{ "best_metric": null, "best_model_checkpoint": null, "epoch": 2.998430141287284, "eval_steps": 100, "global_step": 2865, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0010465724751439038, "grad_norm": 1.480058074617895, "learning_rate": 1.7421602787456448e-07, "logits/chosen": -9.875, "logits/rejected": -9.9375, "logps/chosen": -448.0, "logps/rejected": -396.0, "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": 1.2643614412528625, "learning_rate": 1.7421602787456445e-06, "logits/chosen": -9.75, "logits/rejected": -9.875, "logps/chosen": -298.0, "logps/rejected": -264.0, "loss": 0.6921, "rewards/accuracies": 0.2222222238779068, "rewards/chosen": 0.000370025634765625, "rewards/margins": 0.00061798095703125, "rewards/rejected": -0.0002498626708984375, "step": 10 }, { "epoch": 0.020931449502878074, "grad_norm": 1.4275100659623685, "learning_rate": 3.484320557491289e-06, "logits/chosen": -9.6875, "logits/rejected": -9.5625, "logps/chosen": -324.0, "logps/rejected": -310.0, "loss": 0.6917, "rewards/accuracies": 0.30000001192092896, "rewards/chosen": 5.841255187988281e-05, "rewards/margins": 0.000331878662109375, "rewards/rejected": -0.000270843505859375, "step": 20 }, { "epoch": 0.03139717425431711, "grad_norm": 1.4780024754233048, "learning_rate": 5.226480836236934e-06, "logits/chosen": -10.0625, "logits/rejected": -9.875, "logps/chosen": -316.0, "logps/rejected": -310.0, "loss": 0.6882, "rewards/accuracies": 0.4937500059604645, "rewards/chosen": 0.00347900390625, "rewards/margins": 0.00762939453125, "rewards/rejected": -0.004150390625, "step": 30 }, { "epoch": 0.04186289900575615, "grad_norm": 1.4432628300363817, "learning_rate": 6.968641114982578e-06, "logits/chosen": -10.125, "logits/rejected": -10.25, "logps/chosen": -338.0, "logps/rejected": -314.0, "loss": 0.68, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -0.04736328125, "rewards/margins": 0.0238037109375, "rewards/rejected": -0.0712890625, "step": 40 }, { "epoch": 0.052328623757195186, "grad_norm": 1.4925455989723233, "learning_rate": 8.710801393728225e-06, "logits/chosen": -10.25, "logits/rejected": -10.0625, "logps/chosen": -328.0, "logps/rejected": -286.0, "loss": 0.6671, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -0.1630859375, "rewards/margins": 0.06396484375, "rewards/rejected": -0.2265625, "step": 50 }, { "epoch": 0.06279434850863422, "grad_norm": 1.914942939701622, "learning_rate": 1.0452961672473868e-05, "logits/chosen": -10.0625, "logits/rejected": -10.0, "logps/chosen": -396.0, "logps/rejected": -364.0, "loss": 0.6527, "rewards/accuracies": 0.6000000238418579, "rewards/chosen": -0.447265625, "rewards/margins": 0.0966796875, "rewards/rejected": -0.54296875, "step": 60 }, { "epoch": 0.07326007326007326, "grad_norm": 2.5810244776955176, "learning_rate": 1.2195121951219513e-05, "logits/chosen": -9.125, "logits/rejected": -8.6875, "logps/chosen": -356.0, "logps/rejected": -364.0, "loss": 0.6348, "rewards/accuracies": 0.6812499761581421, "rewards/chosen": -0.57421875, "rewards/margins": 0.224609375, "rewards/rejected": -0.796875, "step": 70 }, { "epoch": 0.0837257980115123, "grad_norm": 2.200873590925822, "learning_rate": 1.3937282229965156e-05, "logits/chosen": -7.5, "logits/rejected": -7.21875, "logps/chosen": -388.0, "logps/rejected": -362.0, "loss": 0.6375, "rewards/accuracies": 0.731249988079071, "rewards/chosen": -0.61328125, "rewards/margins": 0.21875, "rewards/rejected": -0.83203125, "step": 80 }, { "epoch": 0.09419152276295134, "grad_norm": 1.7206863125159313, "learning_rate": 1.56794425087108e-05, "logits/chosen": -8.0625, "logits/rejected": -8.0, "logps/chosen": -374.0, "logps/rejected": -392.0, "loss": 0.6222, "rewards/accuracies": 0.5375000238418579, "rewards/chosen": -0.87109375, "rewards/margins": 0.11962890625, "rewards/rejected": -0.9921875, "step": 90 }, { "epoch": 0.10465724751439037, "grad_norm": 2.088252978650692, "learning_rate": 1.742160278745645e-05, "logits/chosen": -8.75, "logits/rejected": -8.5, "logps/chosen": -458.0, "logps/rejected": -452.0, "loss": 0.6007, "rewards/accuracies": 0.675000011920929, "rewards/chosen": -1.09375, "rewards/margins": 0.34375, "rewards/rejected": -1.4375, "step": 100 }, { "epoch": 0.10465724751439037, "eval_logits/chosen": -8.8125, "eval_logits/rejected": -8.5, "eval_logps/chosen": -438.0, "eval_logps/rejected": -444.0, "eval_loss": 0.6139843463897705, "eval_rewards/accuracies": 0.65625, "eval_rewards/chosen": -1.234375, "eval_rewards/margins": 0.341796875, "eval_rewards/rejected": -1.578125, "eval_runtime": 47.6054, "eval_samples_per_second": 42.012, "eval_steps_per_second": 0.672, "step": 100 }, { "epoch": 0.1151229722658294, "grad_norm": 1.892401324213614, "learning_rate": 1.9163763066202093e-05, "logits/chosen": -8.875, "logits/rejected": -8.75, "logps/chosen": -388.0, "logps/rejected": -402.0, "loss": 0.6178, "rewards/accuracies": 0.637499988079071, "rewards/chosen": -1.0625, "rewards/margins": 0.279296875, "rewards/rejected": -1.34375, "step": 110 }, { "epoch": 0.12558869701726844, "grad_norm": 2.3486436229150693, "learning_rate": 2.0905923344947736e-05, "logits/chosen": -7.5, "logits/rejected": -6.90625, "logps/chosen": -446.0, "logps/rejected": -494.0, "loss": 0.5926, "rewards/accuracies": 0.6875, "rewards/chosen": -1.375, "rewards/margins": 0.376953125, "rewards/rejected": -1.7578125, "step": 120 }, { "epoch": 0.1360544217687075, "grad_norm": 2.9223922944285823, "learning_rate": 2.264808362369338e-05, "logits/chosen": -5.15625, "logits/rejected": -4.375, "logps/chosen": -504.0, "logps/rejected": -504.0, "loss": 0.6106, "rewards/accuracies": 0.675000011920929, "rewards/chosen": -1.7421875, "rewards/margins": 0.447265625, "rewards/rejected": -2.1875, "step": 130 }, { "epoch": 0.14652014652014653, "grad_norm": 2.3103686596900292, "learning_rate": 2.4390243902439026e-05, "logits/chosen": -5.15625, "logits/rejected": -4.75, "logps/chosen": -480.0, "logps/rejected": -496.0, "loss": 0.5821, "rewards/accuracies": 0.668749988079071, "rewards/chosen": -1.453125, "rewards/margins": 0.408203125, "rewards/rejected": -1.859375, "step": 140 }, { "epoch": 0.15698587127158556, "grad_norm": 2.122179532642416, "learning_rate": 2.6132404181184672e-05, "logits/chosen": -6.90625, "logits/rejected": -6.15625, "logps/chosen": -492.0, "logps/rejected": -486.0, "loss": 0.6155, "rewards/accuracies": 0.6812499761581421, "rewards/chosen": -1.5859375, "rewards/margins": 0.46875, "rewards/rejected": -2.046875, "step": 150 }, { "epoch": 0.1674515960230246, "grad_norm": 2.670737384190233, "learning_rate": 2.7874564459930312e-05, "logits/chosen": -8.625, "logits/rejected": -8.0, "logps/chosen": -466.0, "logps/rejected": -442.0, "loss": 0.5891, "rewards/accuracies": 0.6625000238418579, "rewards/chosen": -1.171875, "rewards/margins": 0.431640625, "rewards/rejected": -1.609375, "step": 160 }, { "epoch": 0.17791732077446362, "grad_norm": 1.8820739895930487, "learning_rate": 2.9616724738675962e-05, "logits/chosen": -7.875, "logits/rejected": -7.28125, "logps/chosen": -512.0, "logps/rejected": -524.0, "loss": 0.5927, "rewards/accuracies": 0.7250000238418579, "rewards/chosen": -1.9609375, "rewards/margins": 0.5859375, "rewards/rejected": -2.546875, "step": 170 }, { "epoch": 0.18838304552590268, "grad_norm": 2.1940336238411535, "learning_rate": 3.13588850174216e-05, "logits/chosen": -7.375, "logits/rejected": -6.9375, "logps/chosen": -452.0, "logps/rejected": -492.0, "loss": 0.5865, "rewards/accuracies": 0.65625, "rewards/chosen": -1.6875, "rewards/margins": 0.50390625, "rewards/rejected": -2.203125, "step": 180 }, { "epoch": 0.1988487702773417, "grad_norm": 2.3402818296231365, "learning_rate": 3.310104529616725e-05, "logits/chosen": -7.96875, "logits/rejected": -7.65625, "logps/chosen": -450.0, "logps/rejected": -492.0, "loss": 0.5811, "rewards/accuracies": 0.6875, "rewards/chosen": -1.75, "rewards/margins": 0.423828125, "rewards/rejected": -2.171875, "step": 190 }, { "epoch": 0.20931449502878074, "grad_norm": 2.6464496876672015, "learning_rate": 3.48432055749129e-05, "logits/chosen": -7.375, "logits/rejected": -6.625, "logps/chosen": -588.0, "logps/rejected": -564.0, "loss": 0.591, "rewards/accuracies": 0.65625, "rewards/chosen": -2.28125, "rewards/margins": 0.51953125, "rewards/rejected": -2.796875, "step": 200 }, { "epoch": 0.20931449502878074, "eval_logits/chosen": -7.53125, "eval_logits/rejected": -6.9375, "eval_logps/chosen": -508.0, "eval_logps/rejected": -532.0, "eval_loss": 0.6025000214576721, "eval_rewards/accuracies": 0.689453125, "eval_rewards/chosen": -1.9296875, "eval_rewards/margins": 0.53125, "eval_rewards/rejected": -2.46875, "eval_runtime": 47.0509, "eval_samples_per_second": 42.507, "eval_steps_per_second": 0.68, "step": 200 }, { "epoch": 0.21978021978021978, "grad_norm": 1.991255294820422, "learning_rate": 3.6585365853658535e-05, "logits/chosen": -7.75, "logits/rejected": -7.34375, "logps/chosen": -504.0, "logps/rejected": -510.0, "loss": 0.63, "rewards/accuracies": 0.637499988079071, "rewards/chosen": -1.75, "rewards/margins": 0.26171875, "rewards/rejected": -2.015625, "step": 210 }, { "epoch": 0.2302459445316588, "grad_norm": 2.2351884453500563, "learning_rate": 3.8327526132404185e-05, "logits/chosen": -8.375, "logits/rejected": -7.875, "logps/chosen": -470.0, "logps/rejected": -428.0, "loss": 0.5959, "rewards/accuracies": 0.65625, "rewards/chosen": -1.375, "rewards/margins": 0.359375, "rewards/rejected": -1.7265625, "step": 220 }, { "epoch": 0.24071166928309787, "grad_norm": 2.1006070721837147, "learning_rate": 4.006968641114983e-05, "logits/chosen": -6.875, "logits/rejected": -6.28125, "logps/chosen": -532.0, "logps/rejected": -556.0, "loss": 0.6031, "rewards/accuracies": 0.6937500238418579, "rewards/chosen": -2.015625, "rewards/margins": 0.53515625, "rewards/rejected": -2.546875, "step": 230 }, { "epoch": 0.25117739403453687, "grad_norm": 2.1768415030840447, "learning_rate": 4.181184668989547e-05, "logits/chosen": -8.1875, "logits/rejected": -7.78125, "logps/chosen": -462.0, "logps/rejected": -488.0, "loss": 0.5782, "rewards/accuracies": 0.7124999761581421, "rewards/chosen": -1.5234375, "rewards/margins": 0.5, "rewards/rejected": -2.015625, "step": 240 }, { "epoch": 0.2616431187859759, "grad_norm": 2.3007182472021617, "learning_rate": 4.3554006968641115e-05, "logits/chosen": -6.5625, "logits/rejected": -6.25, "logps/chosen": -588.0, "logps/rejected": -604.0, "loss": 0.5688, "rewards/accuracies": 0.6812499761581421, "rewards/chosen": -2.671875, "rewards/margins": 0.51171875, "rewards/rejected": -3.1875, "step": 250 }, { "epoch": 0.272108843537415, "grad_norm": 2.584978021792629, "learning_rate": 4.529616724738676e-05, "logits/chosen": -8.25, "logits/rejected": -7.75, "logps/chosen": -596.0, "logps/rejected": -612.0, "loss": 0.606, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -2.640625, "rewards/margins": 0.365234375, "rewards/rejected": -3.0, "step": 260 }, { "epoch": 0.282574568288854, "grad_norm": 2.0452124258505244, "learning_rate": 4.703832752613241e-05, "logits/chosen": -8.625, "logits/rejected": -8.25, "logps/chosen": -540.0, "logps/rejected": -516.0, "loss": 0.5927, "rewards/accuracies": 0.675000011920929, "rewards/chosen": -1.890625, "rewards/margins": 0.43359375, "rewards/rejected": -2.328125, "step": 270 }, { "epoch": 0.29304029304029305, "grad_norm": 1.6800532211010257, "learning_rate": 4.878048780487805e-05, "logits/chosen": -9.875, "logits/rejected": -9.125, "logps/chosen": -506.0, "logps/rejected": -528.0, "loss": 0.6002, "rewards/accuracies": 0.699999988079071, "rewards/chosen": -1.6171875, "rewards/margins": 0.484375, "rewards/rejected": -2.109375, "step": 280 }, { "epoch": 0.3035060177917321, "grad_norm": 2.3966464432317496, "learning_rate": 4.999983293481418e-05, "logits/chosen": -10.6875, "logits/rejected": -10.1875, "logps/chosen": -466.0, "logps/rejected": -444.0, "loss": 0.6122, "rewards/accuracies": 0.6937500238418579, "rewards/chosen": -1.4765625, "rewards/margins": 0.322265625, "rewards/rejected": -1.796875, "step": 290 }, { "epoch": 0.3139717425431711, "grad_norm": 1.9703121790532745, "learning_rate": 4.9996862949181595e-05, "logits/chosen": -6.5625, "logits/rejected": -6.1875, "logps/chosen": -568.0, "logps/rejected": -588.0, "loss": 0.6351, "rewards/accuracies": 0.6312500238418579, "rewards/chosen": -2.546875, "rewards/margins": 0.3828125, "rewards/rejected": -2.921875, "step": 300 }, { "epoch": 0.3139717425431711, "eval_logits/chosen": -5.78125, "eval_logits/rejected": -4.90625, "eval_logps/chosen": -540.0, "eval_logps/rejected": -556.0, "eval_loss": 0.5961874723434448, "eval_rewards/accuracies": 0.6875, "eval_rewards/chosen": -2.234375, "eval_rewards/margins": 0.451171875, "eval_rewards/rejected": -2.6875, "eval_runtime": 47.0649, "eval_samples_per_second": 42.495, "eval_steps_per_second": 0.68, "step": 300 }, { "epoch": 0.32443746729461015, "grad_norm": 2.240826262421424, "learning_rate": 4.9990180911526896e-05, "logits/chosen": -7.40625, "logits/rejected": -5.34375, "logps/chosen": -536.0, "logps/rejected": -502.0, "loss": 0.6127, "rewards/accuracies": 0.699999988079071, "rewards/chosen": -2.0, "rewards/margins": 0.515625, "rewards/rejected": -2.515625, "step": 310 }, { "epoch": 0.3349031920460492, "grad_norm": 1.5561698423103294, "learning_rate": 4.997978781413745e-05, "logits/chosen": -3.25, "logits/rejected": -2.453125, "logps/chosen": -496.0, "logps/rejected": -536.0, "loss": 0.6019, "rewards/accuracies": 0.625, "rewards/chosen": -2.0625, "rewards/margins": 0.52734375, "rewards/rejected": -2.59375, "step": 320 }, { "epoch": 0.3453689167974882, "grad_norm": 1.9945042952302237, "learning_rate": 4.996568520039565e-05, "logits/chosen": -2.046875, "logits/rejected": -1.0546875, "logps/chosen": -506.0, "logps/rejected": -520.0, "loss": 0.5868, "rewards/accuracies": 0.7124999761581421, "rewards/chosen": -2.09375, "rewards/margins": 0.51171875, "rewards/rejected": -2.59375, "step": 330 }, { "epoch": 0.35583464154892724, "grad_norm": 1.6621551720054157, "learning_rate": 4.994787516454973e-05, "logits/chosen": -4.875, "logits/rejected": -3.671875, "logps/chosen": -510.0, "logps/rejected": -506.0, "loss": 0.612, "rewards/accuracies": 0.6187499761581421, "rewards/chosen": -2.0, "rewards/margins": 0.404296875, "rewards/rejected": -2.40625, "step": 340 }, { "epoch": 0.3663003663003663, "grad_norm": 2.321476824534754, "learning_rate": 4.992636035140273e-05, "logits/chosen": -4.34375, "logits/rejected": -3.21875, "logps/chosen": -528.0, "logps/rejected": -516.0, "loss": 0.5767, "rewards/accuracies": 0.643750011920929, "rewards/chosen": -1.875, "rewards/margins": 0.453125, "rewards/rejected": -2.328125, "step": 350 }, { "epoch": 