{ "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": 1.9449436475996986, "learning_rate": 1.7421602787456448e-07, "logits/chosen": -8.375, "logits/rejected": -8.25, "logps/chosen": -1552.0, "logps/rejected": -1768.0, "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": 2.1287081335553637, "learning_rate": 1.7421602787456445e-06, "logits/chosen": -9.3125, "logits/rejected": -8.875, "logps/chosen": -1400.0, "logps/rejected": -1592.0, "loss": 0.6912, "rewards/accuracies": 0.0555555559694767, "rewards/chosen": -0.0005035400390625, "rewards/margins": 0.000217437744140625, "rewards/rejected": -0.000720977783203125, "step": 10 }, { "epoch": 0.020931449502878074, "grad_norm": 1.842664394429035, "learning_rate": 3.484320557491289e-06, "logits/chosen": -9.5, "logits/rejected": -9.25, "logps/chosen": -1256.0, "logps/rejected": -1392.0, "loss": 0.6913, "rewards/accuracies": 0.1875, "rewards/chosen": -0.0184326171875, "rewards/margins": -0.00244140625, "rewards/rejected": -0.0159912109375, "step": 20 }, { "epoch": 0.03139717425431711, "grad_norm": 2.089404532239707, "learning_rate": 5.226480836236934e-06, "logits/chosen": -9.6875, "logits/rejected": -9.375, "logps/chosen": -1248.0, "logps/rejected": -1432.0, "loss": 0.6753, "rewards/accuracies": 0.44999998807907104, "rewards/chosen": -0.11279296875, "rewards/margins": 0.0189208984375, "rewards/rejected": -0.1318359375, "step": 30 }, { "epoch": 0.04186289900575615, "grad_norm": 2.2181005251165855, "learning_rate": 6.968641114982578e-06, "logits/chosen": -10.25, "logits/rejected": -9.5625, "logps/chosen": -1464.0, "logps/rejected": -1800.0, "loss": 0.6573, "rewards/accuracies": 0.53125, "rewards/chosen": -0.51953125, "rewards/margins": 0.1455078125, "rewards/rejected": -0.6640625, "step": 40 }, { "epoch": 0.052328623757195186, "grad_norm": 2.2154188207671206, "learning_rate": 8.710801393728225e-06, "logits/chosen": -10.5, "logits/rejected": -10.0, "logps/chosen": -1584.0, "logps/rejected": -1848.0, "loss": 0.6615, "rewards/accuracies": 0.550000011920929, "rewards/chosen": -0.95703125, "rewards/margins": 0.27734375, "rewards/rejected": -1.234375, "step": 50 }, { "epoch": 0.06279434850863422, "grad_norm": 2.297820328759129, "learning_rate": 1.0452961672473868e-05, "logits/chosen": -11.25, "logits/rejected": -10.8125, "logps/chosen": -1408.0, "logps/rejected": -1640.0, "loss": 0.6467, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -0.9140625, "rewards/margins": 0.185546875, "rewards/rejected": -1.1015625, "step": 60 }, { "epoch": 0.07326007326007326, "grad_norm": 2.2376833908437423, "learning_rate": 1.2195121951219513e-05, "logits/chosen": -11.0625, "logits/rejected": -10.5, "logps/chosen": -1528.0, "logps/rejected": -1768.0, "loss": 0.628, "rewards/accuracies": 0.625, "rewards/chosen": -0.94140625, "rewards/margins": 0.255859375, "rewards/rejected": -1.1953125, "step": 70 }, { "epoch": 0.0837257980115123, "grad_norm": 2.59151268574024, "learning_rate": 1.3937282229965156e-05, "logits/chosen": -11.125, "logits/rejected": -10.5625, "logps/chosen": -1352.0, "logps/rejected": -1656.0, "loss": 0.632, "rewards/accuracies": 0.6625000238418579, "rewards/chosen": -0.984375, "rewards/margins": 0.3046875, "rewards/rejected": -1.2890625, "step": 80 }, { "epoch": 0.09419152276295134, "grad_norm": 2.321366257657489, "learning_rate": 1.56794425087108e-05, "logits/chosen": -10.5625, "logits/rejected": -10.3125, "logps/chosen": -1576.0, "logps/rejected": -1752.0, "loss": 0.6368, "rewards/accuracies": 0.6187499761581421, "rewards/chosen": -1.296875, "rewards/margins": 0.302734375, "rewards/rejected": -1.6015625, "step": 90 }, { "epoch": 0.10465724751439037, "grad_norm": 2.5472251208586996, "learning_rate": 1.742160278745645e-05, "logits/chosen": -10.5, "logits/rejected": -9.9375, "logps/chosen": -1448.0, "logps/rejected": -1824.0, "loss": 0.6224, "rewards/accuracies": 0.6187499761581421, "rewards/chosen": -1.09375, "rewards/margins": 0.3203125, "rewards/rejected": -1.4140625, "step": 100 }, { "epoch": 0.10465724751439037, "eval_logits/chosen": -12.5, "eval_logits/rejected": -12.25, "eval_logps/chosen": -362.0, "eval_logps/rejected": -338.0, "eval_loss": 0.6810937523841858, "eval_rewards/accuracies": 0.556640625, "eval_rewards/chosen": -0.4375, "eval_rewards/margins": 0.061767578125, "eval_rewards/rejected": -0.498046875, "eval_runtime": 48.3296, "eval_samples_per_second": 41.383, "eval_steps_per_second": 0.662, "step": 100 }, { "epoch": 0.1151229722658294, "grad_norm": 2.594508523397368, "learning_rate": 1.9163763066202093e-05, "logits/chosen": -10.5, "logits/rejected": -9.75, "logps/chosen": -1440.0, "logps/rejected": -1824.0, "loss": 0.6258, "rewards/accuracies": 0.612500011920929, "rewards/chosen": -0.97265625, "rewards/margins": 0.3828125, "rewards/rejected": -1.3515625, "step": 110 }, { "epoch": 0.12558869701726844, "grad_norm": 2.400662110584477, "learning_rate": 2.0905923344947736e-05, "logits/chosen": -10.6875, "logits/rejected": -10.5, "logps/chosen": -1216.0, "logps/rejected": -1416.0, "loss": 0.6118, "rewards/accuracies": 0.643750011920929, "rewards/chosen": -0.890625, "rewards/margins": 0.375, "rewards/rejected": -1.265625, "step": 120 }, { "epoch": 0.1360544217687075, "grad_norm": 2.7150120473542, "learning_rate": 2.264808362369338e-05, "logits/chosen": -9.9375, "logits/rejected": -9.4375, "logps/chosen": -1504.0, "logps/rejected": -1744.0, "loss": 0.6587, "rewards/accuracies": 0.5562499761581421, "rewards/chosen": -1.0703125, "rewards/margins": 0.2265625, "rewards/rejected": -1.296875, "step": 130 }, { "epoch": 0.14652014652014653, "grad_norm": 2.0952415103960718, "learning_rate": 2.4390243902439026e-05, "logits/chosen": -9.9375, "logits/rejected": -9.5625, "logps/chosen": -1224.0, "logps/rejected": -1496.0, "loss": 0.6451, "rewards/accuracies": 0.6312500238418579, "rewards/chosen": -0.59375, "rewards/margins": 0.251953125, "rewards/rejected": -0.84765625, "step": 140 }, { "epoch": 0.15698587127158556, "grad_norm": 2.0229199321640787, "learning_rate": 2.6132404181184672e-05, "logits/chosen": -9.1875, "logits/rejected": -8.5, "logps/chosen": -1352.0, "logps/rejected": -1640.0, "loss": 0.6101, "rewards/accuracies": 0.643750011920929, "rewards/chosen": -0.71875, "rewards/margins": 0.359375, "rewards/rejected": -1.078125, "step": 150 }, { "epoch": 0.1674515960230246, "grad_norm": 2.5019064047389348, "learning_rate": 2.7874564459930312e-05, "logits/chosen": -8.75, "logits/rejected": -8.0625, "logps/chosen": -1224.0, "logps/rejected": -1584.0, "loss": 0.6287, "rewards/accuracies": 0.625, "rewards/chosen": -1.15625, "rewards/margins": 0.34375, "rewards/rejected": -1.5, "step": 160 }, { "epoch": 0.17791732077446362, "grad_norm": 2.1843745771999252, "learning_rate": 2.9616724738675962e-05, "logits/chosen": -8.6875, "logits/rejected": -8.5, "logps/chosen": -1504.0, "logps/rejected": -1744.0, "loss": 0.6077, "rewards/accuracies": 0.612500011920929, "rewards/chosen": -1.265625, "rewards/margins": 0.330078125, "rewards/rejected": -1.59375, "step": 170 }, { "epoch": 0.18838304552590268, "grad_norm": 2.3895499621561407, "learning_rate": 3.13588850174216e-05, "logits/chosen": -7.40625, "logits/rejected": -7.1875, "logps/chosen": -1608.0, "logps/rejected": -1792.0, "loss": 0.6357, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -1.484375, "rewards/margins": 0.259765625, "rewards/rejected": -1.7421875, "step": 180 }, { "epoch": 0.1988487702773417, "grad_norm": 2.9974118283491356, "learning_rate": 3.310104529616725e-05, "logits/chosen": -7.65625, "logits/rejected": -6.96875, "logps/chosen": -1320.0, "logps/rejected": -1544.0, "loss": 0.6361, "rewards/accuracies": 0.612500011920929, "rewards/chosen": -1.421875, "rewards/margins": 0.3125, "rewards/rejected": -1.734375, "step": 190 }, { "epoch": 0.20931449502878074, "grad_norm": 2.284073470472728, "learning_rate": 3.48432055749129e-05, "logits/chosen": -9.6875, "logits/rejected": -9.0, "logps/chosen": -1432.0, "logps/rejected": -1768.0, "loss": 0.624, "rewards/accuracies": 0.6312500238418579, "rewards/chosen": -1.546875, "rewards/margins": 0.337890625, "rewards/rejected": -1.890625, "step": 200 }, { "epoch": 0.20931449502878074, "eval_logits/chosen": -11.4375, "eval_logits/rejected": -11.0, "eval_logps/chosen": -402.0, "eval_logps/rejected": -378.0, "eval_loss": 0.6957499980926514, "eval_rewards/accuracies": 0.517578125, "eval_rewards/chosen": -0.828125, "eval_rewards/margins": 0.056884765625, "eval_rewards/rejected": -0.88671875, "eval_runtime": 46.6559, "eval_samples_per_second": 42.867, "eval_steps_per_second": 0.686, "step": 200 }, { "epoch": 0.21978021978021978, "grad_norm": 2.0637708839646627, "learning_rate": 3.6585365853658535e-05, "logits/chosen": -9.75, "logits/rejected": -9.0625, "logps/chosen": -1536.0, "logps/rejected": -1888.0, "loss": 0.6358, "rewards/accuracies": 0.6312500238418579, "rewards/chosen": -1.421875, "rewards/margins": 0.359375, "rewards/rejected": -1.78125, "step": 210 }, { "epoch": 0.2302459445316588, "grad_norm": 2.7690161139813894, "learning_rate": 3.8327526132404185e-05, "logits/chosen": -9.5, "logits/rejected": -9.125, "logps/chosen": -1496.0, "logps/rejected": -1640.0, "loss": 0.6239, "rewards/accuracies": 0.5562499761581421, "rewards/chosen": -1.578125, "rewards/margins": 0.203125, "rewards/rejected": -1.78125, "step": 220 }, { "epoch": 0.24071166928309787, "grad_norm": 2.2364252614954485, "learning_rate": 4.006968641114983e-05, "logits/chosen": -9.6875, "logits/rejected": -9.1875, "logps/chosen": -1568.0, "logps/rejected": -1800.0, "loss": 0.6254, "rewards/accuracies": 0.6187499761581421, "rewards/chosen": -1.3203125, "rewards/margins": 0.48046875, "rewards/rejected": -1.8046875, "step": 230 }, { "epoch": 0.25117739403453687, "grad_norm": 2.2924909948920926, "learning_rate": 4.181184668989547e-05, "logits/chosen": -8.8125, "logits/rejected": -8.1875, "logps/chosen": -1480.0, "logps/rejected": -1576.0, "loss": 0.6267, "rewards/accuracies": 0.637499988079071, "rewards/chosen": -1.171875, "rewards/margins": 0.3046875, "rewards/rejected": -1.4765625, "step": 240 }, { "epoch": 0.2616431187859759, "grad_norm": 2.8007558311445377, "learning_rate": 4.3554006968641115e-05, "logits/chosen": -9.375, "logits/rejected": -8.8125, "logps/chosen": -1712.0, "logps/rejected": -1936.0, "loss": 0.6094, "rewards/accuracies": 0.706250011920929, "rewards/chosen": -1.8828125, "rewards/margins": 0.5390625, "rewards/rejected": -2.421875, "step": 250 }, { "epoch": 0.272108843537415, "grad_norm": 2.245403281069764, "learning_rate": 4.529616724738676e-05, "logits/chosen": -9.25, "logits/rejected": -8.9375, "logps/chosen": -1584.0, "logps/rejected": -1744.0, "loss": 0.6453, "rewards/accuracies": 0.637499988079071, "rewards/chosen": -1.8125, "rewards/margins": 0.443359375, "rewards/rejected": -2.25, "step": 260 }, { "epoch": 0.282574568288854, "grad_norm": 2.50122779689848, "learning_rate": 4.703832752613241e-05, "logits/chosen": -8.625, "logits/rejected": -7.65625, "logps/chosen": -1600.0, "logps/rejected": -1984.0, "loss": 0.6162, "rewards/accuracies": 0.612500011920929, "rewards/chosen": -2.078125, "rewards/margins": 0.494140625, "rewards/rejected": -2.5625, "step": 270 }, { "epoch": 0.29304029304029305, "grad_norm": 2.3081408921832125, "learning_rate": 4.878048780487805e-05, "logits/chosen": -8.8125, "logits/rejected": -8.75, "logps/chosen": -1608.0, "logps/rejected": -1688.0, "loss": 0.6456, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -2.1875, "rewards/margins": 0.375, "rewards/rejected": -2.5625, "step": 280 }, { "epoch": 0.3035060177917321, "grad_norm": 2.0060071105494837, "learning_rate": 4.999983293481418e-05, "logits/chosen": -8.9375, "logits/rejected": -8.25, "logps/chosen": -1336.0, "logps/rejected": -1672.0, "loss": 0.6219, "rewards/accuracies": 0.6499999761581421, "rewards/chosen": -1.3515625, "rewards/margins": 0.4375, "rewards/rejected": -1.796875, "step": 290 }, { "epoch": 0.3139717425431711, "grad_norm": 2.1049700688054838, "learning_rate": 4.9996862949181595e-05, "logits/chosen": -10.1875, "logits/rejected": -9.6875, "logps/chosen": -1464.0, "logps/rejected": -1728.0, "loss": 0.6062, "rewards/accuracies": 0.625, "rewards/chosen": -1.359375, "rewards/margins": 0.46484375, "rewards/rejected": -1.8203125, "step": 300 }, { "epoch": 0.3139717425431711, "eval_logits/chosen": -12.4375, "eval_logits/rejected": -12.25, "eval_logps/chosen": -378.0, "eval_logps/rejected": -352.0, "eval_loss": 0.7006562352180481, "eval_rewards/accuracies": 0.5078125, "eval_rewards/chosen": -0.59375, "eval_rewards/margins": 0.03369140625, "eval_rewards/rejected": -0.62890625, "eval_runtime": 46.6418, "eval_samples_per_second": 42.88, "eval_steps_per_second": 0.686, "step": 300 }, { "epoch": 0.32443746729461015, "grad_norm": 2.7572015380023345, "learning_rate": 4.9990180911526896e-05, "logits/chosen": -8.625, "logits/rejected": -7.90625, "logps/chosen": -1448.0, "logps/rejected": -1704.0, "loss": 0.6448, "rewards/accuracies": 0.625, "rewards/chosen": -1.4296875, "rewards/margins": 0.431640625, "rewards/rejected": -1.859375, "step": 310 }, { "epoch": 0.3349031920460492, "grad_norm": 2.271866726351977, "learning_rate": 4.997978781413745e-05, "logits/chosen": -8.1875, "logits/rejected": -7.875, "logps/chosen": -1448.0, "logps/rejected": -1640.0, "loss": 0.6253, "rewards/accuracies": 0.59375, "rewards/chosen": -1.453125, "rewards/margins": 0.33984375, "rewards/rejected": -1.796875, "step": 320 }, { "epoch": 0.3453689167974882, "grad_norm": 2.772047227263434, "learning_rate": 4.996568520039565e-05, "logits/chosen": -10.375, "logits/rejected": -10.0625, "logps/chosen": -1576.0, "logps/rejected": -1752.0, "loss": 0.6282, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.4375, "rewards/margins": 0.38671875, "rewards/rejected": -1.8203125, "step": 330 }, { "epoch": 0.35583464154892724, "grad_norm": 2.5133149058096436, "learning_rate": 4.994787516454973e-05, "logits/chosen": -11.0, "logits/rejected": -10.5, "logps/chosen": -1568.0, "logps/rejected": -1744.0, "loss": 0.6248, "rewards/accuracies": 0.65625, "rewards/chosen": -1.8359375, "rewards/margins": 0.4609375, "rewards/rejected": -2.296875, "step": 340 }, { "epoch": 0.3663003663003663, "grad_norm": 2.0347654544167084, "learning_rate": 4.992636035140273e-05, "logits/chosen": -9.75, "logits/rejected": -9.4375, "logps/chosen": -1592.0, "logps/rejected": -1720.0, "loss": 0.6381, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -1.4453125, "rewards/margins": 0.271484375, "rewards/rejected": -1.71875, "step": 350 }, { "epoch": 0.37676609105180536, "grad_norm": 1.8679372733737207, "learning_rate": 