps/rejected": -496.0, "loss": 0.5781, "rewards/accuracies": 0.71875, "rewards/chosen": -1.90625, "rewards/margins": 0.53125, "rewards/rejected": -2.4375, "step": 500 }, { "epoch": 0.5232862375719518, "eval_logits/chosen": -6.84375, "eval_logits/rejected": -6.0625, "eval_logps/chosen": -486.0, "eval_logps/rejected": -512.0, "eval_loss": 0.5763593912124634, "eval_rewards/accuracies": 0.681640625, "eval_rewards/chosen": -1.71875, "eval_rewards/margins": 0.55859375, "eval_rewards/rejected": -2.265625, "eval_runtime": 52.7734, "eval_samples_per_second": 37.898, "eval_steps_per_second": 0.606, "step": 500 }, { "epoch": 0.533751962323391, "grad_norm": 1.6521111958327657, "learning_rate": 4.999315837607136e-05, "logits/chosen": -7.03125, "logits/rejected": -6.25, "logps/chosen": -524.0, "logps/rejected": -556.0, "loss": 0.572, "rewards/accuracies": 0.7124999761581421, "rewards/chosen": -1.9296875, "rewards/margins": 0.68359375, "rewards/rejected": -2.609375, "step": 510 }, { "epoch": 0.54421768707483, "grad_norm": 2.124354025750924, "learning_rate": 4.998821462226535e-05, "logits/chosen": -4.5, "logits/rejected": -3.90625, "logps/chosen": -672.0, "logps/rejected": -688.0, "loss": 0.5989, "rewards/accuracies": 0.6499999761581421, "rewards/chosen": -3.53125, "rewards/margins": 0.498046875, "rewards/rejected": -4.03125, "step": 520 }, { "epoch": 0.554683411826269, "grad_norm": 2.1800165538121425, "learning_rate": 4.998193518874268e-05, "logits/chosen": -7.125, "logits/rejected": -6.46875, "logps/chosen": -640.0, "logps/rejected": -668.0, "loss": 0.6098, "rewards/accuracies": 0.65625, "rewards/chosen": -3.234375, "rewards/margins": 0.57421875, "rewards/rejected": -3.8125, "step": 530 }, { "epoch": 0.565149136577708, "grad_norm": 1.8923017590968032, "learning_rate": 4.997432041115405e-05, "logits/chosen": -8.5, "logits/rejected": -7.8125, "logps/chosen": -560.0, "logps/rejected": -548.0, "loss": 0.5872, "rewards/accuracies": 0.675000011920929, "rewards/chosen": -2.15625, "rewards/margins": 0.4296875, "rewards/rejected": -2.578125, "step": 540 }, { "epoch": 0.5756148613291471, "grad_norm": 1.507962725186336, "learning_rate": 4.9965370696527504e-05, "logits/chosen": -6.875, "logits/rejected": -6.0, "logps/chosen": -620.0, "logps/rejected": -616.0, "loss": 0.6074, "rewards/accuracies": 0.625, "rewards/chosen": -3.109375, "rewards/margins": 0.55078125, "rewards/rejected": -3.65625, "step": 550 }, { "epoch": 0.5860805860805861, "grad_norm": 1.91935882186746, "learning_rate": 4.9955086523246655e-05, "logits/chosen": -7.84375, "logits/rejected": -7.0625, "logps/chosen": -560.0, "logps/rejected": -576.0, "loss": 0.5955, "rewards/accuracies": 0.6937500238418579, "rewards/chosen": -2.328125, "rewards/margins": 0.5703125, "rewards/rejected": -2.90625, "step": 560 }, { "epoch": 0.5965463108320251, "grad_norm": 1.8238249534958182, "learning_rate": 4.99434684410251e-05, "logits/chosen": -8.4375, "logits/rejected": -7.78125, "logps/chosen": -552.0, "logps/rejected": -572.0, "loss": 0.5758, "rewards/accuracies": 0.706250011920929, "rewards/chosen": -2.25, "rewards/margins": 0.55078125, "rewards/rejected": -2.796875, "step": 570 }, { "epoch": 0.6070120355834642, "grad_norm": 2.0134942953537562, "learning_rate": 4.9930517070877066e-05, "logits/chosen": -7.90625, "logits/rejected": -7.21875, "logps/chosen": -632.0, "logps/rejected": -644.0, "loss": 0.6037, "rewards/accuracies": 0.71875, "rewards/chosen": -3.390625, "rewards/margins": 0.71484375, "rewards/rejected": -4.09375, "step": 580 }, { "epoch": 0.6174777603349032, "grad_norm": 1.8626820784967801, "learning_rate": 4.991623310508422e-05, "logits/chosen": -8.5, "logits/rejected": -7.65625, "logps/chosen": -576.0, "logps/rejected": -616.0, "loss": 0.5652, "rewards/accuracies": 0.668749988079071, "rewards/chosen": -2.546875, "rewards/margins": 0.6796875, "rewards/rejected": -3.234375, "step": 590 }, { "epoch": 0.6279434850863422, "grad_norm": 2.3196683948602286, "learning_rate": 4.990061730715862e-05, "logits/chosen": -5.28125, "logits/rejected": -4.375, "logps/chosen": -644.0, "logps/rejected": -692.0, "loss": 0.5356, "rewards/accuracies": 0.7124999761581421, "rewards/chosen": -3.34375, "rewards/margins": 0.75, "rewards/rejected": -4.09375, "step": 600 }, { "epoch": 0.6279434850863422, "eval_logits/chosen": -4.25, "eval_logits/rejected": -3.328125, "eval_logps/chosen": -704.0, "eval_logps/rejected": -744.0, "eval_loss": 0.5830625295639038, "eval_rewards/accuracies": 0.669921875, "eval_rewards/chosen": -3.890625, "eval_rewards/margins": 0.6796875, "eval_rewards/rejected": -4.5625, "eval_runtime": 52.7702, "eval_samples_per_second": 37.9, "eval_steps_per_second": 0.606, "step": 600 }, { "epoch": 0.6384092098377813, "grad_norm": 1.969214676284523, "learning_rate": 4.9883670511801914e-05, "logits/chosen": -5.03125, "logits/rejected": -3.90625, "logps/chosen": -700.0, "logps/rejected": -688.0, "loss": 0.6028, "rewards/accuracies": 0.65625, "rewards/chosen": -3.4375, "rewards/margins": 0.67578125, "rewards/rejected": -4.09375, "step": 610 }, { "epoch": 0.6488749345892203, "grad_norm": 1.8292251328731444, "learning_rate": 4.98653936248608e-05, "logits/chosen": -5.46875, "logits/rejected": -4.75, "logps/chosen": -556.0, "logps/rejected": -572.0, "loss": 0.5732, "rewards/accuracies": 0.7437499761581421, "rewards/chosen": -2.4375, "rewards/margins": 0.6640625, "rewards/rejected": -3.09375, "step": 620 }, { "epoch": 0.6593406593406593, "grad_norm": 2.0559585908544658, "learning_rate": 4.9845787623278495e-05, "logits/chosen": -6.09375, "logits/rejected": -5.34375, "logps/chosen": -580.0, "logps/rejected": -592.0, "loss": 0.559, "rewards/accuracies": 0.6875, "rewards/chosen": -2.640625, "rewards/margins": 0.65234375, "rewards/rejected": -3.296875, "step": 630 }, { "epoch": 0.6698063840920984, "grad_norm": 2.367176900894739, "learning_rate": 4.982485355504258e-05, "logits/chosen": -5.34375, "logits/rejected": -4.375, "logps/chosen": -616.0, "logps/rejected": -632.0, "loss": 0.5808, "rewards/accuracies": 0.6625000238418579, "rewards/chosen": -3.03125, "rewards/margins": 0.68359375, "rewards/rejected": -3.71875, "step": 640 }, { "epoch": 0.6802721088435374, "grad_norm": 1.9017110444929817, "learning_rate": 4.9802592539128984e-05, "logits/chosen": -6.59375, "logits/rejected": -5.875, "logps/chosen": -544.0, "logps/rejected": -596.0, "loss": 0.5699, "rewards/accuracies": 0.6875, "rewards/chosen": -2.390625, "rewards/margins": 0.55078125, "rewards/rejected": -2.9375, "step": 650 }, { "epoch": 0.6907378335949764, "grad_norm": 1.5844136116169825, "learning_rate": 4.977900576544212e-05, "logits/chosen": -5.15625, "logits/rejected": -4.03125, "logps/chosen": -708.0, "logps/rejected": -736.0, "loss": 0.5568, "rewards/accuracies": 0.75, "rewards/chosen": -3.78125, "rewards/margins": 0.703125, "rewards/rejected": -4.5, "step": 660 }, { "epoch": 0.7012035583464155, "grad_norm": 1.9775753765391848, "learning_rate": 4.9754094494751364e-05, "logits/chosen": -6.125, "logits/rejected": -5.3125, "logps/chosen": -712.0, "logps/rejected": -764.0, "loss": 0.5442, "rewards/accuracies": 0.6875, "rewards/chosen": -3.734375, "rewards/margins": 0.69921875, "rewards/rejected": -4.4375, "step": 670 }, { "epoch": 0.7116692830978545, "grad_norm": 2.080545669185722, "learning_rate": 4.972786005862358e-05, "logits/chosen": -10.125, "logits/rejected": -9.5, "logps/chosen": -608.0, "logps/rejected": -632.0, "loss": 0.598, "rewards/accuracies": 0.6875, "rewards/chosen": -2.4375, "rewards/margins": 0.6171875, "rewards/rejected": -3.046875, "step": 680 }, { "epoch": 0.7221350078492935, "grad_norm": 1.7635314791728216, "learning_rate": 4.9700303859352e-05, "logits/chosen": -9.875, "logits/rejected": -9.3125, "logps/chosen": -536.0, "logps/rejected": -556.0, "loss": 0.537, "rewards/accuracies": 0.731249988079071, "rewards/chosen": -2.171875, "rewards/margins": 0.74609375, "rewards/rejected": -2.921875, "step": 690 }, { "epoch": 0.7326007326007326, "grad_norm": 2.6926500474089856, "learning_rate": 4.967142736988127e-05, "logits/chosen": -9.25, "logits/rejected": -8.75, "logps/chosen": -568.0, "logps/rejected": -596.0, "loss": 0.5756, "rewards/accuracies": 0.706250011920929, "rewards/chosen": -2.71875, "rewards/margins": 0.6015625, "rewards/rejected": -3.3125, "step": 700 }, { "epoch": 0.7326007326007326, "eval_logits/chosen": -9.375, "eval_logits/rejected": -8.8125, "eval_logps/chosen": -656.0, "eval_logps/rejected": -692.0, "eval_loss": 0.5859062671661377, "eval_rewards/accuracies": 0.701171875, "eval_rewards/chosen": -3.421875, "eval_rewards/margins": 0.61328125, "eval_rewards/rejected": -4.03125, "eval_runtime": 52.78, "eval_samples_per_second": 37.893, "eval_steps_per_second": 0.606, "step": 700 }, { "epoch": 0.7430664573521716, "grad_norm": 1.8338694406415879, "learning_rate": 4.964123213372865e-05, "logits/chosen": -8.625, "logits/rejected": -7.8125, "logps/chosen": -652.0, "logps/rejected": -672.0, "loss": 0.6116, "rewards/accuracies": 0.699999988079071, "rewards/chosen": -3.5625, "rewards/margins": 0.50390625, "rewards/rejected": -4.0625, "step": 710 }, { "epoch": 0.7535321821036107, "grad_norm": 1.7715417270349543, "learning_rate": 4.96097197649016e-05, "logits/chosen": -9.25, "logits/rejected": -8.375, "logps/chosen": -624.0, "logps/rejected": -632.0, "loss": 0.5629, "rewards/accuracies": 0.6937500238418579, "rewards/chosen": -2.90625, "rewards/margins": 0.64453125, "rewards/rejected": -3.546875, "step": 720 }, { "epoch": 0.7639979068550498, "grad_norm": 2.09439081053431, "learning_rate": 4.9576891947811455e-05, "logits/chosen": -8.375, "logits/rejected": -7.75, "logps/chosen": -644.0, "logps/rejected": -648.0, "loss": 0.5916, "rewards/accuracies": 0.643750011920929, "rewards/chosen": -3.328125, "rewards/margins": 0.486328125, "rewards/rejected": -3.8125, "step": 730 }, { "epoch": 0.7744636316064888, "grad_norm": 1.84005754928221, "learning_rate": 4.95427504371834e-05, "logits/chosen": -8.875, "logits/rejected": -8.25, "logps/chosen": -608.0, "logps/rejected": -644.0, "loss": 0.5471, "rewards/accuracies": 0.706250011920929, "rewards/chosen": -3.171875, "rewards/margins": 0.65625, "rewards/rejected": -3.828125, "step": 740 }, { "epoch": 0.7849293563579278, "grad_norm": 1.6580778209495195, "learning_rate": 4.9507297057962645e-05, "logits/chosen": -10.375, "logits/rejected": -9.5625, "logps/chosen": -576.0, "logps/rejected": -640.0, "loss": 0.5362, "rewards/accuracies": 0.668749988079071, "rewards/chosen": -2.96875, "rewards/margins": 0.71875, "rewards/rejected": -3.671875, "step": 750 }, { "epoch": 0.7953950811093669, "grad_norm": 1.834684453607076, "learning_rate": 4.947053370521696e-05, "logits/chosen": -9.0625, "logits/rejected": -7.8125, "logps/chosen": -560.0, "logps/rejected": -620.0, "loss": 0.5777, "rewards/accuracies": 0.699999988079071, "rewards/chosen": -2.796875, "rewards/margins": 0.7109375, "rewards/rejected": -3.5, "step": 760 }, { "epoch": 0.8058608058608059, "grad_norm": 1.718132653359686, "learning_rate": 4.9432462344035294e-05, "logits/chosen": -9.25, "logits/rejected": -8.4375, "logps/chosen": -612.0, "logps/rejected": -680.0, "loss": 0.5631, "rewards/accuracies": 0.6499999761581421, "rewards/chosen": -2.65625, "rewards/margins": 0.51171875, "rewards/rejected": -3.171875, "step": 770 }, { "epoch": 0.8163265306122449, "grad_norm": 1.9148197442567063, "learning_rate": 4.939308500942277e-05, "logits/chosen": -5.9375, "logits/rejected": -4.75, "logps/chosen": -692.0, "logps/rejected": -696.0, "loss": 0.5711, "rewards/accuracies": 0.668749988079071, "rewards/chosen": -3.328125, "rewards/margins": 0.68359375, "rewards/rejected": -4.03125, "step": 780 }, { "epoch": 0.826792255363684, "grad_norm": 1.9380251783145208, "learning_rate": 4.9352403806191926e-05, "logits/chosen": -3.96875, "logits/rejected": -2.625, "logps/chosen": -688.0, "logps/rejected": -700.0, "loss": 0.5952, "rewards/accuracies": 0.7124999761581421, "rewards/chosen": -3.609375, "rewards/margins": 0.625, "rewards/rejected": -4.25, "step": 790 }, { "epoch": 0.837257980115123, "grad_norm": 1.980206883814282, "learning_rate": 4.931042090885017e-05, "logits/chosen": -6.9375, "logits/rejected": -5.90625, "logps/chosen": -596.0, "logps/rejected": -644.0, "loss": 0.5528, "rewards/accuracies": 0.7124999761581421, "rewards/chosen": -2.859375, "rewards/margins": 0.6484375, "rewards/rejected": -3.5, "step": 800 }, { "epoch": 0.837257980115123, "eval_logits/chosen": -8.3125, "eval_logits/rejected": -7.4375, "eval_logps/chosen": -604.0, "eval_logps/rejected": -636.0, "eval_loss": 0.5731562376022339, "eval_rewards/accuracies": 0.68359375, "eval_rewards/chosen": -2.890625, "eval_rewards/margins": 0.6015625, "eval_rewards/rejected": -3.5, "eval_runtime": 52.7984, "eval_samples_per_second": 37.88, "eval_steps_per_second": 0.606, "step": 800 }, { "epoch": 0.847723704866562, "grad_norm": 1.970105831115586, "learning_rate": 4.926713856148357e-05, "logits/chosen": -7.8125, "logits/rejected": -6.84375, "logps/chosen": -652.0, "logps/rejected": -652.0, "loss": 0.5817, "rewards/accuracies": 0.6812499761581421, "rewards/chosen": -3.1875, "rewards/margins": 0.56640625, "rewards/rejected": -3.75, "step": 810 }, { "epoch": 0.858189429618001, "grad_norm": 2.163038230104847, "learning_rate": 4.922255907763692e-05, "logits/chosen": -7.46875, "logits/rejected": -6.46875, "logps/chosen": -684.0, "logps/rejected": -720.0, "loss": 0.5878, "rewards/accuracies": 0.6875, "rewards/chosen": -3.3125, "rewards/margins": 0.703125, "rewards/rejected": -4.0, "step": 820 }, { "epoch": 0.8686551543694401, "grad_norm": 1.5500963032625696, "learning_rate": 4.917668484019002e-05, "logits/chosen": -8.125, "logits/rejected": -7.46875, "logps/chosen": -580.0, "logps/rejected": -620.0, "loss": 0.5732, "rewards/accuracies": 0.6812499761581421, "rewards/chosen": -2.828125, "rewards/margins": 0.6015625, "rewards/rejected": -3.421875, "step": 830 }, { "epoch": 0.8791208791208791, "grad_norm": 2.2367725914269956, "learning_rate": 4.9129518301230375e-05, "logits/chosen": -7.28125, "logits/rejected": -5.84375, "logps/chosen": -616.0, "logps/rejected": -668.0, "loss": 0.5445, "rewards/accuracies": 0.6937500238418579, "rewards/chosen": -2.9375, "rewards/margins": 0.8125, "rewards/rejected": -3.75, "step": 840 }, { "epoch": 0.8895866038723181, "grad_norm": 2.4245666537731267, "learning_rate": 4.908106198192207e-05, "logits/chosen": -5.28125, "logits/rejected": -4.15625, "logps/chosen": -692.0, "logps/rejected": -704.0, "loss": 0.5262, "rewards/accuracies": 0.706250011920929, "rewards/chosen": -3.671875, "rewards/margins": 0.828125, "rewards/rejected": -4.5, "step": 850 }, { "epoch": 0.9000523286237572, "grad_norm": 1.8299461987606431, "learning_rate": 4.903131847237104e-05, "logits/chosen": -6.40625, "logits/rejected": -5.09375, "logps/chosen": -736.0, "logps/rejected": -760.0, "loss": 0.5486, "rewards/accuracies": 0.6625000238418579, "rewards/chosen": -4.03125, "rewards/margins": 0.734375, "rewards/rejected": -4.78125, "step": 860 }, { "epoch": 0.9105180533751962, "grad_norm": 1.776519837344144, "learning_rate": 4.8980290431486595e-05, "logits/chosen": -9.0, "logits/rejected": -7.875, "logps/chosen": -612.0, "logps/rejected": -632.0, "loss": 0.6109, "rewards/accuracies": 0.6812499761581421, "rewards/chosen": -2.921875, "rewards/margins": 0.6328125, "rewards/rejected": -3.5625, "step": 870 }, { "epoch": 0.9209837781266352, "grad_norm": 1.5665799264044165, "learning_rate": 4.892798058683933e-05, "logits/chosen": -7.5625, "logits/rejected": -6.71875, "logps/chosen": -636.0, "logps/rejected": -692.0, "loss": 0.5469, "rewards/accuracies": 0.6875, "rewards/chosen": -2.921875, "rewards/margins": 0.6640625, "rewards/rejected": -3.59375, "step": 880 }, { "epoch": 0.9314495028780743, "grad_norm": 1.6938125314327865, "learning_rate": 4.887439173451529e-05, "logits/chosen": -8.3125, "logits/rejected": -7.375, "logps/chosen": -628.0, "logps/rejected": -668.0, "loss": 0.5701, "rewards/accuracies": 0.6625000238418579, "rewards/chosen": -2.875, "rewards/margins": 0.6640625, "rewards/rejected": -3.546875, "step": 890 }, { "epoch": 0.9419152276295133, "grad_norm": 1.9001080952998146, "learning_rate": 4.8819526738966546e-05, "logits/chosen": -8.625, "logits/rejected": -7.9375, "logps/chosen": -616.0, "logps/rejected": -668.0, "loss": 0.5753, "rewards/accuracies": 0.6499999761581421, "rewards/chosen": -2.90625, "rewards/margins": 0.59765625, "rewards/rejected": -3.5, "step": 900 }, { "epoch": 0.9419152276295133, "eval_logits/chosen": -7.90625, "eval_logits/rejected": -7.0, "eval_logps/chosen": -620.0, "eval_logps/rejected": -660.0, "eval_loss": 0.5692968964576721, "eval_rewards/accuracies": 0.716796875, "eval_rewards/chosen": -3.046875, "eval_rewards/margins": 0.6796875, "eval_rewards/rejected": -3.734375, "eval_runtime": 52.8034, "eval_samples_per_second": 37.876, "eval_steps_per_second": 0.606, "step": 900 }, { "epoch": 0.9523809523809523, "grad_norm": 1.8582669591039784, "learning_rate": 4.876338853285809e-05, "logits/chosen": -8.375, "logits/rejected": -7.59375, "logps/chosen": -532.0, "logps/rejected": -572.0, "loss": 0.593, "rewards/accuracies": 0.668749988079071, "rewards/chosen": -2.75, "rewards/margins": 0.57421875, "rewards/rejected": -3.328125, "step": 910 }, { "epoch": 0.9628466771323915, "grad_norm": 1.927805055357823, "learning_rate": 4.870598011691101e-05, "logits/chosen": -9.5, "logits/rejected": -9.0, "logps/chosen": -628.0, "logps/rejected": -636.0, "loss": 0.5746, "rewards/accuracies": 0.706250011920929, "rewards/chosen": -3.046875, "rewards/margins": 0.55078125, "rewards/rejected": -3.59375, "step": 920 }, { "epoch": 0.9733124018838305, "grad_norm": 1.5760630448625366, "learning_rate": 4.864730455974218e-05, "logits/chosen": -7.34375, "logits/rejected": -6.4375, "logps/chosen": -760.0, "logps/rejected": -756.0, "loss": 0.5694, "rewards/accuracies": 0.6812499761581421, "rewards/chosen": -4.0, "rewards/margins": 0.640625, "rewards/rejected": -4.65625, "step": 930 }, { "epoch": 0.9837781266352695, "grad_norm": 1.7670941807840217, "learning_rate": 4.858736499770018e-05, "logits/chosen": -8.4375, "logits/rejected": -7.59375, "logps/chosen": -652.0, "logps/rejected": -692.0, "loss": 0.5349, "rewards/accuracies": 0.675000011920929, "rewards/chosen": -3.484375, "rewards/margins": 0.63671875, "rewards/rejected": -4.125, "step": 940 }, { "epoch": 0.9942438513867086, "grad_norm": 1.8451915529649179, "learning_rate": 4.852616463469768e-05, "logits/chosen": -8.125, "logits/rejected": -7.4375, "logps/chosen": -708.0, "logps/rejected": -744.0, "loss": 0.5689, "rewards/accuracies": 0.706250011920929, "rewards/chosen": -3.578125, "rewards/margins": 0.625, "rewards/rejected": -4.21875, "step": 950 }, { "epoch": 1.0047095761381475, "grad_norm": 1.6173131691447116, "learning_rate": 4.846370674204016e-05, "logits/chosen": -8.5625, "logits/rejected": -7.90625, "logps/chosen": -644.0, "logps/rejected": -692.0, "loss": 0.47, "rewards/accuracies": 0.737500011920929, "rewards/chosen": -3.265625, "rewards/margins": 0.90625, "rewards/rejected": -4.1875, "step": 960 }, { "epoch": 1.0151753008895865, "grad_norm": 1.865923958174982, "learning_rate": 4.839999465825107e-05, "logits/chosen": -6.5, "logits/rejected": -5.125, "logps/chosen": -776.0, "logps/rejected": -964.0, "loss": 0.2891, "rewards/accuracies": 0.90625, "rewards/chosen": -4.1875, "rewards/margins": 2.40625, "rewards/rejected": -6.59375, "step": 970 }, { "epoch": 1.0256410256410255, "grad_norm": 1.9783894181617947, "learning_rate": 4.8335031788893346e-05, "logits/chosen": -5.28125, "logits/rejected": -3.0, "logps/chosen": -752.0, "logps/rejected": -944.0, "loss": 0.2655, "rewards/accuracies": 0.887499988079071, "rewards/chosen": -4.5, "rewards/margins": 2.421875, "rewards/rejected": -6.9375, "step": 980 }, { "epoch": 1.0361067503924646, "grad_norm": 1.6618530076050113, "learning_rate": 4.826882160638744e-05, "logits/chosen": -6.5625, "logits/rejected": -4.15625, "logps/chosen": -728.0, "logps/rejected": -956.0, "loss": 0.2357, "rewards/accuracies": 0.9312499761581421, "rewards/chosen": -3.90625, "rewards/margins": 2.84375, "rewards/rejected": -6.75, "step": 990 }, { "epoch": 1.0465724751439036, "grad_norm": 1.6852553538671544, "learning_rate": 4.820136764982564e-05, "logits/chosen": -5.8125, "logits/rejected": -3.984375, "logps/chosen": -704.0, "logps/rejected": -896.0, "loss": 0.2632, "rewards/accuracies": 0.875, "rewards/chosen": -3.546875, "rewards/margins": 2.375, "rewards/rejected": -5.9375, "step": 1000 }, { "epoch": 1.0465724751439036, "eval_logits/chosen": -4.25, "eval_logits/rejected": -2.875, "eval_logps/chosen": -732.0, "eval_logps/rejected": -808.0, "eval_loss": 0.5880781412124634, "eval_rewards/accuracies": 0.71484375, "eval_rewards/chosen": -4.1875, "eval_rewards/margins": 1.03125, "eval_rewards/rejected": -5.21875, "eval_runtime": 52.7927, "eval_samples_per_second": 37.884, "eval_steps_per_second": 0.606, "step": 1000 }, { "epoch": 1.0570381998953426, "grad_norm": 2.6176318352128414, "learning_rate": 4.813267352478292e-05, "logits/chosen": -4.40625, "logits/rejected": -2.453125, "logps/chosen": -660.0, "logps/rejected": -836.0, "loss": 0.2373, "rewards/accuracies": 0.90625, "rewards/chosen": -3.4375, "rewards/margins": 2.375, "rewards/rejected": -5.8125, "step": 1010 }, { "epoch": 1.0675039246467817, "grad_norm": 4.188123637402671, "learning_rate": 4.8062742903124246e-05, "logits/chosen": -3.53125, "logits/rejected": -2.0625, "logps/chosen": -760.0, "logps/rejected": -996.0, "loss": 0.2263, "rewards/accuracies": 0.918749988079071, "rewards/chosen": -4.125, "rewards/margins": 2.6875, "rewards/rejected": -6.8125, "step": 1020 }, { "epoch": 1.077969649398221, "grad_norm": 