0.37676609105180536, "grad_norm": 1.9232527481051325, "learning_rate": 4.99011439559198e-05, "logits/chosen": -2.953125, "logits/rejected": -1.9375, "logps/chosen": -556.0, "logps/rejected": -584.0, "loss": 0.5874, "rewards/accuracies": 0.7250000238418579, "rewards/chosen": -2.390625, "rewards/margins": 0.59765625, "rewards/rejected": -3.0, "step": 360 }, { "epoch": 0.3872318158032444, "grad_norm": 2.059071175528353, "learning_rate": 4.987222972275369e-05, "logits/chosen": -1.7421875, "logits/rejected": -0.85546875, "logps/chosen": -528.0, "logps/rejected": -532.0, "loss": 0.5819, "rewards/accuracies": 0.6312500238418579, "rewards/chosen": -2.21875, "rewards/margins": 0.404296875, "rewards/rejected": -2.625, "step": 370 }, { "epoch": 0.3976975405546834, "grad_norm": 2.489905658180968, "learning_rate": 4.983962194568872e-05, "logits/chosen": -1.6328125, "logits/rejected": -0.74609375, "logps/chosen": -520.0, "logps/rejected": -552.0, "loss": 0.5955, "rewards/accuracies": 0.6187499761581421, "rewards/chosen": -2.234375, "rewards/margins": 0.45703125, "rewards/rejected": -2.6875, "step": 380 }, { "epoch": 0.40816326530612246, "grad_norm": 1.9522474511777042, "learning_rate": 4.9803325467003084e-05, "logits/chosen": -3.578125, "logits/rejected": -2.828125, "logps/chosen": -544.0, "logps/rejected": -560.0, "loss": 0.5965, "rewards/accuracies": 0.6499999761581421, "rewards/chosen": -2.1875, "rewards/margins": 0.40234375, "rewards/rejected": -2.59375, "step": 390 }, { "epoch": 0.4186289900575615, "grad_norm": 2.1290793667857115, "learning_rate": 4.9763345676749816e-05, "logits/chosen": -5.4375, "logits/rejected": -4.6875, "logps/chosen": -528.0, "logps/rejected": -548.0, "loss": 0.6031, "rewards/accuracies": 0.675000011920929, "rewards/chosen": -2.125, "rewards/margins": 0.416015625, "rewards/rejected": -2.546875, "step": 400 }, { "epoch": 0.4186289900575615, "eval_logits/chosen": -7.21875, "eval_logits/rejected": -6.25, "eval_logps/chosen": -486.0, "eval_logps/rejected": -512.0, "eval_loss": 0.5899843573570251, "eval_rewards/accuracies": 0.6875, "eval_rewards/chosen": -1.7109375, "eval_rewards/margins": 0.5625, "eval_rewards/rejected": -2.28125, "eval_runtime": 47.063, "eval_samples_per_second": 42.496, "eval_steps_per_second": 0.68, "step": 400 }, { "epoch": 0.4290947148090005, "grad_norm": 1.7378739628405557, "learning_rate": 4.9719688511956374e-05, "logits/chosen": -5.78125, "logits/rejected": -5.0, "logps/chosen": -556.0, "logps/rejected": -560.0, "loss": 0.6151, "rewards/accuracies": 0.6187499761581421, "rewards/chosen": -2.359375, "rewards/margins": 0.474609375, "rewards/rejected": -2.84375, "step": 410 }, { "epoch": 0.43956043956043955, "grad_norm": 1.587769886345485, "learning_rate": 4.967236045574294e-05, "logits/chosen": -4.9375, "logits/rejected": -3.390625, "logps/chosen": -584.0, "logps/rejected": -576.0, "loss": 0.6225, "rewards/accuracies": 0.699999988079071, "rewards/chosen": -2.53125, "rewards/margins": 0.40234375, "rewards/rejected": -2.921875, "step": 420 }, { "epoch": 0.4500261643118786, "grad_norm": 1.4569652452770199, "learning_rate": 4.96213685363597e-05, "logits/chosen": -4.03125, "logits/rejected": -3.109375, "logps/chosen": -504.0, "logps/rejected": -532.0, "loss": 0.5956, "rewards/accuracies": 0.675000011920929, "rewards/chosen": -2.046875, "rewards/margins": 0.4296875, "rewards/rejected": -2.484375, "step": 430 }, { "epoch": 0.4604918890633176, "grad_norm": 2.2997074115701093, "learning_rate": 4.956672032614315e-05, "logits/chosen": -4.4375, "logits/rejected": -3.28125, "logps/chosen": -556.0, "logps/rejected": -536.0, "loss": 0.5836, "rewards/accuracies": 0.706250011920929, "rewards/chosen": -2.1875, "rewards/margins": 0.4609375, "rewards/rejected": -2.640625, "step": 440 }, { "epoch": 0.47095761381475665, "grad_norm": 1.978188324469207, "learning_rate": 4.950842394039157e-05, "logits/chosen": -4.09375, "logits/rejected": -2.875, "logps/chosen": -584.0, "logps/rejected": -632.0, "loss": 0.585, "rewards/accuracies": 0.737500011920929, "rewards/chosen": -2.5, "rewards/margins": 0.72265625, "rewards/rejected": -3.21875, "step": 450 }, { "epoch": 0.48142333856619574, "grad_norm": 1.8470214848974822, "learning_rate": 4.944648803615994e-05, "logits/chosen": -2.21875, "logits/rejected": -1.2109375, "logps/chosen": -544.0, "logps/rejected": -564.0, "loss": 0.5893, "rewards/accuracies": 0.6812499761581421, "rewards/chosen": -2.4375, "rewards/margins": 0.490234375, "rewards/rejected": -2.921875, "step": 460 }, { "epoch": 0.49188906331763477, "grad_norm": 1.8216567110345698, "learning_rate": 4.938092181097433e-05, "logits/chosen": -2.859375, "logits/rejected": -1.8046875, "logps/chosen": -540.0, "logps/rejected": -548.0, "loss": 0.533, "rewards/accuracies": 0.699999988079071, "rewards/chosen": -2.15625, "rewards/margins": 0.6015625, "rewards/rejected": -2.75, "step": 470 }, { "epoch": 0.5023547880690737, "grad_norm": 2.1466484011877314, "learning_rate": 4.931173500146607e-05, "logits/chosen": -1.9375, "logits/rejected": -0.58984375, "logps/chosen": -556.0, "logps/rejected": -604.0, "loss": 0.5632, "rewards/accuracies": 0.675000011920929, "rewards/chosen": -2.453125, "rewards/margins": 0.7734375, "rewards/rejected": -3.234375, "step": 480 }, { "epoch": 0.5128205128205128, "grad_norm": 1.6022137986582567, "learning_rate": 4.923893788192582e-05, "logits/chosen": -2.546875, "logits/rejected": 0.14453125, "logps/chosen": -652.0, "logps/rejected": -660.0, "loss": 0.5657, "rewards/accuracies": 0.7562500238418579, "rewards/chosen": -2.859375, "rewards/margins": 0.91796875, "rewards/rejected": -3.765625, "step": 490 }, { "epoch": 0.5232862375719518, "grad_norm": 1.7861878895737113, "learning_rate": 4.9162541262777906e-05, "logits/chosen": -3.96875, "logits/rejected": -2.75, "logps/chosen": -544.0, "logps/rejected": -552.0, "loss": 0.5813, "rewards/accuracies": 0.6875, "rewards/chosen": -2.40625, "rewards/margins": 0.5546875, "rewards/rejected": -2.96875, "step": 500 }, { "epoch": 0.5232862375719518, "eval_logits/chosen": -4.8125, "eval_logits/rejected": -3.640625, "eval_logps/chosen": -540.0, "eval_logps/rejected": -568.0, "eval_loss": 0.5823593735694885, "eval_rewards/accuracies": 0.705078125, "eval_rewards/chosen": -2.25, "eval_rewards/margins": 0.5546875, "eval_rewards/rejected": -2.8125, "eval_runtime": 47.0756, "eval_samples_per_second": 42.485, "eval_steps_per_second": 0.68, "step": 500 }, { "epoch": 0.533751962323391, "grad_norm": 1.3835282652172172, "learning_rate": 4.90825564889749e-05, "logits/chosen": -5.28125, "logits/rejected": -4.25, "logps/chosen": -552.0, "logps/rejected": -580.0, "loss": 0.5798, "rewards/accuracies": 0.6812499761581421, "rewards/chosen": -2.203125, "rewards/margins": 0.6484375, "rewards/rejected": -2.859375, "step": 510 }, { "epoch": 0.54421768707483, "grad_norm": 1.7424721615714878, "learning_rate": 4.899899543831288e-05, "logits/chosen": 0.84765625, "logits/rejected": 1.6328125, "logps/chosen": -640.0, "logps/rejected": -656.0, "loss": 0.5978, "rewards/accuracies": 0.6499999761581421, "rewards/chosen": -3.21875, "rewards/margins": 0.494140625, "rewards/rejected": -3.71875, "step": 520 }, { "epoch": 0.554683411826269, "grad_norm": 1.897646663019382, "learning_rate": 4.891187051966765e-05, "logits/chosen": 1.8359375, "logits/rejected": 2.828125, "logps/chosen": -600.0, "logps/rejected": -632.0, "loss": 0.5815, "rewards/accuracies": 0.6625000238418579, "rewards/chosen": -2.828125, "rewards/margins": 0.63671875, "rewards/rejected": -3.46875, "step": 530 }, { "epoch": 0.565149136577708, "grad_norm": 1.8163340570207902, "learning_rate": 4.882119467115191e-05, "logits/chosen": -0.33203125, "logits/rejected": 2.078125, "logps/chosen": -536.0, "logps/rejected": -532.0, "loss": 0.5944, "rewards/accuracies": 0.6625000238418579, "rewards/chosen": -1.9296875, "rewards/margins": 0.478515625, "rewards/rejected": -2.40625, "step": 540 }, { "epoch": 0.5756148613291471, "grad_norm": 1.7783971838691621, "learning_rate": 4.872698135819402e-05, "logits/chosen": -2.875, "logits/rejected": -0.353515625, "logps/chosen": -520.0, "logps/rejected": -520.0, "loss": 0.5964, "rewards/accuracies": 0.6625000238418579, "rewards/chosen": -2.09375, "rewards/margins": 0.61328125, "rewards/rejected": -2.703125, "step": 550 }, { "epoch": 0.5860805860805861, "grad_norm": 1.7219402241292967, "learning_rate": 4.862924457153831e-05, "logits/chosen": -6.125, "logits/rejected": -4.4375, "logps/chosen": -520.0, "logps/rejected": -540.0, "loss": 0.5926, "rewards/accuracies": 0.6937500238418579, "rewards/chosen": -1.9375, "rewards/margins": 0.59375, "rewards/rejected": -2.53125, "step": 560 }, { "epoch": 0.5965463108320251, "grad_norm": 1.5284344742307574, "learning_rate": 4.8527998825167484e-05, "logits/chosen": -6.40625, "logits/rejected": -4.84375, "logps/chosen": -540.0, "logps/rejected": -552.0, "loss": 0.5875, "rewards/accuracies": 0.637499988079071, "rewards/chosen": -2.140625, "rewards/margins": 0.4296875, "rewards/rejected": -2.578125, "step": 570 }, { "epoch": 0.6070120355834642, "grad_norm": 2.0315339095391844, "learning_rate": 4.842325915414728e-05, "logits/chosen": -2.65625, "logits/rejected": -1.2421875, "logps/chosen": -560.0, "logps/rejected": -568.0, "loss": 0.5696, "rewards/accuracies": 0.71875, "rewards/chosen": -2.640625, "rewards/margins": 0.6953125, "rewards/rejected": -3.34375, "step": 580 }, { "epoch": 0.6174777603349032, "grad_norm": 1.6868968427276214, "learning_rate": 4.831504111239373e-05, "logits/chosen": -4.125, "logits/rejected": -2.421875, "logps/chosen": -572.0, "logps/rejected": -604.0, "loss": 0.5605, "rewards/accuracies": 0.7124999761581421, "rewards/chosen": -2.46875, "rewards/margins": 0.6171875, "rewards/rejected": -3.09375, "step": 590 }, { "epoch": 0.6279434850863422, "grad_norm": 1.725931044564495, "learning_rate": 4.8203360770363425e-05, "logits/chosen": 0.0859375, "logits/rejected": 1.3984375, "logps/chosen": -584.0, "logps/rejected": -624.0, "loss": 0.5376, "rewards/accuracies": 0.71875, "rewards/chosen": -2.75, "rewards/margins": 0.6796875, "rewards/rejected": -3.421875, "step": 600 }, { "epoch": 0.6279434850863422, "eval_logits/chosen": 0.92578125, "eval_logits/rejected": 2.421875, "eval_logps/chosen": -576.0, "eval_logps/rejected": -620.0, "eval_loss": 0.5624374747276306, "eval_rewards/accuracies": 0.701171875, "eval_rewards/chosen": -2.625, "eval_rewards/margins": 0.7109375, "eval_rewards/rejected": -3.328125, "eval_runtime": 47.0673, "eval_samples_per_second": 42.492, "eval_steps_per_second": 0.68, "step": 600 }, { "epoch": 0.6384092098377813, "grad_norm": 1.795238451251697, "learning_rate": 4.808823471266702e-05, "logits/chosen": -0.56640625, "logits/rejected": 1.6640625, "logps/chosen": -596.0, "logps/rejected": -588.0, "loss": 0.5927, "rewards/accuracies": 0.6937500238418579, "rewards/chosen": -2.40625, "rewards/margins": 0.72265625, "rewards/rejected": -3.125, "step": 610 }, { "epoch": 0.6488749345892203, "grad_norm": 1.8922657787391153, "learning_rate": 4.796968003560639e-05, "logits/chosen": 0.58203125, "logits/rejected": 2.0625, "logps/chosen": -516.0, "logps/rejected": -540.0, "loss": 0.5548, "rewards/accuracies": 0.7250000238418579, "rewards/chosen": -2.015625, "rewards/margins": 0.7421875, "rewards/rejected": -2.765625, "step": 620 }, { "epoch": 0.6593406593406593, "grad_norm": 2.0313556109943316, "learning_rate": 4.7847714344635864e-05, "logits/chosen": 0.349609375, "logits/rejected": 2.078125, "logps/chosen": -520.0, "logps/rejected": -544.0, "loss": 0.5494, "rewards/accuracies": 0.7250000238418579, "rewards/chosen": -2.046875, "rewards/margins": 0.79296875, "rewards/rejected": -2.828125, "step": 630 }, { "epoch": 0.6698063840920984, "grad_norm": 2.2640199899515587, "learning_rate": 4.772235575174776e-05, "logits/chosen": 0.46875, "logits/rejected": 2.09375, "logps/chosen": -576.0, "logps/rejected": -600.0, "loss": 0.5698, "rewards/accuracies": 0.699999988079071, "rewards/chosen": -2.65625, "rewards/margins": 0.75390625, "rewards/rejected": -3.40625, "step": 640 }, { "epoch": 0.6802721088435374, "grad_norm": 1.5616049458209416, "learning_rate": 4.7593622872782754e-05, "logits/chosen": -3.078125, "logits/rejected": -1.828125, "logps/chosen": -520.0, "logps/rejected": -572.0, "loss": 0.5782, "rewards/accuracies": 0.6875, "rewards/chosen": -2.140625, "rewards/margins": 0.53125, "rewards/rejected": -2.671875, "step": 650 }, { "epoch": 0.6907378335949764, "grad_norm": 1.5254831991096864, "learning_rate": 4.746153482466542e-05, "logits/chosen": -0.2890625, "logits/rejected": 1.515625, "logps/chosen": -616.0, "logps/rejected": -644.0, "loss": 0.5686, "rewards/accuracies": 0.7437499761581421, "rewards/chosen": -2.828125, "rewards/margins": 0.6953125, "rewards/rejected": -3.53125, "step": 660 }, { "epoch": 0.7012035583464155, "grad_norm": 1.569659001815168, "learning_rate": 4.732611122256537e-05, "logits/chosen": -1.0625, "logits/rejected": -0.060791015625, "logps/chosen": -656.0, "logps/rejected": -696.0, "loss": 0.5369, "rewards/accuracies": 0.6625000238418579, "rewards/chosen": -3.140625, "rewards/margins": 0.62890625, "rewards/rejected": -3.78125, "step": 670 }, { "epoch": 0.7116692830978545, "grad_norm": 2.0633805163659287, "learning_rate": 4.718737217698435e-05, "logits/chosen": -2.8125, "logits/rejected": -1.59375, "logps/chosen": -620.0, "logps/rejected": -648.0, "loss": 0.5975, "rewards/accuracies": 0.699999988079071, "rewards/chosen": -2.53125, "rewards/margins": 0.67578125, "rewards/rejected": -3.21875, "step": 680 }, { "epoch": 0.7221350078492935, "grad_norm": 1.8369759132228272, "learning_rate": 4.704533829076982e-05, "logits/chosen": -3.828125, "logits/rejected": -1.671875, "logps/chosen": -560.0, "logps/rejected": -584.0, "loss": 0.5203, "rewards/accuracies": 0.7562500238418579, "rewards/chosen": -2.40625, "rewards/margins": 0.8046875, "rewards/rejected": -3.21875, "step": 690 }, { "epoch": 0.7326007326007326, "grad_norm": 2.7365611725455747, "learning_rate": 4.6900030656055455e-05, "logits/chosen": -2.390625, "logits/rejected": -1.3203125, "logps/chosen": -588.0, "logps/rejected": -624.0, "loss": 0.5582, "rewards/accuracies": 0.699999988079071, "rewards/chosen": -2.90625, "rewards/margins": 0.70703125, "rewards/rejected": -3.609375, "step": 700 }, { "epoch": 0.7326007326007326, "eval_logits/chosen": -1.7890625, "eval_logits/rejected": -0.328125, "eval_logps/chosen": -644.0, "eval_logps/rejected": -696.0, "eval_loss": 0.5655156373977661, "eval_rewards/accuracies": 0.705078125, "eval_rewards/chosen": -3.28125, "eval_rewards/margins": 0.80078125, "eval_rewards/rejected": -4.09375, "eval_runtime": 47.0735, "eval_samples_per_second": 42.487, "eval_steps_per_second": 