4.99011439559198e-05, "logits/chosen": -10.125, "logits/rejected": -9.625, "logps/chosen": -1392.0, "logps/rejected": -1616.0, "loss": 0.6378, "rewards/accuracies": 0.612500011920929, "rewards/chosen": -1.453125, "rewards/margins": 0.369140625, "rewards/rejected": -1.8203125, "step": 360 }, { "epoch": 0.3872318158032444, "grad_norm": 2.3241639373847387, "learning_rate": 4.987222972275369e-05, "logits/chosen": -9.5625, "logits/rejected": -9.3125, "logps/chosen": -1448.0, "logps/rejected": -1680.0, "loss": 0.6096, "rewards/accuracies": 0.643750011920929, "rewards/chosen": -1.796875, "rewards/margins": 0.4921875, "rewards/rejected": -2.296875, "step": 370 }, { "epoch": 0.3976975405546834, "grad_norm": 2.553554070152953, "learning_rate": 4.983962194568872e-05, "logits/chosen": -10.3125, "logits/rejected": -9.9375, "logps/chosen": -1496.0, "logps/rejected": -1752.0, "loss": 0.6712, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.9140625, "rewards/margins": 0.388671875, "rewards/rejected": -2.3125, "step": 380 }, { "epoch": 0.40816326530612246, "grad_norm": 2.4810245134732396, "learning_rate": 4.9803325467003084e-05, "logits/chosen": -10.875, "logits/rejected": -10.4375, "logps/chosen": -1408.0, "logps/rejected": -1680.0, "loss": 0.6347, "rewards/accuracies": 0.637499988079071, "rewards/chosen": -1.7734375, "rewards/margins": 0.314453125, "rewards/rejected": -2.09375, "step": 390 }, { "epoch": 0.4186289900575615, "grad_norm": 2.462980527764434, "learning_rate": 4.9763345676749816e-05, "logits/chosen": -10.625, "logits/rejected": -10.1875, "logps/chosen": -1552.0, "logps/rejected": -1680.0, "loss": 0.6334, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -2.015625, "rewards/margins": 0.365234375, "rewards/rejected": -2.390625, "step": 400 }, { "epoch": 0.4186289900575615, "eval_logits/chosen": -11.75, "eval_logits/rejected": -11.375, "eval_logps/chosen": -444.0, "eval_logps/rejected": -424.0, "eval_loss": 0.7011250257492065, "eval_rewards/accuracies": 0.517578125, "eval_rewards/chosen": -1.265625, "eval_rewards/margins": 0.08154296875, "eval_rewards/rejected": -1.34375, "eval_runtime": 46.654, "eval_samples_per_second": 42.869, "eval_steps_per_second": 0.686, "step": 400 }, { "epoch": 0.4290947148090005, "grad_norm": 2.3264066212384775, "learning_rate": 4.9719688511956374e-05, "logits/chosen": -10.375, "logits/rejected": -9.75, "logps/chosen": -1480.0, "logps/rejected": -1824.0, "loss": 0.6154, "rewards/accuracies": 0.675000011920929, "rewards/chosen": -1.765625, "rewards/margins": 0.478515625, "rewards/rejected": -2.25, "step": 410 }, { "epoch": 0.43956043956043955, "grad_norm": 1.8546082388528016, "learning_rate": 4.967236045574294e-05, "logits/chosen": -11.0625, "logits/rejected": -10.6875, "logps/chosen": -1368.0, "logps/rejected": -1648.0, "loss": 0.629, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -1.7734375, "rewards/margins": 0.37109375, "rewards/rejected": -2.140625, "step": 420 }, { "epoch": 0.4500261643118786, "grad_norm": 1.9303097839495018, "learning_rate": 4.96213685363597e-05, "logits/chosen": -11.125, "logits/rejected": -10.9375, "logps/chosen": -1544.0, "logps/rejected": -1824.0, "loss": 0.6405, "rewards/accuracies": 0.6187499761581421, "rewards/chosen": -2.09375, "rewards/margins": 0.361328125, "rewards/rejected": -2.453125, "step": 430 }, { "epoch": 0.4604918890633176, "grad_norm": 2.220170811732844, "learning_rate": 4.956672032614315e-05, "logits/chosen": -10.8125, "logits/rejected": -10.625, "logps/chosen": -1664.0, "logps/rejected": -1824.0, "loss": 0.6124, "rewards/accuracies": 0.675000011920929, "rewards/chosen": -2.03125, "rewards/margins": 0.5234375, "rewards/rejected": -2.5625, "step": 440 }, { "epoch": 0.47095761381475665, "grad_norm": 1.8333953586663607, "learning_rate": 4.950842394039157e-05, "logits/chosen": -11.3125, "logits/rejected": -11.25, "logps/chosen": -1592.0, "logps/rejected": -1712.0, "loss": 0.6469, "rewards/accuracies": 0.6312500238418579, "rewards/chosen": -2.109375, "rewards/margins": 0.322265625, "rewards/rejected": -2.4375, "step": 450 }, { "epoch": 0.48142333856619574, "grad_norm": 2.8252442407968665, "learning_rate": 4.944648803615994e-05, "logits/chosen": -11.5, "logits/rejected": -10.8125, "logps/chosen": -1416.0, "logps/rejected": -1760.0, "loss": 0.6457, "rewards/accuracies": 0.637499988079071, "rewards/chosen": -2.015625, "rewards/margins": 0.62109375, "rewards/rejected": -2.625, "step": 460 }, { "epoch": 0.49188906331763477, "grad_norm": 2.4966546461528294, "learning_rate": 4.938092181097433e-05, "logits/chosen": -10.8125, "logits/rejected": -10.375, "logps/chosen": -1552.0, "logps/rejected": -1792.0, "loss": 0.6169, "rewards/accuracies": 0.612500011920929, "rewards/chosen": -2.03125, "rewards/margins": 0.4609375, "rewards/rejected": -2.484375, "step": 470 }, { "epoch": 0.5023547880690737, "grad_norm": 2.036555445116339, "learning_rate": 4.931173500146607e-05, "logits/chosen": -10.375, "logits/rejected": -10.125, "logps/chosen": -1616.0, "logps/rejected": -1784.0, "loss": 0.653, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -2.09375, "rewards/margins": 0.408203125, "rewards/rejected": -2.5, "step": 480 }, { "epoch": 0.5128205128205128, "grad_norm": 1.9990692732991728, "learning_rate": 4.923893788192582e-05, "logits/chosen": -10.625, "logits/rejected": -10.3125, "logps/chosen": -1536.0, "logps/rejected": -1672.0, "loss": 0.6416, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -1.953125, "rewards/margins": 0.3671875, "rewards/rejected": -2.328125, "step": 490 }, { "epoch": 0.5232862375719518, "grad_norm": 2.2044999699025944, "learning_rate": 4.9162541262777906e-05, "logits/chosen": -10.625, "logits/rejected": -10.375, "logps/chosen": -1616.0, "logps/rejected": -1776.0, "loss": 0.6236, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -2.15625, "rewards/margins": 0.40234375, "rewards/rejected": -2.5625, "step": 500 }, { "epoch": 0.5232862375719518, "eval_logits/chosen": -11.8125, "eval_logits/rejected": -11.4375, "eval_logps/chosen": -430.0, "eval_logps/rejected": -408.0, "eval_loss": 0.727343738079071, "eval_rewards/accuracies": 0.552734375, "eval_rewards/chosen": -1.1171875, "eval_rewards/margins": 0.06591796875, "eval_rewards/rejected": -1.1875, "eval_runtime": 46.6655, "eval_samples_per_second": 42.858, "eval_steps_per_second": 0.686, "step": 500 }, { "epoch": 0.533751962323391, "grad_norm": 2.0799866050139117, "learning_rate": 4.90825564889749e-05, "logits/chosen": -10.6875, "logits/rejected": -10.5, "logps/chosen": -1664.0, "logps/rejected": -1840.0, "loss": 0.5922, "rewards/accuracies": 0.625, "rewards/chosen": -2.34375, "rewards/margins": 0.462890625, "rewards/rejected": -2.8125, "step": 510 }, { "epoch": 0.54421768707483, "grad_norm": 2.1622735698169, "learning_rate": 4.899899543831288e-05, "logits/chosen": -11.0625, "logits/rejected": -10.75, "logps/chosen": -1536.0, "logps/rejected": -1808.0, "loss": 0.6454, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -2.1875, "rewards/margins": 0.431640625, "rewards/rejected": -2.625, "step": 520 }, { "epoch": 0.554683411826269, "grad_norm": 1.9476451485558177, "learning_rate": 4.891187051966765e-05, "logits/chosen": -11.0625, "logits/rejected": -10.625, "logps/chosen": -1584.0, "logps/rejected": -1776.0, "loss": 0.6275, "rewards/accuracies": 0.637499988079071, "rewards/chosen": -1.96875, "rewards/margins": 0.41015625, "rewards/rejected": -2.375, "step": 530 }, { "epoch": 0.565149136577708, "grad_norm": 1.8779970859228006, "learning_rate": 4.882119467115191e-05, "logits/chosen": -10.875, "logits/rejected": -10.375, "logps/chosen": -1648.0, "logps/rejected": -1904.0, "loss": 0.6319, "rewards/accuracies": 0.637499988079071, "rewards/chosen": -2.34375, "rewards/margins": 0.53125, "rewards/rejected": -2.875, "step": 540 }, { "epoch": 0.5756148613291471, "grad_norm": 2.251939194440402, "learning_rate": 4.872698135819402e-05, "logits/chosen": -11.0625, "logits/rejected": -10.5625, "logps/chosen": -1624.0, "logps/rejected": -1896.0, "loss": 0.649, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -2.5, "rewards/margins": 0.38671875, "rewards/rejected": -2.890625, "step": 550 }, { "epoch": 0.5860805860805861, "grad_norm": 2.0885024019901866, "learning_rate": 4.862924457153831e-05, "logits/chosen": -10.25, "logits/rejected": -9.875, "logps/chosen": -1672.0, "logps/rejected": -1872.0, "loss": 0.6255, "rewards/accuracies": 0.637499988079071, "rewards/chosen": -2.25, "rewards/margins": 0.435546875, "rewards/rejected": -2.671875, "step": 560 }, { "epoch": 0.5965463108320251, "grad_norm": 1.8491614532665703, "learning_rate": 4.8527998825167484e-05, "logits/chosen": -9.875, "logits/rejected": -9.375, "logps/chosen": -1520.0, "logps/rejected": -1752.0, "loss": 0.6146, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -2.03125, "rewards/margins": 0.3984375, "rewards/rejected": -2.421875, "step": 570 }, { "epoch": 0.6070120355834642, "grad_norm": 2.1805494998833233, "learning_rate": 4.842325915414728e-05, "logits/chosen": -10.4375, "logits/rejected": -9.875, "logps/chosen": -1488.0, "logps/rejected": -1808.0, "loss": 0.6484, "rewards/accuracies": 0.625, "rewards/chosen": -2.328125, "rewards/margins": 0.447265625, "rewards/rejected": -2.78125, "step": 580 }, { "epoch": 0.6174777603349032, "grad_norm": 1.7539205473382076, "learning_rate": 4.831504111239373e-05, "logits/chosen": -11.375, "logits/rejected": -11.3125, "logps/chosen": -1736.0, "logps/rejected": -1880.0, "loss": 0.635, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -2.59375, "rewards/margins": 0.37109375, "rewards/rejected": -2.96875, "step": 590 }, { "epoch": 0.6279434850863422, "grad_norm": 1.9029604113471421, "learning_rate": 4.8203360770363425e-05, "logits/chosen": -12.4375, "logits/rejected": -12.4375, "logps/chosen": -1624.0, "logps/rejected": -1832.0, "loss": 0.648, "rewards/accuracies": 0.6499999761581421, "rewards/chosen": -1.9375, "rewards/margins": 0.427734375, "rewards/rejected": -2.359375, "step": 600 }, { "epoch": 0.6279434850863422, "eval_logits/chosen": -13.625, "eval_logits/rejected": -13.75, "eval_logps/chosen": -452.0, "eval_logps/rejected": -428.0, "eval_loss": 0.6997187733650208, "eval_rewards/accuracies": 0.505859375, "eval_rewards/chosen": -1.34375, "eval_rewards/margins": 0.05078125, "eval_rewards/rejected": -1.3984375, "eval_runtime": 46.6674, "eval_samples_per_second": 42.856, "eval_steps_per_second": 0.686, "step": 600 }, { "epoch": 0.6384092098377813, "grad_norm": 2.0218196241800452, "learning_rate": 4.808823471266702e-05, "logits/chosen": -12.5, "logits/rejected": -12.125, "logps/chosen": -1416.0, "logps/rejected": -1608.0, "loss": 0.6127, "rewards/accuracies": 0.550000011920929, "rewards/chosen": -2.046875, "rewards/margins": 0.302734375, "rewards/rejected": -2.34375, "step": 610 }, { "epoch": 0.6488749345892203, "grad_norm": 1.794412593376868, "learning_rate": 4.796968003560639e-05, "logits/chosen": -11.25, "logits/rejected": -11.0, "logps/chosen": -1680.0, "logps/rejected": -1808.0, "loss": 0.631, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -2.265625, "rewards/margins": 0.412109375, "rewards/rejected": -2.6875, "step": 620 }, { "epoch": 0.6593406593406593, "grad_norm": 2.344223960828674, "learning_rate": 4.7847714344635864e-05, "logits/chosen": -13.125, "logits/rejected": -12.8125, "logps/chosen": -1544.0, "logps/rejected": -1792.0, "loss": 0.6074, "rewards/accuracies": 0.6312500238418579, "rewards/chosen": -2.5625, "rewards/margins": 0.40625, "rewards/rejected": -2.96875, "step": 630 }, { "epoch": 0.6698063840920984, "grad_norm": 2.034838473863574, "learning_rate": 4.772235575174776e-05, "logits/chosen": -12.75, "logits/rejected": -12.0, "logps/chosen": -1464.0, "logps/rejected": -1744.0, "loss": 0.5941, "rewards/accuracies": 0.6875, "rewards/chosen": -2.6875, "rewards/margins": 0.56640625, "rewards/rejected": -3.25, "step": 640 }, { "epoch": 0.6802721088435374, "grad_norm": 2.0690171505599055, "learning_rate": 4.7593622872782754e-05, "logits/chosen": -11.75, "logits/rejected": -10.9375, "logps/chosen": -1632.0, "logps/rejected": -1856.0, "loss": 0.6173, "rewards/accuracies": 0.625, "rewards/chosen": -2.671875, "rewards/margins": 0.470703125, "rewards/rejected": -3.140625, "step": 650 }, { "epoch": 0.6907378335949764, "grad_norm": 2.090337781153315, "learning_rate": 4.746153482466542e-05, "logits/chosen": -11.625, "logits/rejected": -10.4375, "logps/chosen": -1520.0, "logps/rejected": -1840.0, "loss": 0.616, "rewards/accuracies": 0.6875, "rewards/chosen": -2.515625, "rewards/margins": 0.62890625, "rewards/rejected": -3.140625, "step": 660 }, { "epoch": 0.7012035583464155, "grad_norm": 1.9364855120000886, "learning_rate": 4.732611122256537e-05, "logits/chosen": -10.75, "logits/rejected": -9.5625, "logps/chosen": -1576.0, "logps/rejected": -1928.0, "loss": 0.647, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -2.484375, "rewards/margins": 0.455078125, "rewards/rejected": -2.9375, "step": 670 }, { "epoch": 0.7116692830978545, "grad_norm": 1.9269271151799625, "learning_rate": 4.718737217698435e-05, "logits/chosen": -8.5625, "logits/rejected": -8.1875, "logps/chosen": -1680.0, "logps/rejected": -1832.0, "loss": 0.6462, "rewards/accuracies": 0.637499988079071, "rewards/chosen": -2.53125, "rewards/margins": 0.375, "rewards/rejected": -2.90625, "step": 680 }, { "epoch": 0.7221350078492935, "grad_norm": 2.217639479483015, "learning_rate": 4.704533829076982e-05, "logits/chosen": -8.5625, "logits/rejected": -8.25, "logps/chosen": -1496.0, "logps/rejected": -1664.0, "loss": 0.6448, "rewards/accuracies": 0.65625, "rewards/chosen": -2.40625, "rewards/margins": 0.421875, "rewards/rejected": -2.828125, "step": 690 }, { "epoch": 0.7326007326007326, "grad_norm": 2.0803595530806263, "learning_rate": 4.6900030656055455e-05, "logits/chosen": -9.8125, "logits/rejected": -9.3125, "logps/chosen": -1664.0, "logps/rejected": -1816.0, "loss": 0.6131, "rewards/accuracies": 0.6875, "rewards/chosen": -2.421875, "rewards/margins": 0.4921875, "rewards/rejected": -2.90625, "step": 700 }, { "epoch": 0.7326007326007326, "eval_logits/chosen": -12.625, "eval_logits/rejected": -12.8125, "eval_logps/chosen": -468.0, "eval_logps/rejected": -442.0, "eval_loss": 0.710812509059906, "eval_rewards/accuracies": 0.529296875, "eval_rewards/chosen": -1.4921875, "eval_rewards/margins": 0.03955078125, "eval_rewards/rejected": -1.53125, "eval_runtime": 46.6507, "eval_samples_per_second": 42.872, "eval_steps_per_second": 0.686, "step": 700 }, { "epoch": 0.7430664573521716, "grad_norm": 1.871781980290299, "learning_rate": 