1.6827915651039376, "learning_rate": 4.799157952280826e-05, "logits/chosen": -4.8125, "logits/rejected": -2.984375, "logps/chosen": -700.0, "logps/rejected": -904.0, "loss": 0.2376, "rewards/accuracies": 0.875, "rewards/chosen": -3.765625, "rewards/margins": 2.453125, "rewards/rejected": -6.21875, "step": 1030 }, { "epoch": 1.08843537414966, "grad_norm": 1.6536012233098174, "learning_rate": 4.7919187187687496e-05, "logits/chosen": -3.90625, "logits/rejected": -2.03125, "logps/chosen": -672.0, "logps/rejected": -860.0, "loss": 0.2642, "rewards/accuracies": 0.856249988079071, "rewards/chosen": -3.703125, "rewards/margins": 2.28125, "rewards/rejected": -6.0, "step": 1040 }, { "epoch": 1.098901098901099, "grad_norm": 2.774611600374004, "learning_rate": 4.784556976730507e-05, "logits/chosen": -6.0625, "logits/rejected": -4.4375, "logps/chosen": -632.0, "logps/rejected": -868.0, "loss": 0.1976, "rewards/accuracies": 0.918749988079071, "rewards/chosen": -3.3125, "rewards/margins": 2.703125, "rewards/rejected": -6.03125, "step": 1050 }, { "epoch": 1.109366823652538, "grad_norm": 2.4394840573251284, "learning_rate": 4.777073119668782e-05, "logits/chosen": -4.46875, "logits/rejected": -2.1875, "logps/chosen": -760.0, "logps/rejected": -968.0, "loss": 0.236, "rewards/accuracies": 0.893750011920929, "rewards/chosen": -4.40625, "rewards/margins": 2.53125, "rewards/rejected": -6.9375, "step": 1060 }, { "epoch": 1.119832548403977, "grad_norm": 3.093080712317507, "learning_rate": 4.769467547613598e-05, "logits/chosen": -5.21875, "logits/rejected": -2.53125, "logps/chosen": -752.0, "logps/rejected": -972.0, "loss": 0.2397, "rewards/accuracies": 0.9437500238418579, "rewards/chosen": -4.28125, "rewards/margins": 2.6875, "rewards/rejected": -7.0, "step": 1070 }, { "epoch": 1.130298273155416, "grad_norm": 3.8414209437581888, "learning_rate": 4.761740667100934e-05, "logits/chosen": -6.1875, "logits/rejected": -4.53125, "logps/chosen": -708.0, "logps/rejected": -980.0, "loss": 0.2282, "rewards/accuracies": 0.9312499761581421, "rewards/chosen": -4.09375, "rewards/margins": 2.890625, "rewards/rejected": -7.0, "step": 1080 }, { "epoch": 1.1407639979068551, "grad_norm": 1.3714400548829968, "learning_rate": 4.7538928911509986e-05, "logits/chosen": -6.1875, "logits/rejected": -3.953125, "logps/chosen": -748.0, "logps/rejected": -1024.0, "loss": 0.2141, "rewards/accuracies": 0.9125000238418579, "rewards/chosen": -4.28125, "rewards/margins": 3.03125, "rewards/rejected": -7.3125, "step": 1090 }, { "epoch": 1.1512297226582942, "grad_norm": 2.8001835210684174, "learning_rate": 4.745924639246148e-05, "logits/chosen": -5.875, "logits/rejected": -4.125, "logps/chosen": -720.0, "logps/rejected": -920.0, "loss": 0.2283, "rewards/accuracies": 0.893750011920929, "rewards/chosen": -3.90625, "rewards/margins": 2.40625, "rewards/rejected": -6.3125, "step": 1100 }, { "epoch": 1.1512297226582942, "eval_logits/chosen": -7.0625, "eval_logits/rejected": -5.375, "eval_logps/chosen": -768.0, "eval_logps/rejected": -840.0, "eval_loss": 0.6142030954360962, "eval_rewards/accuracies": 0.712890625, "eval_rewards/chosen": -4.53125, "eval_rewards/margins": 0.99609375, "eval_rewards/rejected": -5.53125, "eval_runtime": 52.7843, "eval_samples_per_second": 37.89, "eval_steps_per_second": 0.606, "step": 1100 }, { "epoch": 1.1616954474097332, "grad_norm": 1.8190177892297257, "learning_rate": 4.737836337308465e-05, "logits/chosen": -6.375, "logits/rejected": -3.796875, "logps/chosen": -768.0, "logps/rejected": -1016.0, "loss": 0.1826, "rewards/accuracies": 0.956250011920929, "rewards/chosen": -4.5625, "rewards/margins": 2.875, "rewards/rejected": -7.4375, "step": 1110 }, { "epoch": 1.1721611721611722, "grad_norm": 2.4839194991731746, "learning_rate": 4.729628417676997e-05, "logits/chosen": -4.0625, "logits/rejected": -2.1875, "logps/chosen": -816.0, "logps/rejected": -1072.0, "loss": 0.2546, "rewards/accuracies": 0.90625, "rewards/chosen": -4.9375, "rewards/margins": 2.84375, "rewards/rejected": -7.78125, "step": 1120 }, { "epoch": 1.1826268969126112, "grad_norm": 1.7415070251510811, "learning_rate": 4.7213013190846375e-05, "logits/chosen": -6.1875, "logits/rejected": -3.921875, "logps/chosen": -756.0, "logps/rejected": -996.0, "loss": 0.192, "rewards/accuracies": 0.925000011920929, "rewards/chosen": -4.25, "rewards/margins": 2.671875, "rewards/rejected": -6.90625, "step": 1130 }, { "epoch": 1.1930926216640503, "grad_norm": 2.6883474669097196, "learning_rate": 4.712855486634684e-05, "logits/chosen": -6.15625, "logits/rejected": -4.3125, "logps/chosen": -832.0, "logps/rejected": -1096.0, "loss": 0.2138, "rewards/accuracies": 0.9375, "rewards/chosen": -4.78125, "rewards/margins": 3.078125, "rewards/rejected": -7.84375, "step": 1140 }, { "epoch": 1.2035583464154893, "grad_norm": 3.3034116848079753, "learning_rate": 4.7042913717770385e-05, "logits/chosen": -5.4375, "logits/rejected": -3.953125, "logps/chosen": -800.0, "logps/rejected": -1072.0, "loss": 0.203, "rewards/accuracies": 0.9312499761581421, "rewards/chosen": -4.84375, "rewards/margins": 2.8125, "rewards/rejected": -7.65625, "step": 1150 }, { "epoch": 1.2140240711669283, "grad_norm": 3.083004913157005, "learning_rate": 4.695609432284085e-05, "logits/chosen": -6.0625, "logits/rejected": -3.34375, "logps/chosen": -804.0, "logps/rejected": -1032.0, "loss": 0.2456, "rewards/accuracies": 0.8999999761581421, "rewards/chosen": -4.90625, "rewards/margins": 2.75, "rewards/rejected": -7.65625, "step": 1160 }, { "epoch": 1.2244897959183674, "grad_norm": 2.3706996918578445, "learning_rate": 4.68681013222621e-05, "logits/chosen": -5.59375, "logits/rejected": -3.65625, "logps/chosen": -820.0, "logps/rejected": -1056.0, "loss": 0.2102, "rewards/accuracies": 0.9312499761581421, "rewards/chosen": -5.09375, "rewards/margins": 2.78125, "rewards/rejected": -7.875, "step": 1170 }, { "epoch": 1.2349555206698064, "grad_norm": 3.001358387229265, "learning_rate": 4.677893941947006e-05, "logits/chosen": -6.53125, "logits/rejected": -4.75, "logps/chosen": -736.0, "logps/rejected": -980.0, "loss": 0.2553, "rewards/accuracies": 0.887499988079071, "rewards/chosen": -4.15625, "rewards/margins": 2.859375, "rewards/rejected": -7.03125, "step": 1180 }, { "epoch": 1.2454212454212454, "grad_norm": 2.714225581231745, "learning_rate": 4.668861338038126e-05, "logits/chosen": -5.34375, "logits/rejected": -3.390625, "logps/chosen": -728.0, "logps/rejected": -928.0, "loss": 0.2101, "rewards/accuracies": 0.8999999761581421, "rewards/chosen": -3.953125, "rewards/margins": 2.5, "rewards/rejected": -6.46875, "step": 1190 }, { "epoch": 1.2558869701726845, "grad_norm": 2.7020928838849194, "learning_rate": 4.6597128033138095e-05, "logits/chosen": -3.984375, "logits/rejected": -1.984375, "logps/chosen": -740.0, "logps/rejected": -960.0, "loss": 0.2202, "rewards/accuracies": 0.925000011920929, "rewards/chosen": -3.65625, "rewards/margins": 2.703125, "rewards/rejected": -6.375, "step": 1200 }, { "epoch": 1.2558869701726845, "eval_logits/chosen": -3.375, "eval_logits/rejected": -1.875, "eval_logps/chosen": -724.0, "eval_logps/rejected": -804.0, "eval_loss": 0.5942812561988831, "eval_rewards/accuracies": 0.708984375, "eval_rewards/chosen": -4.09375, "eval_rewards/margins": 1.078125, "eval_rewards/rejected": -5.1875, "eval_runtime": 52.81, "eval_samples_per_second": 37.872, "eval_steps_per_second": 0.606, "step": 1200 }, { "epoch": 1.2663526949241235, "grad_norm": 2.3938000016341037, "learning_rate": 4.650448826785073e-05, "logits/chosen": -4.25, "logits/rejected": -1.9765625, "logps/chosen": -776.0, "logps/rejected": -944.0, "loss": 0.238, "rewards/accuracies": 0.9312499761581421, "rewards/chosen": -3.875, "rewards/margins": 2.59375, "rewards/rejected": -6.46875, "step": 1210 }, { "epoch": 1.2768184196755625, "grad_norm": 2.4095410418117633, "learning_rate": 4.641069903633575e-05, "logits/chosen": -3.921875, "logits/rejected": -2.390625, "logps/chosen": -756.0, "logps/rejected": -956.0, "loss": 0.2538, "rewards/accuracies": 0.8812500238418579, "rewards/chosen": -4.34375, "rewards/margins": 2.5, "rewards/rejected": -6.8125, "step": 1220 }, { "epoch": 1.2872841444270016, "grad_norm": 2.71326099150034, "learning_rate": 4.631576535185146e-05, "logits/chosen": -5.15625, "logits/rejected": -3.375, "logps/chosen": -648.0, "logps/rejected": -864.0, "loss": 0.2384, "rewards/accuracies": 0.90625, "rewards/chosen": -3.5, "rewards/margins": 2.609375, "rewards/rejected": -6.09375, "step": 1230 }, { "epoch": 1.2977498691784406, "grad_norm": 2.743366901868589, "learning_rate": 4.621969228882987e-05, "logits/chosen": -6.25, "logits/rejected": -4.34375, "logps/chosen": -684.0, "logps/rejected": -936.0, "loss": 0.2121, "rewards/accuracies": 0.918749988079071, "rewards/chosen": -3.8125, "rewards/margins": 2.796875, "rewards/rejected": -6.625, "step": 1240 }, { "epoch": 1.3082155939298796, "grad_norm": 3.3400284830383797, "learning_rate": 4.612248498260554e-05, "logits/chosen": -6.6875, "logits/rejected": -4.25, "logps/chosen": -728.0, "logps/rejected": -944.0, "loss": 0.2304, "rewards/accuracies": 0.8999999761581421, "rewards/chosen": -3.890625, "rewards/margins": 2.828125, "rewards/rejected": -6.71875, "step": 1250 }, { "epoch": 1.3186813186813187, "grad_norm": 2.6309946499126577, "learning_rate": 4.602414862914101e-05, "logits/chosen": -5.21875, "logits/rejected": -3.703125, "logps/chosen": -720.0, "logps/rejected": -960.0, "loss": 0.2174, "rewards/accuracies": 0.893750011920929, "rewards/chosen": -4.1875, "rewards/margins": 2.59375, "rewards/rejected": -6.78125, "step": 1260 }, { "epoch": 1.3291470434327577, "grad_norm": 2.2785673557762522, "learning_rate": 4.5924688484749087e-05, "logits/chosen": -4.40625, "logits/rejected": -2.375, "logps/chosen": -764.0, "logps/rejected": -992.0, "loss": 0.2348, "rewards/accuracies": 0.887499988079071, "rewards/chosen": -4.53125, "rewards/margins": 2.609375, "rewards/rejected": -7.125, "step": 1270 }, { "epoch": 1.3396127681841967, "grad_norm": 2.676901490717523, "learning_rate": 4.582410986581191e-05, "logits/chosen": -2.6875, "logits/rejected": -0.134765625, "logps/chosen": -812.0, "logps/rejected": -1064.0, "loss": 0.2292, "rewards/accuracies": 0.8999999761581421, "rewards/chosen": -4.875, "rewards/margins": 3.0625, "rewards/rejected": -7.9375, "step": 1280 }, { "epoch": 1.3500784929356358, "grad_norm": 2.748893098182492, "learning_rate": 4.572241814849672e-05, "logits/chosen": -2.4375, "logits/rejected": -0.58984375, "logps/chosen": -764.0, "logps/rejected": -1004.0, "loss": 0.2677, "rewards/accuracies": 0.9312499761581421, "rewards/chosen": -4.34375, "rewards/margins": 2.84375, "rewards/rejected": -7.1875, "step": 1290 }, { "epoch": 1.3605442176870748, "grad_norm": 2.5611387716985607, "learning_rate": 4.561961876846853e-05, "logits/chosen": -3.96875, "logits/rejected": -2.5, "logps/chosen": -684.0, "logps/rejected": -912.0, "loss": 0.2472, "rewards/accuracies": 0.918749988079071, "rewards/chosen": -3.703125, "rewards/margins": 2.359375, "rewards/rejected": -6.0625, "step": 1300 }, { "epoch": 1.3605442176870748, "eval_logits/chosen": -3.6875, "eval_logits/rejected": -2.21875, "eval_logps/chosen": -760.0, "eval_logps/rejected": -828.0, "eval_loss": 0.5994687676429749, "eval_rewards/accuracies": 0.716796875, "eval_rewards/chosen": -4.4375, "eval_rewards/margins": 0.984375, "eval_rewards/rejected": -5.40625, "eval_runtime": 52.7983, "eval_samples_per_second": 37.88, "eval_steps_per_second": 0.606, "step": 1300 }, { "epoch": 1.3710099424385138, "grad_norm": 1.6282540391036573, "learning_rate": 4.55157172205996e-05, "logits/chosen": -4.25, "logits/rejected": -1.953125, "logps/chosen": -756.0, "logps/rejected": -940.0, "loss": 0.2387, "rewards/accuracies": 0.8999999761581421, "rewards/chosen": -4.25, "rewards/margins": 2.390625, "rewards/rejected": -6.65625, "step": 1310 }, { "epoch": 1.3814756671899528, "grad_norm": 3.4806185138527828, "learning_rate": 4.5410719058675654e-05, "logits/chosen": -3.203125, "logits/rejected": -1.8828125, "logps/chosen": -744.0, "logps/rejected": -972.0, "loss": 0.2337, "rewards/accuracies": 0.8812500238418579, "rewards/chosen": -4.5625, "rewards/margins": 2.5, "rewards/rejected": -7.0625, "step": 1320 }, { "epoch": 1.3919413919413919, "grad_norm": 2.7953097923118806, "learning_rate": 4.5304629895099085e-05, "logits/chosen": -3.828125, "logits/rejected": -2.78125, "logps/chosen": -760.0, "logps/rejected": -1040.0, "loss": 0.2346, "rewards/accuracies": 0.8999999761581421, "rewards/chosen": -4.46875, "rewards/margins": 2.6875, "rewards/rejected": -7.15625, "step": 1330 }, { "epoch": 1.402407116692831, "grad_norm": 2.5716033588233125, "learning_rate": 4.51974554005889e-05, "logits/chosen": -3.9375, "logits/rejected": -1.5703125, "logps/chosen": -804.0, "logps/rejected": -1032.0, "loss": 0.2277, "rewards/accuracies": 0.875, "rewards/chosen": -4.6875, "rewards/margins": 2.8125, "rewards/rejected": -7.5, "step": 1340 }, { "epoch": 1.41287284144427, "grad_norm": 2.672897305019244, "learning_rate": 4.508920130387764e-05, "logits/chosen": -3.953125, "logits/rejected": -1.7421875, "logps/chosen": -724.0, "logps/rejected": -1004.0, "loss": 0.2336, "rewards/accuracies": 0.893750011920929, "rewards/chosen": -4.09375, "rewards/margins": 3.078125, "rewards/rejected": -7.1875, "step": 1350 }, { "epoch": 1.423338566195709, "grad_norm": 2.185839165890125, "learning_rate": 4.497987339140519e-05, "logits/chosen": -5.03125, "logits/rejected": -3.140625, "logps/chosen": -700.0, "logps/rejected": -920.0, "loss": 0.2334, "rewards/accuracies": 0.9125000238418579, "rewards/chosen": -3.796875, "rewards/margins": 2.6875, "rewards/rejected": -6.46875, "step": 1360 }, { "epoch": 1.433804290947148, "grad_norm": 3.9172194922718746, "learning_rate": 4.4869477507009424e-05, "logits/chosen": -4.4375, "logits/rejected": -2.859375, "logps/chosen": -768.0, "logps/rejected": -988.0, "loss": 0.2671, "rewards/accuracies": 0.887499988079071, "rewards/chosen": -4.625, "rewards/margins": 2.546875, "rewards/rejected": -7.1875, "step": 1370 }, { "epoch": 1.4442700156985873, "grad_norm": 1.5309369238147958, "learning_rate": 4.475801955161386e-05, "logits/chosen": -4.96875, "logits/rejected": -2.859375, "logps/chosen": -796.0, "logps/rejected": -1024.0, "loss": 0.2203, "rewards/accuracies": 0.925000011920929, "rewards/chosen": -4.78125, "rewards/margins": 2.71875, "rewards/rejected": -7.5, "step": 1380 }, { "epoch": 1.454735740450026, "grad_norm": 2.400254274178371, "learning_rate": 4.464550548291227e-05, "logits/chosen": -6.1875, "logits/rejected": -3.9375, "logps/chosen": -760.0, "logps/rejected": -976.0, "loss": 0.2565, "rewards/accuracies": 0.875, "rewards/chosen": -4.375, "rewards/margins": 2.734375, "rewards/rejected": -7.09375, "step": 1390 }, { "epoch": 1.4652014652014653, "grad_norm": 3.2947707252238945, "learning_rate": 4.453194131505018e-05, "logits/chosen": -6.1875, "logits/rejected": -5.0, "logps/chosen": -816.0, "logps/rejected": -1056.0, "loss": 0.2406, "rewards/accuracies": 0.887499988079071, "rewards/chosen": -4.75, "rewards/margins": 2.5, "rewards/rejected": -7.25, "step": 1400 }, { "epoch": 1.4652014652014653, "eval_logits/chosen": -5.28125, "eval_logits/rejected": -3.875, "eval_logps/chosen": -836.0, "eval_logps/rejected": -908.0, "eval_loss": 0.5971249938011169, "eval_rewards/accuracies": 0.71875, "eval_rewards/chosen": -5.21875, "eval_rewards/margins": 1.015625, "eval_rewards/rejected": -6.21875, "eval_runtime": 52.8282, "eval_samples_per_second": 37.859, "eval_steps_per_second": 0.606, "step": 1400 }, { "epoch": 1.4756671899529041, "grad_norm": 2.504445176515428, "learning_rate": 4.4417333118303434e-05, "logits/chosen": -4.9375, "logits/rejected": -2.90625, "logps/chosen": -764.0, "logps/rejected": -1016.0, "loss": 0.2169, "rewards/accuracies": 0.9125000238418579, "rewards/chosen": -4.5625, "rewards/margins": 2.734375, "rewards/rejected": -7.28125, "step": 1410 }, { "epoch": 1.4861329147043434, "grad_norm": 1.890385623018933, "learning_rate": 4.430168701875371e-05, "logits/chosen": -1.765625, "logits/rejected": -0.3359375, "logps/chosen": -844.0, "logps/rejected": -1064.0, "loss": 0.227, "rewards/accuracies": 0.887499988079071, "rewards/chosen": -5.03125, "rewards/margins": 2.671875, "rewards/rejected": -7.71875, "step": 1420 }, { "epoch": 1.4965986394557822, "grad_norm": 3.1961582028367768, "learning_rate": 4.418500919796107e-05, "logits/chosen": -2.5625, "logits/rejected": -0.37890625, "logps/chosen": -828.0, "logps/rejected": -1128.0, "loss": 0.2094, "rewards/accuracies": 0.949999988079071, "rewards/chosen": -4.90625, "rewards/margins": 3.328125, "rewards/rejected": -8.25, "step": 1430 }, { "epoch": 1.5070643642072215, "grad_norm": 1.7802682782877906, "learning_rate": 4.4067305892633526e-05, "logits/chosen": -3.1875, "logits/rejected": -1.2734375, "logps/chosen": -844.0, "logps/rejected": -1120.0, "loss": 0.2196, "rewards/accuracies": 0.9125000238418579, "rewards/chosen": -5.03125, "rewards/margins": 3.171875, "rewards/rejected": -8.1875, "step": 1440 }, { "epoch": 1.5175300889586603, "grad_norm": 2.835575157287789, "learning_rate": 4.3948583394293716e-05, "logits/chosen": -3.671875, "logits/rejected": -2.046875, "logps/chosen": -924.0, "logps/rejected": -1168.0, "loss": 0.2255, "rewards/accuracies": 0.893750011920929, "rewards/chosen": -5.5625, "rewards/margins": 2.9375, "rewards/rejected": -8.5, "step": 1450 }, { "epoch": 1.5279958137100995, "grad_norm": 2.68427303811402, "learning_rate": 4.382884804894255e-05, "logits/chosen": -4.4375, "logits/rejected": -2.5625, "logps/chosen": -820.0, "logps/rejected": -1120.0, "loss": 0.1839, "rewards/accuracies": 0.925000011920929, "rewards/chosen": -4.9375, "rewards/margins": 3.078125, "rewards/rejected": -8.0, "step": 1460 }, { "epoch": 1.5384615384615383, "grad_norm": 2.9253750364643425, "learning_rate": 4.3708106256720025e-05, "logits/chosen": -5.59375, "logits/rejected": -4.03125, "logps/chosen": -772.0, "logps/rejected": -1048.0, "loss": 0.2311, "rewards/accuracies": 0.9125000238418579, "rewards/chosen": -4.15625, "rewards/margins": 2.90625, "rewards/rejected": -7.09375, "step": 1470 }, { "epoch": 1.5489272632129776, "grad_norm": 1.9394845104409457, "learning_rate": 4.358636447156316e-05, "logits/chosen": -4.40625, "logits/rejected": -2.3125, "logps/chosen": -744.0, "logps/rejected": -928.0, "loss": 0.2532, "rewards/accuracies": 0.862500011920929, "rewards/chosen": -4.46875, "rewards/margins": 2.46875, "rewards/rejected": -6.9375, "step": 1480 }, { "epoch": 1.5593929879644164, "grad_norm": 2.018780263579091, "learning_rate": 4.346362920086094e-05, "logits/chosen": -5.09375, "logits/rejected": -2.953125, "logps/chosen": -828.0, "logps/rejected": -1016.0, "loss": 0.2116, "rewards/accuracies": 0.862500011920929, "rewards/chosen": -4.90625, "rewards/margins": 2.34375, "rewards/rejected": -7.25, "step": 1490 }, { "epoch": 1.5698587127158556, "grad_norm": 1.9518425840610047, "learning_rate": 4.333990700510656e-05, "logits/chosen": -6.125, "logits/rejected": -4.15625, "logps/chosen": -832.0, "logps/rejected": -1080.0, "loss": 0.2059, "rewards/accuracies": 0.925000011920929, "rewards/chosen": -4.9375, "rewards/margins": 2.703125, "rewards/rejected": -7.65625, "step": 1500 }, { "epoch": 1.5698587127158556, "eval_logits/chosen": -5.78125, "eval_logits/rejected": -4.21875, "eval_logps/chosen": -848.0, "eval_logps/rejected": -940.0, "eval_loss": 0.6052031517028809, "eval_rewards/accuracies": 0.71484375, "eval_rewards/chosen": -5.34375, "eval_rewards/margins": 1.1953125, "eval_rewards/rejected": -6.53125, "eval_runtime": 52.836, "eval_samples_per_second": 37.853, "eval_steps_per_second": 0.606, "step": 1500 }, { "epoch": 1.5803244374672945, "grad_norm": 2.1970202585834695, "learning_rate": 4.321520449754669e-05, "logits/chosen": -6.15625, "logits/rejected": -4.34375, "logps/chosen": -820.0, "logps/rejected": -1072.0, "loss": 0.2231, "rewards/accuracies": 0.949999988079071, "rewards/chosen": -4.71875, "rewards/margins": 2.890625, "rewards/rejected": -7.625, "step": 1510 }, { "epoch": 1.5907901622187337, "grad_norm": 2.7541131093791233, "learning_rate": 4.308952834382802e-05, "logits/chosen": -7.75, "logits/rejected": -5.75, "logps/chosen": -816.0, "logps/rejected": -1048.0, "loss": 0.2373, "rewards/accuracies": 0.8999999761581421, "rewards/chosen": -4.8125, "rewards/margins": 2.640625, "rewards/rejected": -7.4375, "step": 1520 }, { "epoch": 1.6012558869701727, "grad_norm": 2.1173531392884413, "learning_rate": 4.296288526164094e-05, "logits/chosen": -7.1875, "logits/rejected": -4.5625, "logps/chosen": -836.0, "logps/rejected": -1056.0, "loss": 0.2056, "rewards/accuracies": 0.9375, "rewards/chosen": -5.0625, "rewards/margins": 2.9375, "rewards/rejected": -8.0, "step": 1530 }, { "epoch": 1.6117216117216118, "grad_norm": 1.871611457195126, "learning_rate": 4.283528202036049e-05, "logits/chosen": -8.0625, "logits/rejected": -5.78125, "logps/chosen": -808.0, "logps/rejected": -1016.0, "loss": 0.223, "rewards/accuracies": 0.918749988079071, "rewards/chosen": -4.59375, "rewards/margins": 2.640625, "rewards/rejected": -7.25, "step": 1540 }, { "epoch": 1.6221873364730508, "grad_norm": 1.7973566338479199, "learning_rate": 4.270672544068448e-05, "logits/chosen": -8.3125, "logits/rejected": -6.1875, "logps/chosen": -816.0, "logps/rejected": -1080.0, "loss": 0.2537, "rewards/accuracies": 0.8999999761581421, "rewards/chosen": -4.65625, "rewards/margins": 2.890625, "rewards/rejected": -7.5625, "step": 1550 }, { "epoch": 1.6326530612244898, "grad_norm": 2.0569039776175493, "learning_rate": 4.257722239426897e-05, "logits/chosen": -7.75, "logits/rejected": -5.34375, "logps/chosen": -784.0, "logps/rejected": -1024.0, "loss": 