0.68, "step": 700 }, { "epoch": 0.7430664573521716, "grad_norm": 2.1078364385004034, "learning_rate": 4.6751470851128906e-05, "logits/chosen": -2.109375, "logits/rejected": -0.640625, "logps/chosen": -608.0, "logps/rejected": -636.0, "loss": 0.6071, "rewards/accuracies": 0.6875, "rewards/chosen": -3.109375, "rewards/margins": 0.578125, "rewards/rejected": -3.6875, "step": 710 }, { "epoch": 0.7535321821036107, "grad_norm": 1.7026329969170844, "learning_rate": 4.6599680937227454e-05, "logits/chosen": -1.765625, "logits/rejected": -0.022705078125, "logps/chosen": -592.0, "logps/rejected": -600.0, "loss": 0.5473, "rewards/accuracies": 0.6812499761581421, "rewards/chosen": -2.546875, "rewards/margins": 0.66796875, "rewards/rejected": -3.21875, "step": 720 }, { "epoch": 0.7639979068550498, "grad_norm": 2.393442783229528, "learning_rate": 4.644468345526183e-05, "logits/chosen": 0.8203125, "logits/rejected": 2.15625, "logps/chosen": -664.0, "logps/rejected": -664.0, "loss": 0.5962, "rewards/accuracies": 0.612500011920929, "rewards/chosen": -3.546875, "rewards/margins": 0.4453125, "rewards/rejected": -3.984375, "step": 730 }, { "epoch": 0.7744636316064888, "grad_norm": 1.9942076882511888, "learning_rate": 4.628650142246894e-05, "logits/chosen": 1.3515625, "logits/rejected": 2.140625, "logps/chosen": -640.0, "logps/rejected": -664.0, "loss": 0.5576, "rewards/accuracies": 0.643750011920929, "rewards/chosen": -3.484375, "rewards/margins": 0.5546875, "rewards/rejected": -4.03125, "step": 740 }, { "epoch": 0.7849293563579278, "grad_norm": 1.457970055726266, "learning_rate": 4.6125158328993765e-05, "logits/chosen": -0.369140625, "logits/rejected": 0.85546875, "logps/chosen": -600.0, "logps/rejected": -660.0, "loss": 0.5356, "rewards/accuracies": 0.6812499761581421, "rewards/chosen": -3.171875, "rewards/margins": 0.69921875, "rewards/rejected": -3.875, "step": 750 }, { "epoch": 0.7953950811093669, "grad_norm": 1.7876444895156745, "learning_rate": 4.5960678134401064e-05, "logits/chosen": -0.1181640625, "logits/rejected": 1.125, "logps/chosen": -572.0, "logps/rejected": -632.0, "loss": 0.5682, "rewards/accuracies": 0.71875, "rewards/chosen": -2.921875, "rewards/margins": 0.72265625, "rewards/rejected": -3.640625, "step": 760 }, { "epoch": 0.8058608058608059, "grad_norm": 1.6427303240483597, "learning_rate": 4.579308526411734e-05, "logits/chosen": -1.9921875, "logits/rejected": -1.8046875, "logps/chosen": -620.0, "logps/rejected": -696.0, "loss": 0.5393, "rewards/accuracies": 0.668749988079071, "rewards/chosen": -2.734375, "rewards/margins": 0.60546875, "rewards/rejected": -3.34375, "step": 770 }, { "epoch": 0.8163265306122449, "grad_norm": 1.8698958315566019, "learning_rate": 4.562240460580367e-05, "logits/chosen": -2.03125, "logits/rejected": -0.390625, "logps/chosen": -672.0, "logps/rejected": -680.0, "loss": 0.5736, "rewards/accuracies": 0.668749988079071, "rewards/chosen": -3.125, "rewards/margins": 0.69921875, "rewards/rejected": -3.828125, "step": 780 }, { "epoch": 0.826792255363684, "grad_norm": 1.759103474481964, "learning_rate": 4.544866150565985e-05, "logits/chosen": -0.0673828125, "logits/rejected": 1.8515625, "logps/chosen": -684.0, "logps/rejected": -696.0, "loss": 0.5942, "rewards/accuracies": 0.706250011920929, "rewards/chosen": -3.5625, "rewards/margins": 0.66015625, "rewards/rejected": -4.21875, "step": 790 }, { "epoch": 0.837257980115123, "grad_norm": 1.7957856888448038, "learning_rate": 4.527188176466051e-05, "logits/chosen": -1.5078125, "logits/rejected": -0.39453125, "logps/chosen": -604.0, "logps/rejected": -660.0, "loss": 0.5437, "rewards/accuracies": 0.75, "rewards/chosen": -2.953125, "rewards/margins": 0.7109375, "rewards/rejected": -3.671875, "step": 800 }, { "epoch": 0.837257980115123, "eval_logits/chosen": -3.171875, "eval_logits/rejected": -1.671875, "eval_logps/chosen": -596.0, "eval_logps/rejected": -632.0, "eval_loss": 0.570437490940094, "eval_rewards/accuracies": 0.69921875, "eval_rewards/chosen": -2.828125, "eval_rewards/margins": 0.6171875, "eval_rewards/rejected": -3.4375, "eval_runtime": 47.0671, "eval_samples_per_second": 42.493, "eval_steps_per_second": 0.68, "step": 800 }, { "epoch": 0.847723704866562, "grad_norm": 1.7065770791597716, "learning_rate": 4.509209163472357e-05, "logits/chosen": -3.390625, "logits/rejected": -1.6328125, "logps/chosen": -648.0, "logps/rejected": -652.0, "loss": 0.5797, "rewards/accuracies": 0.668749988079071, "rewards/chosen": -3.140625, "rewards/margins": 0.62109375, "rewards/rejected": -3.765625, "step": 810 }, { "epoch": 0.858189429618001, "grad_norm": 1.9124175215045767, "learning_rate": 4.4909317814811896e-05, "logits/chosen": -1.546875, "logits/rejected": -0.3046875, "logps/chosen": -700.0, "logps/rejected": -740.0, "loss": 0.5896, "rewards/accuracies": 0.675000011920929, "rewards/chosen": -3.484375, "rewards/margins": 0.75390625, "rewards/rejected": -4.25, "step": 820 }, { "epoch": 0.8686551543694401, "grad_norm": 1.421390768271575, "learning_rate": 4.472358744696845e-05, "logits/chosen": -2.109375, "logits/rejected": -1.3671875, "logps/chosen": -584.0, "logps/rejected": -628.0, "loss": 0.5743, "rewards/accuracies": 0.699999988079071, "rewards/chosen": -2.875, "rewards/margins": 0.63671875, "rewards/rejected": -3.5, "step": 830 }, { "epoch": 0.8791208791208791, "grad_norm": 1.8482759984277446, "learning_rate": 4.4534928112285645e-05, "logits/chosen": -1.8515625, "logits/rejected": 0.138671875, "logps/chosen": -600.0, "logps/rejected": -664.0, "loss": 0.5316, "rewards/accuracies": 0.706250011920929, "rewards/chosen": -2.796875, "rewards/margins": 0.921875, "rewards/rejected": -3.71875, "step": 840 }, { "epoch": 0.8895866038723181, "grad_norm": 2.2807002191810755, "learning_rate": 4.43433678268096e-05, "logits/chosen": 0.4140625, "logits/rejected": 1.9609375, "logps/chosen": -628.0, "logps/rejected": -652.0, "loss": 0.5323, "rewards/accuracies": 0.71875, "rewards/chosen": -3.046875, "rewards/margins": 0.9375, "rewards/rejected": -3.984375, "step": 850 }, { "epoch": 0.9000523286237572, "grad_norm": 1.5008442995406996, "learning_rate": 4.414893503737968e-05, "logits/chosen": 2.625, "logits/rejected": 4.6875, "logps/chosen": -700.0, "logps/rejected": -712.0, "loss": 0.5615, "rewards/accuracies": 0.6499999761581421, "rewards/chosen": -3.6875, "rewards/margins": 0.60546875, "rewards/rejected": -4.3125, "step": 860 }, { "epoch": 0.9105180533751962, "grad_norm": 1.6532984297840536, "learning_rate": 4.395165861740414e-05, "logits/chosen": 0.25390625, "logits/rejected": 2.0625, "logps/chosen": -604.0, "logps/rejected": -628.0, "loss": 0.5726, "rewards/accuracies": 0.643750011920929, "rewards/chosen": -2.875, "rewards/margins": 0.6328125, "rewards/rejected": -3.5, "step": 870 }, { "epoch": 0.9209837781266352, "grad_norm": 1.7201960312519582, "learning_rate": 4.375156786257241e-05, "logits/chosen": -0.95703125, "logits/rejected": 0.10498046875, "logps/chosen": -636.0, "logps/rejected": -692.0, "loss": 0.5457, "rewards/accuracies": 0.6875, "rewards/chosen": -2.921875, "rewards/margins": 0.66015625, "rewards/rejected": -3.578125, "step": 880 }, { "epoch": 0.9314495028780743, "grad_norm": 1.50258808280136, "learning_rate": 4.354869248650466e-05, "logits/chosen": -3.640625, "logits/rejected": -2.46875, "logps/chosen": -616.0, "logps/rejected": -660.0, "loss": 0.57, "rewards/accuracies": 0.6937500238418579, "rewards/chosen": -2.765625, "rewards/margins": 0.6953125, "rewards/rejected": -3.453125, "step": 890 }, { "epoch": 0.9419152276295133, "grad_norm": 1.9352771451161646, "learning_rate": 4.334306261633931e-05, "logits/chosen": -4.75, "logits/rejected": -3.65625, "logps/chosen": -600.0, "logps/rejected": -652.0, "loss": 0.567, "rewards/accuracies": 0.668749988079071, "rewards/chosen": -2.71875, "rewards/margins": 0.6171875, "rewards/rejected": -3.34375, "step": 900 }, { "epoch": 0.9419152276295133, "eval_logits/chosen": -2.421875, "eval_logits/rejected": -1.078125, "eval_logps/chosen": -628.0, "eval_logps/rejected": -676.0, "eval_loss": 0.5633281469345093, "eval_rewards/accuracies": 0.72265625, "eval_rewards/chosen": -3.140625, "eval_rewards/margins": 0.75390625, "eval_rewards/rejected": -3.90625, "eval_runtime": 47.0604, "eval_samples_per_second": 42.499, "eval_steps_per_second": 0.68, "step": 900 }, { "epoch": 0.9523809523809523, "grad_norm": 1.8626658805171226, "learning_rate": 4.313470878825911e-05, "logits/chosen": -1.734375, "logits/rejected": -0.96484375, "logps/chosen": -568.0, "logps/rejected": -620.0, "loss": 0.5913, "rewards/accuracies": 0.706250011920929, "rewards/chosen": -3.125, "rewards/margins": 0.6953125, "rewards/rejected": -3.828125, "step": 910 }, { "epoch": 0.9628466771323915, "grad_norm": 1.5704308837267742, "learning_rate": 4.292366194295652e-05, "logits/chosen": -2.015625, "logits/rejected": -0.62109375, "logps/chosen": -672.0, "logps/rejected": -684.0, "loss": 0.5588, "rewards/accuracies": 0.699999988079071, "rewards/chosen": -3.46875, "rewards/margins": 0.609375, "rewards/rejected": -4.09375, "step": 920 }, { "epoch": 0.9733124018838305, "grad_norm": 1.6135873239585972, "learning_rate": 4.2709953421038964e-05, "logits/chosen": -1.765625, "logits/rejected": -0.06298828125, "logps/chosen": -716.0, "logps/rejected": -724.0, "loss": 0.5662, "rewards/accuracies": 0.7437499761581421, "rewards/chosen": -3.578125, "rewards/margins": 0.76953125, "rewards/rejected": -4.34375, "step": 930 }, { "epoch": 0.9837781266352695, "grad_norm": 1.5466519777751746, "learning_rate": 4.249361495837478e-05, "logits/chosen": -3.0, "logits/rejected": -1.7734375, "logps/chosen": -584.0, "logps/rejected": -628.0, "loss": 0.5293, "rewards/accuracies": 0.6875, "rewards/chosen": -2.78125, "rewards/margins": 0.69140625, "rewards/rejected": -3.46875, "step": 940 }, { "epoch": 0.9942438513867086, "grad_norm": 1.833921018098829, "learning_rate": 4.227467868138035e-05, "logits/chosen": -1.4609375, "logits/rejected": -0.484375, "logps/chosen": -664.0, "logps/rejected": -708.0, "loss": 0.5552, "rewards/accuracies": 0.7124999761581421, "rewards/chosen": -3.140625, "rewards/margins": 0.70703125, "rewards/rejected": -3.84375, "step": 950 }, { "epoch": 1.0047095761381475, "grad_norm": 1.0393688014221059, "learning_rate": 4.205317710224938e-05, "logits/chosen": -0.13671875, "logits/rejected": 1.078125, "logps/chosen": -616.0, "logps/rejected": -680.0, "loss": 0.429, "rewards/accuracies": 0.7562500238418579, "rewards/chosen": -2.96875, "rewards/margins": 1.0625, "rewards/rejected": -4.03125, "step": 960 }, { "epoch": 1.0151753008895865, "grad_norm": 1.5313318577640207, "learning_rate": 4.1829143114124733e-05, "logits/chosen": -0.11572265625, "logits/rejected": 1.890625, "logps/chosen": -700.0, "logps/rejected": -900.0, "loss": 0.2444, "rewards/accuracies": 0.9125000238418579, "rewards/chosen": -3.40625, "rewards/margins": 2.53125, "rewards/rejected": -5.9375, "step": 970 }, { "epoch": 1.0256410256410255, "grad_norm": 1.7966850720207939, "learning_rate": 4.160260998621387e-05, "logits/chosen": 0.2275390625, "logits/rejected": 2.375, "logps/chosen": -724.0, "logps/rejected": -956.0, "loss": 0.2184, "rewards/accuracies": 0.9125000238418579, "rewards/chosen": -4.25, "rewards/margins": 2.8125, "rewards/rejected": -7.0625, "step": 980 }, { "epoch": 1.0361067503924646, "grad_norm": 2.2245928537162016, "learning_rate": 4.137361135884824e-05, "logits/chosen": -0.86328125, "logits/rejected": 1.578125, "logps/chosen": -692.0, "logps/rejected": -952.0, "loss": 0.2012, "rewards/accuracies": 0.925000011920929, "rewards/chosen": -3.546875, "rewards/margins": 3.171875, "rewards/rejected": -6.71875, "step": 990 }, { "epoch": 1.0465724751439036, "grad_norm": 1.7148996022502598, "learning_rate": 4.114218123848777e-05, "logits/chosen": 0.0191650390625, "logits/rejected": 1.953125, "logps/chosen": -700.0, "logps/rejected": -928.0, "loss": 0.223, "rewards/accuracies": 0.893750011920929, "rewards/chosen": -3.53125, "rewards/margins": 2.71875, "rewards/rejected": -6.25, "step": 1000 }, { "epoch": 1.0465724751439036, "eval_logits/chosen": 1.7734375, "eval_logits/rejected": 3.375, "eval_logps/chosen": -732.0, "eval_logps/rejected": -812.0, "eval_loss": 0.5834531188011169, "eval_rewards/accuracies": 0.74609375, "eval_rewards/chosen": -4.15625, "eval_rewards/margins": 1.0859375, "eval_rewards/rejected": -5.25, "eval_runtime": 47.0162, "eval_samples_per_second": 42.539, "eval_steps_per_second": 0.681, "step": 1000 }, { "epoch": 1.0570381998953426, "grad_norm": 2.7368575422168, "learning_rate": 4.0908353992670824e-05, "logits/chosen": 0.8125, "logits/rejected": 3.34375, "logps/chosen": -636.0, "logps/rejected": -840.0, "loss": 0.2055, "rewards/accuracies": 0.925000011920929, "rewards/chosen": -3.171875, "rewards/margins": 2.6875, "rewards/rejected": -5.875, "step": 1010 }, { "epoch": 1.0675039246467817, "grad_norm": 1.3571231422525123, "learning_rate": 4.067216434491059e-05, "logits/chosen": 1.75, "logits/rejected": 3.40625, "logps/chosen": -728.0, "logps/rejected": -992.0, "loss": 0.1942, "rewards/accuracies": 0.925000011920929, "rewards/chosen": -3.78125, "rewards/margins": 2.984375, "rewards/rejected": -6.78125, "step": 1020 }, { "epoch": 1.077969649398221, "grad_norm": 1.4146938366362345, "learning_rate": 4.043364736953866e-05, "logits/chosen": 0.326171875, "logits/rejected": 2.296875, "logps/chosen": -712.0, "logps/rejected": -948.0, "loss": 0.2023, "rewards/accuracies": 0.9125000238418579, "rewards/chosen": -3.875, "rewards/margins": 2.78125, "rewards/rejected": -6.65625, "step": 1030 }, { "epoch": 1.08843537414966, "grad_norm": 2.531420045083258, "learning_rate": 4.019283848649643e-05, "logits/chosen": 0.62890625, "logits/rejected": 2.96875, "logps/chosen": -640.0, "logps/rejected": -856.0, "loss": 0.2299, "rewards/accuracies": 0.8999999761581421, "rewards/chosen": -3.40625, "rewards/margins": 2.5625, "rewards/rejected": -5.9375, "step": 1040 }, { "epoch": 1.098901098901099, "grad_norm": 1.6597755313425049, "learning_rate": 3.9949773456075215e-05, "logits/chosen": -0.92578125, "logits/rejected": 0.77734375, "logps/chosen": -636.0, "logps/rejected": -880.0, "loss": 0.171, "rewards/accuracies": 0.9312499761581421, "rewards/chosen": -3.328125, "rewards/margins": 2.84375, "rewards/rejected": -6.15625, "step": 1050 }, { "epoch": 1.109366823652538, "grad_norm": 1.7078420244306431, "learning_rate": 3.970448837360585e-05, "logits/chosen": 0.84765625, "logits/rejected": 3.140625, "logps/chosen": -712.0, "logps/rejected": -956.0, "loss": 0.2118, "rewards/accuracies": 0.918749988079071, "rewards/chosen": -3.9375, "rewards/margins": 2.90625, "rewards/rejected": -6.84375, "step": 1060 }, { "epoch": 