4.6751470851128906e-05, "logits/chosen": -11.25, "logits/rejected": -11.0, "logps/chosen": -1472.0, "logps/rejected": -1752.0, "loss": 0.6088, "rewards/accuracies": 0.6625000238418579, "rewards/chosen": -2.234375, "rewards/margins": 0.546875, "rewards/rejected": -2.78125, "step": 710 }, { "epoch": 0.7535321821036107, "grad_norm": 2.8689597128657214, "learning_rate": 4.6599680937227454e-05, "logits/chosen": -11.625, "logits/rejected": -11.25, "logps/chosen": -1568.0, "logps/rejected": -1808.0, "loss": 0.6055, "rewards/accuracies": 0.6625000238418579, "rewards/chosen": -2.796875, "rewards/margins": 0.443359375, "rewards/rejected": -3.25, "step": 720 }, { "epoch": 0.7639979068550498, "grad_norm": 2.262975636115908, "learning_rate": 4.644468345526183e-05, "logits/chosen": -11.1875, "logits/rejected": -9.875, "logps/chosen": -1576.0, "logps/rejected": -1920.0, "loss": 0.6176, "rewards/accuracies": 0.6875, "rewards/chosen": -2.96875, "rewards/margins": 0.6953125, "rewards/rejected": -3.671875, "step": 730 }, { "epoch": 0.7744636316064888, "grad_norm": 1.89103468137331, "learning_rate": 4.628650142246894e-05, "logits/chosen": -10.0625, "logits/rejected": -9.4375, "logps/chosen": -1504.0, "logps/rejected": -1760.0, "loss": 0.6296, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -2.671875, "rewards/margins": 0.439453125, "rewards/rejected": -3.09375, "step": 740 }, { "epoch": 0.7849293563579278, "grad_norm": 2.2462176361305, "learning_rate": 4.6125158328993765e-05, "logits/chosen": -9.6875, "logits/rejected": -9.0, "logps/chosen": -1624.0, "logps/rejected": -1816.0, "loss": 0.638, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -2.734375, "rewards/margins": 0.3125, "rewards/rejected": -3.046875, "step": 750 }, { "epoch": 0.7953950811093669, "grad_norm": 1.8134394471487076, "learning_rate": 4.5960678134401064e-05, "logits/chosen": -11.0, "logits/rejected": -10.375, "logps/chosen": -1432.0, "logps/rejected": -1696.0, "loss": 0.6253, "rewards/accuracies": 0.643750011920929, "rewards/chosen": -2.5, "rewards/margins": 0.52734375, "rewards/rejected": -3.03125, "step": 760 }, { "epoch": 0.8058608058608059, "grad_norm": 1.9985679095917177, "learning_rate": 4.579308526411734e-05, "logits/chosen": -9.625, "logits/rejected": -9.25, "logps/chosen": -1584.0, "logps/rejected": -1840.0, "loss": 0.6018, "rewards/accuracies": 0.6499999761581421, "rewards/chosen": -2.578125, "rewards/margins": 0.6171875, "rewards/rejected": -3.1875, "step": 770 }, { "epoch": 0.8163265306122449, "grad_norm": 1.820860212716631, "learning_rate": 4.562240460580367e-05, "logits/chosen": -9.3125, "logits/rejected": -8.5, "logps/chosen": -1664.0, "logps/rejected": -1952.0, "loss": 0.5959, "rewards/accuracies": 0.668749988079071, "rewards/chosen": -2.671875, "rewards/margins": 0.67578125, "rewards/rejected": -3.359375, "step": 780 }, { "epoch": 0.826792255363684, "grad_norm": 1.7669199391713484, "learning_rate": 4.544866150565985e-05, "logits/chosen": -9.8125, "logits/rejected": -9.375, "logps/chosen": -1464.0, "logps/rejected": -1728.0, "loss": 0.6168, "rewards/accuracies": 0.6625000238418579, "rewards/chosen": -2.875, "rewards/margins": 0.46875, "rewards/rejected": -3.34375, "step": 790 }, { "epoch": 0.837257980115123, "grad_norm": 2.5056212632916908, "learning_rate": 4.527188176466051e-05, "logits/chosen": -11.25, "logits/rejected": -10.9375, "logps/chosen": -1592.0, "logps/rejected": -1864.0, "loss": 0.621, "rewards/accuracies": 0.6625000238418579, "rewards/chosen": -2.96875, "rewards/margins": 0.62109375, "rewards/rejected": -3.59375, "step": 800 }, { "epoch": 0.837257980115123, "eval_logits/chosen": -14.0625, "eval_logits/rejected": -14.0625, "eval_logps/chosen": -454.0, "eval_logps/rejected": -430.0, "eval_loss": 0.7203750014305115, "eval_rewards/accuracies": 0.537109375, "eval_rewards/chosen": -1.3515625, "eval_rewards/margins": 0.05810546875, "eval_rewards/rejected": -1.4140625, "eval_runtime": 46.6455, "eval_samples_per_second": 42.877, "eval_steps_per_second": 0.686, "step": 800 }, { "epoch": 0.847723704866562, "grad_norm": 1.7806877901651768, "learning_rate": 4.509209163472357e-05, "logits/chosen": -11.3125, "logits/rejected": -10.25, "logps/chosen": -1624.0, "logps/rejected": -1992.0, "loss": 0.5963, "rewards/accuracies": 0.643750011920929, "rewards/chosen": -2.765625, "rewards/margins": 0.6953125, "rewards/rejected": -3.46875, "step": 810 }, { "epoch": 0.858189429618001, "grad_norm": 1.9813522991221886, "learning_rate": 4.4909317814811896e-05, "logits/chosen": -10.875, "logits/rejected": -10.0, "logps/chosen": -1552.0, "logps/rejected": -1856.0, "loss": 0.6248, "rewards/accuracies": 0.625, "rewards/chosen": -2.71875, "rewards/margins": 0.63671875, "rewards/rejected": -3.359375, "step": 820 }, { "epoch": 0.8686551543694401, "grad_norm": 1.8816514373034843, "learning_rate": 4.472358744696845e-05, "logits/chosen": -10.6875, "logits/rejected": -10.25, "logps/chosen": -1552.0, "logps/rejected": -1728.0, "loss": 0.6083, "rewards/accuracies": 0.637499988079071, "rewards/chosen": -2.75, "rewards/margins": 0.353515625, "rewards/rejected": -3.109375, "step": 830 }, { "epoch": 0.8791208791208791, "grad_norm": 1.8302637422477892, "learning_rate": 4.4534928112285645e-05, "logits/chosen": -10.875, "logits/rejected": -9.9375, "logps/chosen": -1488.0, "logps/rejected": -1808.0, "loss": 0.5904, "rewards/accuracies": 0.65625, "rewards/chosen": -2.453125, "rewards/margins": 0.5859375, "rewards/rejected": -3.046875, "step": 840 }, { "epoch": 0.8895866038723181, "grad_norm": 1.8022496944979107, "learning_rate": 4.43433678268096e-05, "logits/chosen": -10.0, "logits/rejected": -9.5625, "logps/chosen": -1640.0, "logps/rejected": -1752.0, "loss": 0.6253, "rewards/accuracies": 0.6187499761581421, "rewards/chosen": -2.59375, "rewards/margins": 0.423828125, "rewards/rejected": -3.015625, "step": 850 }, { "epoch": 0.9000523286237572, "grad_norm": 1.6864470191138066, "learning_rate": 4.414893503737968e-05, "logits/chosen": -11.0625, "logits/rejected": -10.5, "logps/chosen": -1472.0, "logps/rejected": -1744.0, "loss": 0.6304, "rewards/accuracies": 0.6187499761581421, "rewards/chosen": -2.40625, "rewards/margins": 0.451171875, "rewards/rejected": -2.84375, "step": 860 }, { "epoch": 0.9105180533751962, "grad_norm": 1.7673964276302028, "learning_rate": 4.395165861740414e-05, "logits/chosen": -10.75, "logits/rejected": -10.25, "logps/chosen": -1584.0, "logps/rejected": -1832.0, "loss": 0.6312, "rewards/accuracies": 0.612500011920929, "rewards/chosen": -2.640625, "rewards/margins": 0.51953125, "rewards/rejected": -3.15625, "step": 870 }, { "epoch": 0.9209837781266352, "grad_norm": 1.8205454151369895, "learning_rate": 4.375156786257241e-05, "logits/chosen": -9.9375, "logits/rejected": -9.0625, "logps/chosen": -1768.0, "logps/rejected": -2008.0, "loss": 0.6084, "rewards/accuracies": 0.699999988079071, "rewards/chosen": -2.75, "rewards/margins": 0.5234375, "rewards/rejected": -3.28125, "step": 880 }, { "epoch": 0.9314495028780743, "grad_norm": 2.006840129140987, "learning_rate": 4.354869248650466e-05, "logits/chosen": -10.5, "logits/rejected": -9.8125, "logps/chosen": -1584.0, "logps/rejected": -1840.0, "loss": 0.6034, "rewards/accuracies": 0.65625, "rewards/chosen": -2.9375, "rewards/margins": 0.59765625, "rewards/rejected": -3.546875, "step": 890 }, { "epoch": 0.9419152276295133, "grad_norm": 2.000234656261563, "learning_rate": 4.334306261633931e-05, "logits/chosen": -11.0, "logits/rejected": -10.3125, "logps/chosen": -1624.0, "logps/rejected": -1832.0, "loss": 0.6114, "rewards/accuracies": 0.6875, "rewards/chosen": -2.765625, "rewards/margins": 0.69140625, "rewards/rejected": -3.453125, "step": 900 }, { "epoch": 0.9419152276295133, "eval_logits/chosen": -13.8125, "eval_logits/rejected": -13.875, "eval_logps/chosen": -486.0, "eval_logps/rejected": -470.0, "eval_loss": 0.7060312628746033, "eval_rewards/accuracies": 0.537109375, "eval_rewards/chosen": -1.6796875, "eval_rewards/margins": 0.1328125, "eval_rewards/rejected": -1.8125, "eval_runtime": 46.6683, "eval_samples_per_second": 42.856, "eval_steps_per_second": 0.686, "step": 900 }, { "epoch": 0.9523809523809523, "grad_norm": 1.9040168922590635, "learning_rate": 4.313470878825911e-05, "logits/chosen": -10.875, "logits/rejected": -10.25, "logps/chosen": -1576.0, "logps/rejected": -1784.0, "loss": 0.5795, "rewards/accuracies": 0.65625, "rewards/chosen": -2.671875, "rewards/margins": 0.59375, "rewards/rejected": -3.265625, "step": 910 }, { "epoch": 0.9628466771323915, "grad_norm": 2.3294693137062623, "learning_rate": 4.292366194295652e-05, "logits/chosen": -10.1875, "logits/rejected": -9.4375, "logps/chosen": -1632.0, "logps/rejected": -1968.0, "loss": 0.6226, "rewards/accuracies": 0.643750011920929, "rewards/chosen": -3.0, "rewards/margins": 0.5390625, "rewards/rejected": -3.53125, "step": 920 }, { "epoch": 0.9733124018838305, "grad_norm": 1.7914828642780467, "learning_rate": 4.2709953421038964e-05, "logits/chosen": -10.1875, "logits/rejected": -9.75, "logps/chosen": -1608.0, "logps/rejected": -1872.0, "loss": 0.6189, "rewards/accuracies": 0.65625, "rewards/chosen": -2.828125, "rewards/margins": 0.408203125, "rewards/rejected": -3.234375, "step": 930 }, { "epoch": 0.9837781266352695, "grad_norm": 1.93496993945409, "learning_rate": 4.249361495837478e-05, "logits/chosen": -10.0, "logits/rejected": -9.5625, "logps/chosen": -1760.0, "logps/rejected": -1904.0, "loss": 0.6125, "rewards/accuracies": 0.6625000238418579, "rewards/chosen": -2.84375, "rewards/margins": 0.4765625, "rewards/rejected": -3.328125, "step": 940 }, { "epoch": 0.9942438513867086, "grad_norm": 1.9866972900374376, "learning_rate": 4.227467868138035e-05, "logits/chosen": -11.0, "logits/rejected": -10.5625, "logps/chosen": -1624.0, "logps/rejected": -1848.0, "loss": 0.6243, "rewards/accuracies": 0.625, "rewards/chosen": -2.859375, "rewards/margins": 0.56640625, "rewards/rejected": -3.421875, "step": 950 }, { "epoch": 1.0047095761381475, "grad_norm": 1.0392851662807159, "learning_rate": 4.205317710224938e-05, "logits/chosen": -10.125, "logits/rejected": -9.75, "logps/chosen": -1704.0, "logps/rejected": -2024.0, "loss": 0.4762, "rewards/accuracies": 0.7250000238418579, "rewards/chosen": -2.734375, "rewards/margins": 0.85546875, "rewards/rejected": -3.578125, "step": 960 }, { "epoch": 1.0151753008895865, "grad_norm": 1.419737450699832, "learning_rate": 4.1829143114124733e-05, "logits/chosen": -10.25, "logits/rejected": -10.0, "logps/chosen": -1784.0, "logps/rejected": -2096.0, "loss": 0.2028, "rewards/accuracies": 0.956250011920929, "rewards/chosen": -3.578125, "rewards/margins": 2.25, "rewards/rejected": -5.8125, "step": 970 }, { "epoch": 1.0256410256410255, "grad_norm": 3.7604063706161224, "learning_rate": 4.160260998621387e-05, "logits/chosen": -9.9375, "logits/rejected": -9.25, "logps/chosen": -1736.0, "logps/rejected": -2240.0, "loss": 0.2024, "rewards/accuracies": 0.90625, "rewards/chosen": -4.46875, "rewards/margins": 2.890625, "rewards/rejected": -7.375, "step": 980 }, { "epoch": 1.0361067503924646, "grad_norm": 1.1660050697832864, "learning_rate": 4.137361135884824e-05, "logits/chosen": -9.6875, "logits/rejected": -9.1875, "logps/chosen": -1792.0, "logps/rejected": -2352.0, "loss": 0.1712, "rewards/accuracies": 0.9375, "rewards/chosen": -3.828125, "rewards/margins": 3.125, "rewards/rejected": -6.96875, "step": 990 }, { "epoch": 1.0465724751439036, "grad_norm": 1.3430939142219194, "learning_rate": 4.114218123848777e-05, "logits/chosen": -9.1875, "logits/rejected": -8.4375, "logps/chosen": -1688.0, "logps/rejected": -2256.0, "loss": 0.1659, "rewards/accuracies": 0.9375, "rewards/chosen": -4.375, "rewards/margins": 3.125, "rewards/rejected": -7.5, "step": 1000 }, { "epoch": 1.0465724751439036, "eval_logits/chosen": -8.5625, "eval_logits/rejected": -7.5, "eval_logps/chosen": -616.0, "eval_logps/rejected": -608.0, "eval_loss": 0.8399999737739563, "eval_rewards/accuracies": 0.564453125, "eval_rewards/chosen": -2.96875, "eval_rewards/margins": 0.251953125, "eval_rewards/rejected": -3.21875, "eval_runtime": 46.6514, "eval_samples_per_second": 42.871, "eval_steps_per_second": 0.686, "step": 1000 }, { "epoch": 1.0570381998953426, "grad_norm": 1.3037333574609864, "learning_rate": 4.0908353992670824e-05, "logits/chosen": -7.5625, "logits/rejected": -7.25, "logps/chosen": -1664.0, "logps/rejected": -2096.0, "loss": 0.1935, "rewards/accuracies": 0.8999999761581421, "rewards/chosen": -4.5625, "rewards/margins": 2.828125, "rewards/rejected": -7.375, "step": 1010 }, { "epoch": 1.0675039246467817, "grad_norm": 1.7570398940893637, "learning_rate": 4.067216434491059e-05, "logits/chosen": -9.125, "logits/rejected": -8.625, "logps/chosen": -1600.0, "logps/rejected": -2112.0, "loss": 0.1614, "rewards/accuracies": 0.918749988079071, "rewards/chosen": -4.09375, "rewards/margins": 2.828125, "rewards/rejected": -6.9375, "step": 1020 }, { "epoch": 1.077969649398221, "grad_norm": 1.228788352958514, "learning_rate": 4.043364736953866e-05, "logits/chosen": -10.875, "logits/rejected": -10.0, "logps/chosen": -1992.0, "logps/rejected": -2416.0, "loss": 0.1634, "rewards/accuracies": 0.9437500238418579, "rewards/chosen": -4.625, "rewards/margins": 3.125, "rewards/rejected": -7.75, "step": 1030 }, { "epoch": 1.08843537414966, "grad_norm": 1.8256673613795895, "learning_rate": 4.019283848649643e-05, "logits/chosen": -11.0, "logits/rejected": -10.375, "logps/chosen": -1856.0, "logps/rejected": -2352.0, "loss": 0.173, "rewards/accuracies": 0.8999999761581421, "rewards/chosen": -5.0625, "rewards/margins": 3.5, "rewards/rejected": -8.5625, "step": 1040 }, { "epoch": 1.098901098901099, "grad_norm": 2.931934677552056, "learning_rate": 3.9949773456075215e-05, "logits/chosen": -11.3125, "logits/rejected": -10.6875, "logps/chosen": -1840.0, "logps/rejected": -2272.0, "loss": 0.1806, "rewards/accuracies": 0.949999988079071, "rewards/chosen": -5.15625, "rewards/margins": 3.328125, "rewards/rejected": -8.5, "step": 1050 }, { "epoch": 1.109366823652538, "grad_norm": 1.53460809373633, "learning_rate": 3.970448837360585e-05, "logits/chosen": -11.4375, "logits/rejected": -11.0, "logps/chosen": -1856.0, "logps/rejected": -2224.0, "loss": 0.167, "rewards/accuracies": 0.9312499761581421, "rewards/chosen": -5.0625, "rewards/margins": 3.03125, "rewards/rejected": -8.0625, "step": 1060 }, { "epoch": 1.119832548403977, "grad_norm": 1.1653076593136338, "learning_rate": 3.945701966409846e-05, "logits/chosen": -10.75, "logits/rejected": -9.8125, "logps/chosen": -1776.0, "logps/rejected": -2224.0, "loss": 0.1703, "rewards/accuracies": 0.9125000238418579, "rewards/chosen": -4.5625, "rewards/margins": 