0.2426, "rewards/accuracies": 0.90625, "rewards/chosen": -4.4375, "rewards/margins": 2.90625, "rewards/rejected": -7.375, "step": 1560 }, { "epoch": 1.6431187859759289, "grad_norm": 2.0178296101250197, "learning_rate": 4.2446779803360895e-05, "logits/chosen": -6.1875, "logits/rejected": -4.03125, "logps/chosen": -772.0, "logps/rejected": -996.0, "loss": 0.2115, "rewards/accuracies": 0.918749988079071, "rewards/chosen": -4.6875, "rewards/margins": 2.6875, "rewards/rejected": -7.375, "step": 1570 }, { "epoch": 1.653584510727368, "grad_norm": 1.79335431596216, "learning_rate": 4.23154046404281e-05, "logits/chosen": -6.0625, "logits/rejected": -3.625, "logps/chosen": -848.0, "logps/rejected": -1080.0, "loss": 0.1949, "rewards/accuracies": 0.9375, "rewards/chosen": -5.1875, "rewards/margins": 2.890625, "rewards/rejected": -8.0625, "step": 1580 }, { "epoch": 1.664050235478807, "grad_norm": 2.290555032802585, "learning_rate": 4.218310392778662e-05, "logits/chosen": -5.125, "logits/rejected": -2.296875, "logps/chosen": -880.0, "logps/rejected": -1144.0, "loss": 0.229, "rewards/accuracies": 0.925000011920929, "rewards/chosen": -5.5625, "rewards/margins": 3.015625, "rewards/rejected": -8.5625, "step": 1590 }, { "epoch": 1.674515960230246, "grad_norm": 2.115801499090191, "learning_rate": 4.204988473722535e-05, "logits/chosen": -5.625, "logits/rejected": -3.40625, "logps/chosen": -832.0, "logps/rejected": -1072.0, "loss": 0.2305, "rewards/accuracies": 0.918749988079071, "rewards/chosen": -4.9375, "rewards/margins": 2.921875, "rewards/rejected": -7.84375, "step": 1600 }, { "epoch": 1.674515960230246, "eval_logits/chosen": -6.6875, "eval_logits/rejected": -5.15625, "eval_logps/chosen": -800.0, "eval_logps/rejected": -876.0, "eval_loss": 0.6067500114440918, "eval_rewards/accuracies": 0.71875, "eval_rewards/chosen": -4.875, "eval_rewards/margins": 1.0390625, "eval_rewards/rejected": -5.90625, "eval_runtime": 52.8557, "eval_samples_per_second": 37.839, "eval_steps_per_second": 0.605, "step": 1600 }, { "epoch": 1.684981684981685, "grad_norm": 1.8649325471458336, "learning_rate": 4.1915754189628e-05, "logits/chosen": -8.0, "logits/rejected": -5.90625, "logps/chosen": -796.0, "logps/rejected": -1040.0, "loss": 0.2418, "rewards/accuracies": 0.90625, "rewards/chosen": -4.375, "rewards/margins": 2.78125, "rewards/rejected": -7.15625, "step": 1610 }, { "epoch": 1.695447409733124, "grad_norm": 2.3340385989698715, "learning_rate": 4.1780719454592496e-05, "logits/chosen": -6.9375, "logits/rejected": -4.6875, "logps/chosen": -744.0, "logps/rejected": -960.0, "loss": 0.2388, "rewards/accuracies": 0.9375, "rewards/chosen": -4.21875, "rewards/margins": 2.609375, "rewards/rejected": -6.84375, "step": 1620 }, { "epoch": 1.705913134484563, "grad_norm": 3.5283350333951633, "learning_rate": 4.164478775004775e-05, "logits/chosen": -5.65625, "logits/rejected": -3.703125, "logps/chosen": -764.0, "logps/rejected": -1012.0, "loss": 0.2165, "rewards/accuracies": 0.893750011920929, "rewards/chosen": -4.6875, "rewards/margins": 2.6875, "rewards/rejected": -7.34375, "step": 1630 }, { "epoch": 1.716378859236002, "grad_norm": 2.8779850397285642, "learning_rate": 4.150796634186784e-05, "logits/chosen": -6.5, "logits/rejected": -4.59375, "logps/chosen": -828.0, "logps/rejected": -1152.0, "loss": 0.218, "rewards/accuracies": 0.9375, "rewards/chosen": -4.71875, "rewards/margins": 3.390625, "rewards/rejected": -8.125, "step": 1640 }, { "epoch": 1.7268445839874411, "grad_norm": 2.481408886713114, "learning_rate": 4.1370262543483595e-05, "logits/chosen": -6.28125, "logits/rejected": -4.25, "logps/chosen": -808.0, "logps/rejected": -1064.0, "loss": 0.2176, "rewards/accuracies": 0.9312499761581421, "rewards/chosen": -4.75, "rewards/margins": 2.9375, "rewards/rejected": -7.6875, "step": 1650 }, { "epoch": 1.7373103087388801, "grad_norm": 2.1094531143461777, "learning_rate": 4.123168371549175e-05, "logits/chosen": -4.3125, "logits/rejected": -3.015625, "logps/chosen": -864.0, "logps/rejected": -1120.0, "loss": 0.2373, "rewards/accuracies": 0.8687499761581421, "rewards/chosen": -5.5, "rewards/margins": 2.890625, "rewards/rejected": -8.375, "step": 1660 }, { "epoch": 1.7477760334903192, "grad_norm": 2.081186434782386, "learning_rate": 4.1092237265261426e-05, "logits/chosen": -5.0625, "logits/rejected": -2.703125, "logps/chosen": -888.0, "logps/rejected": -1128.0, "loss": 0.2195, "rewards/accuracies": 0.8999999761581421, "rewards/chosen": -5.59375, "rewards/margins": 2.828125, "rewards/rejected": -8.4375, "step": 1670 }, { "epoch": 1.7582417582417582, "grad_norm": 3.7374931598597065, "learning_rate": 4.095193064653823e-05, "logits/chosen": -4.46875, "logits/rejected": -2.578125, "logps/chosen": -828.0, "logps/rejected": -1080.0, "loss": 0.2421, "rewards/accuracies": 0.893750011920929, "rewards/chosen": -5.4375, "rewards/margins": 2.8125, "rewards/rejected": -8.25, "step": 1680 }, { "epoch": 1.7687074829931972, "grad_norm": 2.5735549072900583, "learning_rate": 4.081077135904584e-05, "logits/chosen": -6.34375, "logits/rejected": -3.703125, "logps/chosen": -920.0, "logps/rejected": -1104.0, "loss": 0.217, "rewards/accuracies": 0.918749988079071, "rewards/chosen": -5.71875, "rewards/margins": 2.625, "rewards/rejected": -8.375, "step": 1690 }, { "epoch": 1.7791732077446363, "grad_norm": 2.9795017126662375, "learning_rate": 4.066876694808512e-05, "logits/chosen": -6.3125, "logits/rejected": -4.46875, "logps/chosen": -836.0, "logps/rejected": -1072.0, "loss": 0.2327, "rewards/accuracies": 0.856249988079071, "rewards/chosen": -5.25, "rewards/margins": 2.640625, "rewards/rejected": -7.875, "step": 1700 }, { "epoch": 1.7791732077446363, "eval_logits/chosen": -6.0625, "eval_logits/rejected": -4.5, "eval_logps/chosen": -908.0, "eval_logps/rejected": -1000.0, "eval_loss": 0.6141093969345093, "eval_rewards/accuracies": 0.716796875, "eval_rewards/chosen": -5.9375, "eval_rewards/margins": 1.21875, "eval_rewards/rejected": -7.15625, "eval_runtime": 52.8573, "eval_samples_per_second": 37.838, "eval_steps_per_second": 0.605, "step": 1700 }, { "epoch": 1.7896389324960753, "grad_norm": 2.246296295697, "learning_rate": 4.052592500413079e-05, "logits/chosen": -6.53125, "logits/rejected": -4.5625, "logps/chosen": -892.0, "logps/rejected": -1064.0, "loss": 0.2642, "rewards/accuracies": 0.90625, "rewards/chosen": -5.28125, "rewards/margins": 2.578125, "rewards/rejected": -7.875, "step": 1710 }, { "epoch": 1.8001046572475143, "grad_norm": 1.5806277059274703, "learning_rate": 4.0382253162425696e-05, "logits/chosen": -6.96875, "logits/rejected": -5.46875, "logps/chosen": -812.0, "logps/rejected": -1008.0, "loss": 0.2702, "rewards/accuracies": 0.925000011920929, "rewards/chosen": -4.78125, "rewards/margins": 2.421875, "rewards/rejected": -7.1875, "step": 1720 }, { "epoch": 1.8105703819989536, "grad_norm": 2.273534363986886, "learning_rate": 4.023775910257271e-05, "logits/chosen": -6.96875, "logits/rejected": -4.46875, "logps/chosen": -844.0, "logps/rejected": -1080.0, "loss": 0.2014, "rewards/accuracies": 0.9125000238418579, "rewards/chosen": -5.09375, "rewards/margins": 2.890625, "rewards/rejected": -7.96875, "step": 1730 }, { "epoch": 1.8210361067503924, "grad_norm": 3.5238832041647283, "learning_rate": 4.0092450548124235e-05, "logits/chosen": -5.09375, "logits/rejected": -2.28125, "logps/chosen": -1032.0, "logps/rejected": -1288.0, "loss": 0.2676, "rewards/accuracies": 0.887499988079071, "rewards/chosen": -6.84375, "rewards/margins": 3.140625, "rewards/rejected": -10.0, "step": 1740 }, { "epoch": 1.8315018315018317, "grad_norm": 3.6999864745213786, "learning_rate": 3.994633526616936e-05, "logits/chosen": -6.53125, "logits/rejected": -3.46875, "logps/chosen": -944.0, "logps/rejected": -1144.0, "loss": 0.2675, "rewards/accuracies": 0.90625, "rewards/chosen": -6.03125, "rewards/margins": 2.671875, "rewards/rejected": -8.6875, "step": 1750 }, { "epoch": 1.8419675562532705, "grad_norm": 2.2019977227457685, "learning_rate": 3.9799421066918653e-05, "logits/chosen": -7.71875, "logits/rejected": -6.1875, "logps/chosen": -840.0, "logps/rejected": -1056.0, "loss": 0.2259, "rewards/accuracies": 0.9375, "rewards/chosen": -5.28125, "rewards/margins": 2.4375, "rewards/rejected": -7.71875, "step": 1760 }, { "epoch": 1.8524332810047097, "grad_norm": 1.7022096305520797, "learning_rate": 3.965171580328675e-05, "logits/chosen": -6.8125, "logits/rejected": -5.1875, "logps/chosen": -888.0, "logps/rejected": -1104.0, "loss": 0.2178, "rewards/accuracies": 0.9312499761581421, "rewards/chosen": -5.34375, "rewards/margins": 2.734375, "rewards/rejected": -8.0625, "step": 1770 }, { "epoch": 1.8628990057561485, "grad_norm": 2.5728669879495243, "learning_rate": 3.950322737047254e-05, "logits/chosen": -6.5, "logits/rejected": -4.4375, "logps/chosen": -940.0, "logps/rejected": -1208.0, "loss": 0.2442, "rewards/accuracies": 0.9312499761581421, "rewards/chosen": -5.71875, "rewards/margins": 3.046875, "rewards/rejected": -8.8125, "step": 1780 }, { "epoch": 1.8733647305075878, "grad_norm": 3.3021132407979272, "learning_rate": 3.93539637055372e-05, "logits/chosen": -6.0, "logits/rejected": -3.84375, "logps/chosen": -896.0, "logps/rejected": -1120.0, "loss": 0.2241, "rewards/accuracies": 0.893750011920929, "rewards/chosen": -5.71875, "rewards/margins": 2.703125, "rewards/rejected": -8.4375, "step": 1790 }, { "epoch": 1.8838304552590266, "grad_norm": 3.5449002725785066, "learning_rate": 3.9203932786979904e-05, "logits/chosen": -4.875, "logits/rejected": -2.703125, "logps/chosen": -848.0, "logps/rejected": -1104.0, "loss": 0.2221, "rewards/accuracies": 0.9437500238418579, "rewards/chosen": -5.4375, "rewards/margins": 2.9375, "rewards/rejected": -8.375, "step": 1800 }, { "epoch": 1.8838304552590266, "eval_logits/chosen": -3.625, "eval_logits/rejected": -1.984375, "eval_logps/chosen": -960.0, "eval_logps/rejected": -1048.0, "eval_loss": 0.607171893119812, "eval_rewards/accuracies": 0.7265625, "eval_rewards/chosen": -6.46875, "eval_rewards/margins": 1.1875, "eval_rewards/rejected": -7.65625, "eval_runtime": 52.8636, "eval_samples_per_second": 37.833, "eval_steps_per_second": 0.605, "step": 1800 }, { "epoch": 1.8942961800104658, "grad_norm": 2.2325981090247087, "learning_rate": 3.905314263431136e-05, "logits/chosen": -2.609375, "logits/rejected": -0.62890625, "logps/chosen": -912.0, "logps/rejected": -1192.0, "loss": 0.2316, "rewards/accuracies": 0.9375, "rewards/chosen": -6.15625, "rewards/margins": 3.03125, "rewards/rejected": -9.1875, "step": 1810 }, { "epoch": 1.9047619047619047, "grad_norm": 2.755081964679097, "learning_rate": 3.8901601307625144e-05, "logits/chosen": -2.609375, "logits/rejected": -0.63671875, "logps/chosen": -948.0, "logps/rejected": -1176.0, "loss": 0.2322, "rewards/accuracies": 0.893750011920929, "rewards/chosen": -6.4375, "rewards/margins": 2.640625, "rewards/rejected": -9.0625, "step": 1820 }, { "epoch": 1.915227629513344, "grad_norm": 2.81171176700697, "learning_rate": 3.87493169071669e-05, "logits/chosen": -3.78125, "logits/rejected": -2.3125, "logps/chosen": -796.0, "logps/rejected": -1072.0, "loss": 0.2409, "rewards/accuracies": 0.9312499761581421, "rewards/chosen": -5.125, "rewards/margins": 2.640625, "rewards/rejected": -7.78125, "step": 1830 }, { "epoch": 1.9256933542647827, "grad_norm": 2.959505766153652, "learning_rate": 3.859629757290132e-05, "logits/chosen": -5.625, "logits/rejected": -3.234375, "logps/chosen": -848.0, "logps/rejected": -1056.0, "loss": 0.2436, "rewards/accuracies": 0.90625, "rewards/chosen": -4.90625, "rewards/margins": 2.75, "rewards/rejected": -7.65625, "step": 1840 }, { "epoch": 1.936159079016222, "grad_norm": 2.224627550716824, "learning_rate": 3.8442551484077055e-05, "logits/chosen": -5.375, "logits/rejected": -3.21875, "logps/chosen": -836.0, "logps/rejected": -1080.0, "loss": 0.2241, "rewards/accuracies": 0.893750011920929, "rewards/chosen": -5.1875, "rewards/margins": 2.75, "rewards/rejected": -7.9375, "step": 1850 }, { "epoch": 1.9466248037676608, "grad_norm": 2.450233635234575, "learning_rate": 3.828808685878955e-05, "logits/chosen": -4.59375, "logits/rejected": -2.328125, "logps/chosen": -872.0, "logps/rejected": -1088.0, "loss": 0.2398, "rewards/accuracies": 0.887499988079071, "rewards/chosen": -5.625, "rewards/margins": 2.71875, "rewards/rejected": -8.375, "step": 1860 }, { "epoch": 1.9570905285191, "grad_norm": 2.25540433870103, "learning_rate": 3.8132911953541715e-05, "logits/chosen": -4.75, "logits/rejected": -1.8828125, "logps/chosen": -908.0, "logps/rejected": -1144.0, "loss": 0.202, "rewards/accuracies": 0.9375, "rewards/chosen": -5.34375, "rewards/margins": 3.109375, "rewards/rejected": -8.5, "step": 1870 }, { "epoch": 1.9675562532705388, "grad_norm": 2.52329283794622, "learning_rate": 3.7977035062802636e-05, "logits/chosen": -5.21875, "logits/rejected": -3.125, "logps/chosen": -844.0, "logps/rejected": -1080.0, "loss": 0.2225, "rewards/accuracies": 0.90625, "rewards/chosen": -5.1875, "rewards/margins": 2.9375, "rewards/rejected": -8.125, "step": 1880 }, { "epoch": 1.978021978021978, "grad_norm": 2.820473355276973, "learning_rate": 3.782046451856417e-05, "logits/chosen": -6.75, "logits/rejected": -4.59375, "logps/chosen": -816.0, "logps/rejected": -1056.0, "loss": 0.2529, "rewards/accuracies": 0.856249988079071, "rewards/chosen": -4.9375, "rewards/margins": 2.8125, "rewards/rejected": -7.71875, "step": 1890 }, { "epoch": 1.988487702773417, "grad_norm": 2.7800488605542393, "learning_rate": 3.7663208689895654e-05, "logits/chosen": -5.96875, "logits/rejected": -4.25, "logps/chosen": -904.0, "logps/rejected": -1144.0, "loss": 0.2153, "rewards/accuracies": 0.9125000238418579, "rewards/chosen": -5.59375, "rewards/margins": 2.796875, "rewards/rejected": -8.375, "step": 1900 }, { "epoch": 1.988487702773417, "eval_logits/chosen": -4.875, "eval_logits/rejected": -3.3125, "eval_logps/chosen": -964.0, "eval_logps/rejected": -1056.0, "eval_loss": 0.594921886920929, "eval_rewards/accuracies": 0.7265625, "eval_rewards/chosen": -6.5, "eval_rewards/margins": 1.1953125, "eval_rewards/rejected": -7.6875, "eval_runtime": 52.89, "eval_samples_per_second": 37.814, "eval_steps_per_second": 0.605, "step": 1900 }, { "epoch": 1.9989534275248562, "grad_norm": 2.7990258770794254, "learning_rate": 3.750527598249647e-05, "logits/chosen": -5.1875, "logits/rejected": -3.125, "logps/chosen": -928.0, "logps/rejected": -1152.0, "loss": 0.2225, "rewards/accuracies": 0.90625, "rewards/chosen": -5.875, "rewards/margins": 2.828125, "rewards/rejected": -8.6875, "step": 1910 }, { "epoch": 2.009419152276295, "grad_norm": 0.33032701679760734, "learning_rate": 3.734667483824681e-05, "logits/chosen": -6.125, "logits/rejected": -2.78125, "logps/chosen": -892.0, "logps/rejected": -1312.0, "loss": 0.0487, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -5.28125, "rewards/margins": 4.875, "rewards/rejected": -10.1875, "step": 1920 }, { "epoch": 2.0198848770277342, "grad_norm": 0.7845585783913027, "learning_rate": 3.718741373475641e-05, "logits/chosen": -4.21875, "logits/rejected": -0.78515625, "logps/chosen": -900.0, "logps/rejected": -1424.0, "loss": 0.0395, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -6.0625, "rewards/margins": 5.4375, "rewards/rejected": -11.5, "step": 1930 }, { "epoch": 2.030350601779173, "grad_norm": 0.5975494289154536, "learning_rate": 3.7027501184911385e-05, "logits/chosen": -4.03125, "logits/rejected": 0.1279296875, "logps/chosen": -984.0, "logps/rejected": -1552.0, "loss": 0.0278, "rewards/accuracies": 1.0, "rewards/chosen": -6.75, "rewards/margins": 6.09375, "rewards/rejected": -12.875, "step": 1940 }, { "epoch": 2.0408163265306123, "grad_norm": 0.6075478455420598, "learning_rate": 3.6866945736419234e-05, "logits/chosen": -3.75, "logits/rejected": 0.6953125, "logps/chosen": -1004.0, "logps/rejected": -1592.0, "loss": 0.0341, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -6.96875, "rewards/margins": 6.34375, "rewards/rejected": -13.3125, "step": 1950 }, { "epoch": 2.051282051282051, "grad_norm": 0.5887917714033745, "learning_rate": 3.67057559713519e-05, "logits/chosen": -3.1875, "logits/rejected": -0.80859375, "logps/chosen": -924.0, "logps/rejected": -1560.0, "loss": 0.0278, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -6.40625, "rewards/margins": 6.28125, "rewards/rejected": -12.6875, "step": 1960 }, { "epoch": 2.0617477760334904, "grad_norm": 1.663357092082489, "learning_rate": 3.6543940505687096e-05, "logits/chosen": -3.75, "logits/rejected": 0.06201171875, "logps/chosen": -1008.0, "logps/rejected": -1600.0, "loss": 0.0317, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -6.8125, "rewards/margins": 6.34375, "rewards/rejected": -13.125, "step": 1970 }, { "epoch": 2.072213500784929, "grad_norm": 3.371311980994405, "learning_rate": 3.638150798884768e-05, "logits/chosen": -3.515625, "logits/rejected": -0.2392578125, "logps/chosen": -968.0, "logps/rejected": -1568.0, "loss": 0.0309, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -6.4375, "rewards/margins": 6.375, "rewards/rejected": -12.8125, "step": 1980 }, { "epoch": 2.0826792255363684, "grad_norm": 0.9209075646693089, "learning_rate": 3.621846710323941e-05, "logits/chosen": -3.25, "logits/rejected": 0.0576171875, "logps/chosen": -956.0, "logps/rejected": -1528.0, "loss": 0.0295, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -6.25, "rewards/margins": 6.21875, "rewards/rejected": -12.4375, "step": 1990 }, { "epoch": 2.0931449502878072, "grad_norm": 0.3031417723240036, "learning_rate": 3.605482656378677e-05, "logits/chosen": -3.65625, "logits/rejected": 0.36328125, "logps/chosen": -912.0, "logps/rejected": -1528.0, "loss": 0.0215, "rewards/accuracies": 1.0, "rewards/chosen": -6.125, "rewards/margins": 6.5625, "rewards/rejected": -12.6875, "step": 2000 }, { "epoch": 2.0931449502878072, "eval_logits/chosen": -1.9609375, "eval_logits/rejected": -0.107421875, "eval_logps/chosen": -1184.0, "eval_logps/rejected": -1336.0, "eval_loss": 0.746999979019165, "eval_rewards/accuracies": 0.724609375, "eval_rewards/chosen": -8.6875, "eval_rewards/margins": 1.8125, "eval_rewards/rejected": -10.5, "eval_runtime": 52.8881, "eval_samples_per_second": 37.816, "eval_steps_per_second": 0.605, "step": 2000 }, { "epoch": 2.1036106750392465, "grad_norm": 1.9374058291925411, "learning_rate": 3.589059511746722e-05, "logits/chosen": -2.953125, "logits/rejected": 0.369140625, "logps/chosen": -1096.0, "logps/rejected": -1768.0, "loss": 0.0289, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -7.5625, "rewards/margins": 7.09375, "rewards/rejected": -14.625, "step": 2010 }, { "epoch": 2.1140763997906853, "grad_norm": 0.4050763470400383, "learning_rate": 3.572578154284356e-05, "logits/chosen": -3.5, "logits/rejected": 0.047119140625, "logps/chosen": -1096.0, "logps/rejected": -1784.0, "loss": 0.0228, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -7.75, "rewards/margins": 7.21875, "rewards/rejected": -15.0, "step": 2020 }, { "epoch": 2.1245421245421245, "grad_norm": 0.4988691556623063, "learning_rate": 3.556039464959474e-05, "logits/chosen": -4.375, "logits/rejected": -0.333984375, "logps/chosen": -1040.0, "logps/rejected": -1680.0, "loss": 0.0285, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -7.125, "rewards/margins": 6.90625, "rewards/rejected": -14.0, "step": 2030 }, { "epoch": 2.1350078492935634, "grad_norm": 0.3314005859234191, "learning_rate": 3.539444327804498e-05, "logits/chosen": -3.390625, "logits/rejected": -0.671875, "logps/chosen": -1032.0, "logps/rejected": -1696.0, "loss": 0.0305, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -7.3125, "rewards/margins": 6.78125, "rewards/rejected": -14.0625, "step": 2040 }, { "epoch": 2.1454735740450026, "grad_norm": 0.5736427946721916, "learning_rate": 3.5227936298691214e-05, "logits/chosen": -2.546875, "logits/rejected": 0.8984375, "logps/chosen": -1088.0, "logps/rejected": -1704.0, "loss": 0.037, "rewards/accuracies": 0.96875, "rewards/chosen": -7.71875, "rewards/margins": 6.5625, "rewards/rejected": -14.25, "step": 2050 }, { "epoch": 2.155939298796442, "grad_norm": 1.3303976437021603, "learning_rate": 3.5060882611728915e-05, "logits/chosen": -3.5625, "logits/rejected": -0.2734375, "logps/chosen": -976.0, "logps/rejected": -1544.0, "loss": 0.0293, "rewards/accuracies": 1.0, "rewards/chosen": -6.3125, "rewards/margins": 6.1875, "rewards/rejected": -12.5, "step": 2060 }, { "epoch": 2.1664050235478807, "grad_norm": 0.3792335895581482, "learning_rate": 3.4893291146576404e-05, "logits/chosen": -2.8125, "logits/rejected": 0.12890625, "logps/chosen": -940.0, "logps/rejected": -1576.0, "loss": 0.0204, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -6.375, "rewards/margins": 6.5, "rewards/rejected": -12.875, "step": 2070 }, { "epoch": 2.17687074829932, "grad_norm": 1.0854147028131242, "learning_rate": 3.4725170861397524e-05, "logits/chosen": -4.59375, "logits/rejected": -0.474609375, "logps/chosen": -948.0, "logps/rejected": -1584.0, "loss": 0.0327, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -6.09375, "rewards/margins": 6.71875, "rewards/rejected": -12.8125, "step": 2080 }, { "epoch": 2.1873364730507587, "grad_norm": 1.263333596618435, "learning_rate": 3.45565307426228e-05, "logits/chosen": -2.125, "logits/rejected": 1.1484375, "logps/chosen": -1136.0, "logps/rejected": -1840.0, "loss": 0.0321, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -7.90625, "rewards/margins": 7.5, "rewards/rejected": -15.375, "step": 2090 }, { "epoch": 2.197802197802198, "grad_norm": 1.206205770100736, "learning_rate": 3.438737980446912e-05, "logits/chosen": -2.296875, "logits/rejected": 1.296875, "logps/chosen": -1104.0, "logps/rejected": -1768.0, "loss": 