1.119832548403977, "grad_norm": 2.479613467088712, "learning_rate": 3.945701966409846e-05, "logits/chosen": 0.0247802734375, "logits/rejected": 2.5625, "logps/chosen": -676.0, "logps/rejected": -908.0, "loss": 0.205, "rewards/accuracies": 0.9624999761581421, "rewards/chosen": -3.53125, "rewards/margins": 2.828125, "rewards/rejected": -6.34375, "step": 1070 }, { "epoch": 1.130298273155416, "grad_norm": 2.1446144458423495, "learning_rate": 3.9207404076833374e-05, "logits/chosen": 0.7578125, "logits/rejected": 2.359375, "logps/chosen": -716.0, "logps/rejected": -1024.0, "loss": 0.1996, "rewards/accuracies": 0.949999988079071, "rewards/chosen": -4.15625, "rewards/margins": 3.25, "rewards/rejected": -7.40625, "step": 1080 }, { "epoch": 1.1407639979068551, "grad_norm": 1.3256457293542654, "learning_rate": 3.895567867990379e-05, "logits/chosen": 0.1904296875, "logits/rejected": 2.5625, "logps/chosen": -800.0, "logps/rejected": -1136.0, "loss": 0.1791, "rewards/accuracies": 0.9312499761581421, "rewards/chosen": -4.8125, "rewards/margins": 3.625, "rewards/rejected": -8.4375, "step": 1090 }, { "epoch": 1.1512297226582942, "grad_norm": 2.0738567712055413, "learning_rate": 3.870188085471114e-05, "logits/chosen": 0.70703125, "logits/rejected": 2.3125, "logps/chosen": -712.0, "logps/rejected": -976.0, "loss": 0.1774, "rewards/accuracies": 0.9312499761581421, "rewards/chosen": -3.796875, "rewards/margins": 3.03125, "rewards/rejected": -6.84375, "step": 1100 }, { "epoch": 1.1512297226582942, "eval_logits/chosen": 0.984375, "eval_logits/rejected": 2.890625, "eval_logps/chosen": -800.0, "eval_logps/rejected": -884.0, "eval_loss": 0.6000000238418579, "eval_rewards/accuracies": 0.72265625, "eval_rewards/chosen": -4.84375, "eval_rewards/margins": 1.1328125, "eval_rewards/rejected": -5.96875, "eval_runtime": 47.0425, "eval_samples_per_second": 42.515, "eval_steps_per_second": 0.68, "step": 1100 }, { "epoch": 1.1616954474097332, "grad_norm": 1.8066221511385154, "learning_rate": 3.844604829041395e-05, "logits/chosen": 0.52734375, "logits/rejected": 3.234375, "logps/chosen": -772.0, "logps/rejected": -1064.0, "loss": 0.1489, "rewards/accuracies": 0.956250011920929, "rewards/chosen": -4.5625, "rewards/margins": 3.3125, "rewards/rejected": -7.90625, "step": 1110 }, { "epoch": 1.1721611721611722, "grad_norm": 2.3136586918512196, "learning_rate": 3.818821897833095e-05, "logits/chosen": 0.89453125, "logits/rejected": 2.703125, "logps/chosen": -776.0, "logps/rejected": -1072.0, "loss": 0.2125, "rewards/accuracies": 0.925000011920929, "rewards/chosen": -4.5, "rewards/margins": 3.28125, "rewards/rejected": -7.78125, "step": 1120 }, { "epoch": 1.1826268969126112, "grad_norm": 1.397370065146398, "learning_rate": 3.792843120629935e-05, "logits/chosen": -0.48046875, "logits/rejected": 1.703125, "logps/chosen": -752.0, "logps/rejected": -1032.0, "loss": 0.1569, "rewards/accuracies": 0.925000011920929, "rewards/chosen": -4.21875, "rewards/margins": 3.03125, "rewards/rejected": -7.25, "step": 1130 }, { "epoch": 1.1930926216640503, "grad_norm": 2.0480242805637143, "learning_rate": 3.766672355298907e-05, "logits/chosen": -0.0019683837890625, "logits/rejected": 1.5390625, "logps/chosen": -784.0, "logps/rejected": -1072.0, "loss": 0.1745, "rewards/accuracies": 0.949999988079071, "rewards/chosen": -4.3125, "rewards/margins": 3.3125, "rewards/rejected": -7.625, "step": 1140 }, { "epoch": 1.2035583464154893, "grad_norm": 2.6590902823148466, "learning_rate": 3.740313488217373e-05, "logits/chosen": 0.6953125, "logits/rejected": 1.8125, "logps/chosen": -716.0, "logps/rejected": -1032.0, "loss": 0.1654, "rewards/accuracies": 0.9312499761581421, "rewards/chosen": -3.984375, "rewards/margins": 3.296875, "rewards/rejected": -7.28125, "step": 1150 }, { "epoch": 1.2140240711669283, "grad_norm": 1.9471357324296918, "learning_rate": 3.7137704336959464e-05, "logits/chosen": 0.00738525390625, "logits/rejected": 2.640625, "logps/chosen": -712.0, "logps/rejected": -976.0, "loss": 0.1804, "rewards/accuracies": 0.925000011920929, "rewards/chosen": -3.984375, "rewards/margins": 3.078125, "rewards/rejected": -7.0625, "step": 1160 }, { "epoch": 1.2244897959183674, "grad_norm": 1.967447964549196, "learning_rate": 3.687047133397201e-05, "logits/chosen": -0.1123046875, "logits/rejected": 1.5546875, "logps/chosen": -716.0, "logps/rejected": -980.0, "loss": 0.1795, "rewards/accuracies": 0.9375, "rewards/chosen": -4.0625, "rewards/margins": 3.09375, "rewards/rejected": -7.125, "step": 1170 }, { "epoch": 1.2349555206698064, "grad_norm": 1.7291176286779186, "learning_rate": 3.6601475557503415e-05, "logits/chosen": 0.1337890625, "logits/rejected": 1.5, "logps/chosen": -764.0, "logps/rejected": -1072.0, "loss": 0.2074, "rewards/accuracies": 0.925000011920929, "rewards/chosen": -4.4375, "rewards/margins": 3.484375, "rewards/rejected": -7.9375, "step": 1180 }, { "epoch": 1.2454212454212454, "grad_norm": 3.0438406185828804, "learning_rate": 3.633075695361882e-05, "logits/chosen": 0.197265625, "logits/rejected": 2.359375, "logps/chosen": -764.0, "logps/rejected": -1012.0, "loss": 0.1767, "rewards/accuracies": 0.9312499761581421, "rewards/chosen": -4.34375, "rewards/margins": 2.984375, "rewards/rejected": -7.3125, "step": 1190 }, { "epoch": 1.2558869701726845, "grad_norm": 2.4012378637246936, "learning_rate": 3.6058355724224477e-05, "logits/chosen": 0.341796875, "logits/rejected": 2.328125, "logps/chosen": -792.0, "logps/rejected": -1040.0, "loss": 0.1868, "rewards/accuracies": 0.925000011920929, "rewards/chosen": -4.1875, "rewards/margins": 3.0, "rewards/rejected": -7.1875, "step": 1200 }, { "epoch": 1.2558869701726845, "eval_logits/chosen": 1.9609375, "eval_logits/rejected": 3.5, "eval_logps/chosen": -804.0, "eval_logps/rejected": -892.0, "eval_loss": 0.5954218506813049, "eval_rewards/accuracies": 0.71875, "eval_rewards/chosen": -4.90625, "eval_rewards/margins": 1.1484375, "eval_rewards/rejected": -6.0625, "eval_runtime": 47.1673, "eval_samples_per_second": 42.402, "eval_steps_per_second": 0.678, "step": 1200 }, { "epoch": 1.2663526949241235, "grad_norm": 2.4296549845278737, "learning_rate": 3.5784312321097744e-05, "logits/chosen": 0.56640625, "logits/rejected": 3.359375, "logps/chosen": -824.0, "logps/rejected": -1056.0, "loss": 0.1953, "rewards/accuracies": 0.9375, "rewards/chosen": -4.375, "rewards/margins": 3.1875, "rewards/rejected": -7.5625, "step": 1210 }, { "epoch": 1.2768184196755625, "grad_norm": 2.5723549706730253, "learning_rate": 3.550866743987994e-05, "logits/chosen": 1.5546875, "logits/rejected": 3.390625, "logps/chosen": -804.0, "logps/rejected": -1048.0, "loss": 0.2069, "rewards/accuracies": 0.918749988079071, "rewards/chosen": -4.8125, "rewards/margins": 2.96875, "rewards/rejected": -7.75, "step": 1220 }, { "epoch": 1.2872841444270016, "grad_norm": 2.504215400815392, "learning_rate": 3.5231462014033024e-05, "logits/chosen": 0.75390625, "logits/rejected": 2.6875, "logps/chosen": -736.0, "logps/rejected": -1016.0, "loss": 0.1746, "rewards/accuracies": 0.949999988079071, "rewards/chosen": -4.375, "rewards/margins": 3.25, "rewards/rejected": -7.625, "step": 1230 }, { "epoch": 1.2977498691784406, "grad_norm": 2.8504157844640425, "learning_rate": 3.495273720876095e-05, "logits/chosen": 0.1533203125, "logits/rejected": 1.8359375, "logps/chosen": -728.0, "logps/rejected": -1020.0, "loss": 0.1606, "rewards/accuracies": 0.9437500238418579, "rewards/chosen": -4.28125, "rewards/margins": 3.203125, "rewards/rejected": -7.46875, "step": 1240 }, { "epoch": 1.3082155939298796, "grad_norm": 3.7667898045086186, "learning_rate": 3.4672534414896616e-05, "logits/chosen": -1.0546875, "logits/rejected": 2.015625, "logps/chosen": -808.0, "logps/rejected": -1072.0, "loss": 0.1818, "rewards/accuracies": 0.9125000238418579, "rewards/chosen": -4.6875, "rewards/margins": 3.3125, "rewards/rejected": -8.0, "step": 1250 }, { "epoch": 1.3186813186813187, "grad_norm": 2.515350904417584, "learning_rate": 3.439089524275526e-05, "logits/chosen": -0.72265625, "logits/rejected": 1.2890625, "logps/chosen": -792.0, "logps/rejected": -1104.0, "loss": 0.1758, "rewards/accuracies": 0.918749988079071, "rewards/chosen": -4.90625, "rewards/margins": 3.34375, "rewards/rejected": -8.25, "step": 1260 }, { "epoch": 1.3291470434327577, "grad_norm": 2.4225745935536684, "learning_rate": 3.410786151595533e-05, "logits/chosen": -0.96484375, "logits/rejected": 1.8828125, "logps/chosen": -768.0, "logps/rejected": -1056.0, "loss": 0.1916, "rewards/accuracies": 0.925000011920929, "rewards/chosen": -4.5625, "rewards/margins": 3.21875, "rewards/rejected": -7.78125, "step": 1270 }, { "epoch": 1.3396127681841967, "grad_norm": 2.388596466074455, "learning_rate": 3.382347526520765e-05, "logits/chosen": -0.0125732421875, "logits/rejected": 3.015625, "logps/chosen": -748.0, "logps/rejected": -1024.0, "loss": 0.1868, "rewards/accuracies": 0.8999999761581421, "rewards/chosen": -4.25, "rewards/margins": 3.34375, "rewards/rejected": -7.5625, "step": 1280 }, { "epoch": 1.3500784929356358, "grad_norm": 2.3617931451105627, "learning_rate": 3.353777872207381e-05, "logits/chosen": 0.76171875, "logits/rejected": 2.953125, "logps/chosen": -772.0, "logps/rejected": -1064.0, "loss": 0.1842, "rewards/accuracies": 0.956250011920929, "rewards/chosen": -4.4375, "rewards/margins": 3.359375, "rewards/rejected": -7.78125, "step": 1290 }, { "epoch": 1.3605442176870748, "grad_norm": 2.5821544358464643, "learning_rate": 3.325081431269475e-05, "logits/chosen": 0.5078125, "logits/rejected": 2.34375, "logps/chosen": -744.0, "logps/rejected": -1024.0, "loss": 0.1871, "rewards/accuracies": 0.925000011920929, "rewards/chosen": -4.28125, "rewards/margins": 2.9375, "rewards/rejected": -7.25, "step": 1300 }, { "epoch": 1.3605442176870748, "eval_logits/chosen": 1.328125, "eval_logits/rejected": 3.171875, "eval_logps/chosen": -848.0, "eval_logps/rejected": -932.0, "eval_loss": 0.6086093783378601, "eval_rewards/accuracies": 0.732421875, "eval_rewards/chosen": -5.34375, "eval_rewards/margins": 1.15625, "eval_rewards/rejected": -6.5, "eval_runtime": 47.0126, "eval_samples_per_second": 42.542, "eval_steps_per_second": 0.681, "step": 1300 }, { "epoch": 1.3710099424385138, "grad_norm": 1.9097096614091884, "learning_rate": 3.296262465149046e-05, "logits/chosen": 0.29296875, "logits/rejected": 3.234375, "logps/chosen": -828.0, "logps/rejected": -1080.0, "loss": 0.1842, "rewards/accuracies": 0.925000011920929, "rewards/chosen": -4.9375, "rewards/margins": 3.0625, "rewards/rejected": -8.0, "step": 1310 }, { "epoch": 1.3814756671899528, "grad_norm": 2.9115296575383884, "learning_rate": 3.2673252534831687e-05, "logits/chosen": 0.546875, "logits/rejected": 2.0, "logps/chosen": -748.0, "logps/rejected": -1040.0, "loss": 0.1716, "rewards/accuracies": 0.9375, "rewards/chosen": -4.625, "rewards/margins": 3.15625, "rewards/rejected": -7.78125, "step": 1320 }, { "epoch": 1.3919413919413919, "grad_norm": 2.317316332890873, "learning_rate": 3.2382740934684696e-05, "logits/chosen": 0.734375, "logits/rejected": 1.8828125, "logps/chosen": -776.0, "logps/rejected": -1112.0, "loss": 0.2056, "rewards/accuracies": 0.90625, "rewards/chosen": -4.625, "rewards/margins": 3.25, "rewards/rejected": -7.875, "step": 1330 }, { "epoch": 1.402407116692831, "grad_norm": 2.470288302462943, "learning_rate": 3.209113299222982e-05, "logits/chosen": 0.2333984375, "logits/rejected": 3.015625, "logps/chosen": -796.0, "logps/rejected": -1064.0, "loss": 0.1785, "rewards/accuracies": 0.9312499761581421, "rewards/chosen": -4.59375, "rewards/margins": 3.171875, "rewards/rejected": -7.78125, "step": 1340 }, { "epoch": 1.41287284144427, "grad_norm": 2.2648746641185746, "learning_rate": 3.179847201145505e-05, "logits/chosen": 0.48046875, "logits/rejected": 2.828125, "logps/chosen": -760.0, "logps/rejected": -1080.0, "loss": 0.1719, "rewards/accuracies": 0.90625, "rewards/chosen": -4.4375, "rewards/margins": 3.515625, "rewards/rejected": -7.96875, "step": 1350 }, { "epoch": 1.423338566195709, "grad_norm": 2.1550999951506444, "learning_rate": 3.150480145272528e-05, "logits/chosen": -0.66796875, "logits/rejected": 1.640625, "logps/chosen": -728.0, "logps/rejected": -1000.0, "loss": 0.2007, "rewards/accuracies": 0.90625, "rewards/chosen": -4.09375, "rewards/margins": 3.171875, "rewards/rejected": -7.25, "step": 1360 }, { "epoch": 1.433804290947148, "grad_norm": 4.08171408328077, "learning_rate": 3.121016492632848e-05, "logits/chosen": 1.046875, "logits/rejected": 3.3125, "logps/chosen": -784.0, "logps/rejected": -1048.0, "loss": 0.2232, "rewards/accuracies": 0.8999999761581421, "rewards/chosen": -4.78125, "rewards/margins": 3.0, "rewards/rejected": -7.78125, "step": 1370 }, { "epoch": 1.4442700156985873, "grad_norm": 1.7359514420969364, "learning_rate": 3.091460618599951e-05, "logits/chosen": 0.82421875, "logits/rejected": 3.421875, "logps/chosen": -792.0, "logps/rejected": -1072.0, "loss": 0.1771, "rewards/accuracies": 0.949999988079071, "rewards/chosen": -4.75, "rewards/margins": 3.21875, "rewards/rejected": -7.96875, "step": 1380 }, { "epoch": 1.454735740450026, "grad_norm": 2.7102316452529376, "learning_rate": 3.061816912242265e-05, "logits/chosen": -0.6171875, "logits/rejected": 2.4375, "logps/chosen": -740.0, "logps/rejected": -1008.0, "loss": 0.2202, "rewards/accuracies": 0.9125000238418579, "rewards/chosen": -4.15625, "rewards/margins": 3.296875, "rewards/rejected": -7.4375, "step": 1390 }, { "epoch": 1.4652014652014653, "grad_norm": 2.360560074176657, "learning_rate": 3.0320897756713785e-05, "logits/chosen": -0.1650390625, "logits/rejected": 1.1328125, "logps/chosen": -768.0, "logps/rejected": -1072.0, "loss": 0.1651, "rewards/accuracies": 0.925000011920929, "rewards/chosen": -4.28125, "rewards/margins": 3.078125, "rewards/rejected": -7.34375, "step": 1400 }, { "epoch": 1.4652014652014653, "eval_logits/chosen": 1.0625, "eval_logits/rejected": 2.9375, "eval_logps/chosen": -852.0, "eval_logps/rejected": -932.0, "eval_loss": 0.5995468497276306, "eval_rewards/accuracies": 0.708984375, "eval_rewards/chosen": -5.375, "eval_rewards/margins": 1.09375, "eval_rewards/rejected": -6.46875, "eval_runtime": 47.0094, "eval_samples_per_second": 42.545, "eval_steps_per_second": 0.681, "step": 1400 }, { "epoch": 1.4756671899529041, "grad_norm": 2.3017201815751926, "learning_rate": 3.002283623388332e-05, "logits/chosen": 0.0230712890625, "logits/rejected": 2.1875, "logps/chosen": -768.0, "logps/rejected": -1088.0, "loss": 0.1792, "rewards/accuracies": 0.893750011920929, "rewards/chosen": -4.59375, "rewards/margins": 3.4375, "rewards/rejected": -8.0625, "step": 1410 }, { "epoch": 1.4861329147043434, "grad_norm": 2.160300650956657, "learning_rate": 