2.859375, "rewards/rejected": -7.40625, "step": 1070 }, { "epoch": 1.130298273155416, "grad_norm": 1.692881814156669, "learning_rate": 3.9207404076833374e-05, "logits/chosen": -10.0625, "logits/rejected": -9.1875, "logps/chosen": -1768.0, "logps/rejected": -2144.0, "loss": 0.1566, "rewards/accuracies": 0.9312499761581421, "rewards/chosen": -4.28125, "rewards/margins": 3.296875, "rewards/rejected": -7.5625, "step": 1080 }, { "epoch": 1.1407639979068551, "grad_norm": 2.513894764887396, "learning_rate": 3.895567867990379e-05, "logits/chosen": -8.75, "logits/rejected": -7.59375, "logps/chosen": -1896.0, "logps/rejected": -2272.0, "loss": 0.1749, "rewards/accuracies": 0.9312499761581421, "rewards/chosen": -5.375, "rewards/margins": 3.125, "rewards/rejected": -8.5, "step": 1090 }, { "epoch": 1.1512297226582942, "grad_norm": 1.479027429766319, "learning_rate": 3.870188085471114e-05, "logits/chosen": -9.375, "logits/rejected": -7.9375, "logps/chosen": -1616.0, "logps/rejected": -2272.0, "loss": 0.1767, "rewards/accuracies": 0.9437500238418579, "rewards/chosen": -4.59375, "rewards/margins": 3.203125, "rewards/rejected": -7.8125, "step": 1100 }, { "epoch": 1.1512297226582942, "eval_logits/chosen": -13.0, "eval_logits/rejected": -12.125, "eval_logps/chosen": -624.0, "eval_logps/rejected": -612.0, "eval_loss": 0.9193906188011169, "eval_rewards/accuracies": 0.515625, "eval_rewards/chosen": -3.078125, "eval_rewards/margins": 0.140625, "eval_rewards/rejected": -3.21875, "eval_runtime": 46.6608, "eval_samples_per_second": 42.863, "eval_steps_per_second": 0.686, "step": 1100 }, { "epoch": 1.1616954474097332, "grad_norm": 1.7696927762188595, "learning_rate": 3.844604829041395e-05, "logits/chosen": -9.0625, "logits/rejected": -7.53125, "logps/chosen": -1848.0, "logps/rejected": -2368.0, "loss": 0.1433, "rewards/accuracies": 0.9375, "rewards/chosen": -4.875, "rewards/margins": 3.421875, "rewards/rejected": -8.3125, "step": 1110 }, { "epoch": 1.1721611721611722, "grad_norm": 1.7361011902716506, "learning_rate": 3.818821897833095e-05, "logits/chosen": -9.5625, "logits/rejected": -7.84375, "logps/chosen": -1672.0, "logps/rejected": -2240.0, "loss": 0.1703, "rewards/accuracies": 0.9437500238418579, "rewards/chosen": -4.9375, "rewards/margins": 3.640625, "rewards/rejected": -8.5625, "step": 1120 }, { "epoch": 1.1826268969126112, "grad_norm": 2.035519976021059, "learning_rate": 3.792843120629935e-05, "logits/chosen": -10.375, "logits/rejected": -8.625, "logps/chosen": -1696.0, "logps/rejected": -2480.0, "loss": 0.1543, "rewards/accuracies": 0.9437500238418579, "rewards/chosen": -5.09375, "rewards/margins": 3.765625, "rewards/rejected": -8.875, "step": 1130 }, { "epoch": 1.1930926216640503, "grad_norm": 1.22492585694424, "learning_rate": 3.766672355298907e-05, "logits/chosen": -10.6875, "logits/rejected": -9.6875, "logps/chosen": -1824.0, "logps/rejected": -2352.0, "loss": 0.1693, "rewards/accuracies": 0.925000011920929, "rewards/chosen": -5.0625, "rewards/margins": 3.515625, "rewards/rejected": -8.5625, "step": 1140 }, { "epoch": 1.2035583464154893, "grad_norm": 2.757699499667783, "learning_rate": 3.740313488217373e-05, "logits/chosen": -10.4375, "logits/rejected": -9.5625, "logps/chosen": -1864.0, "logps/rejected": -2336.0, "loss": 0.1688, "rewards/accuracies": 0.90625, "rewards/chosen": -5.03125, "rewards/margins": 3.171875, "rewards/rejected": -8.1875, "step": 1150 }, { "epoch": 1.2140240711669283, "grad_norm": 1.2062041900641236, "learning_rate": 3.7137704336959464e-05, "logits/chosen": -9.375, "logits/rejected": -8.75, "logps/chosen": -1864.0, "logps/rejected": -2352.0, "loss": 0.1759, "rewards/accuracies": 0.9437500238418579, "rewards/chosen": -5.15625, "rewards/margins": 3.171875, "rewards/rejected": -8.3125, "step": 1160 }, { "epoch": 1.2244897959183674, "grad_norm": 1.5261485069167071, "learning_rate": 3.687047133397201e-05, "logits/chosen": -10.0625, "logits/rejected": -9.1875, "logps/chosen": -1976.0, "logps/rejected": -2464.0, "loss": 0.1528, "rewards/accuracies": 0.956250011920929, "rewards/chosen": -5.25, "rewards/margins": 3.890625, "rewards/rejected": -9.125, "step": 1170 }, { "epoch": 1.2349555206698064, "grad_norm": 1.8077591896337852, "learning_rate": 3.6601475557503415e-05, "logits/chosen": -9.5625, "logits/rejected": -8.875, "logps/chosen": -2016.0, "logps/rejected": -2528.0, "loss": 0.1569, "rewards/accuracies": 0.90625, "rewards/chosen": -5.46875, "rewards/margins": 3.390625, "rewards/rejected": -8.875, "step": 1180 }, { "epoch": 1.2454212454212454, "grad_norm": 1.9114367768897904, "learning_rate": 3.633075695361882e-05, "logits/chosen": -9.9375, "logits/rejected": -8.6875, "logps/chosen": -1808.0, "logps/rejected": -2464.0, "loss": 0.1633, "rewards/accuracies": 0.9312499761581421, "rewards/chosen": -5.4375, "rewards/margins": 3.265625, "rewards/rejected": -8.6875, "step": 1190 }, { "epoch": 1.2558869701726845, "grad_norm": 2.148357173111322, "learning_rate": 3.6058355724224477e-05, "logits/chosen": -9.1875, "logits/rejected": -7.8125, "logps/chosen": -1656.0, "logps/rejected": -2128.0, "loss": 0.1574, "rewards/accuracies": 0.925000011920929, "rewards/chosen": -5.46875, "rewards/margins": 3.25, "rewards/rejected": -8.75, "step": 1200 }, { "epoch": 1.2558869701726845, "eval_logits/chosen": -12.75, "eval_logits/rejected": -11.9375, "eval_logps/chosen": -700.0, "eval_logps/rejected": -696.0, "eval_loss": 0.9109687209129333, "eval_rewards/accuracies": 0.533203125, "eval_rewards/chosen": -3.8125, "eval_rewards/margins": 0.271484375, "eval_rewards/rejected": -4.09375, "eval_runtime": 46.6584, "eval_samples_per_second": 42.865, "eval_steps_per_second": 0.686, "step": 1200 }, { "epoch": 1.2663526949241235, "grad_norm": 1.8716072521998313, "learning_rate": 3.5784312321097744e-05, "logits/chosen": -10.125, "logits/rejected": -8.9375, "logps/chosen": -1728.0, "logps/rejected": -2304.0, "loss": 0.1649, "rewards/accuracies": 0.9312499761581421, "rewards/chosen": -5.28125, "rewards/margins": 3.734375, "rewards/rejected": -9.0, "step": 1210 }, { "epoch": 1.2768184196755625, "grad_norm": 1.0872648673267589, "learning_rate": 3.550866743987994e-05, "logits/chosen": -10.625, "logits/rejected": -9.5625, "logps/chosen": -1776.0, "logps/rejected": -2288.0, "loss": 0.1452, "rewards/accuracies": 0.9437500238418579, "rewards/chosen": -4.875, "rewards/margins": 2.9375, "rewards/rejected": -7.78125, "step": 1220 }, { "epoch": 1.2872841444270016, "grad_norm": 1.7848741923493812, "learning_rate": 3.5231462014033024e-05, "logits/chosen": -10.25, "logits/rejected": -9.0, "logps/chosen": -1968.0, "logps/rejected": -2496.0, "loss": 0.1526, "rewards/accuracies": 0.9312499761581421, "rewards/chosen": -5.9375, "rewards/margins": 3.5, "rewards/rejected": -9.4375, "step": 1230 }, { "epoch": 1.2977498691784406, "grad_norm": 1.6951943719280478, "learning_rate": 3.495273720876095e-05, "logits/chosen": -8.875, "logits/rejected": -7.625, "logps/chosen": -1952.0, "logps/rejected": -2544.0, "loss": 0.1673, "rewards/accuracies": 0.918749988079071, "rewards/chosen": -5.59375, "rewards/margins": 3.296875, "rewards/rejected": -8.875, "step": 1240 }, { "epoch": 1.3082155939298796, "grad_norm": 2.269797094847673, "learning_rate": 3.4672534414896616e-05, "logits/chosen": -8.0625, "logits/rejected": -6.5, "logps/chosen": -1704.0, "logps/rejected": -2304.0, "loss": 0.1766, "rewards/accuracies": 0.893750011920929, "rewards/chosen": -4.90625, "rewards/margins": 3.078125, "rewards/rejected": -7.96875, "step": 1250 }, { "epoch": 1.3186813186813187, "grad_norm": 2.1595719385122596, "learning_rate": 3.439089524275526e-05, "logits/chosen": -8.125, "logits/rejected": -7.0, "logps/chosen": -1824.0, "logps/rejected": -2304.0, "loss": 0.1496, "rewards/accuracies": 0.9312499761581421, "rewards/chosen": -5.34375, "rewards/margins": 3.421875, "rewards/rejected": -8.75, "step": 1260 }, { "epoch": 1.3291470434327577, "grad_norm": 1.9791902566967785, "learning_rate": 3.410786151595533e-05, "logits/chosen": -8.625, "logits/rejected": -7.90625, "logps/chosen": -2024.0, "logps/rejected": -2400.0, "loss": 0.1765, "rewards/accuracies": 0.9375, "rewards/chosen": -5.34375, "rewards/margins": 3.1875, "rewards/rejected": -8.5, "step": 1270 }, { "epoch": 1.3396127681841967, "grad_norm": 1.8968978463805661, "learning_rate": 3.382347526520765e-05, "logits/chosen": -8.25, "logits/rejected": -7.46875, "logps/chosen": -2024.0, "logps/rejected": -2464.0, "loss": 0.1539, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -5.03125, "rewards/margins": 3.4375, "rewards/rejected": -8.5, "step": 1280 }, { "epoch": 1.3500784929356358, "grad_norm": 2.086984817568916, "learning_rate": 3.353777872207381e-05, "logits/chosen": -9.1875, "logits/rejected": -7.46875, "logps/chosen": -1848.0, "logps/rejected": -2336.0, "loss": 0.1601, "rewards/accuracies": 0.90625, "rewards/chosen": -4.96875, "rewards/margins": 3.078125, "rewards/rejected": -8.0625, "step": 1290 }, { "epoch": 1.3605442176870748, "grad_norm": 1.9167903868949743, "learning_rate": 3.325081431269475e-05, "logits/chosen": -8.5, "logits/rejected": -7.375, "logps/chosen": -2048.0, "logps/rejected": -2528.0, "loss": 0.1637, "rewards/accuracies": 0.9437500238418579, "rewards/chosen": -5.71875, "rewards/margins": 3.71875, "rewards/rejected": -9.4375, "step": 1300 }, { "epoch": 1.3605442176870748, "eval_logits/chosen": -12.0, "eval_logits/rejected": -11.25, "eval_logps/chosen": -672.0, "eval_logps/rejected": -664.0, "eval_loss": 0.8867655992507935, "eval_rewards/accuracies": 0.541015625, "eval_rewards/chosen": -3.53125, "eval_rewards/margins": 0.2314453125, "eval_rewards/rejected": -3.765625, "eval_runtime": 46.6505, "eval_samples_per_second": 42.872, "eval_steps_per_second": 0.686, "step": 1300 }, { "epoch": 1.3710099424385138, "grad_norm": 1.444023722901135, "learning_rate": 3.296262465149046e-05, "logits/chosen": -8.5625, "logits/rejected": -7.1875, "logps/chosen": -1792.0, "logps/rejected": -2496.0, "loss": 0.1373, "rewards/accuracies": 0.956250011920929, "rewards/chosen": -5.21875, "rewards/margins": 3.4375, "rewards/rejected": -8.6875, "step": 1310 }, { "epoch": 1.3814756671899528, "grad_norm": 2.4099467562123045, "learning_rate": 3.2673252534831687e-05, "logits/chosen": -7.03125, "logits/rejected": -5.90625, "logps/chosen": -1752.0, "logps/rejected": -2336.0, "loss": 0.158, "rewards/accuracies": 0.949999988079071, "rewards/chosen": -5.40625, "rewards/margins": 3.46875, "rewards/rejected": -8.875, "step": 1320 }, { "epoch": 1.3919413919413919, "grad_norm": 1.5100573804112183, "learning_rate": 3.2382740934684696e-05, "logits/chosen": -8.9375, "logits/rejected": -7.53125, "logps/chosen": -1888.0, "logps/rejected": -2400.0, "loss": 0.1635, "rewards/accuracies": 0.9312499761581421, "rewards/chosen": -5.40625, "rewards/margins": 3.578125, "rewards/rejected": -9.0, "step": 1330 }, { "epoch": 1.402407116692831, "grad_norm": 1.7491889122077937, "learning_rate": 3.209113299222982e-05, "logits/chosen": -8.25, "logits/rejected": -7.3125, "logps/chosen": -1920.0, "logps/rejected": -2432.0, "loss": 0.157, "rewards/accuracies": 0.956250011920929, "rewards/chosen": -5.4375, "rewards/margins": 3.53125, "rewards/rejected": -9.0, "step": 1340 }, { "epoch": 1.41287284144427, "grad_norm": 2.499325892842199, "learning_rate": 3.179847201145505e-05, "logits/chosen": -9.5, "logits/rejected": -8.6875, "logps/chosen": -1888.0, "logps/rejected": -2352.0, "loss": 0.1382, "rewards/accuracies": 0.925000011920929, "rewards/chosen": -5.59375, "rewards/margins": 3.5, "rewards/rejected": -9.125, "step": 1350 }, { "epoch": 1.423338566195709, "grad_norm": 2.052373375201287, "learning_rate": 3.150480145272528e-05, "logits/chosen": -9.125, "logits/rejected": -7.875, "logps/chosen": -1720.0, "logps/rejected": -2272.0, "loss": 0.1815, "rewards/accuracies": 0.90625, "rewards/chosen": -5.34375, "rewards/margins": 3.53125, "rewards/rejected": -8.875, "step": 1360 }, { "epoch": 1.433804290947148, "grad_norm": 1.4122293217034483, "learning_rate": 3.121016492632848e-05, "logits/chosen": -8.375, "logits/rejected": -7.125, "logps/chosen": -1920.0, "logps/rejected": -2496.0, "loss": 0.1494, "rewards/accuracies": 0.9437500238418579, "rewards/chosen": -5.0625, "rewards/margins": 3.46875, "rewards/rejected": -8.5, "step": 1370 }, { "epoch": 1.4442700156985873, "grad_norm": 1.5646886970951084, "learning_rate": 3.091460618599951e-05, "logits/chosen": -7.40625, "logits/rejected": -5.625, "logps/chosen": -1848.0, "logps/rejected": -2480.0, "loss": 0.1462, "rewards/accuracies": 0.925000011920929, "rewards/chosen": -5.4375, "rewards/margins": 3.671875, "rewards/rejected": -9.0625, "step": 1380 }, { "epoch": 1.454735740450026, "grad_norm": 1.8520081795198564, "learning_rate": 3.061816912242265e-05, "logits/chosen": -6.4375, "logits/rejected": -5.09375, "logps/chosen": -2016.0, "logps/rejected": -2544.0, "loss": 0.1542, "rewards/accuracies": 0.9375, "rewards/chosen": -6.1875, "rewards/margins": 3.453125, "rewards/rejected": -9.625, "step": 1390 }, { "epoch": 1.4652014652014653, "grad_norm": 1.727202818903328, "learning_rate": 3.0320897756713785e-05, "logits/chosen": -6.03125, "logits/rejected": -4.65625, "logps/chosen": -1896.0, "logps/rejected": -2560.0, "loss": 0.1275, "rewards/accuracies": 0.956250011920929, "rewards/chosen": -5.75, "rewards/margins": 3.875, "rewards/rejected": -9.625, "step": 1400 }, { "epoch": 1.4652014652014653, "eval_logits/chosen": -10.1875, "eval_logits/rejected": -9.0625, "eval_logps/chosen": -688.0, "eval_logps/rejected": -688.0, "eval_loss": 0.927609384059906, "eval_rewards/accuracies": 0.548828125, "eval_rewards/chosen": -3.703125, "eval_rewards/margins": 0.275390625, "eval_rewards/rejected": -3.984375, "eval_runtime": 46.6472, "eval_samples_per_second": 42.875, "eval_steps_per_second": 0.686, "step": 1400 }, { "epoch": 1.4756671899529041, "grad_norm": 1.8570824877894925, "learning_rate": 3.002283623388332e-05, "logits/chosen": -8.3125, "logits/rejected": -6.1875, "logps/chosen": -1888.0, "logps/rejected": -2384.0, "loss": 0.1442, "rewards/accuracies": 0.949999988079071, "rewards/chosen": -5.75, "rewards/margins": 3.625, "rewards/rejected": -9.375, "step": 1410 }, { "epoch": 1.4861329147043434, "grad_norm": 1.491134246319967, "learning_rate": 2.9724028816280507e-05, "logits/chosen": -8.125, "logits/rejected": -7.0625, "logps/chosen": -1848.0, "logps/rejected": -2272.0, "loss": 0.1538, "rewards/accuracies": 0.9437500238418579, "rewards/chosen": -5.25, "rewards/margins": 3.578125, "rewards/rejected": -8.8125, "step": 1420 }, { "epoch": 1.4965986394557822, "grad_norm": 