0.0303, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -7.6875, "rewards/margins": 7.125, "rewards/rejected": -14.8125, "step": 2100 }, { "epoch": 2.197802197802198, "eval_logits/chosen": -2.078125, "eval_logits/rejected": -0.1298828125, "eval_logps/chosen": -1144.0, "eval_logps/rejected": -1312.0, "eval_loss": 0.7468593716621399, "eval_rewards/accuracies": 0.703125, "eval_rewards/chosen": -8.3125, "eval_rewards/margins": 1.9453125, "eval_rewards/rejected": -10.25, "eval_runtime": 52.9072, "eval_samples_per_second": 37.802, "eval_steps_per_second": 0.605, "step": 2100 }, { "epoch": 2.208267922553637, "grad_norm": 0.4320527786696419, "learning_rate": 3.4217727088457877e-05, "logits/chosen": -4.0625, "logits/rejected": -0.01708984375, "logps/chosen": -936.0, "logps/rejected": -1536.0, "loss": 0.0349, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -6.0, "rewards/margins": 6.53125, "rewards/rejected": -12.5, "step": 2110 }, { "epoch": 2.218733647305076, "grad_norm": 2.2654140252539956, "learning_rate": 3.40475816629317e-05, "logits/chosen": -3.859375, "logits/rejected": -0.404296875, "logps/chosen": -952.0, "logps/rejected": -1680.0, "loss": 0.0392, "rewards/accuracies": 1.0, "rewards/chosen": -6.3125, "rewards/margins": 7.59375, "rewards/rejected": -13.875, "step": 2120 }, { "epoch": 2.229199372056515, "grad_norm": 0.48033075629807964, "learning_rate": 3.387695262256969e-05, "logits/chosen": -4.8125, "logits/rejected": -0.97265625, "logps/chosen": -920.0, "logps/rejected": -1568.0, "loss": 0.0295, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -5.875, "rewards/margins": 6.8125, "rewards/rejected": -12.6875, "step": 2130 }, { "epoch": 2.239665096807954, "grad_norm": 1.3227057399070403, "learning_rate": 3.3705849087901345e-05, "logits/chosen": -5.625, "logits/rejected": -2.234375, "logps/chosen": -912.0, "logps/rejected": -1576.0, "loss": 0.0248, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -6.03125, "rewards/margins": 6.78125, "rewards/rejected": -12.8125, "step": 2140 }, { "epoch": 2.250130821559393, "grad_norm": 2.6922979427204625, "learning_rate": 3.3534280204819004e-05, "logits/chosen": -5.3125, "logits/rejected": -1.046875, "logps/chosen": -1168.0, "logps/rejected": -1840.0, "loss": 0.0333, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -7.84375, "rewards/margins": 7.40625, "rewards/rejected": -15.25, "step": 2150 }, { "epoch": 2.260596546310832, "grad_norm": 0.5164523934974737, "learning_rate": 3.3362255144088986e-05, "logits/chosen": -4.21875, "logits/rejected": -0.89453125, "logps/chosen": -1048.0, "logps/rejected": -1736.0, "loss": 0.0325, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -7.34375, "rewards/margins": 6.96875, "rewards/rejected": -14.3125, "step": 2160 }, { "epoch": 2.271062271062271, "grad_norm": 2.6614586233503834, "learning_rate": 3.3189783100861416e-05, "logits/chosen": -4.375, "logits/rejected": -0.177734375, "logps/chosen": -1064.0, "logps/rejected": -1672.0, "loss": 0.0356, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -7.15625, "rewards/margins": 6.84375, "rewards/rejected": -14.0, "step": 2170 }, { "epoch": 2.2815279958137102, "grad_norm": 0.4538506307609594, "learning_rate": 3.301687329417866e-05, "logits/chosen": -4.53125, "logits/rejected": -0.486328125, "logps/chosen": -1032.0, "logps/rejected": -1624.0, "loss": 0.0299, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -6.625, "rewards/margins": 6.5625, "rewards/rejected": -13.1875, "step": 2180 }, { "epoch": 2.291993720565149, "grad_norm": 1.358570693850712, "learning_rate": 3.284353496648262e-05, "logits/chosen": -3.96875, "logits/rejected": -0.007110595703125, "logps/chosen": -992.0, "logps/rejected": -1736.0, "loss": 0.0337, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -6.78125, "rewards/margins": 7.34375, "rewards/rejected": -14.125, "step": 2190 }, { "epoch": 2.3024594453165883, "grad_norm": 2.0759035062148037, "learning_rate": 3.2669777383120667e-05, "logits/chosen": -3.53125, "logits/rejected": 0.56640625, "logps/chosen": -1032.0, "logps/rejected": -1640.0, "loss": 0.0322, "rewards/accuracies": 0.96875, "rewards/chosen": -7.15625, "rewards/margins": 6.78125, "rewards/rejected": -13.9375, "step": 2200 }, { "epoch": 2.3024594453165883, "eval_logits/chosen": -2.609375, "eval_logits/rejected": -0.515625, "eval_logps/chosen": -1168.0, "eval_logps/rejected": -1328.0, "eval_loss": 0.7583593726158142, "eval_rewards/accuracies": 0.7109375, "eval_rewards/chosen": -8.5625, "eval_rewards/margins": 1.8828125, "eval_rewards/rejected": -10.4375, "eval_runtime": 52.8854, "eval_samples_per_second": 37.818, "eval_steps_per_second": 0.605, "step": 2200 }, { "epoch": 2.312925170068027, "grad_norm": 1.0386695661847614, "learning_rate": 3.2495609831850355e-05, "logits/chosen": -3.25, "logits/rejected": 1.03125, "logps/chosen": -996.0, "logps/rejected": -1576.0, "loss": 0.0291, "rewards/accuracies": 0.96875, "rewards/chosen": -6.78125, "rewards/margins": 6.375, "rewards/rejected": -13.125, "step": 2210 }, { "epoch": 2.3233908948194664, "grad_norm": 0.9017315734991943, "learning_rate": 3.232104162234305e-05, "logits/chosen": -4.09375, "logits/rejected": -0.050048828125, "logps/chosen": -964.0, "logps/rejected": -1600.0, "loss": 0.0237, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -6.125, "rewards/margins": 6.78125, "rewards/rejected": -12.9375, "step": 2220 }, { "epoch": 2.333856619570905, "grad_norm": 3.5221813476753634, "learning_rate": 3.214608208568621e-05, "logits/chosen": -3.484375, "logits/rejected": 0.294921875, "logps/chosen": -944.0, "logps/rejected": -1568.0, "loss": 0.0343, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -6.25, "rewards/margins": 6.625, "rewards/rejected": -12.875, "step": 2230 }, { "epoch": 2.3443223443223444, "grad_norm": 0.8215146677057751, "learning_rate": 3.197074057388471e-05, "logits/chosen": -5.03125, "logits/rejected": -0.55859375, "logps/chosen": -816.0, "logps/rejected": -1544.0, "loss": 0.0267, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -5.25, "rewards/margins": 7.375, "rewards/rejected": -12.625, "step": 2240 }, { "epoch": 2.3547880690737832, "grad_norm": 1.3572854776003833, "learning_rate": 3.179502645936085e-05, "logits/chosen": -3.25, "logits/rejected": 0.8125, "logps/chosen": -1056.0, "logps/rejected": -1704.0, "loss": 0.0257, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -7.21875, "rewards/margins": 7.0, "rewards/rejected": -14.25, "step": 2250 }, { "epoch": 2.3652537938252225, "grad_norm": 1.5402415847464914, "learning_rate": 3.16189491344535e-05, "logits/chosen": -3.6875, "logits/rejected": -0.07373046875, "logps/chosen": -1012.0, "logps/rejected": -1664.0, "loss": 0.0341, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -6.4375, "rewards/margins": 7.0, "rewards/rejected": -13.4375, "step": 2260 }, { "epoch": 2.3757195185766613, "grad_norm": 0.7612792883410395, "learning_rate": 3.144251801091594e-05, "logits/chosen": -3.921875, "logits/rejected": 0.173828125, "logps/chosen": -916.0, "logps/rejected": -1552.0, "loss": 0.0293, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -5.75, "rewards/margins": 6.875, "rewards/rejected": -12.625, "step": 2270 }, { "epoch": 2.3861852433281006, "grad_norm": 0.36793712218726776, "learning_rate": 3.126574251941288e-05, "logits/chosen": -3.46875, "logits/rejected": -0.1416015625, "logps/chosen": -1020.0, "logps/rejected": -1704.0, "loss": 0.0229, "rewards/accuracies": 1.0, "rewards/chosen": -6.78125, "rewards/margins": 7.0625, "rewards/rejected": -13.8125, "step": 2280 }, { "epoch": 2.3966509680795394, "grad_norm": 0.3368205121436285, "learning_rate": 3.108863210901626e-05, "logits/chosen": -2.453125, "logits/rejected": 0.765625, "logps/chosen": -1032.0, "logps/rejected": -1720.0, "loss": 0.0319, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -6.96875, "rewards/margins": 7.1875, "rewards/rejected": -14.1875, "step": 2290 }, { "epoch": 2.4071166928309786, "grad_norm": 0.3793151914230102, "learning_rate": 3.091119624670027e-05, "logits/chosen": -1.71875, "logits/rejected": 1.4296875, "logps/chosen": -1104.0, "logps/rejected": -1800.0, "loss": 0.0253, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -7.8125, "rewards/margins": 7.28125, "rewards/rejected": -15.0625, "step": 2300 }, { "epoch": 2.4071166928309786, "eval_logits/chosen": -0.75390625, "eval_logits/rejected": 1.265625, "eval_logps/chosen": -1296.0, "eval_logps/rejected": -1480.0, "eval_loss": 0.808718740940094, "eval_rewards/accuracies": 0.712890625, "eval_rewards/chosen": -9.8125, "eval_rewards/margins": 2.125, "eval_rewards/rejected": -11.9375, "eval_runtime": 52.9002, "eval_samples_per_second": 37.807, "eval_steps_per_second": 0.605, "step": 2300 }, { "epoch": 2.4175824175824174, "grad_norm": 1.0436093492415373, "learning_rate": 3.0733444416835294e-05, "logits/chosen": -2.53125, "logits/rejected": 0.76171875, "logps/chosen": -1032.0, "logps/rejected": -1776.0, "loss": 0.0259, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -7.34375, "rewards/margins": 7.4375, "rewards/rejected": -14.75, "step": 2310 }, { "epoch": 2.4280481423338567, "grad_norm": 0.6929838487811482, "learning_rate": 3.0555386120680915e-05, "logits/chosen": -2.96875, "logits/rejected": 0.921875, "logps/chosen": -1080.0, "logps/rejected": -1752.0, "loss": 0.0273, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -7.125, "rewards/margins": 7.34375, "rewards/rejected": -14.4375, "step": 2320 }, { "epoch": 2.4385138670852955, "grad_norm": 0.8425811766791768, "learning_rate": 3.0377030875878064e-05, "logits/chosen": -2.09375, "logits/rejected": 1.390625, "logps/chosen": -1080.0, "logps/rejected": -1768.0, "loss": 0.0352, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -7.3125, "rewards/margins": 7.5, "rewards/rejected": -14.8125, "step": 2330 }, { "epoch": 2.4489795918367347, "grad_norm": 0.23250126941693725, "learning_rate": 3.0198388215940305e-05, "logits/chosen": -3.59375, "logits/rejected": 0.267578125, "logps/chosen": -1104.0, "logps/rejected": -1704.0, "loss": 0.0176, "rewards/accuracies": 1.0, "rewards/chosen": -7.03125, "rewards/margins": 6.78125, "rewards/rejected": -13.8125, "step": 2340 }, { "epoch": 2.4594453165881736, "grad_norm": 0.34696574142214337, "learning_rate": 3.0019467689744217e-05, "logits/chosen": -2.203125, "logits/rejected": 1.7421875, "logps/chosen": -1004.0, "logps/rejected": -1720.0, "loss": 0.0219, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -6.90625, "rewards/margins": 7.5, "rewards/rejected": -14.375, "step": 2350 }, { "epoch": 2.469911041339613, "grad_norm": 0.7548322157415859, "learning_rate": 2.9840278861019e-05, "logits/chosen": -1.5859375, "logits/rejected": 2.0625, "logps/chosen": -1056.0, "logps/rejected": -1736.0, "loss": 0.0308, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -7.625, "rewards/margins": 7.1875, "rewards/rejected": -14.8125, "step": 2360 }, { "epoch": 2.4803767660910516, "grad_norm": 2.0665349230040193, "learning_rate": 2.9660831307835257e-05, "logits/chosen": -0.94140625, "logits/rejected": 2.21875, "logps/chosen": -1096.0, "logps/rejected": -1784.0, "loss": 0.0278, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -7.78125, "rewards/margins": 7.28125, "rewards/rejected": -15.0625, "step": 2370 }, { "epoch": 2.490842490842491, "grad_norm": 3.1198291620615493, "learning_rate": 2.9481134622093032e-05, "logits/chosen": -0.263671875, "logits/rejected": 2.265625, "logps/chosen": -1136.0, "logps/rejected": -1880.0, "loss": 0.0201, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -8.375, "rewards/margins": 7.5625, "rewards/rejected": -15.9375, "step": 2380 }, { "epoch": 2.50130821559393, "grad_norm": 0.5509370310846502, "learning_rate": 2.9301198409009108e-05, "logits/chosen": 0.134765625, "logits/rejected": 2.953125, "logps/chosen": -1112.0, "logps/rejected": -1816.0, "loss": 0.0286, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -8.0625, "rewards/margins": 7.625, "rewards/rejected": -15.6875, "step": 2390 }, { "epoch": 2.511773940345369, "grad_norm": 3.1272795221082443, "learning_rate": 2.9121032286603554e-05, "logits/chosen": -1.1953125, "logits/rejected": 2.5, "logps/chosen": -1112.0, "logps/rejected": -1776.0, "loss": 0.0302, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -7.6875, "rewards/margins": 7.09375, "rewards/rejected": -14.8125, "step": 2400 }, { "epoch": 2.511773940345369, "eval_logits/chosen": 0.439453125, "eval_logits/rejected": 2.28125, "eval_logps/chosen": -1216.0, "eval_logps/rejected": -1392.0, "eval_loss": 0.8032734394073486, "eval_rewards/accuracies": 0.724609375, "eval_rewards/chosen": -9.0, "eval_rewards/margins": 2.03125, "eval_rewards/rejected": -11.0625, "eval_runtime": 52.9062, "eval_samples_per_second": 37.803, "eval_steps_per_second": 0.605, "step": 2400 }, { "epoch": 2.5222396650968077, "grad_norm": 0.3983961715251342, "learning_rate": 2.894064588518568e-05, "logits/chosen": -1.1640625, "logits/rejected": 1.6640625, "logps/chosen": -1056.0, "logps/rejected": -1712.0, "loss": 0.0251, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -7.0625, "rewards/margins": 7.125, "rewards/rejected": -14.1875, "step": 2410 }, { "epoch": 2.532705389848247, "grad_norm": 1.8965194428136705, "learning_rate": 2.8760048846839194e-05, "logits/chosen": -1.7265625, "logits/rejected": 1.3984375, "logps/chosen": -968.0, "logps/rejected": -1632.0, "loss": 0.0283, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -6.4375, "rewards/margins": 7.0, "rewards/rejected": -13.4375, "step": 2420 }, { "epoch": 2.5431711145996863, "grad_norm": 0.7430147021819409, "learning_rate": 2.8579250824906877e-05, "logits/chosen": -1.3984375, "logits/rejected": 1.765625, "logps/chosen": -1088.0, "logps/rejected": -1792.0, "loss": 0.0232, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -7.5, "rewards/margins": 7.5625, "rewards/rejected": -15.0, "step": 2430 }, { "epoch": 2.553636839351125, "grad_norm": 2.56696162913765, "learning_rate": 2.8398261483474547e-05, "logits/chosen": -0.58984375, "logits/rejected": 2.96875, "logps/chosen": -1088.0, "logps/rejected": -1848.0, "loss": 0.0344, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -7.71875, "rewards/margins": 7.875, "rewards/rejected": -15.5625, "step": 2440 }, { "epoch": 2.564102564102564, "grad_norm": 2.730368166501539, "learning_rate": 2.8217090496854515e-05, "logits/chosen": -1.515625, "logits/rejected": 1.2109375, "logps/chosen": -1064.0, "logps/rejected": -1736.0, "loss": 0.0313, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -7.3125, "rewards/margins": 7.09375, "rewards/rejected": -14.4375, "step": 2450 }, { "epoch": 2.574568288854003, "grad_norm": 1.1722855893135804, "learning_rate": 2.8035747549068452e-05, "logits/chosen": -1.625, "logits/rejected": 1.5546875, "logps/chosen": -1088.0, "logps/rejected": -1752.0, "loss": 0.026, "rewards/accuracies": 1.0, "rewards/chosen": -7.25, "rewards/margins": 7.1875, "rewards/rejected": -14.4375, "step": 2460 }, { "epoch": 2.5850340136054424, "grad_norm": 0.6287233493076456, "learning_rate": 2.7854242333329782e-05, "logits/chosen": -1.859375, "logits/rejected": 1.2109375, "logps/chosen": -1048.0, "logps/rejected": -1824.0, "loss": 0.0283, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -7.71875, "rewards/margins": 7.75, "rewards/rejected": -15.5, "step": 2470 }, { "epoch": 2.595499738356881, "grad_norm": 0.8747862515761952, "learning_rate": 2.7672584551525527e-05, "logits/chosen": -0.8359375, "logits/rejected": 1.5078125, "logps/chosen": -1120.0, "logps/rejected": -1896.0, "loss": 0.0321, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -8.0, "rewards/margins": 7.625, "rewards/rejected": -15.6875, "step": 2480 }, { "epoch": 2.60596546310832, "grad_norm": 1.4602269013561426, "learning_rate": 2.749078391369771e-05, "logits/chosen": -2.15625, "logits/rejected": 0.734375, "logps/chosen": -1056.0, "logps/rejected": -1808.0, "loss": 0.0302, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -7.1875, "rewards/margins": 7.5625, "rewards/rejected": -14.75, "step": 2490 }, { "epoch": 2.6164311878597593, "grad_norm": 2.4880564377763768, "learning_rate": 2.730885013752439e-05, "logits/chosen": -1.359375, "logits/rejected": 1.6171875, "logps/chosen": -1056.0, "logps/rejected": -1808.0, "loss": 0.0218, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -7.4375, "rewards/margins": 7.78125, "rewards/rejected": -15.25, "step": 2500 }, { "epoch": 2.6164311878597593, "eval_logits/chosen": 0.51953125, "eval_logits/rejected": 2.296875, "eval_logps/chosen": -1408.0, "eval_logps/rejected": -1616.0, "eval_loss": 0.8602734208106995, "eval_rewards/accuracies": 0.71875, "eval_rewards/chosen": -11.0, "eval_rewards/margins": 2.3125, "eval_rewards/rejected": -13.3125, "eval_runtime": 52.9043, "eval_samples_per_second": 37.804, "eval_steps_per_second": 0.605, "step": 2500 }, { "epoch": 2.6268969126111985, "grad_norm": 0.9370128586176782, "learning_rate": 2.712679294780015e-05, "logits/chosen": -1.09375, "logits/rejected": 2.125, "logps/chosen": -1264.0, "logps/rejected": -1952.0, "loss": 0.0395, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -9.1875, "rewards/margins": 7.625, "rewards/rejected": -16.75, "step": 2510 }, { "epoch": 2.6373626373626373, "grad_norm": 0.6235520573693828, "learning_rate": 2.6944622075916344e-05, "logits/chosen": -0.318359375, "logits/rejected": 2.59375, "logps/chosen": -1136.0, "logps/rejected": -1872.0, "loss": 0.0275, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -8.1875, "rewards/margins": 7.65625, "rewards/rejected": -15.8125, "step": 2520 }, { "epoch": 2.647828362114076, "grad_norm": 1.2117156468852468, "learning_rate": 2.6762347259340907e-05, "logits/chosen": -1.0546875, "logits/rejected": 1.296875, "logps/chosen": -1020.0, "logps/rejected": -1712.0, "loss": 0.0281, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -7.09375, "rewards/margins": 7.09375, "rewards/rejected": -14.1875, "step": 2530 }, { "epoch": 2.6582940868655154, "grad_norm": 0.3622207393417241, "learning_rate": 2.6579978241097865e-05, "logits/chosen": -1.1328125, "logits/rejected": 2.109375, "logps/chosen": -1064.0, "logps/rejected": -1744.0, "loss": 0.0346, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -7.3125, "rewards/margins": 7.3125, "rewards/rejected": -14.625, "step": 2540 }, { "epoch": 2.6687598116169546, "grad_norm": 0.5754950910620171, "learning_rate": 2.639752476924653e-05, "logits/chosen": -0.875, "logits/rejected": 2.046875, "logps/chosen": -1136.0, "logps/rejected": -1856.0, "loss": 0.0327, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -8.0625, "rewards/margins": 7.5625, "rewards/rejected": -15.625, "step": 2550 }, { "epoch": 2.6792255363683934, "grad_norm": 1.7122391719550465, "learning_rate": 2.621499659636049e-05, "logits/chosen": -1.3125, "logits/rejected": 1.4921875, "logps/chosen": -1064.0, "logps/rejected": -1736.0, "loss": 0.0258, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -7.40625, "rewards/margins": 6.90625, "rewards/rejected": -14.3125, "step": 2560 }, { "epoch": 2.6896912611198327, "grad_norm": 0.5733433265622923, "learning_rate": 2.6032403479006263e-05, "logits/chosen": -1.421875, "logits/rejected": 1.640625, "logps/chosen": -1056.0, "logps/rejected": -1728.0, "loss": 0.0202, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -7.03125, "rewards/margins": 7.21875, "rewards/rejected": -14.25, "step": 2570 }, { "epoch": 2.7001569858712715, "grad_norm": 0.3121083597914369, "learning_rate": 2.584975517722181e-05, "logits/chosen": -0.9453125, "logits/rejected": 2.421875, "logps/chosen": -1176.0, "logps/rejected": -1864.0, "loss": 0.0417, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -8.625, "rewards/margins": 7.4375, "rewards/rejected": -16.125, "step": 2580 }, { "epoch": 2.7106227106227108, "grad_norm": 3.6347959815546833, "learning_rate": 2.5667061453994834e-05, "logits/chosen": -0.50390625, "logits/rejected": 2.25, "logps/chosen": -1120.0, "logps/rejected": -1832.0, "loss": 0.0385, "rewards/accuracies": 0.9624999761581421, "rewards/chosen": -8.1875, "rewards/margins": 7.40625, "rewards/rejected": -15.625, "step": 2590 }, { "epoch": 2.7210884353741496, "grad_norm": 0.5266099919412788, "learning_rate": 2.5484332074740913e-05, "logits/chosen": -2.40625, "logits/rejected": 0.5703125, "logps/chosen": -1056.0, "logps/rejected": -1728.0, "loss": 0.027, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -7.375, "rewards/margins": 7.125, "rewards/rejected": -14.5, "step": 2600 }, { "epoch": 2.7210884353741496, "eval_logits/chosen": -0.9609375, "eval_logits/rejected": 1.0703125, "eval_logps/chosen": -1288.0, "eval_logps/rejected": -1488.0, "eval_loss": 0.8161718845367432, "eval_rewards/accuracies": 0.740234375, "eval_rewards/chosen": -9.75, "eval_rewards/margins": 2.21875, "eval_rewards/rejected": -12.0, "eval_runtime": 52.8918, "eval_samples_per_second": 37.813, "eval_steps_per_second": 0.605, "step": 2600 }, { "epoch": 2.731554160125589, "grad_norm": 1.5362016184987102, "learning_rate": 2.5301576806781518e-05, "logits/chosen": -1.84375, "logits/rejected": 1.296875, "logps/chosen": -1072.0, "logps/rejected": -1800.0, "loss": 0.0403, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -7.75, "rewards/margins": 7.4375, "rewards/rejected": -15.1875, "step": 2610 }, { "epoch": 2.7420198848770276, "grad_norm": 1.1533355773019651, "learning_rate": 2.5118805418821967e-05, "logits/chosen": -1.8515625, "logits/rejected": 1.9609375, "logps/chosen": -1088.0, "logps/rejected": -1848.0, "loss": 0.025, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -7.875, "rewards/margins": 7.8125, "rewards/rejected": -15.6875, "step": 2620 }, { "epoch": 2.752485609628467, "grad_norm": 0.6534763596898375, "learning_rate": 