2.9724028816280507e-05, "logits/chosen": 1.6640625, "logits/rejected": 3.40625, "logps/chosen": -816.0, "logps/rejected": -1080.0, "loss": 0.1817, "rewards/accuracies": 0.9125000238418579, "rewards/chosen": -4.75, "rewards/margins": 3.125, "rewards/rejected": -7.875, "step": 1420 }, { "epoch": 1.4965986394557822, "grad_norm": 2.089375000678152, "learning_rate": 2.9424519877020524e-05, "logits/chosen": 0.388671875, "logits/rejected": 2.984375, "logps/chosen": -848.0, "logps/rejected": -1192.0, "loss": 0.1573, "rewards/accuracies": 0.9624999761581421, "rewards/chosen": -5.09375, "rewards/margins": 3.84375, "rewards/rejected": -8.9375, "step": 1430 }, { "epoch": 1.5070643642072215, "grad_norm": 1.4133700052733793, "learning_rate": 2.912435389339504e-05, "logits/chosen": 0.068359375, "logits/rejected": 2.46875, "logps/chosen": -828.0, "logps/rejected": -1168.0, "loss": 0.1653, "rewards/accuracies": 0.9375, "rewards/chosen": -4.84375, "rewards/margins": 3.796875, "rewards/rejected": -8.625, "step": 1440 }, { "epoch": 1.5175300889586603, "grad_norm": 2.703992774885425, "learning_rate": 2.8823575440267226e-05, "logits/chosen": 0.265625, "logits/rejected": 2.609375, "logps/chosen": -896.0, "logps/rejected": -1200.0, "loss": 0.1629, "rewards/accuracies": 0.9312499761581421, "rewards/chosen": -5.28125, "rewards/margins": 3.5625, "rewards/rejected": -8.8125, "step": 1450 }, { "epoch": 1.5279958137100995, "grad_norm": 3.1488826051066447, "learning_rate": 2.8522229183452465e-05, "logits/chosen": 0.291015625, "logits/rejected": 2.421875, "logps/chosen": -832.0, "logps/rejected": -1184.0, "loss": 0.1421, "rewards/accuracies": 0.925000011920929, "rewards/chosen": -5.09375, "rewards/margins": 3.578125, "rewards/rejected": -8.6875, "step": 1460 }, { "epoch": 1.5384615384615383, "grad_norm": 2.3917743301184333, "learning_rate": 2.8220359873085373e-05, "logits/chosen": -1.8671875, "logits/rejected": 0.111328125, "logps/chosen": -832.0, "logps/rejected": -1208.0, "loss": 0.1955, "rewards/accuracies": 0.9312499761581421, "rewards/chosen": -4.75, "rewards/margins": 3.90625, "rewards/rejected": -8.625, "step": 1470 }, { "epoch": 1.5489272632129776, "grad_norm": 1.9670094545327208, "learning_rate": 2.7918012336974388e-05, "logits/chosen": -1.3671875, "logits/rejected": 1.796875, "logps/chosen": -752.0, "logps/rejected": -1004.0, "loss": 0.1924, "rewards/accuracies": 0.8687499761581421, "rewards/chosen": -4.53125, "rewards/margins": 3.125, "rewards/rejected": -7.65625, "step": 1480 }, { "epoch": 1.5593929879644164, "grad_norm": 1.9648972742376194, "learning_rate": 2.7615231473944815e-05, "logits/chosen": -2.484375, "logits/rejected": 0.8984375, "logps/chosen": -744.0, "logps/rejected": -980.0, "loss": 0.1681, "rewards/accuracies": 0.90625, "rewards/chosen": -4.0625, "rewards/margins": 2.84375, "rewards/rejected": -6.90625, "step": 1490 }, { "epoch": 1.5698587127158556, "grad_norm": 1.3556057414859004, "learning_rate": 2.7312062247171327e-05, "logits/chosen": -2.703125, "logits/rejected": 0.1591796875, "logps/chosen": -748.0, "logps/rejected": -1064.0, "loss": 0.1557, "rewards/accuracies": 0.9375, "rewards/chosen": -4.125, "rewards/margins": 3.34375, "rewards/rejected": -7.46875, "step": 1500 }, { "epoch": 1.5698587127158556, "eval_logits/chosen": -0.158203125, "eval_logits/rejected": 1.921875, "eval_logps/chosen": -848.0, "eval_logps/rejected": -944.0, "eval_loss": 0.6073281168937683, "eval_rewards/accuracies": 0.701171875, "eval_rewards/chosen": -5.3125, "eval_rewards/margins": 1.265625, "eval_rewards/rejected": -6.59375, "eval_runtime": 47.0174, "eval_samples_per_second": 42.537, "eval_steps_per_second": 0.681, "step": 1500 }, { "epoch": 1.5803244374672945, "grad_norm": 2.624966359250707, "learning_rate": 2.7008549677500876e-05, "logits/chosen": -0.494140625, "logits/rejected": 1.5625, "logps/chosen": -816.0, "logps/rejected": -1112.0, "loss": 0.159, "rewards/accuracies": 0.956250011920929, "rewards/chosen": -4.6875, "rewards/margins": 3.3125, "rewards/rejected": -8.0, "step": 1510 }, { "epoch": 1.5907901622187337, "grad_norm": 2.857031009198905, "learning_rate": 2.670473883676709e-05, "logits/chosen": -1.2578125, "logits/rejected": 1.3125, "logps/chosen": -824.0, "logps/rejected": -1112.0, "loss": 0.2075, "rewards/accuracies": 0.8999999761581421, "rewards/chosen": -4.875, "rewards/margins": 3.25, "rewards/rejected": -8.125, "step": 1520 }, { "epoch": 1.6012558869701727, "grad_norm": 1.2234386318558186, "learning_rate": 2.6400674841097074e-05, "logits/chosen": -1.65625, "logits/rejected": 1.65625, "logps/chosen": -756.0, "logps/rejected": -1008.0, "loss": 0.1509, "rewards/accuracies": 0.949999988079071, "rewards/chosen": -4.28125, "rewards/margins": 3.25, "rewards/rejected": -7.53125, "step": 1530 }, { "epoch": 1.6117216117216118, "grad_norm": 2.0080209760893233, "learning_rate": 2.609640284421158e-05, "logits/chosen": -1.5234375, "logits/rejected": 1.328125, "logps/chosen": -784.0, "logps/rejected": -1056.0, "loss": 0.1805, "rewards/accuracies": 0.956250011920929, "rewards/chosen": -4.34375, "rewards/margins": 3.265625, "rewards/rejected": -7.625, "step": 1540 }, { "epoch": 1.6221873364730508, "grad_norm": 2.182586326150746, "learning_rate": 2.5791968030719726e-05, "logits/chosen": -1.9296875, "logits/rejected": 0.77734375, "logps/chosen": -788.0, "logps/rejected": -1112.0, "loss": 0.1926, "rewards/accuracies": 0.9125000238418579, "rewards/chosen": -4.375, "rewards/margins": 3.46875, "rewards/rejected": -7.84375, "step": 1550 }, { "epoch": 1.6326530612244898, "grad_norm": 2.3565137106969156, "learning_rate": 2.548741560940902e-05, "logits/chosen": -0.8828125, "logits/rejected": 1.8984375, "logps/chosen": -748.0, "logps/rejected": -1048.0, "loss": 0.1894, "rewards/accuracies": 0.9125000238418579, "rewards/chosen": -4.125, "rewards/margins": 3.453125, "rewards/rejected": -7.5625, "step": 1560 }, { "epoch": 1.6431187859759289, "grad_norm": 2.0899452639512375, "learning_rate": 2.5182790806531787e-05, "logits/chosen": -0.44140625, "logits/rejected": 2.421875, "logps/chosen": -740.0, "logps/rejected": -1004.0, "loss": 0.1692, "rewards/accuracies": 0.9375, "rewards/chosen": -4.375, "rewards/margins": 3.078125, "rewards/rejected": -7.46875, "step": 1570 }, { "epoch": 1.653584510727368, "grad_norm": 1.5081564420764841, "learning_rate": 2.4878138859089074e-05, "logits/chosen": -0.2216796875, "logits/rejected": 2.625, "logps/chosen": -796.0, "logps/rejected": -1080.0, "loss": 0.1554, "rewards/accuracies": 0.9624999761581421, "rewards/chosen": -4.65625, "rewards/margins": 3.40625, "rewards/rejected": -8.0625, "step": 1580 }, { "epoch": 1.664050235478807, "grad_norm": 2.5823582357315504, "learning_rate": 2.4573505008112924e-05, "logits/chosen": 0.57421875, "logits/rejected": 3.59375, "logps/chosen": -840.0, "logps/rejected": -1144.0, "loss": 0.1655, "rewards/accuracies": 0.9437500238418579, "rewards/chosen": -5.1875, "rewards/margins": 3.390625, "rewards/rejected": -8.5625, "step": 1590 }, { "epoch": 1.674515960230246, "grad_norm": 2.87900777388136, "learning_rate": 2.4268934491948032e-05, "logits/chosen": 0.7265625, "logits/rejected": 3.015625, "logps/chosen": -828.0, "logps/rejected": -1120.0, "loss": 0.2145, "rewards/accuracies": 0.90625, "rewards/chosen": -4.90625, "rewards/margins": 3.484375, "rewards/rejected": -8.375, "step": 1600 }, { "epoch": 1.674515960230246, "eval_logits/chosen": 0.95703125, "eval_logits/rejected": 3.046875, "eval_logps/chosen": -832.0, "eval_logps/rejected": -932.0, "eval_loss": 0.6255781054496765, "eval_rewards/accuracies": 0.703125, "eval_rewards/chosen": -5.1875, "eval_rewards/margins": 1.265625, "eval_rewards/rejected": -6.46875, "eval_runtime": 47.0346, "eval_samples_per_second": 42.522, "eval_steps_per_second": 0.68, "step": 1600 }, { "epoch": 1.684981684981685, "grad_norm": 1.3894607970254371, "learning_rate": 2.396447253953385e-05, "logits/chosen": -1.2265625, "logits/rejected": 1.609375, "logps/chosen": -756.0, "logps/rejected": -1080.0, "loss": 0.1968, "rewards/accuracies": 0.9312499761581421, "rewards/chosen": -3.984375, "rewards/margins": 3.578125, "rewards/rejected": -7.5625, "step": 1610 }, { "epoch": 1.695447409733124, "grad_norm": 1.5557700204363718, "learning_rate": 2.3660164363687998e-05, "logits/chosen": -1.71875, "logits/rejected": 1.5234375, "logps/chosen": -684.0, "logps/rejected": -948.0, "loss": 0.1844, "rewards/accuracies": 0.9312499761581421, "rewards/chosen": -3.625, "rewards/margins": 3.15625, "rewards/rejected": -6.78125, "step": 1620 }, { "epoch": 1.705913134484563, "grad_norm": 2.826078804452737, "learning_rate": 2.33560551543922e-05, "logits/chosen": -0.61328125, "logits/rejected": 1.296875, "logps/chosen": -672.0, "logps/rejected": -964.0, "loss": 0.1594, "rewards/accuracies": 0.925000011920929, "rewards/chosen": -3.75, "rewards/margins": 3.109375, "rewards/rejected": -6.875, "step": 1630 }, { "epoch": 1.716378859236002, "grad_norm": 2.8489182369602903, "learning_rate": 2.3052190072081493e-05, "logits/chosen": -0.91796875, "logits/rejected": 1.4921875, "logps/chosen": -780.0, "logps/rejected": -1152.0, "loss": 0.1748, "rewards/accuracies": 0.9437500238418579, "rewards/chosen": -4.25, "rewards/margins": 3.890625, "rewards/rejected": -8.125, "step": 1640 }, { "epoch": 1.7268445839874411, "grad_norm": 2.336490529554783, "learning_rate": 2.2748614240937864e-05, "logits/chosen": -0.953125, "logits/rejected": 1.6796875, "logps/chosen": -760.0, "logps/rejected": -1088.0, "loss": 0.1626, "rewards/accuracies": 0.9624999761581421, "rewards/chosen": -4.25, "rewards/margins": 3.65625, "rewards/rejected": -7.90625, "step": 1650 }, { "epoch": 1.7373103087388801, "grad_norm": 2.785915616368276, "learning_rate": 2.2445372742189334e-05, "logits/chosen": 1.0703125, "logits/rejected": 2.625, "logps/chosen": -784.0, "logps/rejected": -1120.0, "loss": 0.1683, "rewards/accuracies": 0.9125000238418579, "rewards/chosen": -4.65625, "rewards/margins": 3.65625, "rewards/rejected": -8.3125, "step": 1660 }, { "epoch": 1.7477760334903192, "grad_norm": 1.8150326748749686, "learning_rate": 2.214251060741528e-05, "logits/chosen": 0.7890625, "logits/rejected": 3.328125, "logps/chosen": -800.0, "logps/rejected": -1112.0, "loss": 0.1682, "rewards/accuracies": 0.9624999761581421, "rewards/chosen": -4.6875, "rewards/margins": 3.546875, "rewards/rejected": -8.25, "step": 1670 }, { "epoch": 1.7582417582417582, "grad_norm": 1.9014513988706205, "learning_rate": 2.1840072811859302e-05, "logits/chosen": 2.171875, "logits/rejected": 4.125, "logps/chosen": -724.0, "logps/rejected": -1020.0, "loss": 0.1782, "rewards/accuracies": 0.9437500238418579, "rewards/chosen": -4.375, "rewards/margins": 3.234375, "rewards/rejected": -7.625, "step": 1680 }, { "epoch": 1.7687074829931972, "grad_norm": 2.148244435636423, "learning_rate": 2.1538104267750287e-05, "logits/chosen": 0.1474609375, "logits/rejected": 3.59375, "logps/chosen": -784.0, "logps/rejected": -1032.0, "loss": 0.1573, "rewards/accuracies": 0.949999988079071, "rewards/chosen": -4.34375, "rewards/margins": 3.234375, "rewards/rejected": -7.59375, "step": 1690 }, { "epoch": 1.7791732077446363, "grad_norm": 3.520913054067009, "learning_rate": 2.1236649817632952e-05, "logits/chosen": 1.15625, "logits/rejected": 3.25, "logps/chosen": -764.0, "logps/rejected": -1080.0, "loss": 0.1666, "rewards/accuracies": 0.918749988079071, "rewards/chosen": -4.5, "rewards/margins": 3.46875, "rewards/rejected": -7.96875, "step": 1700 }, { "epoch": 1.7791732077446363, "eval_logits/chosen": 1.796875, "eval_logits/rejected": 3.890625, "eval_logps/chosen": -868.0, "eval_logps/rejected": -972.0, "eval_loss": 0.622320294380188, "eval_rewards/accuracies": 0.724609375, "eval_rewards/chosen": -5.53125, "eval_rewards/margins": 1.3046875, "eval_rewards/rejected": -6.84375, "eval_runtime": 47.039, "eval_samples_per_second": 42.518, "eval_steps_per_second": 0.68, "step": 1700 }, { "epoch": 1.7896389324960753, "grad_norm": 2.0900943655769444, "learning_rate": 2.093575422770872e-05, "logits/chosen": 0.53515625, "logits/rejected": 3.5625, "logps/chosen": -836.0, "logps/rejected": -1088.0, "loss": 0.205, "rewards/accuracies": 0.918749988079071, "rewards/chosen": -4.71875, "rewards/margins": 3.328125, "rewards/rejected": -8.0625, "step": 1710 }, { "epoch": 1.8001046572475143, "grad_norm": 1.1325153335122133, "learning_rate": 2.0635462181187827e-05, "logits/chosen": 0.8515625, "logits/rejected": 2.65625, "logps/chosen": -768.0, "logps/rejected": -1020.0, "loss": 0.1829, "rewards/accuracies": 0.949999988079071, "rewards/chosen": -4.34375, "rewards/margins": 2.984375, "rewards/rejected": -7.3125, "step": 1720 }, { "epoch": 1.8105703819989536, "grad_norm": 2.094311290363254, "learning_rate": 2.0335818271653917e-05, "logits/chosen": 0.03466796875, "logits/rejected": 2.984375, "logps/chosen": -744.0, "logps/rejected": -1012.0, "loss": 0.1624, "rewards/accuracies": 0.918749988079071, "rewards/chosen": -4.0625, "rewards/margins": 3.234375, "rewards/rejected": -7.28125, "step": 1730 }, { "epoch": 1.8210361067503924, "grad_norm": 2.763941521398199, "learning_rate": 2.0036866996441815e-05, "logits/chosen": 1.0625, "logits/rejected": 4.1875, "logps/chosen": -796.0, "logps/rejected": -1080.0, "loss": 0.1964, "rewards/accuracies": 0.956250011920929, "rewards/chosen": -4.5, "rewards/margins": 3.390625, "rewards/rejected": -7.875, "step": 1740 }, { "epoch": 1.8315018315018317, "grad_norm": 3.113388734326166, "learning_rate": 1.9738652750029677e-05, "logits/chosen": 0.26171875, "logits/rejected": 3.765625, "logps/chosen": -760.0, "logps/rejected": -1004.0, "loss": 0.1841, "rewards/accuracies": 0.925000011920929, "rewards/chosen": -4.21875, "rewards/margins": 3.078125, "rewards/rejected": -7.28125, "step": 1750 }, { "epoch": 1.8419675562532705, "grad_norm": 1.9351500432815578, "learning_rate": 1.944121981744631e-05, "logits/chosen": 0.6171875, "logits/rejected": 2.765625, "logps/chosen": -724.0, "logps/rejected": -1020.0, "loss": 0.169, "rewards/accuracies": 0.9312499761581421, "rewards/chosen": -4.15625, "rewards/margins": 3.203125, "rewards/rejected": -7.34375, "step": 1760 }, { "epoch": 1.8524332810047097, "grad_norm": 1.808605068216095, "learning_rate": 1.9144612367694887e-05, "logits/chosen": 0.193359375, "logits/rejected": 2.296875, "logps/chosen": -748.0, "logps/rejected": -1032.0, "loss": 0.1549, "rewards/accuracies": 0.949999988079071, "rewards/chosen": -3.921875, "rewards/margins": 3.46875, "rewards/rejected": -7.40625, "step": 1770 }, { "epoch": 1.8628990057561485, "grad_norm": 2.070399514164342, "learning_rate": 1.8848874447193803e-05, "logits/chosen": -0.90625, "logits/rejected": 1.8046875, "logps/chosen": -796.0, "logps/rejected": -1128.0, "loss": 