1.8563146725820714, "learning_rate": 2.9424519877020524e-05, "logits/chosen": -7.21875, "logits/rejected": -5.71875, "logps/chosen": -1976.0, "logps/rejected": -2608.0, "loss": 0.1465, "rewards/accuracies": 0.949999988079071, "rewards/chosen": -5.28125, "rewards/margins": 3.5, "rewards/rejected": -8.8125, "step": 1430 }, { "epoch": 1.5070643642072215, "grad_norm": 2.775833223747887, "learning_rate": 2.912435389339504e-05, "logits/chosen": -6.25, "logits/rejected": -5.90625, "logps/chosen": -2016.0, "logps/rejected": -2464.0, "loss": 0.1688, "rewards/accuracies": 0.9312499761581421, "rewards/chosen": -5.6875, "rewards/margins": 3.796875, "rewards/rejected": -9.5, "step": 1440 }, { "epoch": 1.5175300889586603, "grad_norm": 1.6126717998677436, "learning_rate": 2.8823575440267226e-05, "logits/chosen": -7.75, "logits/rejected": -6.875, "logps/chosen": -1896.0, "logps/rejected": -2336.0, "loss": 0.1402, "rewards/accuracies": 0.9312499761581421, "rewards/chosen": -5.625, "rewards/margins": 3.390625, "rewards/rejected": -9.0, "step": 1450 }, { "epoch": 1.5279958137100995, "grad_norm": 1.5579791522841617, "learning_rate": 2.8522229183452465e-05, "logits/chosen": -8.0, "logits/rejected": -6.59375, "logps/chosen": -1936.0, "logps/rejected": -2448.0, "loss": 0.1382, "rewards/accuracies": 0.9437500238418579, "rewards/chosen": -5.8125, "rewards/margins": 3.65625, "rewards/rejected": -9.5, "step": 1460 }, { "epoch": 1.5384615384615383, "grad_norm": 1.8102118738336272, "learning_rate": 2.8220359873085373e-05, "logits/chosen": -6.875, "logits/rejected": -6.21875, "logps/chosen": -2112.0, "logps/rejected": -2480.0, "loss": 0.1315, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -6.34375, "rewards/margins": 3.59375, "rewards/rejected": -9.9375, "step": 1470 }, { "epoch": 1.5489272632129776, "grad_norm": 2.372475145978383, "learning_rate": 2.7918012336974388e-05, "logits/chosen": -7.5, "logits/rejected": -6.65625, "logps/chosen": -1832.0, "logps/rejected": -2432.0, "loss": 0.1564, "rewards/accuracies": 0.8999999761581421, "rewards/chosen": -5.84375, "rewards/margins": 3.5, "rewards/rejected": -9.375, "step": 1480 }, { "epoch": 1.5593929879644164, "grad_norm": 1.995093127899613, "learning_rate": 2.7615231473944815e-05, "logits/chosen": -7.5, "logits/rejected": -6.46875, "logps/chosen": -1880.0, "logps/rejected": -2432.0, "loss": 0.1899, "rewards/accuracies": 0.887499988079071, "rewards/chosen": -5.59375, "rewards/margins": 3.421875, "rewards/rejected": -9.0, "step": 1490 }, { "epoch": 1.5698587127158556, "grad_norm": 2.3726902084951096, "learning_rate": 2.7312062247171327e-05, "logits/chosen": -8.75, "logits/rejected": -7.1875, "logps/chosen": -1840.0, "logps/rejected": -2464.0, "loss": 0.1468, "rewards/accuracies": 0.9312499761581421, "rewards/chosen": -5.28125, "rewards/margins": 3.453125, "rewards/rejected": -8.75, "step": 1500 }, { "epoch": 1.5698587127158556, "eval_logits/chosen": -11.3125, "eval_logits/rejected": -10.6875, "eval_logps/chosen": -716.0, "eval_logps/rejected": -704.0, "eval_loss": 0.916796863079071, "eval_rewards/accuracies": 0.53515625, "eval_rewards/chosen": -3.96875, "eval_rewards/margins": 0.1943359375, "eval_rewards/rejected": -4.15625, "eval_runtime": 46.6617, "eval_samples_per_second": 42.862, "eval_steps_per_second": 0.686, "step": 1500 }, { "epoch": 1.5803244374672945, "grad_norm": 1.4779827738026858, "learning_rate": 2.7008549677500876e-05, "logits/chosen": -8.6875, "logits/rejected": -7.5, "logps/chosen": -1960.0, "logps/rejected": -2432.0, "loss": 0.1654, "rewards/accuracies": 0.918749988079071, "rewards/chosen": -5.90625, "rewards/margins": 3.515625, "rewards/rejected": -9.4375, "step": 1510 }, { "epoch": 1.5907901622187337, "grad_norm": 2.3734820512470716, "learning_rate": 2.670473883676709e-05, "logits/chosen": -8.3125, "logits/rejected": -7.375, "logps/chosen": -2008.0, "logps/rejected": -2512.0, "loss": 0.1399, "rewards/accuracies": 0.9437500238418579, "rewards/chosen": -6.25, "rewards/margins": 3.578125, "rewards/rejected": -9.8125, "step": 1520 }, { "epoch": 1.6012558869701727, "grad_norm": 1.501354809251434, "learning_rate": 2.6400674841097074e-05, "logits/chosen": -7.6875, "logits/rejected": -7.09375, "logps/chosen": -2016.0, "logps/rejected": -2400.0, "loss": 0.1568, "rewards/accuracies": 0.9125000238418579, "rewards/chosen": -6.28125, "rewards/margins": 3.53125, "rewards/rejected": -9.8125, "step": 1530 }, { "epoch": 1.6117216117216118, "grad_norm": 1.8495802025154207, "learning_rate": 2.609640284421158e-05, "logits/chosen": -7.875, "logits/rejected": -6.96875, "logps/chosen": -1824.0, "logps/rejected": -2304.0, "loss": 0.1648, "rewards/accuracies": 0.9125000238418579, "rewards/chosen": -5.46875, "rewards/margins": 3.125, "rewards/rejected": -8.5625, "step": 1540 }, { "epoch": 1.6221873364730508, "grad_norm": 1.6145507874404676, "learning_rate": 2.5791968030719726e-05, "logits/chosen": -8.0, "logits/rejected": -7.34375, "logps/chosen": -1888.0, "logps/rejected": -2416.0, "loss": 0.1394, "rewards/accuracies": 0.9437500238418579, "rewards/chosen": -5.71875, "rewards/margins": 3.484375, "rewards/rejected": -9.1875, "step": 1550 }, { "epoch": 1.6326530612244898, "grad_norm": 1.5217456479566145, "learning_rate": 2.548741560940902e-05, "logits/chosen": -8.6875, "logits/rejected": -7.5, "logps/chosen": -1864.0, "logps/rejected": -2416.0, "loss": 0.1309, "rewards/accuracies": 0.9624999761581421, "rewards/chosen": -6.0, "rewards/margins": 3.609375, "rewards/rejected": -9.625, "step": 1560 }, { "epoch": 1.6431187859759289, "grad_norm": 2.9392584760314504, "learning_rate": 2.5182790806531787e-05, "logits/chosen": -8.4375, "logits/rejected": -7.90625, "logps/chosen": -1784.0, "logps/rejected": -2400.0, "loss": 0.1612, "rewards/accuracies": 0.90625, "rewards/chosen": -5.96875, "rewards/margins": 3.375, "rewards/rejected": -9.375, "step": 1570 }, { "epoch": 1.653584510727368, "grad_norm": 2.344870994163916, "learning_rate": 2.4878138859089074e-05, "logits/chosen": -8.75, "logits/rejected": -7.28125, "logps/chosen": -1864.0, "logps/rejected": -2400.0, "loss": 0.1662, "rewards/accuracies": 0.925000011920929, "rewards/chosen": -5.40625, "rewards/margins": 3.453125, "rewards/rejected": -8.875, "step": 1580 }, { "epoch": 1.664050235478807, "grad_norm": 2.027528024994614, "learning_rate": 2.4573505008112924e-05, "logits/chosen": -8.4375, "logits/rejected": -7.15625, "logps/chosen": -1952.0, "logps/rejected": -2640.0, "loss": 0.1461, "rewards/accuracies": 0.9437500238418579, "rewards/chosen": -5.59375, "rewards/margins": 3.6875, "rewards/rejected": -9.25, "step": 1590 }, { "epoch": 1.674515960230246, "grad_norm": 2.206460642432455, "learning_rate": 2.4268934491948032e-05, "logits/chosen": -7.75, "logits/rejected": -6.75, "logps/chosen": -1984.0, "logps/rejected": -2480.0, "loss": 0.1427, "rewards/accuracies": 0.949999988079071, "rewards/chosen": -5.75, "rewards/margins": 3.453125, "rewards/rejected": -9.1875, "step": 1600 }, { "epoch": 1.674515960230246, "eval_logits/chosen": -11.0, "eval_logits/rejected": -10.125, "eval_logps/chosen": -748.0, "eval_logps/rejected": -744.0, "eval_loss": 0.918749988079071, "eval_rewards/accuracies": 0.5234375, "eval_rewards/chosen": -4.3125, "eval_rewards/margins": 0.265625, "eval_rewards/rejected": -4.5625, "eval_runtime": 46.6637, "eval_samples_per_second": 42.86, "eval_steps_per_second": 0.686, "step": 1600 }, { "epoch": 1.684981684981685, "grad_norm": 2.141187249268674, "learning_rate": 2.396447253953385e-05, "logits/chosen": -8.4375, "logits/rejected": -7.875, "logps/chosen": -2032.0, "logps/rejected": -2528.0, "loss": 0.1641, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -6.0, "rewards/margins": 3.34375, "rewards/rejected": -9.3125, "step": 1610 }, { "epoch": 1.695447409733124, "grad_norm": 2.200030167712362, "learning_rate": 2.3660164363687998e-05, "logits/chosen": -8.625, "logits/rejected": -7.6875, "logps/chosen": -1840.0, "logps/rejected": -2480.0, "loss": 0.1407, "rewards/accuracies": 0.949999988079071, "rewards/chosen": -5.65625, "rewards/margins": 3.578125, "rewards/rejected": -9.25, "step": 1620 }, { "epoch": 1.705913134484563, "grad_norm": 1.9785813017775973, "learning_rate": 2.33560551543922e-05, "logits/chosen": -8.0625, "logits/rejected": -6.875, "logps/chosen": -2000.0, "logps/rejected": -2496.0, "loss": 0.1461, "rewards/accuracies": 0.9375, "rewards/chosen": -5.59375, "rewards/margins": 3.78125, "rewards/rejected": -9.375, "step": 1630 }, { "epoch": 1.716378859236002, "grad_norm": 2.6883789287684783, "learning_rate": 2.3052190072081493e-05, "logits/chosen": -8.9375, "logits/rejected": -8.0625, "logps/chosen": -1968.0, "logps/rejected": -2512.0, "loss": 0.1408, "rewards/accuracies": 0.956250011920929, "rewards/chosen": -5.65625, "rewards/margins": 3.953125, "rewards/rejected": -9.625, "step": 1640 }, { "epoch": 1.7268445839874411, "grad_norm": 2.195949916913737, "learning_rate": 2.2748614240937864e-05, "logits/chosen": -8.125, "logits/rejected": -7.25, "logps/chosen": -2032.0, "logps/rejected": -2432.0, "loss": 0.1424, "rewards/accuracies": 0.949999988079071, "rewards/chosen": -5.9375, "rewards/margins": 3.875, "rewards/rejected": -9.8125, "step": 1650 }, { "epoch": 1.7373103087388801, "grad_norm": 2.1334713005341617, "learning_rate": 2.2445372742189334e-05, "logits/chosen": -7.1875, "logits/rejected": -6.40625, "logps/chosen": -2080.0, "logps/rejected": -2704.0, "loss": 0.1384, "rewards/accuracies": 0.9437500238418579, "rewards/chosen": -6.15625, "rewards/margins": 4.03125, "rewards/rejected": -10.1875, "step": 1660 }, { "epoch": 1.7477760334903192, "grad_norm": 2.6107898700977343, "learning_rate": 2.214251060741528e-05, "logits/chosen": -6.90625, "logits/rejected": -6.125, "logps/chosen": -2112.0, "logps/rejected": -2592.0, "loss": 0.1455, "rewards/accuracies": 0.918749988079071, "rewards/chosen": -6.84375, "rewards/margins": 3.71875, "rewards/rejected": -10.5625, "step": 1670 }, { "epoch": 1.7582417582417582, "grad_norm": 2.6837390130175023, "learning_rate": 2.1840072811859302e-05, "logits/chosen": -8.0625, "logits/rejected": -6.8125, "logps/chosen": -2032.0, "logps/rejected": -2480.0, "loss": 0.1534, "rewards/accuracies": 0.956250011920929, "rewards/chosen": -6.53125, "rewards/margins": 3.375, "rewards/rejected": -9.875, "step": 1680 }, { "epoch": 1.7687074829931972, "grad_norm": 2.3419728714223584, "learning_rate": 2.1538104267750287e-05, "logits/chosen": -8.4375, "logits/rejected": -7.34375, "logps/chosen": -2032.0, "logps/rejected": -2496.0, "loss": 0.1517, "rewards/accuracies": 0.949999988079071, "rewards/chosen": -6.34375, "rewards/margins": 3.5625, "rewards/rejected": -9.9375, "step": 1690 }, { "epoch": 1.7791732077446363, "grad_norm": 1.7655496158648882, "learning_rate": 2.1236649817632952e-05, "logits/chosen": -8.5, "logits/rejected": -7.46875, "logps/chosen": -1920.0, "logps/rejected": -2432.0, "loss": 0.1592, "rewards/accuracies": 0.9312499761581421, "rewards/chosen": -6.34375, "rewards/margins": 3.65625, "rewards/rejected": -10.0, "step": 1700 }, { "epoch": 1.7791732077446363, "eval_logits/chosen": -11.25, "eval_logits/rejected": -10.4375, "eval_logps/chosen": -784.0, "eval_logps/rejected": -792.0, "eval_loss": 0.8701406121253967, "eval_rewards/accuracies": 0.55859375, "eval_rewards/chosen": -4.6875, "eval_rewards/margins": 0.3515625, "eval_rewards/rejected": -5.03125, "eval_runtime": 46.6557, "eval_samples_per_second": 42.867, "eval_steps_per_second": 0.686, "step": 1700 }, { "epoch": 1.7896389324960753, "grad_norm": 2.7919028041017344, "learning_rate": 2.093575422770872e-05, "logits/chosen": -8.9375, "logits/rejected": -7.5, "logps/chosen": -1816.0, "logps/rejected": -2368.0, "loss": 0.1348, "rewards/accuracies": 0.9437500238418579, "rewards/chosen": -5.90625, "rewards/margins": 3.390625, "rewards/rejected": -9.3125, "step": 1710 }, { "epoch": 1.8001046572475143, "grad_norm": 2.801014939171505, "learning_rate": 2.0635462181187827e-05, "logits/chosen": -8.0625, "logits/rejected": -7.34375, "logps/chosen": -2024.0, "logps/rejected": -2576.0, "loss": 0.1522, "rewards/accuracies": 0.9437500238418579, "rewards/chosen": -6.375, "rewards/margins": 3.96875, "rewards/rejected": -10.3125, "step": 1720 }, { "epoch": 1.8105703819989536, "grad_norm": 1.8007359083117451, "learning_rate": 2.0335818271653917e-05, "logits/chosen": -8.6875, "logits/rejected": -7.8125, "logps/chosen": -1968.0, "logps/rejected": -2464.0, "loss": 0.1286, "rewards/accuracies": 0.96875, "rewards/chosen": -6.28125, "rewards/margins": 3.734375, "rewards/rejected": -10.0, "step": 1730 }, { "epoch": 1.8210361067503924, "grad_norm": 1.6488574603218902, "learning_rate": 2.0036866996441815e-05, "logits/chosen": -8.375, "logits/rejected": -7.625, "logps/chosen": -2008.0, "logps/rejected": -2496.0, "loss": 0.1459, "rewards/accuracies": 0.96875, "rewards/chosen": -6.3125, "rewards/margins": 3.96875, "rewards/rejected": -10.3125, "step": 1740 }, { "epoch": 1.8315018315018317, "grad_norm": 2.568362027435373, "learning_rate": 1.9738652750029677e-05, "logits/chosen": -8.625, "logits/rejected": -7.1875, "logps/chosen": -1896.0, "logps/rejected": -2464.0, "loss": 0.1283, "rewards/accuracies": 0.96875, "rewards/chosen": -6.09375, "rewards/margins": 3.84375, "rewards/rejected": -9.9375, "step": 1750 }, { "epoch": 1.8419675562532705, "grad_norm": 2.1926267745125596, "learning_rate": 1.944121981744631e-05, "logits/chosen": -8.6875, "logits/rejected": -6.96875, "logps/chosen": -2024.0, "logps/rejected": -2640.0, "loss": 0.1346, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -6.28125, "rewards/margins": 4.3125, "rewards/rejected": -10.5625, "step": 1760 }, { "epoch": 1.8524332810047097, "grad_norm": 1.7219817341962491, "learning_rate": 1.9144612367694887e-05, "logits/chosen": -7.875, "logits/rejected": -6.3125, "logps/chosen": -2048.0, "logps/rejected": -2624.0, "loss": 0.1502, "rewards/accuracies": 0.9624999761581421, "rewards/chosen": -6.625, "rewards/margins": 3.6875, "rewards/rejected": -10.3125, "step": 1770 }, { "epoch": 1.8628990057561485, "grad_norm": 2.421157337081247, "learning_rate": 1.8848874447193803e-05, "logits/chosen": -7.5, "logits/rejected": -6.375, "logps/chosen": -1928.0, "logps/rejected": -2560.0, "loss": 0.1414, "rewards/accuracies": 0.949999988079071, "rewards/chosen": -6.0625, "rewards/margins": 3.75, "rewards/rejected": -9.8125, "step": 1780 }, { "epoch": 1.8733647305075878, "grad_norm": 2.227890887251112, "learning_rate": 1.855404997323571e-05, "logits/chosen": -7.6875, "logits/rejected": -6.59375, "logps/chosen": -1960.0, "logps/rejected": -2544.0, "loss": 