2.4936027680429197e-05, "logits/chosen": -1.375, "logits/rejected": 1.40625, "logps/chosen": -1080.0, "logps/rejected": -1848.0, "loss": 0.0375, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -7.875, "rewards/margins": 7.71875, "rewards/rejected": -15.5625, "step": 2630 }, { "epoch": 2.7629513343799057, "grad_norm": 1.2269783671113828, "learning_rate": 2.475325336150961e-05, "logits/chosen": -1.828125, "logits/rejected": 1.9921875, "logps/chosen": -1128.0, "logps/rejected": -1808.0, "loss": 0.0474, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -7.9375, "rewards/margins": 7.3125, "rewards/rejected": -15.25, "step": 2640 }, { "epoch": 2.773417059131345, "grad_norm": 0.42128890294068566, "learning_rate": 2.457049223178681e-05, "logits/chosen": -1.703125, "logits/rejected": 1.9609375, "logps/chosen": -1104.0, "logps/rejected": -1808.0, "loss": 0.0216, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -7.8125, "rewards/margins": 7.4375, "rewards/rejected": -15.25, "step": 2650 }, { "epoch": 2.7838827838827838, "grad_norm": 1.2008555193471955, "learning_rate": 2.438775406027944e-05, "logits/chosen": -3.453125, "logits/rejected": 0.92578125, "logps/chosen": -968.0, "logps/rejected": -1608.0, "loss": 0.0346, "rewards/accuracies": 0.9624999761581421, "rewards/chosen": -6.6875, "rewards/margins": 6.78125, "rewards/rejected": -13.4375, "step": 2660 }, { "epoch": 2.794348508634223, "grad_norm": 1.6461672196616548, "learning_rate": 2.4205048614778926e-05, "logits/chosen": -2.15625, "logits/rejected": 2.375, "logps/chosen": -1012.0, "logps/rejected": -1648.0, "loss": 0.044, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -7.03125, "rewards/margins": 7.0, "rewards/rejected": -14.0625, "step": 2670 }, { "epoch": 2.804814233385662, "grad_norm": 1.0824184309184701, "learning_rate": 2.402238566132744e-05, "logits/chosen": -1.546875, "logits/rejected": 1.7734375, "logps/chosen": -1112.0, "logps/rejected": -1880.0, "loss": 0.0184, "rewards/accuracies": 1.0, "rewards/chosen": -7.59375, "rewards/margins": 8.1875, "rewards/rejected": -15.75, "step": 2680 }, { "epoch": 2.815279958137101, "grad_norm": 1.6308888710356009, "learning_rate": 2.383977496369586e-05, "logits/chosen": -0.0303955078125, "logits/rejected": 2.90625, "logps/chosen": -1128.0, "logps/rejected": -1912.0, "loss": 0.0277, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -8.25, "rewards/margins": 8.1875, "rewards/rejected": -16.375, "step": 2690 }, { "epoch": 2.82574568288854, "grad_norm": 3.6508001389254043, "learning_rate": 2.3657226282861844e-05, "logits/chosen": -1.8828125, "logits/rejected": 2.171875, "logps/chosen": -1128.0, "logps/rejected": -1872.0, "loss": 0.0274, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -7.875, "rewards/margins": 7.84375, "rewards/rejected": -15.75, "step": 2700 }, { "epoch": 2.82574568288854, "eval_logits/chosen": -0.91015625, "eval_logits/rejected": 1.125, "eval_logps/chosen": -1288.0, "eval_logps/rejected": -1480.0, "eval_loss": 0.8295781016349792, "eval_rewards/accuracies": 0.71875, "eval_rewards/chosen": -9.75, "eval_rewards/margins": 2.21875, "eval_rewards/rejected": -12.0, "eval_runtime": 52.8913, "eval_samples_per_second": 37.813, "eval_steps_per_second": 0.605, "step": 2700 }, { "epoch": 2.836211407639979, "grad_norm": 1.2993477133286373, "learning_rate": 2.347474937648811e-05, "logits/chosen": -2.671875, "logits/rejected": 0.71875, "logps/chosen": -1096.0, "logps/rejected": -1824.0, "loss": 0.0206, "rewards/accuracies": 1.0, "rewards/chosen": -7.3125, "rewards/margins": 7.875, "rewards/rejected": -15.1875, "step": 2710 }, { "epoch": 2.846677132391418, "grad_norm": 1.208975326201441, "learning_rate": 2.3292353998400875e-05, "logits/chosen": -2.453125, "logits/rejected": 1.2421875, "logps/chosen": -1152.0, "logps/rejected": -1928.0, "loss": 0.0195, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -8.3125, "rewards/margins": 8.25, "rewards/rejected": -16.5, "step": 2720 }, { "epoch": 2.857142857142857, "grad_norm": 0.2446386189614852, "learning_rate": 2.311004989806845e-05, "logits/chosen": -2.4375, "logits/rejected": 0.9921875, "logps/chosen": -1160.0, "logps/rejected": -2064.0, "loss": 0.0308, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -8.5625, "rewards/margins": 9.0625, "rewards/rejected": -17.625, "step": 2730 }, { "epoch": 2.867608581894296, "grad_norm": 1.6860858115184227, "learning_rate": 2.2927846820080156e-05, "logits/chosen": -3.28125, "logits/rejected": -0.244140625, "logps/chosen": -1072.0, "logps/rejected": -1840.0, "loss": 0.024, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -7.5625, "rewards/margins": 7.90625, "rewards/rejected": -15.4375, "step": 2740 }, { "epoch": 2.8780743066457353, "grad_norm": 1.2652848167896822, "learning_rate": 2.2745754503625407e-05, "logits/chosen": -3.484375, "logits/rejected": 0.8203125, "logps/chosen": -1016.0, "logps/rejected": -1704.0, "loss": 0.0242, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -6.96875, "rewards/margins": 7.40625, "rewards/rejected": -14.375, "step": 2750 }, { "epoch": 2.8885400313971745, "grad_norm": 1.0587944363261317, "learning_rate": 2.256378268197317e-05, "logits/chosen": -2.875, "logits/rejected": 1.1796875, "logps/chosen": -1032.0, "logps/rejected": -1736.0, "loss": 0.0267, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -7.15625, "rewards/margins": 7.34375, "rewards/rejected": -14.5, "step": 2760 }, { "epoch": 2.8990057561486133, "grad_norm": 1.099532685024332, "learning_rate": 2.238194108195167e-05, "logits/chosen": -2.53125, "logits/rejected": 1.609375, "logps/chosen": -1080.0, "logps/rejected": -1776.0, "loss": 0.0311, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -7.21875, "rewards/margins": 7.5, "rewards/rejected": -14.75, "step": 2770 }, { "epoch": 2.909471480900052, "grad_norm": 1.1911395866069077, "learning_rate": 2.220023942342844e-05, "logits/chosen": -3.15625, "logits/rejected": 1.0703125, "logps/chosen": -1144.0, "logps/rejected": -1872.0, "loss": 0.0238, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -7.875, "rewards/margins": 7.75, "rewards/rejected": -15.625, "step": 2780 }, { "epoch": 2.9199372056514914, "grad_norm": 1.6577589564365567, "learning_rate": 2.2018687418790874e-05, "logits/chosen": -3.703125, "logits/rejected": 0.0810546875, "logps/chosen": -1064.0, "logps/rejected": -1864.0, "loss": 0.0235, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -7.3125, "rewards/margins": 8.3125, "rewards/rejected": -15.625, "step": 2790 }, { "epoch": 2.9304029304029307, "grad_norm": 0.8181355263028384, "learning_rate": 2.1837294772426948e-05, "logits/chosen": -2.9375, "logits/rejected": 0.5625, "logps/chosen": -1032.0, "logps/rejected": -1704.0, "loss": 0.0369, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -7.25, "rewards/margins": 7.28125, "rewards/rejected": -14.5, "step": 2800 }, { "epoch": 2.9304029304029307, "eval_logits/chosen": -1.453125, "eval_logits/rejected": 0.62890625, "eval_logps/chosen": -1272.0, "eval_logps/rejected": -1472.0, "eval_loss": 0.8085156083106995, "eval_rewards/accuracies": 0.72265625, "eval_rewards/chosen": -9.5625, "eval_rewards/margins": 2.3125, "eval_rewards/rejected": -11.875, "eval_runtime": 52.8836, "eval_samples_per_second": 37.819, "eval_steps_per_second": 0.605, "step": 2800 }, { "epoch": 2.9408686551543695, "grad_norm": 0.9435732480557755, "learning_rate": 2.1656071180206593e-05, "logits/chosen": -2.625, "logits/rejected": 1.0546875, "logps/chosen": -1120.0, "logps/rejected": -1952.0, "loss": 0.0278, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -7.90625, "rewards/margins": 8.5625, "rewards/rejected": -16.5, "step": 2810 }, { "epoch": 2.9513343799058083, "grad_norm": 1.2048992857915675, "learning_rate": 2.1475026328963376e-05, "logits/chosen": -2.953125, "logits/rejected": 1.390625, "logps/chosen": -1096.0, "logps/rejected": -1872.0, "loss": 0.0232, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -7.84375, "rewards/margins": 8.0, "rewards/rejected": -15.875, "step": 2820 }, { "epoch": 2.9618001046572475, "grad_norm": 1.5340965910243034, "learning_rate": 2.1294169895976746e-05, "logits/chosen": -3.078125, "logits/rejected": 0.890625, "logps/chosen": -1072.0, "logps/rejected": -1768.0, "loss": 0.0229, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -7.40625, "rewards/margins": 7.34375, "rewards/rejected": -14.75, "step": 2830 }, { "epoch": 2.9722658294086868, "grad_norm": 0.3265272820820435, "learning_rate": 2.111351154845472e-05, "logits/chosen": -1.546875, "logits/rejected": 2.71875, "logps/chosen": -1088.0, "logps/rejected": -1816.0, "loss": 0.0265, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -7.53125, "rewards/margins": 7.84375, "rewards/rejected": -15.375, "step": 2840 }, { "epoch": 2.9827315541601256, "grad_norm": 0.5998260117426013, "learning_rate": 2.0933060943017215e-05, "logits/chosen": -1.9453125, "logits/rejected": 0.703125, "logps/chosen": -1020.0, "logps/rejected": -1784.0, "loss": 0.0278, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -7.0625, "rewards/margins": 7.75, "rewards/rejected": -14.8125, "step": 2850 }, { "epoch": 2.9931972789115644, "grad_norm": 0.54910635876851, "learning_rate": 2.0752827725179793e-05, "logits/chosen": -2.40625, "logits/rejected": 0.50390625, "logps/chosen": -1040.0, "logps/rejected": -1704.0, "loss": 0.0286, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -7.1875, "rewards/margins": 6.90625, "rewards/rejected": -14.0625, "step": 2860 }, { "epoch": 3.0036630036630036, "grad_norm": 0.146418403562972, "learning_rate": 2.0572821528838175e-05, "logits/chosen": -4.34375, "logits/rejected": 0.859375, "logps/chosen": -988.0, "logps/rejected": -1608.0, "loss": 0.019, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -6.375, "rewards/margins": 7.28125, "rewards/rejected": -13.6875, "step": 2870 }, { "epoch": 3.014128728414443, "grad_norm": 2.895012944047952, "learning_rate": 2.0393051975753196e-05, "logits/chosen": -3.90625, "logits/rejected": 0.2197265625, "logps/chosen": -860.0, "logps/rejected": -1616.0, "loss": 0.0131, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -5.625, "rewards/margins": 7.9375, "rewards/rejected": -13.5625, "step": 2880 }, { "epoch": 3.0245944531658817, "grad_norm": 0.1423538198589034, "learning_rate": 2.021352867503657e-05, "logits/chosen": -3.75, "logits/rejected": 1.3046875, "logps/chosen": -1008.0, "logps/rejected": -1824.0, "loss": 0.0114, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -6.75, "rewards/margins": 8.6875, "rewards/rejected": -15.5, "step": 2890 }, { "epoch": 3.035060177917321, "grad_norm": 0.2962179324568394, "learning_rate": 2.003426122263723e-05, "logits/chosen": -3.375, "logits/rejected": 0.69140625, "logps/chosen": -1032.0, "logps/rejected": -1896.0, "loss": 0.0154, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -7.09375, "rewards/margins": 8.9375, "rewards/rejected": -16.0, "step": 2900 }, { "epoch": 3.035060177917321, "eval_logits/chosen": -1.3125, "eval_logits/rejected": 0.9609375, "eval_logps/chosen": -1296.0, "eval_logps/rejected": -1520.0, "eval_loss": 0.8778906464576721, "eval_rewards/accuracies": 0.7265625, "eval_rewards/chosen": -9.875, "eval_rewards/margins": 2.5, "eval_rewards/rejected": -12.375, "eval_runtime": 52.8723, "eval_samples_per_second": 37.827, "eval_steps_per_second": 0.605, "step": 2900 }, { "epoch": 3.0455259026687598, "grad_norm": 0.03778973429382476, "learning_rate": 1.9855259200828412e-05, "logits/chosen": -2.96875, "logits/rejected": 1.5234375, "logps/chosen": -1080.0, "logps/rejected": -1992.0, "loss": 0.01, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -7.40625, "rewards/margins": 9.4375, "rewards/rejected": -16.75, "step": 2910 }, { "epoch": 3.055991627420199, "grad_norm": 0.05371775375140043, "learning_rate": 1.9676532177695442e-05, "logits/chosen": -2.78125, "logits/rejected": 1.546875, "logps/chosen": -1080.0, "logps/rejected": -2000.0, "loss": 0.0085, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -7.84375, "rewards/margins": 9.25, "rewards/rejected": -17.125, "step": 2920 }, { "epoch": 3.066457352171638, "grad_norm": 0.025662869341906102, "learning_rate": 1.9498089706624316e-05, "logits/chosen": -2.8125, "logits/rejected": 1.390625, "logps/chosen": -1112.0, "logps/rejected": -1976.0, "loss": 0.0112, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -7.84375, "rewards/margins": 9.0, "rewards/rejected": -16.75, "step": 2930 }, { "epoch": 3.076923076923077, "grad_norm": 0.04698152913607542, "learning_rate": 1.9319941325791043e-05, "logits/chosen": -3.140625, "logits/rejected": 1.9921875, "logps/chosen": -1112.0, "logps/rejected": -2032.0, "loss": 0.0136, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -7.75, "rewards/margins": 9.75, "rewards/rejected": -17.5, "step": 2940 }, { "epoch": 3.087388801674516, "grad_norm": 0.2216407872137013, "learning_rate": 1.9142096557651808e-05, "logits/chosen": -2.90625, "logits/rejected": 2.34375, "logps/chosen": -1120.0, "logps/rejected": -2008.0, "loss": 0.0076, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -7.875, "rewards/margins": 9.625, "rewards/rejected": -17.5, "step": 2950 }, { "epoch": 3.097854526425955, "grad_norm": 0.033181262564329926, "learning_rate": 1.8964564908433965e-05, "logits/chosen": -1.5546875, "logits/rejected": 2.046875, "logps/chosen": -1216.0, "logps/rejected": -2176.0, "loss": 0.008, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -8.6875, "rewards/margins": 9.875, "rewards/rejected": -18.625, "step": 2960 }, { "epoch": 3.108320251177394, "grad_norm": 0.034234981051473136, "learning_rate": 1.8787355867627933e-05, "logits/chosen": -1.5234375, "logits/rejected": 2.890625, "logps/chosen": -1104.0, "logps/rejected": -2176.0, "loss": 0.0053, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -8.125, "rewards/margins": 10.75, "rewards/rejected": -18.875, "step": 2970 }, { "epoch": 3.118785975928833, "grad_norm": 0.058508823478964314, "learning_rate": 1.8610478907479916e-05, "logits/chosen": -1.78125, "logits/rejected": 3.484375, "logps/chosen": -1240.0, "logps/rejected": -2400.0, "loss": 0.0057, "rewards/accuracies": 1.0, "rewards/chosen": -9.125, "rewards/margins": 11.8125, "rewards/rejected": -20.875, "step": 2980 }, { "epoch": 3.129251700680272, "grad_norm": 2.6421404482627393, "learning_rate": 1.8433943482485624e-05, "logits/chosen": -1.6015625, "logits/rejected": 1.6875, "logps/chosen": -1264.0, "logps/rejected": -2256.0, "loss": 0.0079, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -9.25, "rewards/margins": 10.0625, "rewards/rejected": -19.375, "step": 2990 }, { "epoch": 3.1397174254317113, "grad_norm": 0.272455042334525, "learning_rate": 1.8257759028884912e-05, "logits/chosen": -1.28125, "logits/rejected": 2.421875, "logps/chosen": -1200.0, "logps/rejected": -2288.0, "loss": 0.007, "rewards/accuracies": 1.0, "rewards/chosen": -8.875, "rewards/margins": 10.9375, "rewards/rejected": -19.75, "step": 3000 }, { "epoch": 3.1397174254317113, "eval_logits/chosen": 0.68359375, "eval_logits/rejected": 2.796875, "eval_logps/chosen": -1464.0, "eval_logps/rejected": -1728.0, "eval_loss": 0.9780234098434448, "eval_rewards/accuracies": 0.720703125, "eval_rewards/chosen": -11.5, "eval_rewards/margins": 2.875, "eval_rewards/rejected": -14.375, "eval_runtime": 52.8667, "eval_samples_per_second": 37.831, "eval_steps_per_second": 0.605, "step": 3000 }, { "epoch": 3.15018315018315, "grad_norm": 0.03150523012206682, "learning_rate": 1.8081934964157353e-05, "logits/chosen": -1.0234375, "logits/rejected": 3.34375, "logps/chosen": -1232.0, "logps/rejected": -2128.0, "loss": 0.0101, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -8.6875, "rewards/margins": 9.6875, "rewards/rejected": -18.375, "step": 3010 }, { "epoch": 3.1606488749345893, "grad_norm": 0.1206717172365664, "learning_rate": 1.7906480686518903e-05, "logits/chosen": -0.359375, "logits/rejected": 3.546875, "logps/chosen": -1184.0, "logps/rejected": -2112.0, "loss": 0.0083, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -8.5, "rewards/margins": 9.8125, "rewards/rejected": -18.25, "step": 3020 }, { "epoch": 3.171114599686028, "grad_norm": 0.033920179683535764, "learning_rate": 1.773140557441948e-05, "logits/chosen": -1.5078125, "logits/rejected": 4.125, "logps/chosen": -1152.0, "logps/rejected": -2064.0, "loss": 0.008, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -8.1875, "rewards/margins": 10.0625, "rewards/rejected": -18.25, "step": 3030 }, { "epoch": 3.1815803244374674, "grad_norm": 0.024285830587125388, "learning_rate": 1.7556718986041733e-05, "logits/chosen": -0.79296875, "logits/rejected": 3.734375, "logps/chosen": -1144.0, "logps/rejected": -2064.0, "loss": 0.0108, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -8.25, "rewards/margins": 9.6875, "rewards/rejected": -18.0, "step": 3040 }, { "epoch": 3.192046049188906, "grad_norm": 1.4925565821536846, "learning_rate": 1.7382430258800763e-05, "logits/chosen": -0.78125, "logits/rejected": 3.359375, "logps/chosen": -1112.0, "logps/rejected": -2128.0, "loss": 0.009, "rewards/accuracies": 1.0, "rewards/chosen": -8.25, "rewards/margins": 10.375, "rewards/rejected": -18.625, "step": 3050 }, { "epoch": 3.2025117739403455, "grad_norm": 1.0235211261758, "learning_rate": 1.720854870884507e-05, "logits/chosen": 0.1494140625, "logits/rejected": 3.8125, "logps/chosen": -1208.0, "logps/rejected": -2192.0, "loss": 0.0152, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -9.0625, "rewards/margins": 10.0625, "rewards/rejected": -19.125, "step": 3060 }, { "epoch": 3.2129774986917843, "grad_norm": 0.2888409223728506, "learning_rate": 1.7035083630558525e-05, "logits/chosen": -1.640625, "logits/rejected": 3.0625, "logps/chosen": -1152.0, "logps/rejected": -2032.0, "loss": 0.0048, "rewards/accuracies": 1.0, "rewards/chosen": -8.125, "rewards/margins": 9.4375, "rewards/rejected": -17.5, "step": 3070 }, { "epoch": 3.2234432234432235, "grad_norm": 0.09681539714463304, "learning_rate": 1.6862044296063635e-05, "logits/chosen": -1.375, "logits/rejected": 2.46875, "logps/chosen": -1160.0, "logps/rejected": -2176.0, "loss": 0.006, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -8.625, "rewards/margins": 10.0, "rewards/rejected": -18.625, "step": 3080 }, { "epoch": 3.2339089481946623, "grad_norm": 0.611624799453792, "learning_rate": 1.6689439954725834e-05, "logits/chosen": -1.0234375, "logits/rejected": 2.546875, "logps/chosen": -1192.0, "logps/rejected": -2192.0, "loss": 0.0081, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -8.5625, "rewards/margins": 10.1875, "rewards/rejected": -18.75, "step": 3090 }, { "epoch": 3.2443746729461016, "grad_norm": 0.034054919296362146, "learning_rate": 1.651727983265918e-05, "logits/chosen": -1.9765625, "logits/rejected": 1.890625, "logps/chosen": -1192.0, "logps/rejected": -2096.0, "loss": 0.0059, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -8.1875, "rewards/margins": 9.6875, "rewards/rejected": -17.875, "step": 3100 }, { "epoch": 3.2443746729461016, "eval_logits/chosen": 0.0257568359375, "eval_logits/rejected": 2.21875, "eval_logps/chosen": -1424.0, "eval_logps/rejected": -1688.0, "eval_loss": 0.9792890548706055, "eval_rewards/accuracies": 0.708984375, "eval_rewards/chosen": -11.125, "eval_rewards/margins": 2.875, "eval_rewards/rejected": -14.0, "eval_runtime": 52.8828, "eval_samples_per_second": 37.819, "eval_steps_per_second": 0.605, "step": 3100 }, { "epoch": 3.2548403976975404, "grad_norm": 1.2632359318208952, "learning_rate": 1.6345573132233116e-05, "logits/chosen": -2.203125, "logits/rejected": 2.40625, "logps/chosen": -1144.0, "logps/rejected": -2096.0, "loss": 0.0088, "rewards/accuracies": 1.0, "rewards/chosen": -8.4375, "rewards/margins": 9.9375, "rewards/rejected": -18.375, "step": 3110 }, { "epoch": 3.2653061224489797, "grad_norm": 0.17830708741981413, "learning_rate": 1.617432903158065e-05, "logits/chosen": -0.85546875, "logits/rejected": 2.734375, "logps/chosen": -1176.0, "logps/rejected": -2144.0, "loss": 0.0109, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -8.625, "rewards/margins": 9.9375, "rewards/rejected": -18.5, "step": 3120 }, { "epoch": 3.2757718472004185, "grad_norm": 0.015030610448317385, "learning_rate": 1.600355668410769e-05, "logits/chosen": -1.109375, "logits/rejected": 3.125, "logps/chosen": -1232.0, "logps/rejected": -2208.0, "loss": 0.0035, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -8.9375, "rewards/margins": 10.25, "rewards/rejected": -19.125, "step": 3130 }, { "epoch": 3.2862375719518577, "grad_norm": 0.27883925569127904, "learning_rate": 1.5833265218003852e-05, "logits/chosen": -0.90234375, "logits/rejected": 3.203125, "logps/chosen": -1208.0, "logps/rejected": -2240.0, "loss": 0.0115, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -8.9375, "rewards/margins": 10.5625, "rewards/rejected": -19.5, "step": 3140 }, { "epoch": 3.2967032967032965, "grad_norm": 0.7036112100678287, "learning_rate": 1.566346373575445e-05, "logits/chosen": -0.443359375, "logits/rejected": 2.828125, "logps/chosen": -1192.0, "logps/rejected": -2208.0, "loss": 0.0089, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -9.0, "rewards/margins": 10.3125, "rewards/rejected": -19.375, "step": 3150 }, { "epoch": 3.307169021454736, "grad_norm": 0.036521152147152444, "learning_rate": 1.5494161313654032e-05, "logits/chosen": -0.7421875, "logits/rejected": 3.390625, "logps/chosen": -1224.0, "logps/rejected": -2160.0, "loss": 0.0088, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -8.875, "rewards/margins": 10.0, "rewards/rejected": -18.875, "step": 3160 }, { "epoch": 3.3176347462061746, "grad_norm": 0.21403998437602492, "learning_rate": 1.5325367001321162e-05, "logits/chosen": -1.3125, "logits/rejected": 3.46875, "logps/chosen": -1176.0, "logps/rejected": -2128.0, "loss": 