| { | |
| "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": 215.13811928435538, | |
| "learning_rate": 1.7421602787456448e-07, | |
| "logits/chosen": -11.5, | |
| "logits/rejected": -11.25, | |
| "logps/chosen": -0.99609375, | |
| "logps/rejected": -1.15625, | |
| "loss": 2496.0, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.010465724751439037, | |
| "grad_norm": 282.49171265868245, | |
| "learning_rate": 1.7421602787456445e-06, | |
| "logits/chosen": -10.8125, | |
| "logits/rejected": -10.875, | |
| "logps/chosen": -1.15625, | |
| "logps/rejected": -1.171875, | |
| "loss": 2496.0, | |
| "rewards/accuracies": 0.1458333283662796, | |
| "rewards/chosen": -3.7997961044311523e-06, | |
| "rewards/margins": -3.933906555175781e-06, | |
| "rewards/rejected": 1.387670636177063e-07, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.020931449502878074, | |
| "grad_norm": 268.3862161558683, | |
| "learning_rate": 3.484320557491289e-06, | |
| "logits/chosen": -10.625, | |
| "logits/rejected": -10.4375, | |
| "logps/chosen": -1.171875, | |
| "logps/rejected": -1.28125, | |
| "loss": 2496.0, | |
| "rewards/accuracies": 0.29374998807907104, | |
| "rewards/chosen": -4.649162292480469e-05, | |
| "rewards/margins": 1.4185905456542969e-05, | |
| "rewards/rejected": -6.079673767089844e-05, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.03139717425431711, | |
| "grad_norm": 179.149164839683, | |
| "learning_rate": 5.226480836236934e-06, | |
| "logits/chosen": -11.125, | |
| "logits/rejected": -10.875, | |
| "logps/chosen": -1.0859375, | |
| "logps/rejected": -1.1796875, | |
| "loss": 2496.0, | |
| "rewards/accuracies": 0.5062500238418579, | |
| "rewards/chosen": -6.29425048828125e-05, | |
| "rewards/margins": 1.6689300537109375e-05, | |
| "rewards/rejected": -7.963180541992188e-05, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.04186289900575615, | |
| "grad_norm": 385.104889833728, | |
| "learning_rate": 6.968641114982578e-06, | |
| "logits/chosen": -11.5, | |
| "logits/rejected": -11.5, | |
| "logps/chosen": -1.1640625, | |
| "logps/rejected": -1.3125, | |
| "loss": 2495.4, | |
| "rewards/accuracies": 0.581250011920929, | |
| "rewards/chosen": -0.000530242919921875, | |
| "rewards/margins": 0.00020313262939453125, | |
| "rewards/rejected": -0.000732421875, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.052328623757195186, | |
| "grad_norm": 592.4763184076282, | |
| "learning_rate": 8.710801393728225e-06, | |
| "logits/chosen": -12.5, | |
| "logits/rejected": -12.25, | |
| "logps/chosen": -1.390625, | |
| "logps/rejected": -1.625, | |
| "loss": 2489.2, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": -0.0023651123046875, | |
| "rewards/margins": 0.001617431640625, | |
| "rewards/rejected": -0.003997802734375, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.06279434850863422, | |
| "grad_norm": 556.5033730462756, | |
| "learning_rate": 1.0452961672473868e-05, | |
| "logits/chosen": -11.5, | |
| "logits/rejected": -11.25, | |
| "logps/chosen": -1.734375, | |
| "logps/rejected": -1.8125, | |
| "loss": 2486.2, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.00567626953125, | |
| "rewards/margins": 0.00115203857421875, | |
| "rewards/rejected": -0.0068359375, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.07326007326007326, | |
| "grad_norm": 982.7196234375095, | |
| "learning_rate": 1.2195121951219513e-05, | |
| "logits/chosen": -10.375, | |
| "logits/rejected": -9.875, | |
| "logps/chosen": -1.765625, | |
| "logps/rejected": -2.125, | |
| "loss": 2476.6, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": -0.006927490234375, | |