0.1561, "rewards/accuracies": 0.9125000238418579, "rewards/chosen": -6.28125, "rewards/margins": 3.40625, "rewards/rejected": -9.6875, "step": 1790 }, { "epoch": 1.8838304552590266, "grad_norm": 2.250974301169281, "learning_rate": 1.82601827274658e-05, "logits/chosen": -7.53125, "logits/rejected": -6.5625, "logps/chosen": -1904.0, "logps/rejected": -2448.0, "loss": 0.1341, "rewards/accuracies": 0.9437500238418579, "rewards/chosen": -5.8125, "rewards/margins": 3.6875, "rewards/rejected": -9.5, "step": 1800 }, { "epoch": 1.8838304552590266, "eval_logits/chosen": -10.5625, "eval_logits/rejected": -9.625, "eval_logps/chosen": -712.0, "eval_logps/rejected": -708.0, "eval_loss": 0.9226093888282776, "eval_rewards/accuracies": 0.5390625, "eval_rewards/chosen": -3.953125, "eval_rewards/margins": 0.259765625, "eval_rewards/rejected": -4.21875, "eval_runtime": 46.6473, "eval_samples_per_second": 42.875, "eval_steps_per_second": 0.686, "step": 1800 }, { "epoch": 1.8942961800104658, "grad_norm": 1.75982085773068, "learning_rate": 1.7967316349380213e-05, "logits/chosen": -7.90625, "logits/rejected": -6.75, "logps/chosen": -1888.0, "logps/rejected": -2608.0, "loss": 0.141, "rewards/accuracies": 0.918749988079071, "rewards/chosen": -5.96875, "rewards/margins": 3.96875, "rewards/rejected": -9.9375, "step": 1810 }, { "epoch": 1.9047619047619047, "grad_norm": 1.620338116722854, "learning_rate": 1.7675494329845516e-05, "logits/chosen": -6.84375, "logits/rejected": -6.0625, "logps/chosen": -2112.0, "logps/rejected": -2528.0, "loss": 0.1235, "rewards/accuracies": 0.96875, "rewards/chosen": -5.84375, "rewards/margins": 3.53125, "rewards/rejected": -9.375, "step": 1820 }, { "epoch": 1.915227629513344, "grad_norm": 1.3312582782740559, "learning_rate": 1.738476000464026e-05, "logits/chosen": -7.65625, "logits/rejected": -6.78125, "logps/chosen": -1960.0, "logps/rejected": -2464.0, "loss": 0.1384, "rewards/accuracies": 0.956250011920929, "rewards/chosen": -5.78125, "rewards/margins": 3.328125, "rewards/rejected": -9.125, "step": 1830 }, { "epoch": 1.9256933542647827, "grad_norm": 2.7193423755357036, "learning_rate": 1.7095156548019646e-05, "logits/chosen": -7.90625, "logits/rejected": -6.625, "logps/chosen": -1880.0, "logps/rejected": -2528.0, "loss": 0.1474, "rewards/accuracies": 0.9437500238418579, "rewards/chosen": -5.78125, "rewards/margins": 3.609375, "rewards/rejected": -9.375, "step": 1840 }, { "epoch": 1.936159079016222, "grad_norm": 1.549630007618533, "learning_rate": 1.6806726966304064e-05, "logits/chosen": -8.3125, "logits/rejected": -7.25, "logps/chosen": -1776.0, "logps/rejected": -2288.0, "loss": 0.1328, "rewards/accuracies": 0.9312499761581421, "rewards/chosen": -5.375, "rewards/margins": 3.734375, "rewards/rejected": -9.125, "step": 1850 }, { "epoch": 1.9466248037676608, "grad_norm": 2.6228707192847165, "learning_rate": 1.6519514091492625e-05, "logits/chosen": -7.96875, "logits/rejected": -6.25, "logps/chosen": -1848.0, "logps/rejected": -2480.0, "loss": 0.1405, "rewards/accuracies": 0.9312499761581421, "rewards/chosen": -5.625, "rewards/margins": 3.8125, "rewards/rejected": -9.4375, "step": 1860 }, { "epoch": 1.9570905285191, "grad_norm": 2.0521613289538174, "learning_rate": 1.6233560574902627e-05, "logits/chosen": -7.96875, "logits/rejected": -6.875, "logps/chosen": -1888.0, "logps/rejected": -2368.0, "loss": 0.1294, "rewards/accuracies": 0.918749988079071, "rewards/chosen": -5.75, "rewards/margins": 3.53125, "rewards/rejected": -9.25, "step": 1870 }, { "epoch": 1.9675562532705388, "grad_norm": 1.910323793391485, "learning_rate": 1.5948908880835753e-05, "logits/chosen": -7.78125, "logits/rejected": -6.65625, "logps/chosen": -1960.0, "logps/rejected": -2480.0, "loss": 0.1518, "rewards/accuracies": 0.918749988079071, "rewards/chosen": -6.0, "rewards/margins": 3.734375, "rewards/rejected": -9.75, "step": 1880 }, { "epoch": 1.978021978021978, "grad_norm": 2.2761965166521, "learning_rate": 1.5665601280272124e-05, "logits/chosen": -8.1875, "logits/rejected": -6.96875, "logps/chosen": -1856.0, "logps/rejected": -2432.0, "loss": 0.1562, "rewards/accuracies": 0.925000011920929, "rewards/chosen": -5.75, "rewards/margins": 3.78125, "rewards/rejected": -9.5625, "step": 1890 }, { "epoch": 1.988487702773417, "grad_norm": 1.6844059195966268, "learning_rate": 1.538367984459298e-05, "logits/chosen": -7.1875, "logits/rejected": -6.1875, "logps/chosen": -1872.0, "logps/rejected": -2368.0, "loss": 0.1366, "rewards/accuracies": 0.949999988079071, "rewards/chosen": -5.96875, "rewards/margins": 3.546875, "rewards/rejected": -9.5, "step": 1900 }, { "epoch": 1.988487702773417, "eval_logits/chosen": -10.75, "eval_logits/rejected": -9.9375, "eval_logps/chosen": -736.0, "eval_logps/rejected": -732.0, "eval_loss": 0.9103437662124634, "eval_rewards/accuracies": 0.5234375, "eval_rewards/chosen": -4.15625, "eval_rewards/margins": 0.275390625, "eval_rewards/rejected": -4.4375, "eval_runtime": 46.6756, "eval_samples_per_second": 42.849, "eval_steps_per_second": 0.686, "step": 1900 }, { "epoch": 1.9989534275248562, "grad_norm": 1.8633417882516954, "learning_rate": 1.5103186439333133e-05, "logits/chosen": -7.34375, "logits/rejected": -5.75, "logps/chosen": -2000.0, "logps/rejected": -2656.0, "loss": 0.1451, "rewards/accuracies": 0.9624999761581421, "rewards/chosen": -5.875, "rewards/margins": 4.09375, "rewards/rejected": -9.9375, "step": 1910 }, { "epoch": 2.009419152276295, "grad_norm": 0.277682644847787, "learning_rate": 1.4824162717963829e-05, "logits/chosen": -6.78125, "logits/rejected": -5.9375, "logps/chosen": -1824.0, "logps/rejected": -2576.0, "loss": 0.0242, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -5.5625, "rewards/margins": 5.46875, "rewards/rejected": -11.0625, "step": 1920 }, { "epoch": 2.0198848770277342, "grad_norm": 0.4025019584640195, "learning_rate": 1.4546650115707192e-05, "logits/chosen": -6.71875, "logits/rejected": -5.25, "logps/chosen": -1856.0, "logps/rejected": -2736.0, "loss": 0.024, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -5.8125, "rewards/margins": 6.09375, "rewards/rejected": -11.875, "step": 1930 }, { "epoch": 2.030350601779173, "grad_norm": 0.24375097319910244, "learning_rate": 1.4270689843383111e-05, "logits/chosen": -6.15625, "logits/rejected": -5.0, "logps/chosen": -2040.0, "logps/rejected": -2976.0, "loss": 0.0285, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -6.71875, "rewards/margins": 7.03125, "rewards/rejected": -13.75, "step": 1940 }, { "epoch": 2.0408163265306123, "grad_norm": 0.2933940125304641, "learning_rate": 1.3996322881289347e-05, "logits/chosen": -5.46875, "logits/rejected": -3.65625, "logps/chosen": -2032.0, "logps/rejected": -2832.0, "loss": 0.0213, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -7.4375, "rewards/margins": 6.65625, "rewards/rejected": -14.125, "step": 1950 }, { "epoch": 2.051282051282051, "grad_norm": 0.23534363751244602, "learning_rate": 1.3723589973115963e-05, "logits/chosen": -6.09375, "logits/rejected": -4.65625, "logps/chosen": -2208.0, "logps/rejected": -3104.0, "loss": 0.0253, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -7.6875, "rewards/margins": 7.09375, "rewards/rejected": -14.75, "step": 1960 }, { "epoch": 2.0617477760334904, "grad_norm": 0.16742778643742806, "learning_rate": 1.3452531619894837e-05, "logits/chosen": -6.125, "logits/rejected": -4.9375, "logps/chosen": -2080.0, "logps/rejected": -2912.0, "loss": 0.0235, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -7.40625, "rewards/margins": 6.625, "rewards/rejected": -14.0, "step": 1970 }, { "epoch": 2.072213500784929, "grad_norm": 0.17466031624269704, "learning_rate": 1.3183188073985245e-05, "logits/chosen": -5.96875, "logits/rejected": -4.25, "logps/chosen": -2064.0, "logps/rejected": -3040.0, "loss": 0.019, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -7.6875, "rewards/margins": 6.75, "rewards/rejected": -14.4375, "step": 1980 }, { "epoch": 2.0826792255363684, "grad_norm": 0.6485065382632995, "learning_rate": 1.2915599333096351e-05, "logits/chosen": -5.625, "logits/rejected": -4.0625, "logps/chosen": -2096.0, "logps/rejected": -2944.0, "loss": 0.0196, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -7.5, "rewards/margins": 6.6875, "rewards/rejected": -14.1875, "step": 1990 }, { "epoch": 2.0931449502878072, "grad_norm": 1.50160790436125, "learning_rate": 1.264980513434752e-05, "logits/chosen": -6.28125, "logits/rejected": -4.46875, "logps/chosen": -2080.0, "logps/rejected": -2848.0, "loss": 0.026, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -7.75, "rewards/margins": 6.46875, "rewards/rejected": -14.1875, "step": 2000 }, { "epoch": 2.0931449502878072, "eval_logits/chosen": -7.53125, "eval_logits/rejected": -6.5, "eval_logps/chosen": -896.0, "eval_logps/rejected": -900.0, "eval_loss": 1.0973438024520874, "eval_rewards/accuracies": 0.525390625, "eval_rewards/chosen": -5.78125, "eval_rewards/margins": 0.337890625, "eval_rewards/rejected": -6.125, "eval_runtime": 46.6578, "eval_samples_per_second": 42.865, "eval_steps_per_second": 0.686, "step": 2000 }, { "epoch": 2.1036106750392465, "grad_norm": 0.2624217994198531, "learning_rate": 1.2385844948367324e-05, "logits/chosen": -6.1875, "logits/rejected": -4.28125, "logps/chosen": -2016.0, "logps/rejected": -2976.0, "loss": 0.0166, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -7.625, "rewards/margins": 6.6875, "rewards/rejected": -14.3125, "step": 2010 }, { "epoch": 2.1140763997906853, "grad_norm": 0.24705345947451288, "learning_rate": 1.2123757973432114e-05, "logits/chosen": -5.75, "logits/rejected": -4.90625, "logps/chosen": -2096.0, "logps/rejected": -2880.0, "loss": 0.0208, "rewards/accuracies": 1.0, "rewards/chosen": -7.5625, "rewards/margins": 6.84375, "rewards/rejected": -14.4375, "step": 2020 }, { "epoch": 2.1245421245421245, "grad_norm": 0.3868941507936348, "learning_rate": 1.1863583129645106e-05, "logits/chosen": -5.53125, "logits/rejected": -4.4375, "logps/chosen": -2160.0, "logps/rejected": -3024.0, "loss": 0.0197, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -8.125, "rewards/margins": 6.875, "rewards/rejected": -15.0, "step": 2030 }, { "epoch": 2.1350078492935634, "grad_norm": 0.5650704373445229, "learning_rate": 1.1605359053156607e-05, "logits/chosen": -5.03125, "logits/rejected": -3.625, "logps/chosen": -2096.0, "logps/rejected": -3104.0, "loss": 0.0148, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -7.9375, "rewards/margins": 7.1875, "rewards/rejected": -15.125, "step": 2040 }, { "epoch": 2.1454735740450026, "grad_norm": 0.3029992870153881, "learning_rate": 1.1349124090426644e-05, "logits/chosen": -5.125, "logits/rejected": -3.8125, "logps/chosen": -1944.0, "logps/rejected": -2752.0, "loss": 0.0233, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -7.6875, "rewards/margins": 6.25, "rewards/rejected": -13.9375, "step": 2050 }, { "epoch": 2.155939298796442, "grad_norm": 0.4295373627647284, "learning_rate": 1.1094916292530406e-05, "logits/chosen": -5.71875, "logits/rejected": -4.375, "logps/chosen": -2192.0, "logps/rejected": -3136.0, "loss": 0.0182, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -8.375, "rewards/margins": 6.9375, "rewards/rejected": -15.3125, "step": 2060 }, { "epoch": 2.1664050235478807, "grad_norm": 0.14958016863050283, "learning_rate": 1.0842773409507623e-05, "logits/chosen": -5.625, "logits/rejected": -4.1875, "logps/chosen": -2112.0, "logps/rejected": -3120.0, "loss": 0.0114, "rewards/accuracies": 1.0, "rewards/chosen": -8.25, "rewards/margins": 7.375, "rewards/rejected": -15.625, "step": 2070 }, { "epoch": 2.17687074829932, "grad_norm": 0.14157603693757284, "learning_rate": 1.0592732884756754e-05, "logits/chosen": -5.03125, "logits/rejected": -3.984375, "logps/chosen": -2240.0, "logps/rejected": -3008.0, "loss": 0.0154, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -8.4375, "rewards/margins": 7.03125, "rewards/rejected": -15.4375, "step": 2080 }, { "epoch": 2.1873364730507587, "grad_norm": 0.288850597594533, "learning_rate": 1.0344831849474507e-05, "logits/chosen": -5.96875, "logits/rejected": -4.3125, "logps/chosen": -2112.0, "logps/rejected": -2960.0, "loss": 0.0242, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -8.25, "rewards/margins": 6.8125, "rewards/rejected": -15.0, "step": 2090 }, { "epoch": 2.197802197802198, "grad_norm": 0.2546861375121904, "learning_rate": 1.0099107117141879e-05, "logits/chosen": -5.6875, "logits/rejected": -4.875, "logps/chosen": -2144.0, "logps/rejected": -3088.0, "loss": 0.0178, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -8.1875, "rewards/margins": 7.03125, "rewards/rejected": -15.25, "step": 2100 }, { "epoch": 2.197802197802198, "eval_logits/chosen": -7.28125, "eval_logits/rejected": -6.21875, "eval_logps/chosen": -924.0, "eval_logps/rejected": -932.0, "eval_loss": 1.170328140258789, "eval_rewards/accuracies": 0.529296875, "eval_rewards/chosen": -6.03125, "eval_rewards/margins": 0.38671875, "eval_rewards/rejected": -6.4375, "eval_runtime": 46.6718, "eval_samples_per_second": 42.852, "eval_steps_per_second": 0.686, "step": 2100 }, { "epoch": 2.208267922553637, "grad_norm": 0.2843205319655988, "learning_rate": 9.855595178057334e-06, "logits/chosen": -6.8125, "logits/rejected": -4.9375, "logps/chosen": -2000.0, "logps/rejected": -2976.0, "loss": 0.0184, "rewards/accuracies": 1.0, "rewards/chosen": -7.5, "rewards/margins": 7.28125, "rewards/rejected": -14.75, "step": 2110 }, { "epoch": 2.218733647305076, "grad_norm": 0.15832772511189233, "learning_rate": 9.614332193917949e-06, "logits/chosen": -6.0625, "logits/rejected": -4.03125, "logps/chosen": -1912.0, "logps/rejected": -2912.0, "loss": 0.019, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -7.8125, "rewards/margins": 7.21875, "rewards/rejected": -15.0, "step": 2120 }, { "epoch": 2.229199372056515, "grad_norm": 0.4346185218919547, "learning_rate": 9.375353992449385e-06, "logits/chosen": -5.125, "logits/rejected": -3.671875, "logps/chosen": -2032.0, "logps/rejected": -2960.0, "loss": 0.0186, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -8.1875, "rewards/margins": 7.46875, "rewards/rejected": -15.6875, "step": 2130 }, { "epoch": 2.239665096807954, "grad_norm": 0.8115871820751419, "learning_rate": 9.138696062085442e-06, "logits/chosen": -5.375, "logits/rejected": -4.28125, "logps/chosen": -2432.0, "logps/rejected": -3264.0, "loss": 0.0226, "rewards/accuracies": 1.0, "rewards/chosen": -8.9375, "rewards/margins": 7.40625, "rewards/rejected": -16.375, "step": 2140 }, { "epoch": 2.250130821559393, "grad_norm": 0.10399775571156004, "learning_rate": 