0.0206, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -8.5, "rewards/margins": 10.125, "rewards/rejected": -18.625, "step": 3170 }, { "epoch": 3.328100470957614, "grad_norm": 0.4443822779711363, "learning_rate": 1.5157089821214745e-05, "logits/chosen": -0.78515625, "logits/rejected": 2.703125, "logps/chosen": -1136.0, "logps/rejected": -2064.0, "loss": 0.0102, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -8.0, "rewards/margins": 9.75, "rewards/rejected": -17.75, "step": 3180 }, { "epoch": 3.3385661957090527, "grad_norm": 0.08603805919540032, "learning_rate": 1.4989338768151714e-05, "logits/chosen": -0.78515625, "logits/rejected": 2.40625, "logps/chosen": -1112.0, "logps/rejected": -2064.0, "loss": 0.0091, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -7.90625, "rewards/margins": 9.6875, "rewards/rejected": -17.5, "step": 3190 }, { "epoch": 3.349031920460492, "grad_norm": 0.14729072438320479, "learning_rate": 1.482212280882626e-05, "logits/chosen": -0.734375, "logits/rejected": 2.3125, "logps/chosen": -1088.0, "logps/rejected": -2040.0, "loss": 0.0102, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -7.8125, "rewards/margins": 9.8125, "rewards/rejected": -17.625, "step": 3200 }, { "epoch": 3.349031920460492, "eval_logits/chosen": 0.75390625, "eval_logits/rejected": 2.765625, "eval_logps/chosen": -1424.0, "eval_logps/rejected": -1672.0, "eval_loss": 0.9822812676429749, "eval_rewards/accuracies": 0.71484375, "eval_rewards/chosen": -11.0625, "eval_rewards/margins": 2.828125, "eval_rewards/rejected": -13.875, "eval_runtime": 52.8816, "eval_samples_per_second": 37.82, "eval_steps_per_second": 0.605, "step": 3200 }, { "epoch": 3.359497645211931, "grad_norm": 0.08429283775876696, "learning_rate": 1.4655450881330557e-05, "logits/chosen": -1.078125, "logits/rejected": 2.90625, "logps/chosen": -1128.0, "logps/rejected": -2048.0, "loss": 0.0095, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -8.0625, "rewards/margins": 9.5625, "rewards/rejected": -17.625, "step": 3210 }, { "epoch": 3.36996336996337, "grad_norm": 0.20188899432806157, "learning_rate": 1.4489331894676955e-05, "logits/chosen": -0.765625, "logits/rejected": 3.015625, "logps/chosen": -1192.0, "logps/rejected": -2176.0, "loss": 0.0062, "rewards/accuracies": 1.0, "rewards/chosen": -8.8125, "rewards/margins": 10.0, "rewards/rejected": -18.75, "step": 3220 }, { "epoch": 3.380429094714809, "grad_norm": 0.4400190726675296, "learning_rate": 1.4323774728321799e-05, "logits/chosen": 0.0186767578125, "logits/rejected": 3.4375, "logps/chosen": -1216.0, "logps/rejected": -2256.0, "loss": 0.0121, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -9.0625, "rewards/margins": 10.6875, "rewards/rejected": -19.75, "step": 3230 }, { "epoch": 3.390894819466248, "grad_norm": 0.1780644502144959, "learning_rate": 1.4158788231690817e-05, "logits/chosen": -0.07080078125, "logits/rejected": 3.375, "logps/chosen": -1192.0, "logps/rejected": -2256.0, "loss": 0.0097, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -8.875, "rewards/margins": 10.875, "rewards/rejected": -19.75, "step": 3240 }, { "epoch": 3.4013605442176873, "grad_norm": 0.08336600862936035, "learning_rate": 1.399438122370606e-05, "logits/chosen": -0.302734375, "logits/rejected": 2.8125, "logps/chosen": -1136.0, "logps/rejected": -2080.0, "loss": 0.0069, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -8.125, "rewards/margins": 9.875, "rewards/rejected": -18.0, "step": 3250 }, { "epoch": 3.411826268969126, "grad_norm": 0.2844580711193598, "learning_rate": 1.383056249231453e-05, "logits/chosen": -0.296875, "logits/rejected": 2.546875, "logps/chosen": -1144.0, "logps/rejected": -2112.0, "loss": 0.0049, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -8.1875, "rewards/margins": 9.875, "rewards/rejected": -18.0, "step": 3260 }, { "epoch": 3.422291993720565, "grad_norm": 0.09143723360965565, "learning_rate": 1.366734079401844e-05, "logits/chosen": -0.171875, "logits/rejected": 3.078125, "logps/chosen": -1152.0, "logps/rejected": -2048.0, "loss": 0.005, "rewards/accuracies": 1.0, "rewards/chosen": -8.125, "rewards/margins": 9.375, "rewards/rejected": -17.5, "step": 3270 }, { "epoch": 3.432757718472004, "grad_norm": 0.1015527005173294, "learning_rate": 1.3504724853407186e-05, "logits/chosen": -0.32421875, "logits/rejected": 2.6875, "logps/chosen": -1192.0, "logps/rejected": -2112.0, "loss": 0.008, "rewards/accuracies": 1.0, "rewards/chosen": -8.5, "rewards/margins": 9.5625, "rewards/rejected": -18.125, "step": 3280 }, { "epoch": 3.4432234432234434, "grad_norm": 0.061434527839048896, "learning_rate": 1.3342723362690915e-05, "logits/chosen": -0.220703125, "logits/rejected": 2.75, "logps/chosen": -1176.0, "logps/rejected": -2176.0, "loss": 0.008, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -8.625, "rewards/margins": 10.25, "rewards/rejected": -18.875, "step": 3290 }, { "epoch": 3.4536891679748822, "grad_norm": 0.24663657756925686, "learning_rate": 1.3181344981236008e-05, "logits/chosen": 0.02490234375, "logits/rejected": 3.234375, "logps/chosen": -1168.0, "logps/rejected": -2224.0, "loss": 0.0082, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -8.5, "rewards/margins": 10.9375, "rewards/rejected": -19.375, "step": 3300 }, { "epoch": 3.4536891679748822, "eval_logits/chosen": 1.4453125, "eval_logits/rejected": 3.328125, "eval_logps/chosen": -1528.0, "eval_logps/rejected": -1800.0, "eval_loss": 1.0422968864440918, "eval_rewards/accuracies": 0.705078125, "eval_rewards/chosen": -12.1875, "eval_rewards/margins": 3.0, "eval_rewards/rejected": -15.1875, "eval_runtime": 52.8855, "eval_samples_per_second": 37.818, "eval_steps_per_second": 0.605, "step": 3300 }, { "epoch": 3.4641548927263215, "grad_norm": 0.07007637393090188, "learning_rate": 1.3020598335102141e-05, "logits/chosen": 0.453125, "logits/rejected": 3.671875, "logps/chosen": -1224.0, "logps/rejected": -2320.0, "loss": 0.0087, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -9.0625, "rewards/margins": 11.4375, "rewards/rejected": -20.5, "step": 3310 }, { "epoch": 3.4746206174777603, "grad_norm": 0.028916860560453693, "learning_rate": 1.2860492016581254e-05, "logits/chosen": -0.1982421875, "logits/rejected": 3.28125, "logps/chosen": -1168.0, "logps/rejected": -2176.0, "loss": 0.0159, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -8.75, "rewards/margins": 10.4375, "rewards/rejected": -19.25, "step": 3320 }, { "epoch": 3.4850863422291996, "grad_norm": 0.08385675983762503, "learning_rate": 1.2701034583738236e-05, "logits/chosen": -0.6484375, "logits/rejected": 3.0, "logps/chosen": -1192.0, "logps/rejected": -2176.0, "loss": 0.0048, "rewards/accuracies": 1.0, "rewards/chosen": -8.5625, "rewards/margins": 10.25, "rewards/rejected": -18.75, "step": 3330 }, { "epoch": 3.4955520669806384, "grad_norm": 0.03573166830757387, "learning_rate": 1.2542234559953481e-05, "logits/chosen": -0.55078125, "logits/rejected": 2.359375, "logps/chosen": -1120.0, "logps/rejected": -2192.0, "loss": 0.006, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -8.25, "rewards/margins": 10.75, "rewards/rejected": -19.0, "step": 3340 }, { "epoch": 3.506017791732077, "grad_norm": 0.2541379015084323, "learning_rate": 1.2384100433467292e-05, "logits/chosen": -0.498046875, "logits/rejected": 2.6875, "logps/chosen": -1128.0, "logps/rejected": -2112.0, "loss": 0.0089, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -8.125, "rewards/margins": 10.0625, "rewards/rejected": -18.125, "step": 3350 }, { "epoch": 3.5164835164835164, "grad_norm": 0.2175200998178627, "learning_rate": 1.2226640656926206e-05, "logits/chosen": -1.4296875, "logits/rejected": 2.65625, "logps/chosen": -1256.0, "logps/rejected": -2352.0, "loss": 0.0079, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -9.0625, "rewards/margins": 11.3125, "rewards/rejected": -20.375, "step": 3360 }, { "epoch": 3.5269492412349557, "grad_norm": 0.9318061276340431, "learning_rate": 1.2069863646931112e-05, "logits/chosen": -0.345703125, "logits/rejected": 3.46875, "logps/chosen": -1264.0, "logps/rejected": -2336.0, "loss": 0.0064, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -9.375, "rewards/margins": 11.0625, "rewards/rejected": -20.5, "step": 3370 }, { "epoch": 3.5374149659863945, "grad_norm": 1.028069173008785, "learning_rate": 1.1913777783587418e-05, "logits/chosen": -0.91796875, "logits/rejected": 3.40625, "logps/chosen": -1200.0, "logps/rejected": -2272.0, "loss": 0.0108, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -8.625, "rewards/margins": 11.25, "rewards/rejected": -19.875, "step": 3380 }, { "epoch": 3.5478806907378337, "grad_norm": 0.10508544826965832, "learning_rate": 1.1758391410057079e-05, "logits/chosen": -1.09375, "logits/rejected": 2.171875, "logps/chosen": -1152.0, "logps/rejected": -2192.0, "loss": 0.0066, "rewards/accuracies": 1.0, "rewards/chosen": -8.3125, "rewards/margins": 10.5625, "rewards/rejected": -18.875, "step": 3390 }, { "epoch": 3.5583464154892726, "grad_norm": 0.08953477427818697, "learning_rate": 1.1603712832112657e-05, "logits/chosen": -0.984375, "logits/rejected": 2.8125, "logps/chosen": -1184.0, "logps/rejected": -2192.0, "loss": 0.0109, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -8.75, "rewards/margins": 10.4375, "rewards/rejected": -19.125, "step": 3400 }, { "epoch": 3.5583464154892726, "eval_logits/chosen": 0.8671875, "eval_logits/rejected": 2.875, "eval_logps/chosen": -1456.0, "eval_logps/rejected": -1720.0, "eval_loss": 1.0224688053131104, "eval_rewards/accuracies": 0.716796875, "eval_rewards/chosen": -11.375, "eval_rewards/margins": 2.96875, "eval_rewards/rejected": -14.375, "eval_runtime": 52.9002, "eval_samples_per_second": 37.807, "eval_steps_per_second": 0.605, "step": 3400 }, { "epoch": 3.568812140240712, "grad_norm": 0.5504935902471659, "learning_rate": 1.1449750317693353e-05, "logits/chosen": -1.0078125, "logits/rejected": 3.015625, "logps/chosen": -1168.0, "logps/rejected": -2240.0, "loss": 0.0134, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -8.375, "rewards/margins": 11.0, "rewards/rejected": -19.375, "step": 3410 }, { "epoch": 3.5792778649921506, "grad_norm": 0.06561132538339516, "learning_rate": 1.1296512096463047e-05, "logits/chosen": -0.95703125, "logits/rejected": 2.796875, "logps/chosen": -1072.0, "logps/rejected": -2080.0, "loss": 0.0078, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -7.9375, "rewards/margins": 10.1875, "rewards/rejected": -18.125, "step": 3420 }, { "epoch": 3.58974358974359, "grad_norm": 0.3476124297679733, "learning_rate": 1.1144006359370453e-05, "logits/chosen": -1.2890625, "logits/rejected": 1.7734375, "logps/chosen": -1176.0, "logps/rejected": -2208.0, "loss": 0.01, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -8.5, "rewards/margins": 10.5625, "rewards/rejected": -19.0, "step": 3430 }, { "epoch": 3.6002093144950287, "grad_norm": 0.1391743907904667, "learning_rate": 1.0992241258211233e-05, "logits/chosen": -1.4765625, "logits/rejected": 2.421875, "logps/chosen": -1160.0, "logps/rejected": -2208.0, "loss": 0.0063, "rewards/accuracies": 1.0, "rewards/chosen": -8.375, "rewards/margins": 10.5625, "rewards/rejected": -18.875, "step": 3440 }, { "epoch": 3.610675039246468, "grad_norm": 0.05180346161416574, "learning_rate": 1.0841224905192294e-05, "logits/chosen": -1.3515625, "logits/rejected": 3.6875, "logps/chosen": -1112.0, "logps/rejected": -2080.0, "loss": 0.0049, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -7.90625, "rewards/margins": 10.25, "rewards/rejected": -18.25, "step": 3450 }, { "epoch": 3.6211407639979067, "grad_norm": 0.06659517905006358, "learning_rate": 1.069096537249816e-05, "logits/chosen": -1.1796875, "logits/rejected": 2.859375, "logps/chosen": -1064.0, "logps/rejected": -2080.0, "loss": 0.007, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -7.46875, "rewards/margins": 10.4375, "rewards/rejected": -17.875, "step": 3460 }, { "epoch": 3.631606488749346, "grad_norm": 0.2334302110267224, "learning_rate": 1.0541470691859532e-05, "logits/chosen": -0.7578125, "logits/rejected": 3.578125, "logps/chosen": -1200.0, "logps/rejected": -2208.0, "loss": 0.0117, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -8.5625, "rewards/margins": 10.5625, "rewards/rejected": -19.125, "step": 3470 }, { "epoch": 3.642072213500785, "grad_norm": 1.057650688155039, "learning_rate": 1.0392748854123913e-05, "logits/chosen": -0.76171875, "logits/rejected": 3.078125, "logps/chosen": -1104.0, "logps/rejected": -2096.0, "loss": 0.0117, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -7.71875, "rewards/margins": 10.4375, "rewards/rejected": -18.125, "step": 3480 }, { "epoch": 3.652537938252224, "grad_norm": 0.20649118172648792, "learning_rate": 1.0244807808828527e-05, "logits/chosen": -0.412109375, "logits/rejected": 3.765625, "logps/chosen": -1192.0, "logps/rejected": -2160.0, "loss": 0.007, "rewards/accuracies": 1.0, "rewards/chosen": -8.5625, "rewards/margins": 10.4375, "rewards/rejected": -19.0, "step": 3490 }, { "epoch": 3.663003663003663, "grad_norm": 0.04027065560164829, "learning_rate": 1.0097655463775347e-05, "logits/chosen": -0.9765625, "logits/rejected": 3.4375, "logps/chosen": -1168.0, "logps/rejected": -2112.0, "loss": 0.0098, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -8.25, "rewards/margins": 10.0, "rewards/rejected": -18.25, "step": 3500 }, { "epoch": 3.663003663003663, "eval_logits/chosen": 1.171875, "eval_logits/rejected": 3.1875, "eval_logps/chosen": -1456.0, "eval_logps/rejected": -1712.0, "eval_loss": 1.0070234537124634, "eval_rewards/accuracies": 0.7109375, "eval_rewards/chosen": -11.4375, "eval_rewards/margins": 2.84375, "eval_rewards/rejected": -14.25, "eval_runtime": 52.8862, "eval_samples_per_second": 37.817, "eval_steps_per_second": 0.605, "step": 3500 }, { "epoch": 3.673469387755102, "grad_norm": 0.3978887479848816, "learning_rate": 9.951299684608479e-06, "logits/chosen": -0.3984375, "logits/rejected": 3.53125, "logps/chosen": -1168.0, "logps/rejected": -2176.0, "loss": 0.0074, "rewards/accuracies": 1.0, "rewards/chosen": -8.6875, "rewards/margins": 10.3125, "rewards/rejected": -19.0, "step": 3510 }, { "epoch": 3.683935112506541, "grad_norm": 0.040155524578011734, "learning_rate": 9.805748294393658e-06, "logits/chosen": -0.10498046875, "logits/rejected": 3.484375, "logps/chosen": -1168.0, "logps/rejected": -2224.0, "loss": 0.0116, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -8.6875, "rewards/margins": 10.75, "rewards/rejected": -19.375, "step": 3520 }, { "epoch": 3.69440083725798, "grad_norm": 0.06256794091530021, "learning_rate": 9.661009073200105e-06, "logits/chosen": 0.6171875, "logits/rejected": 3.90625, "logps/chosen": -1304.0, "logps/rejected": -2288.0, "loss": 0.0092, "rewards/accuracies": 0.96875, "rewards/chosen": -9.5, "rewards/margins": 10.3125, "rewards/rejected": -19.75, "step": 3530 }, { "epoch": 3.704866562009419, "grad_norm": 0.13259944802218135, "learning_rate": 9.517089757684675e-06, "logits/chosen": -0.546875, "logits/rejected": 2.96875, "logps/chosen": -1200.0, "logps/rejected": -2256.0, "loss": 0.0098, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -8.8125, "rewards/margins": 10.9375, "rewards/rejected": -19.75, "step": 3540 }, { "epoch": 3.7153322867608582, "grad_norm": 0.282203368505097, "learning_rate": 9.373998040678317e-06, "logits/chosen": 0.2578125, "logits/rejected": 4.1875, "logps/chosen": -1152.0, "logps/rejected": -2192.0, "loss": 0.0081, "rewards/accuracies": 1.0, "rewards/chosen": -8.625, "rewards/margins": 10.6875, "rewards/rejected": -19.375, "step": 3550 }, { "epoch": 3.725798011512297, "grad_norm": 0.5695312875065408, "learning_rate": 9.231741570774851e-06, "logits/chosen": -0.3359375, "logits/rejected": 4.3125, "logps/chosen": -1240.0, "logps/rejected": -2320.0, "loss": 0.0062, "rewards/accuracies": 1.0, "rewards/chosen": -9.375, "rewards/margins": 11.3125, "rewards/rejected": -20.625, "step": 3560 }, { "epoch": 3.7362637362637363, "grad_norm": 0.01268718846210354, "learning_rate": 9.090327951922147e-06, "logits/chosen": -0.12890625, "logits/rejected": 3.140625, "logps/chosen": -1240.0, "logps/rejected": -2464.0, "loss": 0.0069, "rewards/accuracies": 1.0, "rewards/chosen": -9.4375, "rewards/margins": 12.3125, "rewards/rejected": -21.75, "step": 3570 }, { "epoch": 3.7467294610151756, "grad_norm": 0.16298241858505716, "learning_rate": 8.949764743015668e-06, "logits/chosen": -0.41796875, "logits/rejected": 2.421875, "logps/chosen": -1216.0, "logps/rejected": -2320.0, "loss": 0.0106, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -8.75, "rewards/margins": 11.3125, "rewards/rejected": -20.125, "step": 3580 }, { "epoch": 3.7571951857666144, "grad_norm": 0.21812828668959902, "learning_rate": 8.810059457494474e-06, "logits/chosen": -0.466796875, "logits/rejected": 3.5625, "logps/chosen": -1248.0, "logps/rejected": -2304.0, "loss": 0.0065, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -9.1875, "rewards/margins": 10.8125, "rewards/rejected": -20.0, "step": 3590 }, { "epoch": 3.767660910518053, "grad_norm": 0.256677383911918, "learning_rate": 8.671219562939514e-06, "logits/chosen": -0.69921875, "logits/rejected": 3.234375, "logps/chosen": -1192.0, "logps/rejected": -2256.0, "loss": 0.007, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -9.0, "rewards/margins": 10.8125, "rewards/rejected": -19.875, "step": 3600 }, { "epoch": 3.767660910518053, "eval_logits/chosen": 0.80859375, "eval_logits/rejected": 2.859375, "eval_logps/chosen": -1512.0, "eval_logps/rejected": -1776.0, "eval_loss": 1.0389765501022339, "eval_rewards/accuracies": 0.71484375, "eval_rewards/chosen": -11.9375, "eval_rewards/margins": 3.0, "eval_rewards/rejected": -14.9375, "eval_runtime": 52.8868, "eval_samples_per_second": 37.817, "eval_steps_per_second": 0.605, "step": 3600 }, { "epoch": 3.7781266352694924, "grad_norm": 0.3973905986468847, "learning_rate": 8.533252480674552e-06, "logits/chosen": -1.3515625, "logits/rejected": 2.84375, "logps/chosen": -1232.0, "logps/rejected": -2224.0, "loss": 0.013, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -9.125, "rewards/margins": 10.1875, "rewards/rejected": -19.25, "step": 3610 }, { "epoch": 3.7885923600209317, "grad_norm": 0.017282421705160138, "learning_rate": 8.396165585369473e-06, "logits/chosen": -0.9296875, "logits/rejected": 3.1875, "logps/chosen": -1248.0, "logps/rejected": -2256.0, "loss": 0.0053, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -8.875, "rewards/margins": 10.6875, "rewards/rejected": -19.625, "step": 3620 }, { "epoch": 3.7990580847723705, "grad_norm": 0.06936587201416461, "learning_rate": 8.259966204646036e-06, "logits/chosen": -0.78515625, "logits/rejected": 3.34375, "logps/chosen": -1152.0, "logps/rejected": -2272.0, "loss": 0.0076, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -8.125, "rewards/margins": 11.625, "rewards/rejected": -19.75, "step": 3630 }, { "epoch": 3.8095238095238093, "grad_norm": 0.14962645684813214, "learning_rate": 8.124661618686241e-06, "logits/chosen": -0.4296875, "logits/rejected": 3.015625, "logps/chosen": -1176.0, "logps/rejected": -2160.0, "loss": 0.0151, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -8.625, "rewards/margins": 10.125, "rewards/rejected": -18.75, "step": 3640 }, { "epoch": 3.8199895342752486, "grad_norm": 0.6158368813659576, "learning_rate": 7.99025905984316e-06, "logits/chosen": -0.1220703125, "logits/rejected": 3.234375, "logps/chosen": -1192.0, "logps/rejected": -2240.0, "loss": 0.0104, "rewards/accuracies": 1.0, "rewards/chosen": -8.9375, "rewards/margins": 10.625, "rewards/rejected": -19.5, "step": 3650 }, { "epoch": 3.830455259026688, "grad_norm": 0.047097837629874094, "learning_rate": 7.856765712254393e-06, "logits/chosen": 0.2431640625, "logits/rejected": 3.875, "logps/chosen": -1288.0, "logps/rejected": -2256.0, "loss": 0.0066, "rewards/accuracies": 1.0, "rewards/chosen": -9.25, "rewards/margins": 10.375, "rewards/rejected": -19.625, "step": 3660 }, { "epoch": 3.8409209837781266, "grad_norm": 0.025503821495670966, "learning_rate": 7.724188711457997e-06, "logits/chosen": -0.5546875, "logits/rejected": 2.84375, "logps/chosen": -1224.0, "logps/rejected": -2304.0, "loss": 0.0082, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -8.9375, "rewards/margins": 11.0625, "rewards/rejected": -20.0, "step": 3670 }, { "epoch": 3.8513867085295654, "grad_norm": 0.09958802810357707, "learning_rate": 7.5925351440111175e-06, "logits/chosen": -0.5703125, "logits/rejected": 3.96875, "logps/chosen": -1256.0, "logps/rejected": -2320.0, "loss": 0.0051, "rewards/accuracies": 1.0, "rewards/chosen": -9.1875, "rewards/margins": 11.1875, "rewards/rejected": -20.375, "step": 3680 }, { "epoch": 3.8618524332810047, "grad_norm": 0.34872245324217366, "learning_rate": 7.461812047111172e-06, "logits/chosen": -0.314453125, "logits/rejected": 3.890625, "logps/chosen": -1296.0, "logps/rejected": -2288.0, "loss": 0.0052, "rewards/accuracies": 1.0, "rewards/chosen": -9.5, "rewards/margins": 10.6875, "rewards/rejected": -20.125, "step": 3690 }, { "epoch": 3.872318158032444, "grad_norm": 0.1344362341791477, "learning_rate": 7.332026408219742e-06, "logits/chosen": -0.2373046875, "logits/rejected": 4.03125, "logps/chosen": -1288.0, "logps/rejected": -2336.0, "loss": 0.0057, "rewards/accuracies": 1.0, "rewards/chosen": -9.625, "rewards/margins": 11.125, "rewards/rejected": -20.75, "step": 3700 }, { "epoch": 3.872318158032444, "eval_logits/chosen": 1.5, "eval_logits/rejected": 3.453125, "eval_logps/chosen": -1584.0, "eval_logps/rejected": -1864.0, "eval_loss": 1.0701953172683716, "eval_rewards/accuracies": 0.703125, "eval_rewards/chosen": -12.75, "eval_rewards/margins": 