| "rewards/margins": 0.0020751953125, | |
| "rewards/rejected": -0.00897216796875, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.0837257980115123, | |
| "grad_norm": 1105.7218154015563, | |
| "learning_rate": 1.3937282229965156e-05, | |
| "logits/chosen": -9.375, | |
| "logits/rejected": -9.0, | |
| "logps/chosen": -3.421875, | |
| "logps/rejected": -4.09375, | |
| "loss": 2442.2, | |
| "rewards/accuracies": 0.643750011920929, | |
| "rewards/chosen": -0.0224609375, | |
| "rewards/margins": 0.00640869140625, | |
| "rewards/rejected": -0.02880859375, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.09419152276295134, | |
| "grad_norm": 1565.3972832188163, | |
| "learning_rate": 1.56794425087108e-05, | |
| "logits/chosen": -10.6875, | |
| "logits/rejected": -10.6875, | |
| "logps/chosen": -6.5625, | |
| "logps/rejected": -7.1875, | |
| "loss": 2410.2, | |
| "rewards/accuracies": 0.5562499761581421, | |
| "rewards/chosen": -0.0546875, | |
| "rewards/margins": 0.005615234375, | |
| "rewards/rejected": -0.060302734375, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.10465724751439037, | |
| "grad_norm": 2001.830237669671, | |
| "learning_rate": 1.742160278745645e-05, | |
| "logits/chosen": -11.5, | |
| "logits/rejected": -11.3125, | |
| "logps/chosen": -10.3125, | |
| "logps/rejected": -12.875, | |
| "loss": 2322.6, | |
| "rewards/accuracies": 0.606249988079071, | |
| "rewards/chosen": -0.09130859375, | |
| "rewards/margins": 0.0252685546875, | |
| "rewards/rejected": -0.11669921875, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.10465724751439037, | |
| "eval_logits/chosen": -8.875, | |
| "eval_logits/rejected": -8.375, | |
| "eval_logps/chosen": -13.1875, | |
| "eval_logps/rejected": -15.875, | |
| "eval_loss": 2317.632080078125, | |
| "eval_rewards/accuracies": 0.625, | |
| "eval_rewards/chosen": -0.12109375, | |
| "eval_rewards/margins": 0.0257568359375, | |
| "eval_rewards/rejected": -0.146484375, | |
| "eval_runtime": 47.3716, | |
| "eval_samples_per_second": 42.219, | |
| "eval_steps_per_second": 0.676, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.1151229722658294, | |
| "grad_norm": 5358.817830045398, | |
| "learning_rate": 1.9163763066202093e-05, | |
| "logits/chosen": -9.0625, | |
| "logits/rejected": -8.8125, | |
| "logps/chosen": -14.5625, | |
| "logps/rejected": -17.125, | |
| "loss": 2343.0, | |
| "rewards/accuracies": 0.581250011920929, | |
| "rewards/chosen": -0.134765625, | |
| "rewards/margins": 0.0247802734375, | |
| "rewards/rejected": -0.1591796875, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.12558869701726844, | |
| "grad_norm": 4128.665962688759, | |
| "learning_rate": 2.0905923344947736e-05, | |
| "logits/chosen": -13.25, | |
| "logits/rejected": -13.0625, | |
| "logps/chosen": -17.875, | |
| "logps/rejected": -21.25, | |
| "loss": 2254.8, | |
| "rewards/accuracies": 0.59375, | |
| "rewards/chosen": -0.1669921875, | |
| "rewards/margins": 0.031494140625, | |
| "rewards/rejected": -0.1982421875, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.1360544217687075, | |
| "grad_norm": 4136.6386977272, | |
| "learning_rate": 2.264808362369338e-05, | |
| "logits/chosen": -13.4375, | |
| "logits/rejected": -13.25, | |
| "logps/chosen": -18.625, | |
| "logps/rejected": -24.0, | |
| "loss": 2224.4, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.1748046875, | |
| "rewards/margins": 0.052978515625, | |
| "rewards/rejected": -0.228515625, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.14652014652014653, | |
| "grad_norm": 5293.887021486947, | |
| "learning_rate": 2.4390243902439026e-05, | |
| "logits/chosen": -11.4375, | |