0.1712, "rewards/accuracies": 0.9312499761581421, "rewards/chosen": -4.3125, "rewards/margins": 3.71875, "rewards/rejected": -8.0, "step": 1780 }, { "epoch": 1.8733647305075878, "grad_norm": 2.7236343268183334, "learning_rate": 1.855404997323571e-05, "logits/chosen": -0.22265625, "logits/rejected": 2.71875, "logps/chosen": -744.0, "logps/rejected": -1040.0, "loss": 0.1569, "rewards/accuracies": 0.96875, "rewards/chosen": -4.21875, "rewards/margins": 3.40625, "rewards/rejected": -7.59375, "step": 1790 }, { "epoch": 1.8838304552590266, "grad_norm": 3.839855613087665, "learning_rate": 1.82601827274658e-05, "logits/chosen": 0.404296875, "logits/rejected": 2.96875, "logps/chosen": -700.0, "logps/rejected": -1012.0, "loss": 0.164, "rewards/accuracies": 0.949999988079071, "rewards/chosen": -3.953125, "rewards/margins": 3.53125, "rewards/rejected": -7.5, "step": 1800 }, { "epoch": 1.8838304552590266, "eval_logits/chosen": 0.51171875, "eval_logits/rejected": 2.65625, "eval_logps/chosen": -784.0, "eval_logps/rejected": -880.0, "eval_loss": 0.6083593964576721, "eval_rewards/accuracies": 0.73828125, "eval_rewards/chosen": -4.6875, "eval_rewards/margins": 1.2265625, "eval_rewards/rejected": -5.9375, "eval_runtime": 47.0326, "eval_samples_per_second": 42.524, "eval_steps_per_second": 0.68, "step": 1800 }, { "epoch": 1.8942961800104658, "grad_norm": 1.2341823772867553, "learning_rate": 1.7967316349380213e-05, "logits/chosen": 0.2021484375, "logits/rejected": 2.28125, "logps/chosen": -668.0, "logps/rejected": -996.0, "loss": 0.1753, "rewards/accuracies": 0.9437500238418579, "rewards/chosen": -3.71875, "rewards/margins": 3.53125, "rewards/rejected": -7.25, "step": 1810 }, { "epoch": 1.9047619047619047, "grad_norm": 2.956856829149623, "learning_rate": 1.7675494329845516e-05, "logits/chosen": 0.515625, "logits/rejected": 2.703125, "logps/chosen": -724.0, "logps/rejected": -1024.0, "loss": 0.1747, "rewards/accuracies": 0.925000011920929, "rewards/chosen": -4.1875, "rewards/margins": 3.375, "rewards/rejected": -7.5625, "step": 1820 }, { "epoch": 1.915227629513344, "grad_norm": 2.740726932270945, "learning_rate": 1.738476000464026e-05, "logits/chosen": 1.1953125, "logits/rejected": 2.359375, "logps/chosen": -696.0, "logps/rejected": -1032.0, "loss": 0.1788, "rewards/accuracies": 0.925000011920929, "rewards/chosen": -4.125, "rewards/margins": 3.265625, "rewards/rejected": -7.375, "step": 1830 }, { "epoch": 1.9256933542647827, "grad_norm": 2.578142028618044, "learning_rate": 1.7095156548019646e-05, "logits/chosen": 0.0927734375, "logits/rejected": 2.921875, "logps/chosen": -776.0, "logps/rejected": -1048.0, "loss": 0.1872, "rewards/accuracies": 0.9125000238418579, "rewards/chosen": -4.21875, "rewards/margins": 3.375, "rewards/rejected": -7.59375, "step": 1840 }, { "epoch": 1.936159079016222, "grad_norm": 1.8615725035269717, "learning_rate": 1.6806726966304064e-05, "logits/chosen": -0.1025390625, "logits/rejected": 2.421875, "logps/chosen": -724.0, "logps/rejected": -1012.0, "loss": 0.1695, "rewards/accuracies": 0.925000011920929, "rewards/chosen": -4.0625, "rewards/margins": 3.21875, "rewards/rejected": -7.25, "step": 1850 }, { "epoch": 1.9466248037676608, "grad_norm": 2.072529359177459, "learning_rate": 1.6519514091492625e-05, "logits/chosen": -0.0086669921875, "logits/rejected": 3.0, "logps/chosen": -720.0, "logps/rejected": -1004.0, "loss": 0.1726, "rewards/accuracies": 0.9375, "rewards/chosen": -4.125, "rewards/margins": 3.359375, "rewards/rejected": -7.46875, "step": 1860 }, { "epoch": 1.9570905285191, "grad_norm": 2.3954834529937434, "learning_rate": 1.6233560574902627e-05, "logits/chosen": -1.0390625, "logits/rejected": 2.59375, "logps/chosen": -776.0, "logps/rejected": -1080.0, "loss": 0.1487, "rewards/accuracies": 0.96875, "rewards/chosen": -4.0625, "rewards/margins": 3.796875, "rewards/rejected": -7.84375, "step": 1870 }, { "epoch": 1.9675562532705388, "grad_norm": 2.516162364567967, "learning_rate": 1.5948908880835753e-05, "logits/chosen": -1.0, "logits/rejected": 1.671875, "logps/chosen": -748.0, "logps/rejected": -1032.0, "loss": 0.1688, "rewards/accuracies": 0.956250011920929, "rewards/chosen": -4.21875, "rewards/margins": 3.4375, "rewards/rejected": -7.65625, "step": 1880 }, { "epoch": 1.978021978021978, "grad_norm": 2.5316096017853753, "learning_rate": 1.5665601280272124e-05, "logits/chosen": -1.1171875, "logits/rejected": 1.859375, "logps/chosen": -760.0, "logps/rejected": -1072.0, "loss": 0.1753, "rewards/accuracies": 0.8999999761581421, "rewards/chosen": -4.40625, "rewards/margins": 3.484375, "rewards/rejected": -7.875, "step": 1890 }, { "epoch": 1.988487702773417, "grad_norm": 2.2281936156160187, "learning_rate": 1.538367984459298e-05, "logits/chosen": -0.94921875, "logits/rejected": 1.375, "logps/chosen": -812.0, "logps/rejected": -1152.0, "loss": 0.1552, "rewards/accuracies": 0.956250011920929, "rewards/chosen": -4.6875, "rewards/margins": 3.796875, "rewards/rejected": -8.4375, "step": 1900 }, { "epoch": 1.988487702773417, "eval_logits/chosen": 0.400390625, "eval_logits/rejected": 2.546875, "eval_logps/chosen": -856.0, "eval_logps/rejected": -964.0, "eval_loss": 0.6211249828338623, "eval_rewards/accuracies": 0.736328125, "eval_rewards/chosen": -5.4375, "eval_rewards/margins": 1.3359375, "eval_rewards/rejected": -6.78125, "eval_runtime": 47.0568, "eval_samples_per_second": 42.502, "eval_steps_per_second": 0.68, "step": 1900 }, { "epoch": 1.9989534275248562, "grad_norm": 2.7109243047929468, "learning_rate": 1.5103186439333133e-05, "logits/chosen": 0.287109375, "logits/rejected": 3.265625, "logps/chosen": -800.0, "logps/rejected": -1096.0, "loss": 0.1688, "rewards/accuracies": 0.9312499761581421, "rewards/chosen": -4.59375, "rewards/margins": 3.59375, "rewards/rejected": -8.1875, "step": 1910 }, { "epoch": 2.009419152276295, "grad_norm": 0.3742897544807913, "learning_rate": 1.4824162717963829e-05, "logits/chosen": -0.318359375, "logits/rejected": 3.5625, "logps/chosen": -828.0, "logps/rejected": -1264.0, "loss": 0.033, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -4.625, "rewards/margins": 5.09375, "rewards/rejected": -9.6875, "step": 1920 }, { "epoch": 2.0198848770277342, "grad_norm": 0.73463535676306, "learning_rate": 1.4546650115707192e-05, "logits/chosen": 1.8984375, "logits/rejected": 4.46875, "logps/chosen": -764.0, "logps/rejected": -1256.0, "loss": 0.0336, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -4.6875, "rewards/margins": 5.15625, "rewards/rejected": -9.875, "step": 1930 }, { "epoch": 2.030350601779173, "grad_norm": 0.42468899663958704, "learning_rate": 1.4270689843383111e-05, "logits/chosen": 0.7578125, "logits/rejected": 4.3125, "logps/chosen": -824.0, "logps/rejected": -1352.0, "loss": 0.0209, "rewards/accuracies": 1.0, "rewards/chosen": -5.15625, "rewards/margins": 5.65625, "rewards/rejected": -10.8125, "step": 1940 }, { "epoch": 2.0408163265306123, "grad_norm": 0.644972955693028, "learning_rate": 1.3996322881289347e-05, "logits/chosen": 0.216796875, "logits/rejected": 4.53125, "logps/chosen": -856.0, "logps/rejected": -1384.0, "loss": 0.0194, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -5.46875, "rewards/margins": 5.8125, "rewards/rejected": -11.3125, "step": 1950 }, { "epoch": 2.051282051282051, "grad_norm": 0.5905377469606202, "learning_rate": 1.3723589973115963e-05, "logits/chosen": 1.0234375, "logits/rejected": 2.8125, "logps/chosen": -804.0, "logps/rejected": -1456.0, "loss": 0.02, "rewards/accuracies": 1.0, "rewards/chosen": -5.21875, "rewards/margins": 6.46875, "rewards/rejected": -11.6875, "step": 1960 }, { "epoch": 2.0617477760334904, "grad_norm": 0.3727754493238121, "learning_rate": 1.3452531619894837e-05, "logits/chosen": 1.03125, "logits/rejected": 4.1875, "logps/chosen": -888.0, "logps/rejected": -1480.0, "loss": 0.0174, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -5.59375, "rewards/margins": 6.375, "rewards/rejected": -11.9375, "step": 1970 }, { "epoch": 2.072213500784929, "grad_norm": 0.12090120311845674, "learning_rate": 1.3183188073985245e-05, "logits/chosen": 1.46875, "logits/rejected": 3.6875, "logps/chosen": -904.0, "logps/rejected": -1536.0, "loss": 0.0214, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -5.8125, "rewards/margins": 6.75, "rewards/rejected": -12.5625, "step": 1980 }, { "epoch": 2.0826792255363684, "grad_norm": 0.4212449729084653, "learning_rate": 1.2915599333096351e-05, "logits/chosen": 1.5859375, "logits/rejected": 4.125, "logps/chosen": -900.0, "logps/rejected": -1456.0, "loss": 0.0219, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -5.6875, "rewards/margins": 6.09375, "rewards/rejected": -11.75, "step": 1990 }, { "epoch": 2.0931449502878072, "grad_norm": 0.37252403041001486, "learning_rate": 1.264980513434752e-05, "logits/chosen": 1.15625, "logits/rejected": 4.65625, "logps/chosen": -752.0, "logps/rejected": -1360.0, "loss": 0.0204, "rewards/accuracies": 1.0, "rewards/chosen": -4.5, "rewards/margins": 6.46875, "rewards/rejected": -11.0, "step": 2000 }, { "epoch": 2.0931449502878072, "eval_logits/chosen": 2.171875, "eval_logits/rejected": 4.15625, "eval_logps/chosen": -952.0, "eval_logps/rejected": -1088.0, "eval_loss": 0.682992160320282, "eval_rewards/accuracies": 0.73828125, "eval_rewards/chosen": -6.40625, "eval_rewards/margins": 1.6328125, "eval_rewards/rejected": -8.0, "eval_runtime": 47.0235, "eval_samples_per_second": 42.532, "eval_steps_per_second": 0.681, "step": 2000 }, { "epoch": 2.1036106750392465, "grad_norm": 1.3140050895974096, "learning_rate": 1.2385844948367324e-05, "logits/chosen": 0.890625, "logits/rejected": 3.828125, "logps/chosen": -828.0, "logps/rejected": -1432.0, "loss": 0.0212, "rewards/accuracies": 1.0, "rewards/chosen": -4.84375, "rewards/margins": 6.4375, "rewards/rejected": -11.3125, "step": 2010 }, { "epoch": 2.1140763997906853, "grad_norm": 0.30381922287438445, "learning_rate": 1.2123757973432114e-05, "logits/chosen": 1.515625, "logits/rejected": 4.21875, "logps/chosen": -848.0, "logps/rejected": -1448.0, "loss": 0.0172, "rewards/accuracies": 1.0, "rewards/chosen": -5.3125, "rewards/margins": 6.34375, "rewards/rejected": -11.625, "step": 2020 }, { "epoch": 2.1245421245421245, "grad_norm": 0.400745509622522, "learning_rate": 1.1863583129645106e-05, "logits/chosen": 1.4609375, "logits/rejected": 4.6875, "logps/chosen": -876.0, "logps/rejected": -1464.0, "loss": 0.0199, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -5.46875, "rewards/margins": 6.375, "rewards/rejected": -11.875, "step": 2030 }, { "epoch": 2.1350078492935634, "grad_norm": 0.11987580713159669, "learning_rate": 1.1605359053156607e-05, "logits/chosen": 2.09375, "logits/rejected": 3.59375, "logps/chosen": -900.0, "logps/rejected": -1568.0, "loss": 0.0182, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -6.03125, "rewards/margins": 6.8125, "rewards/rejected": -12.8125, "step": 2040 }, { "epoch": 2.1454735740450026, "grad_norm": 0.40932429459220543, "learning_rate": 1.1349124090426644e-05, "logits/chosen": 2.59375, "logits/rejected": 5.1875, "logps/chosen": -964.0, "logps/rejected": -1616.0, "loss": 0.0228, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -6.46875, "rewards/margins": 6.875, "rewards/rejected": -13.375, "step": 2050 }, { "epoch": 2.155939298796442, "grad_norm": 0.5394316804086853, "learning_rate": 1.1094916292530406e-05, "logits/chosen": 1.5, "logits/rejected": 4.34375, "logps/chosen": -908.0, "logps/rejected": -1552.0, "loss": 0.0186, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -5.65625, "rewards/margins": 6.96875, "rewards/rejected": -12.625, "step": 2060 }, { "epoch": 2.1664050235478807, "grad_norm": 1.4407057440919926, "learning_rate": 1.0842773409507623e-05, "logits/chosen": 2.171875, "logits/rejected": 4.59375, "logps/chosen": -872.0, "logps/rejected": -1552.0, "loss": 0.014, "rewards/accuracies": 1.0, "rewards/chosen": -5.6875, "rewards/margins": 6.96875, "rewards/rejected": -12.6875, "step": 2070 }, { "epoch": 2.17687074829932, "grad_norm": 0.5794945281322136, "learning_rate": 1.0592732884756754e-05, "logits/chosen": 1.015625, "logits/rejected": 4.3125, "logps/chosen": -912.0, "logps/rejected": -1552.0, "loss": 0.0219, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -5.71875, "rewards/margins": 6.8125, "rewards/rejected": -12.5625, "step": 2080 }, { "epoch": 2.1873364730507587, "grad_norm": 0.29136820000228214, "learning_rate": 1.0344831849474507e-05, "logits/chosen": 1.3515625, "logits/rejected": 4.5625, "logps/chosen": -952.0, "logps/rejected": -1616.0, "loss": 0.0176, "rewards/accuracies": 1.0, "rewards/chosen": -6.03125, "rewards/margins": 7.125, "rewards/rejected": -13.125, "step": 2090 }, { "epoch": 2.197802197802198, "grad_norm": 0.4624353204627437, "learning_rate": 1.0099107117141879e-05, "logits/chosen": 2.265625, "logits/rejected": 5.1875, "logps/chosen": -976.0, "logps/rejected": -1656.0, "loss": 0.0205, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -6.40625, "rewards/margins": 7.21875, "rewards/rejected": -13.625, "step": 2100 }, { "epoch": 2.197802197802198, "eval_logits/chosen": 3.546875, "eval_logits/rejected": 5.4375, "eval_logps/chosen": -1216.0, "eval_logps/rejected": -1400.0, "eval_loss": 0.8096015453338623, "eval_rewards/accuracies": 0.716796875, "eval_rewards/chosen": -9.0, "eval_rewards/margins": 2.109375, "eval_rewards/rejected": -11.125, "eval_runtime": 47.018, "eval_samples_per_second": 42.537, "eval_steps_per_second": 0.681, "step": 2100 }, { "epoch": 2.208267922553637, "grad_norm": 0.4817519893553948, "learning_rate": 9.855595178057334e-06, "logits/chosen": 1.78125, "logits/rejected": 4.625, "logps/chosen": -1016.0, "logps/rejected": -1672.0, "loss": 0.02, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -6.78125, "rewards/margins": 7.15625, "rewards/rejected": -13.9375, "step": 2110 }, { "epoch": 2.218733647305076, "grad_norm": 0.3270725664597558, "learning_rate": 9.614332193917949e-06, "logits/chosen": 2.25, "logits/rejected": 4.15625, "logps/chosen": -976.0, "logps/rejected": -1696.0, "loss": 0.0195, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -6.5625, "rewards/margins": 7.53125, "rewards/rejected": -14.125, "step": 2120 }, { "epoch": 2.229199372056515, "grad_norm": 0.3099628228672198, "learning_rate": 9.375353992449385e-06, "logits/chosen": 2.4375, "logits/rejected": 4.75, "logps/chosen": -972.0, "logps/rejected": -1664.0, "loss": 0.0202, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -6.40625, "rewards/margins": 7.25, "rewards/rejected": -13.625, "step": 2130 }, { "epoch": 2.239665096807954, "grad_norm": 0.6242321639179261, "learning_rate": 9.138696062085442e-06, "logits/chosen": 