3.0625, "eval_rewards/rejected": -15.8125, "eval_runtime": 52.8941, "eval_samples_per_second": 37.811, "eval_steps_per_second": 0.605, "step": 3700 }, { "epoch": 3.8827838827838828, "grad_norm": 0.020992160364998343, "learning_rate": 7.203185164688992e-06, "logits/chosen": 0.263671875, "logits/rejected": 3.671875, "logps/chosen": -1272.0, "logps/rejected": -2464.0, "loss": 0.0035, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -9.8125, "rewards/margins": 12.0, "rewards/rejected": -21.875, "step": 3710 }, { "epoch": 3.8932496075353216, "grad_norm": 0.07356976842643481, "learning_rate": 7.075295203390931e-06, "logits/chosen": -0.1162109375, "logits/rejected": 3.015625, "logps/chosen": -1248.0, "logps/rejected": -2320.0, "loss": 0.0078, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -9.375, "rewards/margins": 11.0, "rewards/rejected": -20.375, "step": 3720 }, { "epoch": 3.903715332286761, "grad_norm": 0.203253833608076, "learning_rate": 6.948363360349247e-06, "logits/chosen": -1.078125, "logits/rejected": 3.578125, "logps/chosen": -1280.0, "logps/rejected": -2336.0, "loss": 0.0058, "rewards/accuracies": 1.0, "rewards/chosen": -9.5625, "rewards/margins": 11.0, "rewards/rejected": -20.5, "step": 3730 }, { "epoch": 3.9141810570382, "grad_norm": 0.31356668119037906, "learning_rate": 6.822396420373942e-06, "logits/chosen": -0.380859375, "logits/rejected": 2.90625, "logps/chosen": -1240.0, "logps/rejected": -2352.0, "loss": 0.0085, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -9.0625, "rewards/margins": 11.1875, "rewards/rejected": -20.25, "step": 3740 }, { "epoch": 3.924646781789639, "grad_norm": 0.07165119195640822, "learning_rate": 6.697401116698626e-06, "logits/chosen": -1.015625, "logits/rejected": 3.03125, "logps/chosen": -1224.0, "logps/rejected": -2320.0, "loss": 0.0065, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -9.125, "rewards/margins": 11.25, "rewards/rejected": -20.375, "step": 3750 }, { "epoch": 3.9351125065410777, "grad_norm": 0.08827057592402958, "learning_rate": 6.573384130620638e-06, "logits/chosen": 0.9453125, "logits/rejected": 3.609375, "logps/chosen": -1248.0, "logps/rejected": -2368.0, "loss": 0.0094, "rewards/accuracies": 1.0, "rewards/chosen": -9.375, "rewards/margins": 11.4375, "rewards/rejected": -20.875, "step": 3760 }, { "epoch": 3.945578231292517, "grad_norm": 0.048809836235047845, "learning_rate": 6.450352091143891e-06, "logits/chosen": -0.443359375, "logits/rejected": 4.21875, "logps/chosen": -1256.0, "logps/rejected": -2304.0, "loss": 0.0097, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -9.25, "rewards/margins": 11.0, "rewards/rejected": -20.25, "step": 3770 }, { "epoch": 3.956043956043956, "grad_norm": 0.019544562581198487, "learning_rate": 6.328311574624582e-06, "logits/chosen": -0.671875, "logits/rejected": 2.90625, "logps/chosen": -1272.0, "logps/rejected": -2352.0, "loss": 0.0059, "rewards/accuracies": 1.0, "rewards/chosen": -9.4375, "rewards/margins": 10.875, "rewards/rejected": -20.375, "step": 3780 }, { "epoch": 3.966509680795395, "grad_norm": 0.13614961147873206, "learning_rate": 6.20726910441961e-06, "logits/chosen": 0.09619140625, "logits/rejected": 3.078125, "logps/chosen": -1208.0, "logps/rejected": -2256.0, "loss": 0.0061, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -9.0625, "rewards/margins": 10.625, "rewards/rejected": -19.625, "step": 3790 }, { "epoch": 3.9769754055468343, "grad_norm": 0.10878858635322676, "learning_rate": 6.087231150537933e-06, "logits/chosen": -0.041015625, "logits/rejected": 3.71875, "logps/chosen": -1248.0, "logps/rejected": -2288.0, "loss": 0.0054, "rewards/accuracies": 1.0, "rewards/chosen": -9.0625, "rewards/margins": 11.25, "rewards/rejected": -20.25, "step": 3800 }, { "epoch": 3.9769754055468343, "eval_logits/chosen": 1.46875, "eval_logits/rejected": 3.40625, "eval_logps/chosen": -1560.0, "eval_logps/rejected": -1832.0, "eval_loss": 1.0484999418258667, "eval_rewards/accuracies": 0.703125, "eval_rewards/chosen": -12.4375, "eval_rewards/margins": 3.0, "eval_rewards/rejected": -15.4375, "eval_runtime": 52.8974, "eval_samples_per_second": 37.809, "eval_steps_per_second": 0.605, "step": 3800 }, { "epoch": 3.987441130298273, "grad_norm": 2.2928026898675373, "learning_rate": 5.968204129294705e-06, "logits/chosen": -0.140625, "logits/rejected": 3.0625, "logps/chosen": -1232.0, "logps/rejected": -2288.0, "loss": 0.0063, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -9.125, "rewards/margins": 10.875, "rewards/rejected": -20.0, "step": 3810 }, { "epoch": 3.9979068550497123, "grad_norm": 0.08119257378657181, "learning_rate": 5.850194402968317e-06, "logits/chosen": -0.76953125, "logits/rejected": 3.109375, "logps/chosen": -1216.0, "logps/rejected": -2288.0, "loss": 0.006, "rewards/accuracies": 1.0, "rewards/chosen": -9.0, "rewards/margins": 10.9375, "rewards/rejected": -20.0, "step": 3820 }, { "epoch": 4.008372579801152, "grad_norm": 0.014981417991040193, "learning_rate": 5.733208279460331e-06, "logits/chosen": -0.47265625, "logits/rejected": 3.59375, "logps/chosen": -1248.0, "logps/rejected": -2304.0, "loss": 0.0037, "rewards/accuracies": 1.0, "rewards/chosen": -9.0625, "rewards/margins": 10.9375, "rewards/rejected": -20.0, "step": 3830 }, { "epoch": 4.01883830455259, "grad_norm": 0.02009579096970186, "learning_rate": 5.617252011958268e-06, "logits/chosen": -0.2734375, "logits/rejected": 2.921875, "logps/chosen": -1224.0, "logps/rejected": -2352.0, "loss": 0.0053, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -9.0, "rewards/margins": 11.3125, "rewards/rejected": -20.25, "step": 3840 }, { "epoch": 4.029304029304029, "grad_norm": 0.03084951999687856, "learning_rate": 5.502331798601429e-06, "logits/chosen": -0.87109375, "logits/rejected": 3.0625, "logps/chosen": -1224.0, "logps/rejected": -2240.0, "loss": 0.005, "rewards/accuracies": 1.0, "rewards/chosen": -9.0, "rewards/margins": 10.5625, "rewards/rejected": -19.625, "step": 3850 }, { "epoch": 4.0397697540554685, "grad_norm": 0.015498781392122992, "learning_rate": 5.388453782149514e-06, "logits/chosen": -0.03173828125, "logits/rejected": 3.171875, "logps/chosen": -1304.0, "logps/rejected": -2368.0, "loss": 0.0059, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -9.4375, "rewards/margins": 11.0, "rewards/rejected": -20.5, "step": 3860 }, { "epoch": 4.050235478806908, "grad_norm": 0.02232794620336459, "learning_rate": 5.275624049654318e-06, "logits/chosen": 0.3125, "logits/rejected": 3.578125, "logps/chosen": -1248.0, "logps/rejected": -2448.0, "loss": 0.0059, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -9.375, "rewards/margins": 12.375, "rewards/rejected": -21.75, "step": 3870 }, { "epoch": 4.060701203558346, "grad_norm": 0.020459293734781368, "learning_rate": 5.163848632134358e-06, "logits/chosen": -0.5703125, "logits/rejected": 3.40625, "logps/chosen": -1288.0, "logps/rejected": -2480.0, "loss": 0.007, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -9.5625, "rewards/margins": 12.1875, "rewards/rejected": -21.75, "step": 3880 }, { "epoch": 4.071166928309785, "grad_norm": 0.01373231281615159, "learning_rate": 5.053133504252511e-06, "logits/chosen": -0.49609375, "logits/rejected": 3.6875, "logps/chosen": -1296.0, "logps/rejected": -2400.0, "loss": 0.0095, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -9.5, "rewards/margins": 11.6875, "rewards/rejected": -21.125, "step": 3890 }, { "epoch": 4.081632653061225, "grad_norm": 0.08307094209547573, "learning_rate": 4.94348458399663e-06, "logits/chosen": -0.47265625, "logits/rejected": 3.1875, "logps/chosen": -1168.0, "logps/rejected": -2288.0, "loss": 0.0037, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -8.875, "rewards/margins": 11.375, "rewards/rejected": -20.25, "step": 3900 }, { "epoch": 4.081632653061225, "eval_logits/chosen": 1.6171875, "eval_logits/rejected": 3.546875, "eval_logps/chosen": -1600.0, "eval_logps/rejected": -1880.0, "eval_loss": 1.0904687643051147, "eval_rewards/accuracies": 0.703125, "eval_rewards/chosen": -12.8125, "eval_rewards/margins": 3.140625, "eval_rewards/rejected": -15.9375, "eval_runtime": 52.877, "eval_samples_per_second": 37.824, "eval_steps_per_second": 0.605, "step": 3900 }, { "epoch": 4.092098377812664, "grad_norm": 0.03337438011237956, "learning_rate": 4.834907732363217e-06, "logits/chosen": 7.343292236328125e-05, "logits/rejected": 3.859375, "logps/chosen": -1248.0, "logps/rejected": -2352.0, "loss": 0.0059, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -9.4375, "rewards/margins": 11.5, "rewards/rejected": -21.0, "step": 3910 }, { "epoch": 4.102564102564102, "grad_norm": 0.019973325709137745, "learning_rate": 4.727408753044155e-06, "logits/chosen": 0.248046875, "logits/rejected": 3.765625, "logps/chosen": -1352.0, "logps/rejected": -2464.0, "loss": 0.0067, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -10.0, "rewards/margins": 11.75, "rewards/rejected": -21.75, "step": 3920 }, { "epoch": 4.1130298273155415, "grad_norm": 0.3108228360137749, "learning_rate": 4.620993392116493e-06, "logits/chosen": -0.109375, "logits/rejected": 3.6875, "logps/chosen": -1288.0, "logps/rejected": -2416.0, "loss": 0.0043, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -9.75, "rewards/margins": 11.375, "rewards/rejected": -21.125, "step": 3930 }, { "epoch": 4.123495552066981, "grad_norm": 0.013055671038019035, "learning_rate": 4.515667337735277e-06, "logits/chosen": -0.640625, "logits/rejected": 3.921875, "logps/chosen": -1288.0, "logps/rejected": -2416.0, "loss": 0.0058, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -9.4375, "rewards/margins": 11.75, "rewards/rejected": -21.25, "step": 3940 }, { "epoch": 4.13396127681842, "grad_norm": 0.014303222620106174, "learning_rate": 4.4114362198295155e-06, "logits/chosen": -0.251953125, "logits/rejected": 3.65625, "logps/chosen": -1328.0, "logps/rejected": -2480.0, "loss": 0.009, "rewards/accuracies": 1.0, "rewards/chosen": -9.9375, "rewards/margins": 12.0, "rewards/rejected": -21.875, "step": 3950 }, { "epoch": 4.144427001569858, "grad_norm": 0.0427935909815543, "learning_rate": 4.308305609801244e-06, "logits/chosen": -0.73046875, "logits/rejected": 3.390625, "logps/chosen": -1272.0, "logps/rejected": -2320.0, "loss": 0.0054, "rewards/accuracies": 1.0, "rewards/chosen": -9.5625, "rewards/margins": 10.9375, "rewards/rejected": -20.5, "step": 3960 }, { "epoch": 4.154892726321298, "grad_norm": 0.025518900183695518, "learning_rate": 4.206281020227754e-06, "logits/chosen": -0.08056640625, "logits/rejected": 3.65625, "logps/chosen": -1344.0, "logps/rejected": -2432.0, "loss": 0.0047, "rewards/accuracies": 1.0, "rewards/chosen": -10.0, "rewards/margins": 11.3125, "rewards/rejected": -21.25, "step": 3970 }, { "epoch": 4.165358451072737, "grad_norm": 0.010020871741228683, "learning_rate": 4.105367904566873e-06, "logits/chosen": 0.8125, "logits/rejected": 3.953125, "logps/chosen": -1288.0, "logps/rejected": -2464.0, "loss": 0.0037, "rewards/accuracies": 1.0, "rewards/chosen": -9.625, "rewards/margins": 11.875, "rewards/rejected": -21.5, "step": 3980 }, { "epoch": 4.175824175824176, "grad_norm": 0.011681294736082001, "learning_rate": 4.005571656865503e-06, "logits/chosen": -0.0732421875, "logits/rejected": 4.03125, "logps/chosen": -1320.0, "logps/rejected": -2432.0, "loss": 0.0069, "rewards/accuracies": 1.0, "rewards/chosen": -9.8125, "rewards/margins": 11.5625, "rewards/rejected": -21.375, "step": 3990 }, { "epoch": 4.1862899005756145, "grad_norm": 0.26505203952414647, "learning_rate": 3.906897611471291e-06, "logits/chosen": 0.1298828125, "logits/rejected": 3.890625, "logps/chosen": -1320.0, "logps/rejected": -2528.0, "loss": 0.0031, "rewards/accuracies": 1.0, "rewards/chosen": -9.9375, "rewards/margins": 12.5625, "rewards/rejected": -22.5, "step": 4000 }, { "epoch": 4.1862899005756145, "eval_logits/chosen": 1.65625, "eval_logits/rejected": 3.609375, "eval_logps/chosen": -1616.0, "eval_logps/rejected": -1912.0, "eval_loss": 1.1163359880447388, "eval_rewards/accuracies": 0.701171875, "eval_rewards/chosen": -13.0625, "eval_rewards/margins": 3.21875, "eval_rewards/rejected": -16.25, "eval_runtime": 52.8881, "eval_samples_per_second": 37.816, "eval_steps_per_second": 0.605, "step": 4000 }, { "epoch": 4.196755625327054, "grad_norm": 0.06020278446533552, "learning_rate": 3.8093510427475027e-06, "logits/chosen": 0.47265625, "logits/rejected": 3.859375, "logps/chosen": -1280.0, "logps/rejected": -2448.0, "loss": 0.0059, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -9.625, "rewards/margins": 11.8125, "rewards/rejected": -21.5, "step": 4010 }, { "epoch": 4.207221350078493, "grad_norm": 0.013675313695423432, "learning_rate": 3.712937164791061e-06, "logits/chosen": 0.1328125, "logits/rejected": 2.875, "logps/chosen": -1280.0, "logps/rejected": -2464.0, "loss": 0.0068, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -9.6875, "rewards/margins": 11.8125, "rewards/rejected": -21.5, "step": 4020 }, { "epoch": 4.217687074829932, "grad_norm": 0.015820780338319115, "learning_rate": 3.617661131153871e-06, "logits/chosen": -0.984375, "logits/rejected": 3.09375, "logps/chosen": -1312.0, "logps/rejected": -2384.0, "loss": 0.0057, "rewards/accuracies": 1.0, "rewards/chosen": -9.625, "rewards/margins": 11.25, "rewards/rejected": -20.875, "step": 4030 }, { "epoch": 4.228152799581371, "grad_norm": 1.2290489445593125, "learning_rate": 3.523528034567361e-06, "logits/chosen": 0.0400390625, "logits/rejected": 3.71875, "logps/chosen": -1360.0, "logps/rejected": -2528.0, "loss": 0.0071, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -10.1875, "rewards/margins": 12.0, "rewards/rejected": -22.125, "step": 4040 }, { "epoch": 4.23861852433281, "grad_norm": 0.014275682691672148, "learning_rate": 3.4305429066702284e-06, "logits/chosen": 0.0120849609375, "logits/rejected": 3.40625, "logps/chosen": -1392.0, "logps/rejected": -2560.0, "loss": 0.0025, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -10.375, "rewards/margins": 12.0, "rewards/rejected": -22.375, "step": 4050 }, { "epoch": 4.249084249084249, "grad_norm": 0.01693962235310755, "learning_rate": 3.3387107177395044e-06, "logits/chosen": -0.23046875, "logits/rejected": 3.578125, "logps/chosen": -1304.0, "logps/rejected": -2416.0, "loss": 0.0049, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -9.875, "rewards/margins": 11.5, "rewards/rejected": -21.375, "step": 4060 }, { "epoch": 4.259549973835688, "grad_norm": 0.015628704589392063, "learning_rate": 3.2480363764248843e-06, "logits/chosen": -0.61328125, "logits/rejected": 3.390625, "logps/chosen": -1296.0, "logps/rejected": -2432.0, "loss": 0.0075, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -9.75, "rewards/margins": 11.625, "rewards/rejected": -21.375, "step": 4070 }, { "epoch": 4.270015698587127, "grad_norm": 0.023587527509885093, "learning_rate": 3.158524729486351e-06, "logits/chosen": -0.43359375, "logits/rejected": 3.203125, "logps/chosen": -1264.0, "logps/rejected": -2336.0, "loss": 0.0042, "rewards/accuracies": 1.0, "rewards/chosen": -9.3125, "rewards/margins": 11.0625, "rewards/rejected": -20.375, "step": 4080 }, { "epoch": 4.280481423338566, "grad_norm": 0.05923026570392513, "learning_rate": 3.070180561535088e-06, "logits/chosen": 0.33203125, "logits/rejected": 4.0, "logps/chosen": -1344.0, "logps/rejected": -2496.0, "loss": 0.0025, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -10.125, "rewards/margins": 11.8125, "rewards/rejected": -22.0, "step": 4090 }, { "epoch": 4.290947148090005, "grad_norm": 0.017270659248406598, "learning_rate": 2.983008594777745e-06, "logits/chosen": -0.1416015625, "logits/rejected": 3.296875, "logps/chosen": -1344.0, "logps/rejected": -2544.0, "loss": 0.0037, "rewards/accuracies": 1.0, "rewards/chosen": -9.75, "rewards/margins": 12.25, "rewards/rejected": -22.0, "step": 4100 }, { "epoch": 4.290947148090005, "eval_logits/chosen": 1.65625, "eval_logits/rejected": 3.609375, "eval_logps/chosen": -1624.0, "eval_logps/rejected": -1920.0, "eval_loss": 1.1256483793258667, "eval_rewards/accuracies": 0.708984375, "eval_rewards/chosen": -13.125, "eval_rewards/margins": 3.265625, "eval_rewards/rejected": -16.375, "eval_runtime": 52.8983, "eval_samples_per_second": 37.808, "eval_steps_per_second": 0.605, "step": 4100 }, { "epoch": 4.3014128728414445, "grad_norm": 0.03873780177102751, "learning_rate": 2.8970134887640137e-06, "logits/chosen": -0.19140625, "logits/rejected": 3.5, "logps/chosen": -1304.0, "logps/rejected": -2432.0, "loss": 0.005, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -9.5625, "rewards/margins": 11.875, "rewards/rejected": -21.5, "step": 4110 }, { "epoch": 4.311878597592884, "grad_norm": 0.01075555055581885, "learning_rate": 2.8121998401375804e-06, "logits/chosen": 0.1796875, "logits/rejected": 3.84375, "logps/chosen": -1304.0, "logps/rejected": -2432.0, "loss": 0.0108, "rewards/accuracies": 1.0, "rewards/chosen": -9.8125, "rewards/margins": 11.8125, "rewards/rejected": -21.625, "step": 4120 }, { "epoch": 4.322344322344322, "grad_norm": 0.014744483880713274, "learning_rate": 2.7285721823904044e-06, "logits/chosen": 0.0302734375, "logits/rejected": 3.453125, "logps/chosen": -1296.0, "logps/rejected": -2512.0, "loss": 0.0061, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -9.875, "rewards/margins": 12.1875, "rewards/rejected": -22.0, "step": 4130 }, { "epoch": 4.332810047095761, "grad_norm": 0.050764409494126105, "learning_rate": 2.6461349856204077e-06, "logits/chosen": 0.1064453125, "logits/rejected": 3.125, "logps/chosen": -1240.0, "logps/rejected": -2480.0, "loss": 0.008, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -9.5, "rewards/margins": 12.375, "rewards/rejected": -21.875, "step": 4140 }, { "epoch": 4.343275771847201, "grad_norm": 0.12487661904822162, "learning_rate": 2.5648926562925222e-06, "logits/chosen": 0.169921875, "logits/rejected": 4.3125, "logps/chosen": -1288.0, "logps/rejected": -2432.0, "loss": 0.0049, "rewards/accuracies": 1.0, "rewards/chosen": -9.625, "rewards/margins": 12.0, "rewards/rejected": -21.625, "step": 4150 }, { "epoch": 4.35374149659864, "grad_norm": 0.018877951979946545, "learning_rate": 2.484849537003184e-06, "logits/chosen": -0.7578125, "logits/rejected": 2.765625, "logps/chosen": -1288.0, "logps/rejected": -2464.0, "loss": 0.008, "rewards/accuracies": 1.0, "rewards/chosen": -9.75, "rewards/margins": 11.8125, "rewards/rejected": -21.625, "step": 4160 }, { "epoch": 4.364207221350078, "grad_norm": 0.012654241651534965, "learning_rate": 2.406009906248169e-06, "logits/chosen": -0.208984375, "logits/rejected": 3.640625, "logps/chosen": -1416.0, "logps/rejected": -2544.0, "loss": 0.0055, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -10.4375, "rewards/margins": 12.0625, "rewards/rejected": -22.5, "step": 4170 }, { "epoch": 4.3746729461015175, "grad_norm": 0.015622471752332513, "learning_rate": 2.328377978193924e-06, "logits/chosen": -0.020751953125, "logits/rejected": 3.328125, "logps/chosen": -1304.0, "logps/rejected": -2480.0, "loss": 0.0035, "rewards/accuracies": 1.0, "rewards/chosen": -9.75, "rewards/margins": 12.0625, "rewards/rejected": -21.75, "step": 4180 }, { "epoch": 4.385138670852957, "grad_norm": 0.030115198260313527, "learning_rate": 2.2519579024523035e-06, "logits/chosen": 0.2890625, "logits/rejected": 4.21875, "logps/chosen": -1264.0, "logps/rejected": -2496.0, "loss": 0.008, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -9.6875, "rewards/margins": 12.5625, "rewards/rejected": -22.25, "step": 4190 }, { "epoch": 4.395604395604396, "grad_norm": 0.010975380556523135, "learning_rate": 2.1767537638587743e-06, "logits/chosen": 0.78515625, "logits/rejected": 4.15625, "logps/chosen": -1376.0, "logps/rejected": -2560.0, "loss": 0.0089, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -10.375, "rewards/margins": 12.1875, "rewards/rejected": -22.625, "step": 4200 }, { "epoch": 4.395604395604396, "eval_logits/chosen": 1.8125, "eval_logits/rejected": 3.75, "eval_logps/chosen": -1648.0, "eval_logps/rejected": -1952.0, "eval_loss": 1.1394531726837158, "eval_rewards/accuracies": 0.70703125, "eval_rewards/chosen": -13.3125, "eval_rewards/margins": 3.3125, "eval_rewards/rejected": -16.625, "eval_runtime": 52.8795, "eval_samples_per_second": 37.822, "eval_steps_per_second": 0.605, "step": 4200 }, { "epoch": 4.406070120355834, "grad_norm": 0.010336791984861466, "learning_rate": 2.1027695822540493e-06, "logits/chosen": 0.291015625, "logits/rejected": 4.03125, "logps/chosen": -1280.0, "logps/rejected": -2464.0, "loss": 0.0051, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -9.6875, "rewards/margins": 12.0625, "rewards/rejected": -21.75, "step": 4210 }, { "epoch": 4.416535845107274, "grad_norm": 0.022113022114078094, "learning_rate": 2.0300093122692184e-06, "logits/chosen": 1.0390625, "logits/rejected": 4.71875, "logps/chosen": -1320.0, "logps/rejected": -2496.0, "loss": 0.0047, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -10.0625, "rewards/margins": 12.125, "rewards/rejected": -22.25, "step": 4220 }, { "epoch": 4.427001569858713, "grad_norm": 0.046000333165739614, "learning_rate": 1.9584768431143955e-06, "logits/chosen": -0.1279296875, "logits/rejected": 3.765625, "logps/chosen": -1320.0, "logps/rejected": -2432.0, "loss": 0.0058, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -10.0, "rewards/margins": 11.8125, "rewards/rejected": -21.875, "step": 4230 }, { "epoch": 4.437467294610152, "grad_norm": 0.03154170065295428, "learning_rate": 1.888175998370803e-06, "logits/chosen": 0.43359375, "logits/rejected": 4.09375, "logps/chosen": -1264.0, "logps/rejected": -2368.0, "loss": 0.0092, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -9.6875, "rewards/margins": 11.375, "rewards/rejected": -21.0, "step": 4240 }, { "epoch": 