| "logits/rejected": -11.125, | |
| "logps/chosen": -18.875, | |
| "logps/rejected": -23.625, | |
| "loss": 2219.8, | |
| "rewards/accuracies": 0.6937500238418579, | |
| "rewards/chosen": -0.177734375, | |
| "rewards/margins": 0.04638671875, | |
| "rewards/rejected": -0.2236328125, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.15698587127158556, | |
| "grad_norm": 4199.794152859155, | |
| "learning_rate": 2.6132404181184672e-05, | |
| "logits/chosen": -11.0, | |
| "logits/rejected": -10.4375, | |
| "logps/chosen": -19.875, | |
| "logps/rejected": -25.25, | |
| "loss": 2189.5, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": -0.1865234375, | |
| "rewards/margins": 0.054931640625, | |
| "rewards/rejected": -0.2421875, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.1674515960230246, | |
| "grad_norm": 4027.997724074377, | |
| "learning_rate": 2.7874564459930312e-05, | |
| "logits/chosen": -12.75, | |
| "logits/rejected": -12.5, | |
| "logps/chosen": -18.25, | |
| "logps/rejected": -25.0, | |
| "loss": 2117.25, | |
| "rewards/accuracies": 0.65625, | |
| "rewards/chosen": -0.169921875, | |
| "rewards/margins": 0.0673828125, | |
| "rewards/rejected": -0.2373046875, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.17791732077446362, | |
| "grad_norm": 4807.126207344637, | |
| "learning_rate": 2.9616724738675962e-05, | |
| "logits/chosen": -10.75, | |
| "logits/rejected": -10.375, | |
| "logps/chosen": -24.75, | |
| "logps/rejected": -32.5, | |
| "loss": 2108.4, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -0.236328125, | |
| "rewards/margins": 0.07763671875, | |
| "rewards/rejected": -0.314453125, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.18838304552590268, | |
| "grad_norm": 5572.297461201154, | |
| "learning_rate": 3.13588850174216e-05, | |
| "logits/chosen": -6.71875, | |
| "logits/rejected": -6.0625, | |
| "logps/chosen": -28.75, | |
| "logps/rejected": -36.0, | |
| "loss": 2089.3, | |
| "rewards/accuracies": 0.668749988079071, | |
| "rewards/chosen": -0.275390625, | |
| "rewards/margins": 0.0712890625, | |
| "rewards/rejected": -0.34765625, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.1988487702773417, | |
| "grad_norm": 4586.0442038109295, | |
| "learning_rate": 3.310104529616725e-05, | |
| "logits/chosen": -6.65625, | |
| "logits/rejected": -6.25, | |
| "logps/chosen": -26.625, | |
| "logps/rejected": -33.0, | |
| "loss": 2066.3, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.25390625, | |
| "rewards/margins": 0.06298828125, | |
| "rewards/rejected": -0.31640625, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.20931449502878074, | |
| "grad_norm": 6150.086006568706, | |
| "learning_rate": 3.48432055749129e-05, | |
| "logits/chosen": -9.5625, | |
| "logits/rejected": -8.8125, | |
| "logps/chosen": -27.375, | |
| "logps/rejected": -34.25, | |
| "loss": 2118.6, | |
| "rewards/accuracies": 0.6312500238418579, | |
| "rewards/chosen": -0.26171875, | |
| "rewards/margins": 0.0693359375, | |
| "rewards/rejected": -0.330078125, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.20931449502878074, | |
| "eval_logits/chosen": -10.125, | |
| "eval_logits/rejected": -9.75, | |
| "eval_logps/chosen": -30.5, | |
| "eval_logps/rejected": -37.25, | |
| "eval_loss": 2150.97607421875, | |
| "eval_rewards/accuracies": 0.673828125, | |
| "eval_rewards/chosen": -0.29296875, | |
| "eval_rewards/margins": 0.06640625, | |
| "eval_rewards/rejected": -0.359375, | |
| "eval_runtime": 46.6329, | |
| "eval_samples_per_second": 42.888, | |
| "eval_steps_per_second": 0.686, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.21978021978021978, | |