2.390625, "logits/rejected": 4.6875, "logps/chosen": -936.0, "logps/rejected": -1640.0, "loss": 0.0157, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -6.28125, "rewards/margins": 7.1875, "rewards/rejected": -13.4375, "step": 2140 }, { "epoch": 2.250130821559393, "grad_norm": 0.5589326595772912, "learning_rate": 8.904393546698007e-06, "logits/chosen": 1.453125, "logits/rejected": 4.3125, "logps/chosen": -1048.0, "logps/rejected": -1728.0, "loss": 0.0174, "rewards/accuracies": 1.0, "rewards/chosen": -6.71875, "rewards/margins": 7.34375, "rewards/rejected": -14.0625, "step": 2150 }, { "epoch": 2.260596546310832, "grad_norm": 0.48649645354929805, "learning_rate": 8.672481240378142e-06, "logits/chosen": 2.578125, "logits/rejected": 4.40625, "logps/chosen": -948.0, "logps/rejected": -1656.0, "loss": 0.0175, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -6.3125, "rewards/margins": 7.21875, "rewards/rejected": -13.5625, "step": 2160 }, { "epoch": 2.271062271062271, "grad_norm": 1.6300827095250636, "learning_rate": 8.44299358226919e-06, "logits/chosen": 1.2421875, "logits/rejected": 4.875, "logps/chosen": -968.0, "logps/rejected": -1624.0, "loss": 0.0162, "rewards/accuracies": 1.0, "rewards/chosen": -6.1875, "rewards/margins": 7.3125, "rewards/rejected": -13.5, "step": 2170 }, { "epoch": 2.2815279958137102, "grad_norm": 0.5535797674220645, "learning_rate": 8.215964651452457e-06, "logits/chosen": 1.671875, "logits/rejected": 4.875, "logps/chosen": -1012.0, "logps/rejected": -1696.0, "loss": 0.0165, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -6.4375, "rewards/margins": 7.5, "rewards/rejected": -13.9375, "step": 2180 }, { "epoch": 2.291993720565149, "grad_norm": 0.20429205560967575, "learning_rate": 7.991428161886502e-06, "logits/chosen": 2.359375, "logits/rejected": 4.40625, "logps/chosen": -928.0, "logps/rejected": -1672.0, "loss": 0.0188, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -6.15625, "rewards/margins": 7.34375, "rewards/rejected": -13.5, "step": 2190 }, { "epoch": 2.3024594453165883, "grad_norm": 0.14115453910902034, "learning_rate": 7.76941745740061e-06, "logits/chosen": 2.25, "logits/rejected": 5.15625, "logps/chosen": -936.0, "logps/rejected": -1560.0, "loss": 0.0228, "rewards/accuracies": 0.956250011920929, "rewards/chosen": -6.1875, "rewards/margins": 6.9375, "rewards/rejected": -13.125, "step": 2200 }, { "epoch": 2.3024594453165883, "eval_logits/chosen": 3.328125, "eval_logits/rejected": 5.25, "eval_logps/chosen": -1176.0, "eval_logps/rejected": -1368.0, "eval_loss": 0.8076953291893005, "eval_rewards/accuracies": 0.73046875, "eval_rewards/chosen": -8.625, "eval_rewards/margins": 2.15625, "eval_rewards/rejected": -10.8125, "eval_runtime": 47.0215, "eval_samples_per_second": 42.534, "eval_steps_per_second": 0.681, "step": 2200 }, { "epoch": 2.312925170068027, "grad_norm": 0.9036259447276891, "learning_rate": 7.549965506743123e-06, "logits/chosen": 2.609375, "logits/rejected": 5.8125, "logps/chosen": -1004.0, "logps/rejected": -1664.0, "loss": 0.0238, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -6.875, "rewards/margins": 7.1875, "rewards/rejected": -14.0625, "step": 2210 }, { "epoch": 2.3233908948194664, "grad_norm": 1.1547160634824984, "learning_rate": 7.333104898685631e-06, "logits/chosen": 1.6640625, "logits/rejected": 4.21875, "logps/chosen": -984.0, "logps/rejected": -1688.0, "loss": 0.0198, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -6.3125, "rewards/margins": 7.46875, "rewards/rejected": -13.8125, "step": 2220 }, { "epoch": 2.333856619570905, "grad_norm": 1.1014137122380974, "learning_rate": 7.118867837183518e-06, "logits/chosen": 2.015625, "logits/rejected": 4.53125, "logps/chosen": -916.0, "logps/rejected": -1560.0, "loss": 0.0183, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -5.96875, "rewards/margins": 6.8125, "rewards/rejected": -12.75, "step": 2230 }, { "epoch": 2.3443223443223444, "grad_norm": 0.2697038291737216, "learning_rate": 6.9072861365936045e-06, "logits/chosen": 2.265625, "logits/rejected": 4.4375, "logps/chosen": -820.0, "logps/rejected": -1568.0, "loss": 0.0123, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -5.3125, "rewards/margins": 7.59375, "rewards/rejected": -12.9375, "step": 2240 }, { "epoch": 2.3547880690737832, "grad_norm": 0.846980134015214, "learning_rate": 6.6983912169497005e-06, "logits/chosen": 1.171875, "logits/rejected": 4.78125, "logps/chosen": -944.0, "logps/rejected": -1648.0, "loss": 0.0161, "rewards/accuracies": 1.0, "rewards/chosen": -6.09375, "rewards/margins": 7.59375, "rewards/rejected": -13.6875, "step": 2250 }, { "epoch": 2.3652537938252225, "grad_norm": 0.0947537323547321, "learning_rate": 6.49221409929677e-06, "logits/chosen": 0.6953125, "logits/rejected": 3.796875, "logps/chosen": -952.0, "logps/rejected": -1624.0, "loss": 0.0207, "rewards/accuracies": 1.0, "rewards/chosen": -5.84375, "rewards/margins": 7.15625, "rewards/rejected": -13.0, "step": 2260 }, { "epoch": 2.3757195185766613, "grad_norm": 0.9734425082681136, "learning_rate": 6.2887854010842075e-06, "logits/chosen": 2.53125, "logits/rejected": 5.5625, "logps/chosen": -940.0, "logps/rejected": -1656.0, "loss": 0.0192, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -5.96875, "rewards/margins": 7.75, "rewards/rejected": -13.6875, "step": 2270 }, { "epoch": 2.3861852433281006, "grad_norm": 0.11032058111589124, "learning_rate": 6.0881353316191385e-06, "logits/chosen": 1.546875, "logits/rejected": 4.0, "logps/chosen": -948.0, "logps/rejected": -1656.0, "loss": 0.015, "rewards/accuracies": 1.0, "rewards/chosen": -6.0625, "rewards/margins": 7.34375, "rewards/rejected": -13.4375, "step": 2280 }, { "epoch": 2.3966509680795394, "grad_norm": 0.388145440707301, "learning_rate": 5.890293687580381e-06, "logits/chosen": 2.171875, "logits/rejected": 4.28125, "logps/chosen": -932.0, "logps/rejected": -1608.0, "loss": 0.0315, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -5.96875, "rewards/margins": 7.125, "rewards/rejected": -13.0625, "step": 2290 }, { "epoch": 2.4071166928309786, "grad_norm": 0.7106698808158642, "learning_rate": 5.695289848593532e-06, "logits/chosen": 2.5625, "logits/rejected": 4.65625, "logps/chosen": -924.0, "logps/rejected": -1608.0, "loss": 0.0148, "rewards/accuracies": 1.0, "rewards/chosen": -6.03125, "rewards/margins": 7.125, "rewards/rejected": -13.125, "step": 2300 }, { "epoch": 2.4071166928309786, "eval_logits/chosen": 3.390625, "eval_logits/rejected": 5.25, "eval_logps/chosen": -1128.0, "eval_logps/rejected": -1304.0, "eval_loss": 0.7832265496253967, "eval_rewards/accuracies": 0.72265625, "eval_rewards/chosen": -8.1875, "eval_rewards/margins": 2.046875, "eval_rewards/rejected": -10.1875, "eval_runtime": 47.0258, "eval_samples_per_second": 42.53, "eval_steps_per_second": 0.68, "step": 2300 }, { "epoch": 2.4175824175824174, "grad_norm": 0.1613343461074541, "learning_rate": 5.50315277286812e-06, "logits/chosen": 2.046875, "logits/rejected": 4.4375, "logps/chosen": -896.0, "logps/rejected": -1672.0, "loss": 0.0152, "rewards/accuracies": 1.0, "rewards/chosen": -5.96875, "rewards/margins": 7.75, "rewards/rejected": -13.6875, "step": 2310 }, { "epoch": 2.4280481423338567, "grad_norm": 0.30748111934325995, "learning_rate": 5.313910992897281e-06, "logits/chosen": 1.4609375, "logits/rejected": 4.5625, "logps/chosen": -932.0, "logps/rejected": -1592.0, "loss": 0.0168, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -5.625, "rewards/margins": 7.28125, "rewards/rejected": -12.9375, "step": 2320 }, { "epoch": 2.4385138670852955, "grad_norm": 0.15554523502121337, "learning_rate": 5.127592611220658e-06, "logits/chosen": 1.4765625, "logits/rejected": 4.46875, "logps/chosen": -928.0, "logps/rejected": -1616.0, "loss": 0.0182, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -5.78125, "rewards/margins": 7.5, "rewards/rejected": -13.25, "step": 2330 }, { "epoch": 2.4489795918367347, "grad_norm": 0.24453647237148785, "learning_rate": 4.944225296251159e-06, "logits/chosen": 0.033447265625, "logits/rejected": 3.5625, "logps/chosen": -1020.0, "logps/rejected": -1680.0, "loss": 0.0147, "rewards/accuracies": 1.0, "rewards/chosen": -6.25, "rewards/margins": 7.28125, "rewards/rejected": -13.5625, "step": 2340 }, { "epoch": 2.4594453165881736, "grad_norm": 0.1810410192413809, "learning_rate": 4.763836278166172e-06, "logits/chosen": 1.765625, "logits/rejected": 4.75, "logps/chosen": -900.0, "logps/rejected": -1616.0, "loss": 0.0118, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -5.875, "rewards/margins": 7.5, "rewards/rejected": -13.375, "step": 2350 }, { "epoch": 2.469911041339613, "grad_norm": 0.4869760242727057, "learning_rate": 4.586452344863876e-06, "logits/chosen": 1.3125, "logits/rejected": 4.6875, "logps/chosen": -920.0, "logps/rejected": -1672.0, "loss": 0.0169, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -6.25, "rewards/margins": 7.9375, "rewards/rejected": -14.1875, "step": 2360 }, { "epoch": 2.4803767660910516, "grad_norm": 1.5286287577826605, "learning_rate": 4.412099837985192e-06, "logits/chosen": 1.9765625, "logits/rejected": 4.75, "logps/chosen": -956.0, "logps/rejected": -1640.0, "loss": 0.0209, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -6.375, "rewards/margins": 7.28125, "rewards/rejected": -13.625, "step": 2370 }, { "epoch": 2.490842490842491, "grad_norm": 0.4057149625264462, "learning_rate": 4.240804649002089e-06, "logits/chosen": 2.703125, "logits/rejected": 4.6875, "logps/chosen": -956.0, "logps/rejected": -1712.0, "loss": 0.015, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -6.5625, "rewards/margins": 7.75, "rewards/rejected": -14.3125, "step": 2380 }, { "epoch": 2.50130821559393, "grad_norm": 0.5074520314340354, "learning_rate": 4.072592215372598e-06, "logits/chosen": 2.609375, "logits/rejected": 5.40625, "logps/chosen": -916.0, "logps/rejected": -1584.0, "loss": 0.0174, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -6.09375, "rewards/margins": 7.21875, "rewards/rejected": -13.3125, "step": 2390 }, { "epoch": 2.511773940345369, "grad_norm": 0.7449706869417931, "learning_rate": 3.907487516763389e-06, "logits/chosen": 1.6953125, "logits/rejected": 5.21875, "logps/chosen": -1000.0, "logps/rejected": -1688.0, "loss": 0.0202, "rewards/accuracies": 1.0, "rewards/chosen": -6.5625, "rewards/margins": 7.375, "rewards/rejected": -13.9375, "step": 2400 }, { "epoch": 2.511773940345369, "eval_logits/chosen": 3.375, "eval_logits/rejected": 5.3125, "eval_logps/chosen": -1136.0, "eval_logps/rejected": -1312.0, "eval_loss": 0.7835468649864197, "eval_rewards/accuracies": 0.734375, "eval_rewards/chosen": -8.1875, "eval_rewards/margins": 2.078125, "eval_rewards/rejected": -10.25, "eval_runtime": 47.0405, "eval_samples_per_second": 42.517, "eval_steps_per_second": 0.68, "step": 2400 }, { "epoch": 2.5222396650968077, "grad_norm": 0.3839181521366344, "learning_rate": 3.745515071340286e-06, "logits/chosen": 1.3671875, "logits/rejected": 4.21875, "logps/chosen": -940.0, "logps/rejected": -1600.0, "loss": 0.014, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -5.875, "rewards/margins": 7.1875, "rewards/rejected": -13.0625, "step": 2410 }, { "epoch": 2.532705389848247, "grad_norm": 0.286336933236265, "learning_rate": 3.586698932127236e-06, "logits/chosen": 1.6640625, "logits/rejected": 4.21875, "logps/chosen": -868.0, "logps/rejected": -1520.0, "loss": 0.0184, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -5.40625, "rewards/margins": 6.9375, "rewards/rejected": -12.375, "step": 2420 }, { "epoch": 2.5431711145996863, "grad_norm": 0.6795843267159898, "learning_rate": 3.4310626834344744e-06, "logits/chosen": 2.09375, "logits/rejected": 4.625, "logps/chosen": -928.0, "logps/rejected": -1600.0, "loss": 0.0132, "rewards/accuracies": 1.0, "rewards/chosen": -5.9375, "rewards/margins": 7.15625, "rewards/rejected": -13.0625, "step": 2430 }, { "epoch": 2.553636839351125, "grad_norm": 0.36631545753041345, "learning_rate": 3.2786294373562343e-06, "logits/chosen": 2.359375, "logits/rejected": 5.78125, "logps/chosen": -880.0, "logps/rejected": -1640.0, "loss": 0.0182, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -5.6875, "rewards/margins": 7.78125, "rewards/rejected": -13.5, "step": 2440 }, { "epoch": 2.564102564102564, "grad_norm": 0.7545403140288724, "learning_rate": 3.1294218303385526e-06, "logits/chosen": 1.46875, "logits/rejected": 3.96875, "logps/chosen": -940.0, "logps/rejected": -1632.0, "loss": 0.0213, "rewards/accuracies": 0.96875, "rewards/chosen": -6.03125, "rewards/margins": 7.34375, "rewards/rejected": -13.375, "step": 2450 }, { "epoch": 2.574568288854003, "grad_norm": 0.47752709388827486, "learning_rate": 2.9834620198177483e-06, "logits/chosen": 1.6875, "logits/rejected": 4.84375, "logps/chosen": -948.0, "logps/rejected": -1640.0, "loss": 0.0179, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -5.875, "rewards/margins": 7.5, "rewards/rejected": -13.375, "step": 2460 }, { "epoch": 2.5850340136054424, "grad_norm": 0.2112932435338889, "learning_rate": 2.840771680930068e-06, "logits/chosen": 1.3046875, "logits/rejected": 3.890625, "logps/chosen": -844.0, "logps/rejected": -1584.0, "loss": 0.0227, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -5.65625, "rewards/margins": 7.375, "rewards/rejected": -13.0, "step": 2470 }, { "epoch": 2.595499738356881, "grad_norm": 0.5447959847546988, "learning_rate": 2.701372003292868e-06, "logits/chosen": 2.734375, "logits/rejected": 4.09375, "logps/chosen": -936.0, "logps/rejected": -1672.0, "loss": 0.0186, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -6.1875, "rewards/margins": 7.3125, "rewards/rejected": -13.5, "step": 2480 }, { "epoch": 2.60596546310832, "grad_norm": 0.8434268251160756, "learning_rate": 2.56528368785795e-06, "logits/chosen": 2.15625, "logits/rejected": 4.125, "logps/chosen": -936.0, "logps/rejected": -1744.0, "loss": 0.0201, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -5.96875, "rewards/margins": 8.125, "rewards/rejected": -14.0625, "step": 2490 }, { "epoch": 2.6164311878597593, "grad_norm": 0.22878591607426096, "learning_rate": 2.432526943837471e-06, "logits/chosen": 2.171875, "logits/rejected": 4.71875, "logps/chosen": -868.0, "logps/rejected": -1632.0, "loss": 0.01, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -5.59375, "rewards/margins": 7.875, "rewards/rejected": -13.4375, "step": 2500 }, { "epoch": 2.6164311878597593, "eval_logits/chosen": 3.53125, "eval_logits/rejected": 5.46875, "eval_logps/chosen": -1136.0, "eval_logps/rejected": -1320.0, "eval_loss": 0.7940078377723694, "eval_rewards/accuracies": 0.736328125, "eval_rewards/chosen": -8.1875, "eval_rewards/margins": 2.109375, "eval_rewards/rejected": -10.3125, "eval_runtime": 