4.4479330193615905, "grad_norm": 0.02632161963106536, "learning_rate": 1.819110535786389e-06, "logits/chosen": 0.1044921875, "logits/rejected": 3.890625, "logps/chosen": -1328.0, "logps/rejected": -2448.0, "loss": 0.0069, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -10.0625, "rewards/margins": 11.625, "rewards/rejected": -21.75, "step": 4250 }, { "epoch": 4.45839874411303, "grad_norm": 0.01806533303901749, "learning_rate": 1.751284147074983e-06, "logits/chosen": 0.7109375, "logits/rejected": 3.96875, "logps/chosen": -1376.0, "logps/rejected": -2544.0, "loss": 0.0075, "rewards/accuracies": 1.0, "rewards/chosen": -10.625, "rewards/margins": 11.8125, "rewards/rejected": -22.375, "step": 4260 }, { "epoch": 4.468864468864469, "grad_norm": 0.01073139614027261, "learning_rate": 1.6847004577189656e-06, "logits/chosen": 0.1904296875, "logits/rejected": 3.515625, "logps/chosen": -1408.0, "logps/rejected": -2576.0, "loss": 0.0039, "rewards/accuracies": 1.0, "rewards/chosen": -10.5625, "rewards/margins": 12.0, "rewards/rejected": -22.625, "step": 4270 }, { "epoch": 4.479330193615908, "grad_norm": 0.05597611228727731, "learning_rate": 1.6193630267754555e-06, "logits/chosen": -0.0179443359375, "logits/rejected": 3.359375, "logps/chosen": -1296.0, "logps/rejected": -2480.0, "loss": 0.0047, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -9.8125, "rewards/margins": 12.25, "rewards/rejected": -22.125, "step": 4280 }, { "epoch": 4.489795918367347, "grad_norm": 0.014723482512688289, "learning_rate": 1.5552753466860998e-06, "logits/chosen": 0.6328125, "logits/rejected": 4.40625, "logps/chosen": -1304.0, "logps/rejected": -2448.0, "loss": 0.0024, "rewards/accuracies": 1.0, "rewards/chosen": -10.0, "rewards/margins": 11.8125, "rewards/rejected": -21.875, "step": 4290 }, { "epoch": 4.500261643118786, "grad_norm": 0.21716600651481588, "learning_rate": 1.4924408430903648e-06, "logits/chosen": -0.421875, "logits/rejected": 2.34375, "logps/chosen": -1360.0, "logps/rejected": -2544.0, "loss": 0.0042, "rewards/accuracies": 1.0, "rewards/chosen": -10.3125, "rewards/margins": 11.9375, "rewards/rejected": -22.25, "step": 4300 }, { "epoch": 4.500261643118786, "eval_logits/chosen": 1.8671875, "eval_logits/rejected": 3.796875, "eval_logps/chosen": -1664.0, "eval_logps/rejected": -1968.0, "eval_loss": 1.1511952877044678, "eval_rewards/accuracies": 0.705078125, "eval_rewards/chosen": -13.4375, "eval_rewards/margins": 3.34375, "eval_rewards/rejected": -16.75, "eval_runtime": 52.8769, "eval_samples_per_second": 37.824, "eval_steps_per_second": 0.605, "step": 4300 }, { "epoch": 4.510727367870225, "grad_norm": 0.026913537093199838, "learning_rate": 1.4308628746424496e-06, "logits/chosen": 0.263671875, "logits/rejected": 3.5625, "logps/chosen": -1296.0, "logps/rejected": -2464.0, "loss": 0.0081, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -10.125, "rewards/margins": 11.75, "rewards/rejected": -21.875, "step": 4310 }, { "epoch": 4.521193092621664, "grad_norm": 0.00939530700261345, "learning_rate": 1.3705447328317579e-06, "logits/chosen": 0.154296875, "logits/rejected": 3.953125, "logps/chosen": -1368.0, "logps/rejected": -2496.0, "loss": 0.0089, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -10.375, "rewards/margins": 11.75, "rewards/rejected": -22.125, "step": 4320 }, { "epoch": 4.531658817373103, "grad_norm": 0.0364511094649996, "learning_rate": 1.3114896418069372e-06, "logits/chosen": -0.23046875, "logits/rejected": 3.609375, "logps/chosen": -1360.0, "logps/rejected": -2432.0, "loss": 0.0049, "rewards/accuracies": 1.0, "rewards/chosen": -9.8125, "rewards/margins": 11.5, "rewards/rejected": -21.25, "step": 4330 }, { "epoch": 4.542124542124542, "grad_norm": 0.02113594650850916, "learning_rate": 1.2537007582035631e-06, "logits/chosen": 0.6875, "logits/rejected": 4.40625, "logps/chosen": -1400.0, "logps/rejected": -2560.0, "loss": 0.0091, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -10.6875, "rewards/margins": 12.0625, "rewards/rejected": -22.75, "step": 4340 }, { "epoch": 4.552590266875981, "grad_norm": 0.05586788112085024, "learning_rate": 1.1971811709754089e-06, "logits/chosen": -0.056396484375, "logits/rejected": 4.0625, "logps/chosen": -1304.0, "logps/rejected": -2496.0, "loss": 0.0047, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -10.0625, "rewards/margins": 12.125, "rewards/rejected": -22.25, "step": 4350 }, { "epoch": 4.5630559916274205, "grad_norm": 0.05353120963273352, "learning_rate": 1.1419339012293112e-06, "logits/chosen": 0.1982421875, "logits/rejected": 4.03125, "logps/chosen": -1336.0, "logps/rejected": -2432.0, "loss": 0.0084, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -10.125, "rewards/margins": 11.4375, "rewards/rejected": -21.625, "step": 4360 }, { "epoch": 4.573521716378859, "grad_norm": 0.018632606756551547, "learning_rate": 1.0879619020637167e-06, "logits/chosen": 0.51953125, "logits/rejected": 4.59375, "logps/chosen": -1280.0, "logps/rejected": -2448.0, "loss": 0.007, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -9.8125, "rewards/margins": 11.875, "rewards/rejected": -21.625, "step": 4370 }, { "epoch": 4.583987441130298, "grad_norm": 0.35582959051096275, "learning_rate": 1.0352680584108042e-06, "logits/chosen": 0.3046875, "logits/rejected": 3.921875, "logps/chosen": -1304.0, "logps/rejected": -2448.0, "loss": 0.0024, "rewards/accuracies": 1.0, "rewards/chosen": -10.0625, "rewards/margins": 11.8125, "rewards/rejected": -21.875, "step": 4380 }, { "epoch": 4.594453165881737, "grad_norm": 0.024948404025571833, "learning_rate": 9.838551868822931e-07, "logits/chosen": 0.255859375, "logits/rejected": 4.4375, "logps/chosen": -1344.0, "logps/rejected": -2528.0, "loss": 0.0048, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -10.125, "rewards/margins": 12.4375, "rewards/rejected": -22.625, "step": 4390 }, { "epoch": 4.604918890633177, "grad_norm": 0.015692333357294058, "learning_rate": 9.337260356188899e-07, "logits/chosen": 0.330078125, "logits/rejected": 4.40625, "logps/chosen": -1328.0, "logps/rejected": -2512.0, "loss": 0.0094, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -10.0625, "rewards/margins": 12.4375, "rewards/rejected": -22.5, "step": 4400 }, { "epoch": 4.604918890633177, "eval_logits/chosen": 1.8828125, "eval_logits/rejected": 3.8125, "eval_logps/chosen": -1664.0, "eval_logps/rejected": -1976.0, "eval_loss": 1.157953143119812, "eval_rewards/accuracies": 0.70703125, "eval_rewards/chosen": -13.5, "eval_rewards/margins": 3.359375, "eval_rewards/rejected": -16.875, "eval_runtime": 52.8911, "eval_samples_per_second": 37.814, "eval_steps_per_second": 0.605, "step": 4400 }, { "epoch": 4.615384615384615, "grad_norm": 0.014058491846490884, "learning_rate": 8.848832841433813e-07, "logits/chosen": 0.0361328125, "logits/rejected": 4.09375, "logps/chosen": -1328.0, "logps/rejected": -2464.0, "loss": 0.0061, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -9.9375, "rewards/margins": 12.0, "rewards/rejected": -21.875, "step": 4410 }, { "epoch": 4.625850340136054, "grad_norm": 0.02452867875576879, "learning_rate": 8.373295432174161e-07, "logits/chosen": 0.181640625, "logits/rejected": 3.640625, "logps/chosen": -1264.0, "logps/rejected": -2400.0, "loss": 0.0049, "rewards/accuracies": 1.0, "rewards/chosen": -9.625, "rewards/margins": 11.4375, "rewards/rejected": -21.0, "step": 4420 }, { "epoch": 4.6363160648874935, "grad_norm": 0.014604552527745927, "learning_rate": 7.910673547019576e-07, "logits/chosen": 0.06689453125, "logits/rejected": 3.453125, "logps/chosen": -1352.0, "logps/rejected": -2480.0, "loss": 0.0076, "rewards/accuracies": 1.0, "rewards/chosen": -10.25, "rewards/margins": 11.5625, "rewards/rejected": -21.875, "step": 4430 }, { "epoch": 4.646781789638933, "grad_norm": 0.14235746342352668, "learning_rate": 7.460991914214066e-07, "logits/chosen": 0.4921875, "logits/rejected": 4.09375, "logps/chosen": -1360.0, "logps/rejected": -2544.0, "loss": 0.006, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -10.1875, "rewards/margins": 12.4375, "rewards/rejected": -22.625, "step": 4440 }, { "epoch": 4.657247514390372, "grad_norm": 0.18018234370057737, "learning_rate": 7.024274570314183e-07, "logits/chosen": 0.1396484375, "logits/rejected": 3.828125, "logps/chosen": -1280.0, "logps/rejected": -2528.0, "loss": 0.0106, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -9.75, "rewards/margins": 12.8125, "rewards/rejected": -22.625, "step": 4450 }, { "epoch": 4.66771323914181, "grad_norm": 0.022964972142676853, "learning_rate": 6.600544858904489e-07, "logits/chosen": 0.2392578125, "logits/rejected": 4.625, "logps/chosen": -1328.0, "logps/rejected": -2384.0, "loss": 0.0057, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -9.875, "rewards/margins": 11.1875, "rewards/rejected": -21.125, "step": 4460 }, { "epoch": 4.67817896389325, "grad_norm": 0.028318816060594693, "learning_rate": 6.189825429349399e-07, "logits/chosen": 0.04638671875, "logits/rejected": 3.75, "logps/chosen": -1400.0, "logps/rejected": -2496.0, "loss": 0.0058, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -10.375, "rewards/margins": 11.625, "rewards/rejected": -22.0, "step": 4470 }, { "epoch": 4.688644688644689, "grad_norm": 0.024191997780339095, "learning_rate": 5.792138235582834e-07, "logits/chosen": 0.1025390625, "logits/rejected": 3.328125, "logps/chosen": -1280.0, "logps/rejected": -2448.0, "loss": 0.0051, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -9.875, "rewards/margins": 11.9375, "rewards/rejected": -21.75, "step": 4480 }, { "epoch": 4.699110413396127, "grad_norm": 0.009190935628145188, "learning_rate": 5.407504534934493e-07, "logits/chosen": 0.443359375, "logits/rejected": 3.59375, "logps/chosen": -1352.0, "logps/rejected": -2576.0, "loss": 0.0051, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -10.125, "rewards/margins": 12.375, "rewards/rejected": -22.5, "step": 4490 }, { "epoch": 4.7095761381475665, "grad_norm": 0.016416213341888905, "learning_rate": 5.03594488699377e-07, "logits/chosen": 0.333984375, "logits/rejected": 3.828125, "logps/chosen": -1376.0, "logps/rejected": -2592.0, "loss": 0.006, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -10.4375, "rewards/margins": 12.375, "rewards/rejected": -22.875, "step": 4500 }, { "epoch": 4.7095761381475665, "eval_logits/chosen": 1.921875, "eval_logits/rejected": 3.84375, "eval_logps/chosen": -1672.0, "eval_logps/rejected": -1984.0, "eval_loss": 1.1593124866485596, "eval_rewards/accuracies": 0.705078125, "eval_rewards/chosen": -13.5625, "eval_rewards/margins": 3.375, "eval_rewards/rejected": -17.0, "eval_runtime": 52.8695, "eval_samples_per_second": 37.829, "eval_steps_per_second": 0.605, "step": 4500 }, { "epoch": 4.720041862899006, "grad_norm": 0.01899683797190987, "learning_rate": 4.677479152510794e-07, "logits/chosen": 0.45703125, "logits/rejected": 3.828125, "logps/chosen": -1280.0, "logps/rejected": -2592.0, "loss": 0.0037, "rewards/accuracies": 1.0, "rewards/chosen": -9.875, "rewards/margins": 13.1875, "rewards/rejected": -23.0, "step": 4510 }, { "epoch": 4.730507587650445, "grad_norm": 0.01647864082043874, "learning_rate": 4.3321264923346925e-07, "logits/chosen": -0.22265625, "logits/rejected": 3.90625, "logps/chosen": -1320.0, "logps/rejected": -2544.0, "loss": 0.0026, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -10.0625, "rewards/margins": 12.5625, "rewards/rejected": -22.625, "step": 4520 }, { "epoch": 4.740973312401884, "grad_norm": 0.013445849354382783, "learning_rate": 3.9999053663893893e-07, "logits/chosen": 0.0693359375, "logits/rejected": 4.3125, "logps/chosen": -1384.0, "logps/rejected": -2608.0, "loss": 0.0046, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -10.5, "rewards/margins": 12.75, "rewards/rejected": -23.25, "step": 4530 }, { "epoch": 4.751439037153323, "grad_norm": 0.022597491720675638, "learning_rate": 3.6808335326871944e-07, "logits/chosen": 0.1591796875, "logits/rejected": 4.625, "logps/chosen": -1360.0, "logps/rejected": -2512.0, "loss": 0.0058, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -10.375, "rewards/margins": 12.0, "rewards/rejected": -22.375, "step": 4540 }, { "epoch": 4.761904761904762, "grad_norm": 0.02691200162294267, "learning_rate": 3.374928046379122e-07, "logits/chosen": 0.625, "logits/rejected": 4.78125, "logps/chosen": -1320.0, "logps/rejected": -2400.0, "loss": 0.0072, "rewards/accuracies": 1.0, "rewards/chosen": -10.0, "rewards/margins": 11.5625, "rewards/rejected": -21.625, "step": 4550 }, { "epoch": 4.772370486656201, "grad_norm": 0.06724491422656781, "learning_rate": 3.082205258843618e-07, "logits/chosen": -0.020263671875, "logits/rejected": 3.859375, "logps/chosen": -1336.0, "logps/rejected": -2432.0, "loss": 0.009, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -10.25, "rewards/margins": 11.625, "rewards/rejected": -21.875, "step": 4560 }, { "epoch": 4.7828362114076395, "grad_norm": 0.013820088346344305, "learning_rate": 2.8026808168123973e-07, "logits/chosen": 0.318359375, "logits/rejected": 4.09375, "logps/chosen": -1304.0, "logps/rejected": -2480.0, "loss": 0.0115, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -9.875, "rewards/margins": 12.0625, "rewards/rejected": -21.875, "step": 4570 }, { "epoch": 4.793301936159079, "grad_norm": 0.015977450255060588, "learning_rate": 2.5363696615340883e-07, "logits/chosen": -0.28125, "logits/rejected": 3.765625, "logps/chosen": -1384.0, "logps/rejected": -2512.0, "loss": 0.0076, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -10.4375, "rewards/margins": 11.625, "rewards/rejected": -22.125, "step": 4580 }, { "epoch": 4.803767660910518, "grad_norm": 0.19866355825297102, "learning_rate": 2.2832860279756174e-07, "logits/chosen": 0.4921875, "logits/rejected": 4.34375, "logps/chosen": -1352.0, "logps/rejected": -2528.0, "loss": 0.0055, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -10.3125, "rewards/margins": 12.0625, "rewards/rejected": -22.375, "step": 4590 }, { "epoch": 4.814233385661957, "grad_norm": 0.14021973363064155, "learning_rate": 2.0434434440612381e-07, "logits/chosen": 0.6953125, "logits/rejected": 4.53125, "logps/chosen": -1280.0, "logps/rejected": -2464.0, "loss": 0.0029, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -10.0625, "rewards/margins": 12.0, "rewards/rejected": -22.125, "step": 4600 }, { "epoch": 4.814233385661957, "eval_logits/chosen": 1.9375, "eval_logits/rejected": 3.859375, "eval_logps/chosen": -1680.0, "eval_logps/rejected": -1984.0, "eval_loss": 1.1617499589920044, "eval_rewards/accuracies": 0.705078125, "eval_rewards/chosen": -13.625, "eval_rewards/margins": 3.375, "eval_rewards/rejected": -17.0, "eval_runtime": 52.8851, "eval_samples_per_second": 37.818, "eval_steps_per_second": 0.605, "step": 4600 }, { "epoch": 4.8246991104133965, "grad_norm": 0.013472860117799914, "learning_rate": 1.8168547299496076e-07, "logits/chosen": -0.25, "logits/rejected": 3.8125, "logps/chosen": -1344.0, "logps/rejected": -2448.0, "loss": 0.0029, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -10.0, "rewards/margins": 11.75, "rewards/rejected": -21.75, "step": 4610 }, { "epoch": 4.835164835164835, "grad_norm": 0.022340399767170387, "learning_rate": 1.6035319973483342e-07, "logits/chosen": -0.67578125, "logits/rejected": 3.90625, "logps/chosen": -1352.0, "logps/rejected": -2544.0, "loss": 0.0037, "rewards/accuracies": 1.0, "rewards/chosen": -10.25, "rewards/margins": 12.4375, "rewards/rejected": -22.625, "step": 4620 }, { "epoch": 4.845630559916274, "grad_norm": 0.05257551803550983, "learning_rate": 1.4034866488666364e-07, "logits/chosen": -0.255859375, "logits/rejected": 3.453125, "logps/chosen": -1352.0, "logps/rejected": -2560.0, "loss": 0.0037, "rewards/accuracies": 1.0, "rewards/chosen": -10.0, "rewards/margins": 12.6875, "rewards/rejected": -22.75, "step": 4630 }, { "epoch": 4.856096284667713, "grad_norm": 0.02207698637345573, "learning_rate": 1.2167293774059118e-07, "logits/chosen": 1.2421875, "logits/rejected": 4.65625, "logps/chosen": -1328.0, "logps/rejected": -2544.0, "loss": 0.0041, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -10.3125, "rewards/margins": 12.4375, "rewards/rejected": -22.75, "step": 4640 }, { "epoch": 4.866562009419153, "grad_norm": 0.02412749963928652, "learning_rate": 1.0432701655880838e-07, "logits/chosen": 0.5859375, "logits/rejected": 3.5625, "logps/chosen": -1352.0, "logps/rejected": -2560.0, "loss": 0.009, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -10.0625, "rewards/margins": 12.1875, "rewards/rejected": -22.25, "step": 4650 }, { "epoch": 4.877027734170591, "grad_norm": 0.025434985246309914, "learning_rate": 8.831182852220565e-08, "logits/chosen": -0.8046875, "logits/rejected": 3.25, "logps/chosen": -1384.0, "logps/rejected": -2544.0, "loss": 0.0036, "rewards/accuracies": 1.0, "rewards/chosen": -10.25, "rewards/margins": 12.25, "rewards/rejected": -22.5, "step": 4660 }, { "epoch": 4.88749345892203, "grad_norm": 0.016137357714564624, "learning_rate": 7.362822968081107e-08, "logits/chosen": 0.7109375, "logits/rejected": 4.15625, "logps/chosen": -1264.0, "logps/rejected": -2432.0, "loss": 0.0075, "rewards/accuracies": 1.0, "rewards/chosen": -9.75, "rewards/margins": 12.0, "rewards/rejected": -21.75, "step": 4670 }, { "epoch": 4.8979591836734695, "grad_norm": 0.015816330428665797, "learning_rate": 6.02770049080298e-08, "logits/chosen": 0.5859375, "logits/rejected": 4.125, "logps/chosen": -1344.0, "logps/rejected": -2528.0, "loss": 0.0081, "rewards/accuracies": 1.0, "rewards/chosen": -10.1875, "rewards/margins": 12.125, "rewards/rejected": -22.25, "step": 4680 }, { "epoch": 4.908424908424909, "grad_norm": 0.01587861482983314, "learning_rate": 4.8258867858697044e-08, "logits/chosen": 0.58203125, "logits/rejected": 3.21875, "logps/chosen": -1272.0, "logps/rejected": -2528.0, "loss": 0.0027, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -10.0625, "rewards/margins": 12.4375, "rewards/rejected": -22.5, "step": 4690 }, { "epoch": 4.918890633176347, "grad_norm": 0.01701742299215144, "learning_rate": 3.757446093092254e-08, "logits/chosen": 0.93359375, "logits/rejected": 4.4375, "logps/chosen": -1432.0, "logps/rejected": -2576.0, "loss": 0.0059, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -10.9375, "rewards/margins": 11.875, "rewards/rejected": -22.875, "step": 4700 }, { "epoch": 4.918890633176347, "eval_logits/chosen": 1.9453125, "eval_logits/rejected": 3.859375, "eval_logps/chosen": -1680.0, "eval_logps/rejected": -1984.0, "eval_loss": 1.1636874675750732, "eval_rewards/accuracies": 0.705078125, "eval_rewards/chosen": -13.625, "eval_rewards/margins": 3.375, "eval_rewards/rejected": -17.0, "eval_runtime": 52.896, "eval_samples_per_second": 37.81, "eval_steps_per_second": 0.605, "step": 4700 }, { "epoch": 4.929356357927786, "grad_norm": 0.009136301419989667, "learning_rate": 2.8224355231762368e-08, "logits/chosen": 0.54296875, "logits/rejected": 4.34375, "logps/chosen": -1320.0, "logps/rejected": -2544.0, "loss": 0.006, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -10.0625, "rewards/margins": 12.5, "rewards/rejected": -22.625, "step": 4710 }, { "epoch": 4.939822082679226, "grad_norm": 0.016109187019904162, "learning_rate": 2.020905054668232e-08, "logits/chosen": 0.478515625, "logits/rejected": 3.921875, "logps/chosen": -1360.0, "logps/rejected": -2544.0, "loss": 0.0079, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -10.125, "rewards/margins": 12.1875, "rewards/rejected": -22.375, "step": 4720 }, { "epoch": 4.950287807430665, "grad_norm": 0.08635472918390723, "learning_rate": 1.3528975312848691e-08, "logits/chosen": 0.72265625, "logits/rejected": 3.78125, "logps/chosen": -1360.0, "logps/rejected": -2464.0, "loss": 0.0028, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -10.3125, "rewards/margins": 11.375, "rewards/rejected": -21.625, "step": 4730 }, { "epoch": 4.960753532182103, "grad_norm": 0.13889460429320133, "learning_rate": 8.18448659622717e-09, "logits/chosen": -0.216796875, "logits/rejected": 3.5, "logps/chosen": -1360.0, "logps/rejected": -2528.0, "loss": 0.0038, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -10.5, "rewards/margins": 11.9375, "rewards/rejected": -22.375, "step": 4740 }, { "epoch": 4.9712192569335425, "grad_norm": 0.054086673578341196, "learning_rate": 4.175870072500865e-09, "logits/chosen": -0.1767578125, "logits/rejected": 3.125, "logps/chosen": -1336.0, "logps/rejected": -2400.0, "loss": 0.0048, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -10.0, "rewards/margins": 11.1875, "rewards/rejected": -21.25, "step": 4750 }, { "epoch": 4.981684981684982, "grad_norm": 0.021898955068035898, "learning_rate": 1.5033400117853146e-09, "logits/chosen": 0.3046875, "logits/rejected": 3.90625, "logps/chosen": -1296.0, "logps/rejected": -2416.0, "loss": 0.0035, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -9.875, "rewards/margins": 11.5625, "rewards/rejected": -21.5, "step": 4760 }, { "epoch": 4.992150706436421, "grad_norm": 0.02157469228444244, "learning_rate": 1.670392671959675e-10, "logits/chosen": 0.64453125, "logits/rejected": 4.34375, "logps/chosen": -1336.0, "logps/rejected": -2512.0, "loss": 0.0064, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -10.1875, "rewards/margins": 12.125, "rewards/rejected": -22.25, "step": 4770 }, { "epoch": 4.99738356881214, "step": 4775, "total_flos": 0.0, "train_loss": 0.1729966444779597, "train_runtime": 22065.583, "train_samples_per_second": 13.853, "train_steps_per_second": 0.216 } ], "logging_steps": 10, "max_steps": 4775, "num_input_tokens_seen": 0, "num_train_epochs": 5, "save_steps": 500, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 0.0, "train_batch_size": 8, "trial_name": null, "trial_params": null }