| "grad_norm": 4497.030212594633, | |
| "learning_rate": 3.6585365853658535e-05, | |
| "logits/chosen": -10.9375, | |
| "logits/rejected": -10.5625, | |
| "logps/chosen": -30.0, | |
| "logps/rejected": -35.5, | |
| "loss": 2255.7, | |
| "rewards/accuracies": 0.643750011920929, | |
| "rewards/chosen": -0.2890625, | |
| "rewards/margins": 0.0546875, | |
| "rewards/rejected": -0.34375, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.2302459445316588, | |
| "grad_norm": 4380.229587941468, | |
| "learning_rate": 3.8327526132404185e-05, | |
| "logits/chosen": -10.6875, | |
| "logits/rejected": -10.0, | |
| "logps/chosen": -27.75, | |
| "logps/rejected": -34.75, | |
| "loss": 2138.2, | |
| "rewards/accuracies": 0.643750011920929, | |
| "rewards/chosen": -0.265625, | |
| "rewards/margins": 0.0712890625, | |
| "rewards/rejected": -0.3359375, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.24071166928309787, | |
| "grad_norm": 4540.498400177958, | |
| "learning_rate": 4.006968641114983e-05, | |
| "logits/chosen": -8.5625, | |
| "logits/rejected": -7.875, | |
| "logps/chosen": -28.625, | |
| "logps/rejected": -35.25, | |
| "loss": 2088.25, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.275390625, | |
| "rewards/margins": 0.06640625, | |
| "rewards/rejected": -0.341796875, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.25117739403453687, | |
| "grad_norm": 4299.420665688717, | |
| "learning_rate": 4.181184668989547e-05, | |
| "logits/chosen": -8.8125, | |
| "logits/rejected": -8.375, | |
| "logps/chosen": -29.5, | |
| "logps/rejected": -37.5, | |
| "loss": 2021.4, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.283203125, | |
| "rewards/margins": 0.07861328125, | |
| "rewards/rejected": -0.361328125, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.2616431187859759, | |
| "grad_norm": 6587.738017249876, | |
| "learning_rate": 4.3554006968641115e-05, | |
| "logits/chosen": -10.5625, | |
| "logits/rejected": -10.0625, | |
| "logps/chosen": -26.875, | |
| "logps/rejected": -34.0, | |
| "loss": 2050.0, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.255859375, | |
| "rewards/margins": 0.07080078125, | |
| "rewards/rejected": -0.328125, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.272108843537415, | |
| "grad_norm": 4496.205758738408, | |
| "learning_rate": 4.529616724738676e-05, | |
| "logits/chosen": -8.5625, | |
| "logits/rejected": -7.84375, | |
| "logps/chosen": -30.375, | |
| "logps/rejected": -37.5, | |
| "loss": 2124.3, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": -0.29296875, | |
| "rewards/margins": 0.06982421875, | |
| "rewards/rejected": -0.36328125, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.282574568288854, | |
| "grad_norm": 5080.015991423051, | |
| "learning_rate": 4.703832752613241e-05, | |
| "logits/chosen": -4.9375, | |
| "logits/rejected": -4.09375, | |
| "logps/chosen": -28.875, | |
| "logps/rejected": -35.75, | |
| "loss": 2048.7, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": -0.27734375, | |
| "rewards/margins": 0.068359375, | |
| "rewards/rejected": -0.345703125, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.29304029304029305, | |
| "grad_norm": 7516.788561659381, | |
| "learning_rate": 4.878048780487805e-05, | |
| "logits/chosen": -6.65625, | |
| "logits/rejected": -6.3125, | |
| "logps/chosen": -33.0, | |
| "logps/rejected": -39.5, | |
| "loss": 2132.6, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": -0.3203125, | |
| "rewards/margins": 0.062255859375, | |
| "rewards/rejected": -0.3828125, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.3035060177917321, | |
| "grad_norm": 3838.403617427855, | |