47.0152, "eval_samples_per_second": 42.539, "eval_steps_per_second": 0.681, "step": 2500 }, { "epoch": 2.6268969126111985, "grad_norm": 0.31513554517105563, "learning_rate": 2.3031214857028446e-06, "logits/chosen": 1.3359375, "logits/rejected": 4.5625, "logps/chosen": -952.0, "logps/rejected": -1600.0, "loss": 0.0171, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -6.0625, "rewards/margins": 7.1875, "rewards/rejected": -13.25, "step": 2510 }, { "epoch": 2.6373626373626373, "grad_norm": 0.5230849946209687, "learning_rate": 2.177086530257139e-06, "logits/chosen": 2.453125, "logits/rejected": 5.28125, "logps/chosen": -916.0, "logps/rejected": -1704.0, "loss": 0.0199, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -5.96875, "rewards/margins": 8.125, "rewards/rejected": -14.125, "step": 2520 }, { "epoch": 2.647828362114076, "grad_norm": 0.5986774135034991, "learning_rate": 2.054440793781354e-06, "logits/chosen": 1.6640625, "logits/rejected": 4.0, "logps/chosen": -892.0, "logps/rejected": -1608.0, "loss": 0.0153, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -5.8125, "rewards/margins": 7.375, "rewards/rejected": -13.1875, "step": 2530 }, { "epoch": 2.6582940868655154, "grad_norm": 0.32682974514669016, "learning_rate": 1.9352024892550587e-06, "logits/chosen": 1.8359375, "logits/rejected": 5.1875, "logps/chosen": -908.0, "logps/rejected": -1624.0, "loss": 0.0193, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -5.78125, "rewards/margins": 7.6875, "rewards/rejected": -13.5, "step": 2540 }, { "epoch": 2.6687598116169546, "grad_norm": 0.8318020545836274, "learning_rate": 1.8193893236517379e-06, "logits/chosen": 2.078125, "logits/rejected": 5.09375, "logps/chosen": -936.0, "logps/rejected": -1616.0, "loss": 0.0218, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -6.0625, "rewards/margins": 7.125, "rewards/rejected": -13.1875, "step": 2550 }, { "epoch": 2.6792255363683934, "grad_norm": 0.3829560202905927, "learning_rate": 1.7070184953092956e-06, "logits/chosen": 2.40625, "logits/rejected": 4.84375, "logps/chosen": -936.0, "logps/rejected": -1680.0, "loss": 0.0143, "rewards/accuracies": 1.0, "rewards/chosen": -6.15625, "rewards/margins": 7.65625, "rewards/rejected": -13.8125, "step": 2560 }, { "epoch": 2.6896912611198327, "grad_norm": 0.8627400353031706, "learning_rate": 1.5981066913760918e-06, "logits/chosen": 2.34375, "logits/rejected": 4.96875, "logps/chosen": -964.0, "logps/rejected": -1664.0, "loss": 0.0136, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -6.15625, "rewards/margins": 7.46875, "rewards/rejected": -13.625, "step": 2570 }, { "epoch": 2.7001569858712715, "grad_norm": 0.22809486271219293, "learning_rate": 1.492670085332884e-06, "logits/chosen": 1.671875, "logits/rejected": 4.78125, "logps/chosen": -952.0, "logps/rejected": -1656.0, "loss": 0.0154, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -6.46875, "rewards/margins": 7.46875, "rewards/rejected": -13.9375, "step": 2580 }, { "epoch": 2.7106227106227108, "grad_norm": 2.9771758000310133, "learning_rate": 1.3907243345910787e-06, "logits/chosen": 3.234375, "logits/rejected": 5.375, "logps/chosen": -928.0, "logps/rejected": -1632.0, "loss": 0.0231, "rewards/accuracies": 0.96875, "rewards/chosen": -6.28125, "rewards/margins": 7.3125, "rewards/rejected": -13.5625, "step": 2590 }, { "epoch": 2.7210884353741496, "grad_norm": 0.26887339380558295, "learning_rate": 1.292284578167538e-06, "logits/chosen": 1.9921875, "logits/rejected": 4.03125, "logps/chosen": -936.0, "logps/rejected": -1656.0, "loss": 0.0153, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -6.1875, "rewards/margins": 7.53125, "rewards/rejected": -13.75, "step": 2600 }, { "epoch": 2.7210884353741496, "eval_logits/chosen": 3.75, "eval_logits/rejected": 5.625, "eval_logps/chosen": -1168.0, "eval_logps/rejected": -1360.0, "eval_loss": 0.803640604019165, "eval_rewards/accuracies": 0.732421875, "eval_rewards/chosen": -8.5625, "eval_rewards/margins": 2.171875, "eval_rewards/rejected": -10.75, "eval_runtime": 47.0358, "eval_samples_per_second": 42.521, "eval_steps_per_second": 0.68, "step": 2600 }, { "epoch": 2.731554160125589, "grad_norm": 0.7684642075858851, "learning_rate": 1.1973654344364925e-06, "logits/chosen": 2.453125, "logits/rejected": 4.9375, "logps/chosen": -900.0, "logps/rejected": -1632.0, "loss": 0.0232, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -6.03125, "rewards/margins": 7.4375, "rewards/rejected": -13.4375, "step": 2610 }, { "epoch": 2.7420198848770276, "grad_norm": 0.5907075144781423, "learning_rate": 1.1059809989586583e-06, "logits/chosen": 2.828125, "logits/rejected": 5.5625, "logps/chosen": -924.0, "logps/rejected": -1672.0, "loss": 0.0171, "rewards/accuracies": 1.0, "rewards/chosen": -6.25, "rewards/margins": 7.71875, "rewards/rejected": -13.9375, "step": 2620 }, { "epoch": 2.752485609628467, "grad_norm": 0.5323295222044857, "learning_rate": 1.018144842388044e-06, "logits/chosen": 3.578125, "logits/rejected": 5.21875, "logps/chosen": -912.0, "logps/rejected": -1656.0, "loss": 0.0165, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -6.125, "rewards/margins": 7.5625, "rewards/rejected": -13.6875, "step": 2630 }, { "epoch": 2.7629513343799057, "grad_norm": 0.9352067222509411, "learning_rate": 9.338700084567109e-07, "logits/chosen": 2.546875, "logits/rejected": 5.9375, "logps/chosen": -944.0, "logps/rejected": -1656.0, "loss": 0.0189, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -6.125, "rewards/margins": 7.59375, "rewards/rejected": -13.75, "step": 2640 }, { "epoch": 2.773417059131345, "grad_norm": 0.6003672593422914, "learning_rate": 8.531690120377606e-07, "logits/chosen": 2.75, "logits/rejected": 4.96875, "logps/chosen": -936.0, "logps/rejected": -1616.0, "loss": 0.0141, "rewards/accuracies": 1.0, "rewards/chosen": -6.125, "rewards/margins": 7.1875, "rewards/rejected": -13.3125, "step": 2650 }, { "epoch": 2.7838827838827838, "grad_norm": 0.6029531250013518, "learning_rate": 7.760538372868636e-07, "logits/chosen": 1.53125, "logits/rejected": 5.0, "logps/chosen": -908.0, "logps/rejected": -1576.0, "loss": 0.0189, "rewards/accuracies": 0.9624999761581421, "rewards/chosen": -6.0625, "rewards/margins": 7.0625, "rewards/rejected": -13.125, "step": 2660 }, { "epoch": 2.794348508634223, "grad_norm": 0.4703879006590967, "learning_rate": 7.025359358626166e-07, "logits/chosen": 2.75, "logits/rejected": 5.875, "logps/chosen": -896.0, "logps/rejected": -1536.0, "loss": 0.0193, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -5.90625, "rewards/margins": 6.96875, "rewards/rejected": -12.875, "step": 2670 }, { "epoch": 2.804814233385662, "grad_norm": 0.21943506526520473, "learning_rate": 6.326262252259463e-07, "logits/chosen": 2.203125, "logits/rejected": 4.65625, "logps/chosen": -968.0, "logps/rejected": -1712.0, "loss": 0.0157, "rewards/accuracies": 1.0, "rewards/chosen": -6.125, "rewards/margins": 7.96875, "rewards/rejected": -14.125, "step": 2680 }, { "epoch": 2.815279958137101, "grad_norm": 0.3371863028580141, "learning_rate": 5.66335087018871e-07, "logits/chosen": 3.421875, "logits/rejected": 5.59375, "logps/chosen": -920.0, "logps/rejected": -1656.0, "loss": 0.0144, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -6.21875, "rewards/margins": 7.625, "rewards/rejected": -13.8125, "step": 2690 }, { "epoch": 2.82574568288854, "grad_norm": 2.033784593840589, "learning_rate": 5.036723655228226e-07, "logits/chosen": 2.09375, "logits/rejected": 5.28125, "logps/chosen": -936.0, "logps/rejected": -1632.0, "loss": 0.0205, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -5.9375, "rewards/margins": 7.40625, "rewards/rejected": -13.375, "step": 2700 }, { "epoch": 2.82574568288854, "eval_logits/chosen": 3.640625, "eval_logits/rejected": 5.53125, "eval_logps/chosen": -1152.0, "eval_logps/rejected": -1336.0, "eval_loss": 0.7960546612739563, "eval_rewards/accuracies": 0.734375, "eval_rewards/chosen": -8.375, "eval_rewards/margins": 2.15625, "eval_rewards/rejected": -10.5, "eval_runtime": 47.0138, "eval_samples_per_second": 42.541, "eval_steps_per_second": 0.681, "step": 2700 }, { "epoch": 2.836211407639979, "grad_norm": 1.8964391465555892, "learning_rate": 4.446473661967432e-07, "logits/chosen": 2.234375, "logits/rejected": 4.5625, "logps/chosen": -952.0, "logps/rejected": -1648.0, "loss": 0.0153, "rewards/accuracies": 1.0, "rewards/chosen": -5.90625, "rewards/margins": 7.59375, "rewards/rejected": -13.5, "step": 2710 }, { "epoch": 2.846677132391418, "grad_norm": 0.25310788631230446, "learning_rate": 3.892688542952411e-07, "logits/chosen": 2.03125, "logits/rejected": 4.8125, "logps/chosen": -920.0, "logps/rejected": -1624.0, "loss": 0.0126, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -6.0, "rewards/margins": 7.5, "rewards/rejected": -13.5, "step": 2720 }, { "epoch": 2.857142857142857, "grad_norm": 0.19045071900798055, "learning_rate": 3.3754505356694e-07, "logits/chosen": 2.546875, "logits/rejected": 4.5625, "logps/chosen": -892.0, "logps/rejected": -1704.0, "loss": 0.0191, "rewards/accuracies": 1.0, "rewards/chosen": -5.875, "rewards/margins": 8.1875, "rewards/rejected": -14.0, "step": 2730 }, { "epoch": 2.867608581894296, "grad_norm": 0.523698027683699, "learning_rate": 2.8948364503322276e-07, "logits/chosen": 2.5, "logits/rejected": 4.1875, "logps/chosen": -944.0, "logps/rejected": -1680.0, "loss": 0.0157, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -6.28125, "rewards/margins": 7.59375, "rewards/rejected": -13.875, "step": 2740 }, { "epoch": 2.8780743066457353, "grad_norm": 0.2968881317781756, "learning_rate": 2.4509176584762706e-07, "logits/chosen": 2.421875, "logits/rejected": 5.625, "logps/chosen": -936.0, "logps/rejected": -1624.0, "loss": 0.0151, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -6.1875, "rewards/margins": 7.40625, "rewards/rejected": -13.625, "step": 2750 }, { "epoch": 2.8885400313971745, "grad_norm": 0.23800590987399034, "learning_rate": 2.0437600823595694e-07, "logits/chosen": 2.65625, "logits/rejected": 5.46875, "logps/chosen": -952.0, "logps/rejected": -1664.0, "loss": 0.0141, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -6.34375, "rewards/margins": 7.40625, "rewards/rejected": -13.75, "step": 2760 }, { "epoch": 2.8990057561486133, "grad_norm": 0.28568231662519095, "learning_rate": 1.6734241851733e-07, "logits/chosen": 1.5859375, "logits/rejected": 5.375, "logps/chosen": -948.0, "logps/rejected": -1664.0, "loss": 0.0141, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -5.875, "rewards/margins": 7.71875, "rewards/rejected": -13.5625, "step": 2770 }, { "epoch": 2.909471480900052, "grad_norm": 0.2170341707289819, "learning_rate": 1.339964962062984e-07, "logits/chosen": 1.3828125, "logits/rejected": 5.0, "logps/chosen": -980.0, "logps/rejected": -1680.0, "loss": 0.0187, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -6.28125, "rewards/margins": 7.375, "rewards/rejected": -13.6875, "step": 2780 }, { "epoch": 2.9199372056514914, "grad_norm": 0.27256651321985215, "learning_rate": 1.0434319319617136e-07, "logits/chosen": 2.765625, "logits/rejected": 5.4375, "logps/chosen": -924.0, "logps/rejected": -1680.0, "loss": 0.0165, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -5.9375, "rewards/margins": 7.8125, "rewards/rejected": -13.75, "step": 2790 }, { "epoch": 2.9304029304029307, "grad_norm": 0.2883893658865348, "learning_rate": 7.838691302365363e-08, "logits/chosen": 2.421875, "logits/rejected": 5.46875, "logps/chosen": -904.0, "logps/rejected": -1616.0, "loss": 0.0184, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -5.96875, "rewards/margins": 7.6875, "rewards/rejected": -13.6875, "step": 2800 }, { "epoch": 2.9304029304029307, "eval_logits/chosen": 3.59375, "eval_logits/rejected": 5.5, "eval_logps/chosen": -1144.0, "eval_logps/rejected": -1336.0, "eval_loss": 0.7946875095367432, "eval_rewards/accuracies": 0.732421875, "eval_rewards/chosen": -8.3125, "eval_rewards/margins": 2.15625, "eval_rewards/rejected": -10.5, "eval_runtime": 47.0083, "eval_samples_per_second": 42.546, "eval_steps_per_second": 0.681, "step": 2800 }, { "epoch": 2.9408686551543695, "grad_norm": 2.2439369485224674, "learning_rate": 5.613151021490459e-08, "logits/chosen": 2.203125, "logits/rejected": 4.8125, "logps/chosen": -944.0, "logps/rejected": -1720.0, "loss": 0.0178, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -6.15625, "rewards/margins": 8.0, "rewards/rejected": -14.1875, "step": 2810 }, { "epoch": 2.9513343799058083, "grad_norm": 0.1532388828730624, "learning_rate": 3.7580289713159965e-08, "logits/chosen": 2.15625, "logits/rejected": 5.09375, "logps/chosen": -916.0, "logps/rejected": -1664.0, "loss": 0.0173, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -6.125, "rewards/margins": 7.6875, "rewards/rejected": -13.8125, "step": 2820 }, { "epoch": 2.9618001046572475, "grad_norm": 0.32708647520260203, "learning_rate": 2.2736006387927277e-08, "logits/chosen": 1.375, "logits/rejected": 4.40625, "logps/chosen": -956.0, "logps/rejected": -1640.0, "loss": 0.0143, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -6.21875, "rewards/margins": 7.25, "rewards/rejected": -13.5, "step": 2830 }, { "epoch": 2.9722658294086868, "grad_norm": 0.7034563360463139, "learning_rate": 1.1600864625893604e-08, "logits/chosen": 2.4375, "logits/rejected": 5.84375, "logps/chosen": -928.0, "logps/rejected": -1648.0, "loss": 0.0152, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -5.96875, "rewards/margins": 7.71875, "rewards/rejected": -13.6875, "step": 2840 }, { "epoch": 2.9827315541601256, "grad_norm": 0.26142178485637924, "learning_rate": 4.176518003573549e-09, "logits/chosen": 2.265625, "logits/rejected": 4.09375, "logps/chosen": -920.0, "logps/rejected": -1688.0, "loss": 0.019, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -6.03125, "rewards/margins": 7.84375, "rewards/rejected": -13.875, "step": 2850 }, { "epoch": 2.9931972789115644, "grad_norm": 0.6250332917811459, "learning_rate": 4.640690417528903e-10, "logits/chosen": 3.03125, "logits/rejected": 4.8125, "logps/chosen": -932.0, "logps/rejected": -1624.0, "loss": 0.0124, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -6.09375, "rewards/margins": 7.1875, "rewards/rejected": -13.3125, "step": 2860 }, { "epoch": 2.998430141287284, "step": 2865, "total_flos": 0.0, "train_loss": 0.26257153015070145, "train_runtime": 12376.5138, "train_samples_per_second": 14.819, "train_steps_per_second": 0.231 } ], "logging_steps": 10, "max_steps": 2865, "num_input_tokens_seen": 0, "num_train_epochs": 3, "save_steps": 500, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 0.0, "train_batch_size": 8, "trial_name": null, "trial_params": null }