8.904393546698007e-06, "logits/chosen": -4.40625, "logits/rejected": -3.1875, "logps/chosen": -2144.0, "logps/rejected": -3120.0, "loss": 0.0169, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -8.375, "rewards/margins": 7.46875, "rewards/rejected": -15.8125, "step": 2150 }, { "epoch": 2.260596546310832, "grad_norm": 0.5412846873701266, "learning_rate": 8.672481240378142e-06, "logits/chosen": -4.5, "logits/rejected": -3.34375, "logps/chosen": -2192.0, "logps/rejected": -2976.0, "loss": 0.02, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -8.5, "rewards/margins": 6.6875, "rewards/rejected": -15.1875, "step": 2160 }, { "epoch": 2.271062271062271, "grad_norm": 1.524454352589602, "learning_rate": 8.44299358226919e-06, "logits/chosen": -4.78125, "logits/rejected": -3.75, "logps/chosen": -2304.0, "logps/rejected": -3184.0, "loss": 0.0126, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -8.375, "rewards/margins": 7.59375, "rewards/rejected": -16.0, "step": 2170 }, { "epoch": 2.2815279958137102, "grad_norm": 0.6201299271322791, "learning_rate": 8.215964651452457e-06, "logits/chosen": -5.5, "logits/rejected": -3.734375, "logps/chosen": -2000.0, "logps/rejected": -3040.0, "loss": 0.02, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -8.0, "rewards/margins": 7.1875, "rewards/rejected": -15.1875, "step": 2180 }, { "epoch": 2.291993720565149, "grad_norm": 0.38900663168461175, "learning_rate": 7.991428161886502e-06, "logits/chosen": -6.0, "logits/rejected": -4.09375, "logps/chosen": -2128.0, "logps/rejected": -3088.0, "loss": 0.0131, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -8.125, "rewards/margins": 7.28125, "rewards/rejected": -15.375, "step": 2190 }, { "epoch": 2.3024594453165883, "grad_norm": 0.7863109935866738, "learning_rate": 7.76941745740061e-06, "logits/chosen": -5.0625, "logits/rejected": -3.953125, "logps/chosen": -2128.0, "logps/rejected": -3008.0, "loss": 0.019, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -8.25, "rewards/margins": 6.78125, "rewards/rejected": -15.0625, "step": 2200 }, { "epoch": 2.3024594453165883, "eval_logits/chosen": -6.96875, "eval_logits/rejected": -5.90625, "eval_logps/chosen": -960.0, "eval_logps/rejected": -968.0, "eval_loss": 1.1799999475479126, "eval_rewards/accuracies": 0.53125, "eval_rewards/chosen": -6.40625, "eval_rewards/margins": 0.400390625, "eval_rewards/rejected": -6.8125, "eval_runtime": 46.6644, "eval_samples_per_second": 42.859, "eval_steps_per_second": 0.686, "step": 2200 }, { "epoch": 2.312925170068027, "grad_norm": 0.3844258774787843, "learning_rate": 7.549965506743123e-06, "logits/chosen": -6.0625, "logits/rejected": -4.28125, "logps/chosen": -2096.0, "logps/rejected": -3008.0, "loss": 0.0208, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -8.125, "rewards/margins": 7.0, "rewards/rejected": -15.1875, "step": 2210 }, { "epoch": 2.3233908948194664, "grad_norm": 0.33688786266363263, "learning_rate": 7.333104898685631e-06, "logits/chosen": -6.09375, "logits/rejected": -5.0, "logps/chosen": -2096.0, "logps/rejected": -2960.0, "loss": 0.014, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -8.375, "rewards/margins": 7.25, "rewards/rejected": -15.625, "step": 2220 }, { "epoch": 2.333856619570905, "grad_norm": 0.2239022121248555, "learning_rate": 7.118867837183518e-06, "logits/chosen": -5.59375, "logits/rejected": -4.59375, "logps/chosen": -2208.0, "logps/rejected": -3104.0, "loss": 0.0181, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -8.6875, "rewards/margins": 7.375, "rewards/rejected": -16.0, "step": 2230 }, { "epoch": 2.3443223443223444, "grad_norm": 0.17472468591321672, "learning_rate": 6.9072861365936045e-06, "logits/chosen": -6.3125, "logits/rejected": -4.15625, "logps/chosen": -2096.0, "logps/rejected": -3152.0, "loss": 0.0148, "rewards/accuracies": 1.0, "rewards/chosen": -8.75, "rewards/margins": 7.09375, "rewards/rejected": -15.875, "step": 2240 }, { "epoch": 2.3547880690737832, "grad_norm": 0.21709004882693772, "learning_rate": 6.6983912169497005e-06, "logits/chosen": -5.125, "logits/rejected": -3.90625, "logps/chosen": -1872.0, "logps/rejected": -2832.0, "loss": 0.0182, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -7.78125, "rewards/margins": 7.375, "rewards/rejected": -15.1875, "step": 2250 }, { "epoch": 2.3652537938252225, "grad_norm": 1.3384505153222697, "learning_rate": 6.49221409929677e-06, "logits/chosen": -4.375, "logits/rejected": -3.4375, "logps/chosen": -2272.0, "logps/rejected": -3184.0, "loss": 0.0174, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -9.1875, "rewards/margins": 7.0, "rewards/rejected": -16.25, "step": 2260 }, { "epoch": 2.3757195185766613, "grad_norm": 1.0443969915643345, "learning_rate": 6.2887854010842075e-06, "logits/chosen": -4.75, "logits/rejected": -3.375, "logps/chosen": -2192.0, "logps/rejected": -3120.0, "loss": 0.0196, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -8.8125, "rewards/margins": 7.4375, "rewards/rejected": -16.25, "step": 2270 }, { "epoch": 2.3861852433281006, "grad_norm": 0.2606980513774009, "learning_rate": 6.0881353316191385e-06, "logits/chosen": -5.46875, "logits/rejected": -4.15625, "logps/chosen": -2288.0, "logps/rejected": -3296.0, "loss": 0.0199, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -8.8125, "rewards/margins": 7.375, "rewards/rejected": -16.25, "step": 2280 }, { "epoch": 2.3966509680795394, "grad_norm": 0.2820149419508757, "learning_rate": 5.890293687580381e-06, "logits/chosen": -5.125, "logits/rejected": -4.1875, "logps/chosen": -2256.0, "logps/rejected": -3120.0, "loss": 0.0156, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -8.625, "rewards/margins": 7.15625, "rewards/rejected": -15.75, "step": 2290 }, { "epoch": 2.4071166928309786, "grad_norm": 0.24830425434669212, "learning_rate": 5.695289848593532e-06, "logits/chosen": -5.34375, "logits/rejected": -3.78125, "logps/chosen": -2128.0, "logps/rejected": -3040.0, "loss": 0.0173, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -7.96875, "rewards/margins": 7.9375, "rewards/rejected": -15.9375, "step": 2300 }, { "epoch": 2.4071166928309786, "eval_logits/chosen": -6.78125, "eval_logits/rejected": -5.71875, "eval_logps/chosen": -952.0, "eval_logps/rejected": -964.0, "eval_loss": 1.18931245803833, "eval_rewards/accuracies": 0.529296875, "eval_rewards/chosen": -6.34375, "eval_rewards/margins": 0.396484375, "eval_rewards/rejected": -6.75, "eval_runtime": 46.6717, "eval_samples_per_second": 42.852, "eval_steps_per_second": 0.686, "step": 2300 }, { "epoch": 2.4175824175824174, "grad_norm": 0.36896517836434706, "learning_rate": 5.50315277286812e-06, "logits/chosen": -6.15625, "logits/rejected": -4.34375, "logps/chosen": -2176.0, "logps/rejected": -3104.0, "loss": 0.0153, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -8.625, "rewards/margins": 7.125, "rewards/rejected": -15.75, "step": 2310 }, { "epoch": 2.4280481423338567, "grad_norm": 0.2410560292541919, "learning_rate": 5.313910992897281e-06, "logits/chosen": -5.40625, "logits/rejected": -4.125, "logps/chosen": -2160.0, "logps/rejected": -3056.0, "loss": 0.0163, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -8.5625, "rewards/margins": 7.40625, "rewards/rejected": -16.0, "step": 2320 }, { "epoch": 2.4385138670852955, "grad_norm": 0.43003763779389487, "learning_rate": 5.127592611220658e-06, "logits/chosen": -5.4375, "logits/rejected": -3.8125, "logps/chosen": -2224.0, "logps/rejected": -3248.0, "loss": 0.0171, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -8.75, "rewards/margins": 7.3125, "rewards/rejected": -16.125, "step": 2330 }, { "epoch": 2.4489795918367347, "grad_norm": 0.39282158644605636, "learning_rate": 4.944225296251159e-06, "logits/chosen": -5.34375, "logits/rejected": -4.09375, "logps/chosen": -2096.0, "logps/rejected": -2928.0, "loss": 0.0188, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -8.5, "rewards/margins": 7.09375, "rewards/rejected": -15.5625, "step": 2340 }, { "epoch": 2.4594453165881736, "grad_norm": 0.4230485555587435, "learning_rate": 4.763836278166172e-06, "logits/chosen": -4.90625, "logits/rejected": -4.09375, "logps/chosen": -2288.0, "logps/rejected": -3232.0, "loss": 0.0169, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -8.6875, "rewards/margins": 7.59375, "rewards/rejected": -16.25, "step": 2350 }, { "epoch": 2.469911041339613, "grad_norm": 0.7788132432473915, "learning_rate": 4.586452344863876e-06, "logits/chosen": -5.71875, "logits/rejected": -4.03125, "logps/chosen": -2112.0, "logps/rejected": -3008.0, "loss": 0.018, "rewards/accuracies": 1.0, "rewards/chosen": -8.5, "rewards/margins": 7.34375, "rewards/rejected": -15.8125, "step": 2360 }, { "epoch": 2.4803767660910516, "grad_norm": 1.2571577025728566, "learning_rate": 4.412099837985192e-06, "logits/chosen": -5.5, "logits/rejected": -4.0625, "logps/chosen": -2096.0, "logps/rejected": -3024.0, "loss": 0.0205, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -8.3125, "rewards/margins": 7.375, "rewards/rejected": -15.6875, "step": 2370 }, { "epoch": 2.490842490842491, "grad_norm": 1.603989333762598, "learning_rate": 4.240804649002089e-06, "logits/chosen": -4.875, "logits/rejected": -3.609375, "logps/chosen": -2304.0, "logps/rejected": -3264.0, "loss": 0.0213, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -8.6875, "rewards/margins": 7.5, "rewards/rejected": -16.125, "step": 2380 }, { "epoch": 2.50130821559393, "grad_norm": 0.13185286903148416, "learning_rate": 4.072592215372598e-06, "logits/chosen": -5.78125, "logits/rejected": -4.0, "logps/chosen": -2144.0, "logps/rejected": -3104.0, "loss": 0.0108, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -8.5, "rewards/margins": 7.625, "rewards/rejected": -16.125, "step": 2390 }, { "epoch": 2.511773940345369, "grad_norm": 0.269116610412395, "learning_rate": 3.907487516763389e-06, "logits/chosen": -5.59375, "logits/rejected": -3.734375, "logps/chosen": -2096.0, "logps/rejected": -3088.0, "loss": 0.0147, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -8.625, "rewards/margins": 7.25, "rewards/rejected": -15.875, "step": 2400 }, { "epoch": 2.511773940345369, "eval_logits/chosen": -6.46875, "eval_logits/rejected": -5.375, "eval_logps/chosen": -992.0, "eval_logps/rejected": -1000.0, "eval_loss": 1.2634999752044678, "eval_rewards/accuracies": 0.517578125, "eval_rewards/chosen": -6.71875, "eval_rewards/margins": 0.392578125, "eval_rewards/rejected": -7.125, "eval_runtime": 46.6596, "eval_samples_per_second": 42.864, "eval_steps_per_second": 0.686, "step": 2400 }, { "epoch": 2.5222396650968077, "grad_norm": 0.7311570362718807, "learning_rate": 3.745515071340286e-06, "logits/chosen": -5.03125, "logits/rejected": -3.921875, "logps/chosen": -2304.0, "logps/rejected": -3184.0, "loss": 0.0208, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -9.0625, "rewards/margins": 7.3125, "rewards/rejected": -16.375, "step": 2410 }, { "epoch": 2.532705389848247, "grad_norm": 0.5026685654349168, "learning_rate": 3.586698932127236e-06, "logits/chosen": -5.21875, "logits/rejected": -4.0625, "logps/chosen": -2304.0, "logps/rejected": -3168.0, "loss": 0.0244, "rewards/accuracies": 0.956250011920929, "rewards/chosen": -9.0625, "rewards/margins": 7.0625, "rewards/rejected": -16.125, "step": 2420 }, { "epoch": 2.5431711145996863, "grad_norm": 0.43835435762035, "learning_rate": 3.4310626834344744e-06, "logits/chosen": -5.375, "logits/rejected": -4.0625, "logps/chosen": -2176.0, "logps/rejected": -3088.0, "loss": 0.0171, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -9.0, "rewards/margins": 7.125, "rewards/rejected": -16.125, "step": 2430 }, { "epoch": 2.553636839351125, "grad_norm": 0.18314346237540968, "learning_rate": 3.2786294373562343e-06, "logits/chosen": -5.5, "logits/rejected": -3.828125, "logps/chosen": -2040.0, "logps/rejected": -2880.0, "loss": 0.0174, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -8.375, "rewards/margins": 6.90625, "rewards/rejected": -15.3125, "step": 2440 }, { "epoch": 2.564102564102564, "grad_norm": 0.15766734523419065, "learning_rate": 3.1294218303385526e-06, "logits/chosen": -4.90625, "logits/rejected": -4.0, "logps/chosen": -2224.0, "logps/rejected": -3040.0, "loss": 0.0184, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -8.4375, "rewards/margins": 7.15625, "rewards/rejected": -15.625, "step": 2450 }, { "epoch": 2.574568288854003, "grad_norm": 0.22388255505679452, "learning_rate": 2.9834620198177483e-06, "logits/chosen": -5.5625, "logits/rejected": -3.3125, "logps/chosen": -2176.0, "logps/rejected": -3184.0, "loss": 0.0111, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -8.75, "rewards/margins": 7.84375, "rewards/rejected": -16.5, "step": 2460 }, { "epoch": 2.5850340136054424, "grad_norm": 0.34921891566518637, "learning_rate": 2.840771680930068e-06, "logits/chosen": -5.0625, "logits/rejected": -3.703125, "logps/chosen": -2160.0, "logps/rejected": -3104.0, "loss": 0.0119, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -8.9375, "rewards/margins": 7.375, "rewards/rejected": -16.25, "step": 2470 }, { "epoch": 2.595499738356881, "grad_norm": 1.2565658028627182, "learning_rate": 2.701372003292868e-06, "logits/chosen": -4.375, "logits/rejected": -3.234375, "logps/chosen": -2512.0, "logps/rejected": -3392.0, "loss": 0.0121, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -9.5625, "rewards/margins": 7.625, "rewards/rejected": -17.25, "step": 2480 }, { "epoch": 2.60596546310832, "grad_norm": 0.13368716069850076, "learning_rate": 2.56528368785795e-06, "logits/chosen": -5.21875, "logits/rejected": -3.125, "logps/chosen": -1976.0, "logps/rejected": -2976.0, "loss": 0.0169, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -8.375, "rewards/margins": 7.3125, "rewards/rejected": -15.75, "step": 2490 }, { "epoch": 2.6164311878597593, "grad_norm": 0.3777786705357668, "learning_rate": 2.432526943837471e-06, "logits/chosen": -5.375, "logits/rejected": -3.953125, "logps/chosen": -2064.0, "logps/rejected": -3008.0, "loss": 0.016, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -8.5625, "rewards/margins": 7.3125, "rewards/rejected": -15.875, "step": 2500 }, { "epoch": 2.6164311878597593, "eval_logits/chosen": -6.40625, "eval_logits/rejected": -5.3125, "eval_logps/chosen": -992.0, "eval_logps/rejected": -1000.0, "eval_loss": 1.2629061937332153, "eval_rewards/accuracies": 0.51953125, "eval_rewards/chosen": -6.75, "eval_rewards/margins": 0.375, "eval_rewards/rejected": -7.125, "eval_runtime": 46.6688, "eval_samples_per_second": 42.855, "eval_steps_per_second": 0.686, "step": 2500 }, { "epoch": 2.6268969126111985, "grad_norm": 0.26117429629497846, "learning_rate": 