| "learning_rate": 4.999983293481418e-05, | |
| "logits/chosen": -6.6875, | |
| "logits/rejected": -5.78125, | |
| "logps/chosen": -30.75, | |
| "logps/rejected": -38.0, | |
| "loss": 2073.1, | |
| "rewards/accuracies": 0.6937500238418579, | |
| "rewards/chosen": -0.296875, | |
| "rewards/margins": 0.07275390625, | |
| "rewards/rejected": -0.369140625, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.3139717425431711, | |
| "grad_norm": 3882.954254457922, | |
| "learning_rate": 4.9996862949181595e-05, | |
| "logits/chosen": -8.5625, | |
| "logits/rejected": -8.125, | |
| "logps/chosen": -27.5, | |
| "logps/rejected": -33.0, | |
| "loss": 2172.1, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": -0.263671875, | |
| "rewards/margins": 0.055419921875, | |
| "rewards/rejected": -0.318359375, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.3139717425431711, | |
| "eval_logits/chosen": -8.4375, | |
| "eval_logits/rejected": -7.84375, | |
| "eval_logps/chosen": -37.0, | |
| "eval_logps/rejected": -44.0, | |
| "eval_loss": 2062.592041015625, | |
| "eval_rewards/accuracies": 0.689453125, | |
| "eval_rewards/chosen": -0.359375, | |
| "eval_rewards/margins": 0.0673828125, | |
| "eval_rewards/rejected": -0.427734375, | |
| "eval_runtime": 46.6318, | |
| "eval_samples_per_second": 42.889, | |
| "eval_steps_per_second": 0.686, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.32443746729461015, | |
| "grad_norm": 3738.820080222671, | |
| "learning_rate": 4.9990180911526896e-05, | |
| "logits/chosen": -9.4375, | |
| "logits/rejected": -8.25, | |
| "logps/chosen": -35.5, | |
| "logps/rejected": -45.5, | |
| "loss": 2019.2, | |
| "rewards/accuracies": 0.731249988079071, | |
| "rewards/chosen": -0.34375, | |
| "rewards/margins": 0.10107421875, | |
| "rewards/rejected": -0.443359375, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.3349031920460492, | |
| "grad_norm": 4080.4236527935195, | |
| "learning_rate": 4.997978781413745e-05, | |
| "logits/chosen": -6.8125, | |
| "logits/rejected": -6.5, | |
| "logps/chosen": -30.5, | |
| "logps/rejected": -38.5, | |
| "loss": 2085.9, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.29296875, | |
| "rewards/margins": 0.07958984375, | |
| "rewards/rejected": -0.373046875, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.3453689167974882, | |
| "grad_norm": 4209.069959248639, | |
| "learning_rate": 4.996568520039565e-05, | |
| "logits/chosen": -6.8125, | |
| "logits/rejected": -6.40625, | |
| "logps/chosen": -38.25, | |
| "logps/rejected": -45.0, | |
| "loss": 2117.45, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": -0.37109375, | |
| "rewards/margins": 0.06787109375, | |
| "rewards/rejected": -0.4375, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.35583464154892724, | |
| "grad_norm": 4096.816258949507, | |
| "learning_rate": 4.994787516454973e-05, | |
| "logits/chosen": -6.625, | |
| "logits/rejected": -5.875, | |
| "logps/chosen": -34.0, | |
| "logps/rejected": -40.5, | |
| "loss": 2052.6, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": -0.330078125, | |
| "rewards/margins": 0.0625, | |
| "rewards/rejected": -0.392578125, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.3663003663003663, | |
| "grad_norm": 4551.162722660577, | |
| "learning_rate": 4.992636035140273e-05, | |
| "logits/chosen": -7.4375, | |
| "logits/rejected": -6.8125, | |
| "logps/chosen": -33.75, | |
| "logps/rejected": -40.25, | |
| "loss": 2070.85, | |
| "rewards/accuracies": 0.6312500238418579, | |
| "rewards/chosen": -0.326171875, | |
| "rewards/margins": 0.06640625, | |
| "rewards/rejected": -0.390625, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.37676609105180536, | |
| "grad_norm": 4424.717034806246, | |
| "learning_rate": 4.99011439559198e-05, | |
| "logits/chosen": -7.34375, | |
| "logits/rejected": -6.96875, | |
| "logps/chosen": -35.75, | |
| "logps/rejected": -44.5, | |
| "loss": 2005.2, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.345703125, | |
| "rewards/margins": 0.08544921875, | |
| "rewards/rejected": -0.431640625, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.3872318158032444, | |
| "grad_norm": 4273.477870627681, | |
| "learning_rate": 4.987222972275369e-05, | |
| "logits/chosen": -7.125, | |
| "logits/rejected": -7.15625, | |
| "logps/chosen": -41.25, | |
| "logps/rejected": -48.75, | |
| "loss": 2034.95, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.400390625, | |
| "rewards/margins": 0.0751953125, | |
| "rewards/rejected": -0.4765625, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.3976975405546834, | |
| "grad_norm": 4396.096717974299, | |
| "learning_rate": 4.983962194568872e-05, | |
| "logits/chosen": -6.5, | |
| "logits/rejected": -6.03125, | |
| "logps/chosen": -26.125, | |
| "logps/rejected": -34.75, | |
| "loss": 2126.2, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": -0.25, | |
| "rewards/margins": 0.08447265625, | |
| "rewards/rejected": -0.333984375, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.40816326530612246, | |
| "grad_norm": 3698.6015042908152, | |
| "learning_rate": 4.9803325467003084e-05, | |
| "logits/chosen": -5.4375, | |
| "logits/rejected": -5.125, | |
| "logps/chosen": -35.5, | |
| "logps/rejected": -43.0, | |
| "loss": 2197.05, | |
| "rewards/accuracies": 0.668749988079071, | |
| "rewards/chosen": -0.34375, | |
| "rewards/margins": 0.0732421875, | |
| "rewards/rejected": -0.41796875, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.4186289900575615, | |
| "grad_norm": 4315.730476551106, | |
| "learning_rate": 4.9763345676749816e-05, | |
| "logits/chosen": -5.4375, | |
| "logits/rejected": -4.9375, | |
| "logps/chosen": -33.0, | |
| "logps/rejected": -41.0, | |
| "loss": 2039.3, | |
| "rewards/accuracies": 0.65625, | |
| "rewards/chosen": -0.3203125, | |
| "rewards/margins": 0.0791015625, | |
| "rewards/rejected": -0.3984375, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.4186289900575615, | |
| "eval_logits/chosen": -6.09375, | |
| "eval_logits/rejected": -5.4375, | |
| "eval_logps/chosen": -28.5, | |
| "eval_logps/rejected": -37.0, | |
| "eval_loss": 1999.0400390625, | |
| "eval_rewards/accuracies": 0.69140625, | |
| "eval_rewards/chosen": -0.2734375, | |
| "eval_rewards/margins": 0.0849609375, | |
| "eval_rewards/rejected": -0.359375, | |
| "eval_runtime": 46.6409, | |
| "eval_samples_per_second": 42.881, | |
| "eval_steps_per_second": 0.686, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.4290947148090005, | |
| "grad_norm": 4621.08537010144, | |
| "learning_rate": 4.9719688511956374e-05, | |
| "logits/chosen": -6.53125, | |
| "logits/rejected": -6.3125, | |
| "logps/chosen": -34.0, | |
| "logps/rejected": -41.5, | |
| "loss": 2140.55, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.330078125, | |
| "rewards/margins": 0.07421875, | |
| "rewards/rejected": -0.40234375, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.43956043956043955, | |
| "grad_norm": 4399.057265050762, | |
| "learning_rate": 4.967236045574294e-05, | |
| "logits/chosen": -8.3125, | |
| "logits/rejected": -7.59375, | |
| "logps/chosen": -34.25, | |
| "logps/rejected": -43.5, | |
| "loss": 2078.35, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -0.33203125, | |
| "rewards/margins": 0.09130859375, | |