2.3031214857028446e-06, "logits/chosen": -4.9375, "logits/rejected": -3.890625, "logps/chosen": -2288.0, "logps/rejected": -3200.0, "loss": 0.0115, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -9.1875, "rewards/margins": 7.46875, "rewards/rejected": -16.625, "step": 2510 }, { "epoch": 2.6373626373626373, "grad_norm": 0.1801618018699949, "learning_rate": 2.177086530257139e-06, "logits/chosen": -4.53125, "logits/rejected": -3.390625, "logps/chosen": -2256.0, "logps/rejected": -3216.0, "loss": 0.0227, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -9.125, "rewards/margins": 6.75, "rewards/rejected": -15.875, "step": 2520 }, { "epoch": 2.647828362114076, "grad_norm": 0.44696511965577107, "learning_rate": 2.054440793781354e-06, "logits/chosen": -4.78125, "logits/rejected": -3.59375, "logps/chosen": -2192.0, "logps/rejected": -2992.0, "loss": 0.0157, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -8.75, "rewards/margins": 7.25, "rewards/rejected": -16.0, "step": 2530 }, { "epoch": 2.6582940868655154, "grad_norm": 0.5252010330824208, "learning_rate": 1.9352024892550587e-06, "logits/chosen": -5.0, "logits/rejected": -4.0625, "logps/chosen": -2304.0, "logps/rejected": -3152.0, "loss": 0.0125, "rewards/accuracies": 1.0, "rewards/chosen": -9.1875, "rewards/margins": 7.75, "rewards/rejected": -16.875, "step": 2540 }, { "epoch": 2.6687598116169546, "grad_norm": 0.1677008538363281, "learning_rate": 1.8193893236517379e-06, "logits/chosen": -6.0, "logits/rejected": -4.5625, "logps/chosen": -2224.0, "logps/rejected": -3056.0, "loss": 0.0156, "rewards/accuracies": 1.0, "rewards/chosen": -8.8125, "rewards/margins": 7.3125, "rewards/rejected": -16.125, "step": 2550 }, { "epoch": 2.6792255363683934, "grad_norm": 0.2296587628865339, "learning_rate": 1.7070184953092956e-06, "logits/chosen": -5.40625, "logits/rejected": -4.21875, "logps/chosen": -2160.0, "logps/rejected": -3024.0, "loss": 0.0142, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -8.4375, "rewards/margins": 7.46875, "rewards/rejected": -15.875, "step": 2560 }, { "epoch": 2.6896912611198327, "grad_norm": 0.9918769788992808, "learning_rate": 1.5981066913760918e-06, "logits/chosen": -4.9375, "logits/rejected": -3.6875, "logps/chosen": -2288.0, "logps/rejected": -3248.0, "loss": 0.0169, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -9.0, "rewards/margins": 7.3125, "rewards/rejected": -16.25, "step": 2570 }, { "epoch": 2.7001569858712715, "grad_norm": 0.46849644344748476, "learning_rate": 1.492670085332884e-06, "logits/chosen": -5.28125, "logits/rejected": -3.1875, "logps/chosen": -2128.0, "logps/rejected": -3184.0, "loss": 0.0192, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -8.875, "rewards/margins": 7.53125, "rewards/rejected": -16.375, "step": 2580 }, { "epoch": 2.7106227106227108, "grad_norm": 0.5312446323303043, "learning_rate": 1.3907243345910787e-06, "logits/chosen": -4.875, "logits/rejected": -3.90625, "logps/chosen": -2192.0, "logps/rejected": -3120.0, "loss": 0.0181, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -8.75, "rewards/margins": 7.6875, "rewards/rejected": -16.375, "step": 2590 }, { "epoch": 2.7210884353741496, "grad_norm": 0.3316576735424589, "learning_rate": 1.292284578167538e-06, "logits/chosen": -5.3125, "logits/rejected": -3.6875, "logps/chosen": -2160.0, "logps/rejected": -3200.0, "loss": 0.0171, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -9.0625, "rewards/margins": 7.625, "rewards/rejected": -16.75, "step": 2600 }, { "epoch": 2.7210884353741496, "eval_logits/chosen": -6.21875, "eval_logits/rejected": -5.125, "eval_logps/chosen": -1004.0, "eval_logps/rejected": -1012.0, "eval_loss": 1.271578073501587, "eval_rewards/accuracies": 0.517578125, "eval_rewards/chosen": -6.84375, "eval_rewards/margins": 0.380859375, "eval_rewards/rejected": -7.21875, "eval_runtime": 46.6643, "eval_samples_per_second": 42.859, "eval_steps_per_second": 0.686, "step": 2600 }, { "epoch": 2.731554160125589, "grad_norm": 0.21238047212119857, "learning_rate": 1.1973654344364925e-06, "logits/chosen": -5.09375, "logits/rejected": -4.0, "logps/chosen": -2416.0, "logps/rejected": -3216.0, "loss": 0.0177, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -9.3125, "rewards/margins": 7.6875, "rewards/rejected": -17.0, "step": 2610 }, { "epoch": 2.7420198848770276, "grad_norm": 0.5685677051301709, "learning_rate": 1.1059809989586583e-06, "logits/chosen": -5.40625, "logits/rejected": -4.53125, "logps/chosen": -2208.0, "logps/rejected": -3104.0, "loss": 0.0116, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -8.75, "rewards/margins": 7.65625, "rewards/rejected": -16.375, "step": 2620 }, { "epoch": 2.752485609628467, "grad_norm": 0.39974755890450553, "learning_rate": 1.018144842388044e-06, "logits/chosen": -5.3125, "logits/rejected": -3.796875, "logps/chosen": -2160.0, "logps/rejected": -3136.0, "loss": 0.0153, "rewards/accuracies": 1.0, "rewards/chosen": -8.625, "rewards/margins": 7.84375, "rewards/rejected": -16.5, "step": 2630 }, { "epoch": 2.7629513343799057, "grad_norm": 0.23267087584191581, "learning_rate": 9.338700084567109e-07, "logits/chosen": -4.25, "logits/rejected": -3.609375, "logps/chosen": -2496.0, "logps/rejected": -3248.0, "loss": 0.0176, "rewards/accuracies": 1.0, "rewards/chosen": -9.1875, "rewards/margins": 7.75, "rewards/rejected": -17.0, "step": 2640 }, { "epoch": 2.773417059131345, "grad_norm": 0.15652631355369256, "learning_rate": 8.531690120377606e-07, "logits/chosen": -5.0, "logits/rejected": -3.96875, "logps/chosen": -2384.0, "logps/rejected": -3392.0, "loss": 0.0155, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -9.5625, "rewards/margins": 7.71875, "rewards/rejected": -17.25, "step": 2650 }, { "epoch": 2.7838827838827838, "grad_norm": 0.5862718861853639, "learning_rate": 7.760538372868636e-07, "logits/chosen": -4.6875, "logits/rejected": -3.265625, "logps/chosen": -2272.0, "logps/rejected": -3248.0, "loss": 0.0151, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -8.9375, "rewards/margins": 7.46875, "rewards/rejected": -16.375, "step": 2660 }, { "epoch": 2.794348508634223, "grad_norm": 0.2792289933229418, "learning_rate": 7.025359358626166e-07, "logits/chosen": -5.3125, "logits/rejected": -3.453125, "logps/chosen": -2080.0, "logps/rejected": -3056.0, "loss": 0.0184, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -8.6875, "rewards/margins": 7.28125, "rewards/rejected": -15.9375, "step": 2670 }, { "epoch": 2.804814233385662, "grad_norm": 0.3445176084354875, "learning_rate": 6.326262252259463e-07, "logits/chosen": -5.375, "logits/rejected": -3.75, "logps/chosen": -2208.0, "logps/rejected": -3152.0, "loss": 0.0165, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -8.8125, "rewards/margins": 7.5625, "rewards/rejected": -16.375, "step": 2680 }, { "epoch": 2.815279958137101, "grad_norm": 0.46600789282556987, "learning_rate": 5.66335087018871e-07, "logits/chosen": -5.78125, "logits/rejected": -3.84375, "logps/chosen": -2064.0, "logps/rejected": -3088.0, "loss": 0.0129, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -8.25, "rewards/margins": 7.71875, "rewards/rejected": -15.9375, "step": 2690 }, { "epoch": 2.82574568288854, "grad_norm": 0.6226081849126031, "learning_rate": 5.036723655228226e-07, "logits/chosen": -4.25, "logits/rejected": -3.328125, "logps/chosen": -2304.0, "logps/rejected": -3184.0, "loss": 0.0123, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -9.5, "rewards/margins": 7.21875, "rewards/rejected": -16.75, "step": 2700 }, { "epoch": 2.82574568288854, "eval_logits/chosen": -6.09375, "eval_logits/rejected": -5.0, "eval_logps/chosen": -1008.0, "eval_logps/rejected": -1016.0, "eval_loss": 1.2614531517028809, "eval_rewards/accuracies": 0.51953125, "eval_rewards/chosen": -6.875, "eval_rewards/margins": 0.38671875, "eval_rewards/rejected": -7.25, "eval_runtime": 46.657, "eval_samples_per_second": 42.866, "eval_steps_per_second": 0.686, "step": 2700 }, { "epoch": 2.836211407639979, "grad_norm": 0.243501759920171, "learning_rate": 4.446473661967432e-07, "logits/chosen": -4.78125, "logits/rejected": -4.03125, "logps/chosen": -2352.0, "logps/rejected": -3312.0, "loss": 0.0156, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -9.25, "rewards/margins": 7.84375, "rewards/rejected": -17.125, "step": 2710 }, { "epoch": 2.846677132391418, "grad_norm": 0.19821099525046068, "learning_rate": 3.892688542952411e-07, "logits/chosen": -5.3125, "logits/rejected": -3.828125, "logps/chosen": -2272.0, "logps/rejected": -3248.0, "loss": 0.0141, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -9.1875, "rewards/margins": 7.40625, "rewards/rejected": -16.625, "step": 2720 }, { "epoch": 2.857142857142857, "grad_norm": 0.10986700650386924, "learning_rate": 3.3754505356694e-07, "logits/chosen": -4.90625, "logits/rejected": -2.90625, "logps/chosen": -2144.0, "logps/rejected": -3088.0, "loss": 0.0153, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -8.875, "rewards/margins": 7.25, "rewards/rejected": -16.125, "step": 2730 }, { "epoch": 2.867608581894296, "grad_norm": 0.17902952345643447, "learning_rate": 2.8948364503322276e-07, "logits/chosen": -4.40625, "logits/rejected": -3.640625, "logps/chosen": -2336.0, "logps/rejected": -3264.0, "loss": 0.0148, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -9.1875, "rewards/margins": 7.65625, "rewards/rejected": -16.875, "step": 2740 }, { "epoch": 2.8780743066457353, "grad_norm": 0.19417084814988497, "learning_rate": 2.4509176584762706e-07, "logits/chosen": -4.5, "logits/rejected": -3.125, "logps/chosen": -2208.0, "logps/rejected": -3104.0, "loss": 0.0155, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -8.8125, "rewards/margins": 7.6875, "rewards/rejected": -16.5, "step": 2750 }, { "epoch": 2.8885400313971745, "grad_norm": 1.1753787852031794, "learning_rate": 2.0437600823595694e-07, "logits/chosen": -5.4375, "logits/rejected": -3.625, "logps/chosen": -2272.0, "logps/rejected": -3328.0, "loss": 0.0188, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -9.125, "rewards/margins": 7.625, "rewards/rejected": -16.75, "step": 2760 }, { "epoch": 2.8990057561486133, "grad_norm": 0.4669399867533652, "learning_rate": 1.6734241851733e-07, "logits/chosen": -4.78125, "logits/rejected": -3.75, "logps/chosen": -2304.0, "logps/rejected": -3296.0, "loss": 0.0182, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -9.0, "rewards/margins": 8.1875, "rewards/rejected": -17.25, "step": 2770 }, { "epoch": 2.909471480900052, "grad_norm": 0.26395047075631983, "learning_rate": 1.339964962062984e-07, "logits/chosen": -5.15625, "logits/rejected": -3.734375, "logps/chosen": -2288.0, "logps/rejected": -3136.0, "loss": 0.0142, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -9.0, "rewards/margins": 7.34375, "rewards/rejected": -16.25, "step": 2780 }, { "epoch": 2.9199372056514914, "grad_norm": 0.3821641507267963, "learning_rate": 1.0434319319617136e-07, "logits/chosen": -5.375, "logits/rejected": -3.953125, "logps/chosen": -2368.0, "logps/rejected": -3168.0, "loss": 0.0198, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -9.125, "rewards/margins": 7.65625, "rewards/rejected": -16.75, "step": 2790 }, { "epoch": 2.9304029304029307, "grad_norm": 1.3958374139567116, "learning_rate": 7.838691302365363e-08, "logits/chosen": -4.46875, "logits/rejected": -3.40625, "logps/chosen": -2192.0, "logps/rejected": -3136.0, "loss": 0.0198, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -9.125, "rewards/margins": 7.46875, "rewards/rejected": -16.625, "step": 2800 }, { "epoch": 2.9304029304029307, "eval_logits/chosen": -6.125, "eval_logits/rejected": -5.0, "eval_logps/chosen": -1004.0, "eval_logps/rejected": -1012.0, "eval_loss": 1.2575000524520874, "eval_rewards/accuracies": 0.521484375, "eval_rewards/chosen": -6.84375, "eval_rewards/margins": 0.388671875, "eval_rewards/rejected": -7.25, "eval_runtime": 46.6471, "eval_samples_per_second": 42.875, "eval_steps_per_second": 0.686, "step": 2800 }, { "epoch": 2.9408686551543695, "grad_norm": 0.7297566662329728, "learning_rate": 5.613151021490459e-08, "logits/chosen": -5.375, "logits/rejected": -3.8125, "logps/chosen": -2144.0, "logps/rejected": -3168.0, "loss": 0.0187, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -8.875, "rewards/margins": 7.65625, "rewards/rejected": -16.5, "step": 2810 }, { "epoch": 2.9513343799058083, "grad_norm": 0.29567951105656753, "learning_rate": 3.7580289713159965e-08, "logits/chosen": -4.75, "logits/rejected": -4.09375, "logps/chosen": -2224.0, "logps/rejected": -3072.0, "loss": 0.0141, "rewards/accuracies": 1.0, "rewards/chosen": -8.75, "rewards/margins": 7.9375, "rewards/rejected": -16.75, "step": 2820 }, { "epoch": 2.9618001046572475, "grad_norm": 0.41443918852288303, "learning_rate": 2.2736006387927277e-08, "logits/chosen": -5.40625, "logits/rejected": -3.421875, "logps/chosen": -2160.0, "logps/rejected": -3312.0, "loss": 0.0202, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -9.5, "rewards/margins": 7.875, "rewards/rejected": -17.375, "step": 2830 }, { "epoch": 2.9722658294086868, "grad_norm": 1.2984170367130654, "learning_rate": 1.1600864625893604e-08, "logits/chosen": -4.03125, "logits/rejected": -2.984375, "logps/chosen": -2320.0, "logps/rejected": -3280.0, "loss": 0.0184, "rewards/accuracies": 0.987500011920929, "rewards/chosen": -8.875, "rewards/margins": 7.75, "rewards/rejected": -16.625, "step": 2840 }, { "epoch": 2.9827315541601256, "grad_norm": 0.4467076270260822, "learning_rate": 4.176518003573549e-09, "logits/chosen": -4.25, "logits/rejected": -3.03125, "logps/chosen": -2336.0, "logps/rejected": -3296.0, "loss": 0.0184, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -9.4375, "rewards/margins": 7.59375, "rewards/rejected": -17.0, "step": 2850 }, { "epoch": 2.9931972789115644, "grad_norm": 0.17786436231008027, "learning_rate": 4.640690417528903e-10, "logits/chosen": -4.46875, "logits/rejected": -2.953125, "logps/chosen": -2368.0, "logps/rejected": -3280.0, "loss": 0.0113, "rewards/accuracies": 0.9937499761581421, "rewards/chosen": -9.375, "rewards/margins": 7.4375, "rewards/rejected": -16.75, "step": 2860 }, { "epoch": 2.998430141287284, "step": 2865, "total_flos": 0.0, "train_loss": 0.2668540406601591, "train_runtime": 12367.4943, "train_samples_per_second": 14.826, "train_steps_per_second": 0.232 } ], "logging_steps": 10, "max_steps": 2865, "num_input_tokens_seen": 0, "num_train_epochs": 3, "save_steps": 500, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": false, "should_training_stop": false }, "attributes": {} } }, "total_flos": 0.0, "train_batch_size": 8, "trial_name": null, "trial_params": null }