| "rewards/rejected": -0.423828125, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.4500261643118786, | |
| "grad_norm": 5881.76727416646, | |
| "learning_rate": 4.96213685363597e-05, | |
| "logits/chosen": -5.6875, | |
| "logits/rejected": -5.21875, | |
| "logps/chosen": -37.0, | |
| "logps/rejected": -45.5, | |
| "loss": 2085.8, | |
| "rewards/accuracies": 0.706250011920929, | |
| "rewards/chosen": -0.359375, | |
| "rewards/margins": 0.08349609375, | |
| "rewards/rejected": -0.443359375, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.4604918890633176, | |
| "grad_norm": 5302.142690350491, | |
| "learning_rate": 4.956672032614315e-05, | |
| "logits/chosen": -4.21875, | |
| "logits/rejected": -3.390625, | |
| "logps/chosen": -35.0, | |
| "logps/rejected": -43.25, | |
| "loss": 1978.2, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": -0.33984375, | |
| "rewards/margins": 0.08203125, | |
| "rewards/rejected": -0.421875, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.47095761381475665, | |
| "grad_norm": 3892.1233021442763, | |
| "learning_rate": 4.950842394039157e-05, | |
| "logits/chosen": -4.03125, | |
| "logits/rejected": -3.09375, | |
| "logps/chosen": -33.0, | |
| "logps/rejected": -44.5, | |
| "loss": 2068.85, | |
| "rewards/accuracies": 0.706250011920929, | |
| "rewards/chosen": -0.318359375, | |
| "rewards/margins": 0.11376953125, | |
| "rewards/rejected": -0.43359375, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.48142333856619574, | |
| "grad_norm": 4064.0274629697965, | |
| "learning_rate": 4.944648803615994e-05, | |
| "logits/chosen": -3.125, | |
| "logits/rejected": -2.359375, | |
| "logps/chosen": -28.75, | |
| "logps/rejected": -36.75, | |
| "loss": 2093.3, | |
| "rewards/accuracies": 0.643750011920929, | |
| "rewards/chosen": -0.27734375, | |
| "rewards/margins": 0.07861328125, | |
| "rewards/rejected": -0.35546875, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.49188906331763477, | |
| "grad_norm": 4756.786723999879, | |
| "learning_rate": 4.938092181097433e-05, | |
| "logits/chosen": -4.65625, | |
| "logits/rejected": -3.78125, | |
| "logps/chosen": -34.5, | |
| "logps/rejected": -45.25, | |
| "loss": 1862.3, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.33203125, | |
| "rewards/margins": 0.1083984375, | |
| "rewards/rejected": -0.44140625, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.5023547880690737, | |
| "grad_norm": 3449.5347406499186, | |
| "learning_rate": 4.931173500146607e-05, | |
| "logits/chosen": -4.75, | |
| "logits/rejected": -3.765625, | |
| "logps/chosen": -31.625, | |
| "logps/rejected": -43.25, | |
| "loss": 1977.2, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.3046875, | |
| "rewards/margins": 0.11474609375, | |
| "rewards/rejected": -0.419921875, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.5128205128205128, | |
| "grad_norm": 3775.744763449993, | |
| "learning_rate": 4.923893788192582e-05, | |
| "logits/chosen": -4.96875, | |
| "logits/rejected": -3.171875, | |
| "logps/chosen": -34.0, | |
| "logps/rejected": -48.75, | |
| "loss": 1942.55, | |
| "rewards/accuracies": 0.768750011920929, | |
| "rewards/chosen": -0.330078125, | |
| "rewards/margins": 0.1435546875, | |
| "rewards/rejected": -0.47265625, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.5232862375719518, | |
| "grad_norm": 5821.3919913744785, | |
| "learning_rate": 4.9162541262777906e-05, | |
| "logits/chosen": -5.34375, | |
| "logits/rejected": -4.40625, | |
| "logps/chosen": -32.5, | |
| "logps/rejected": -44.5, | |
| "loss": 1938.55, | |
| "rewards/accuracies": 0.7437499761581421, | |
| "rewards/chosen": -0.3125, | |
| "rewards/margins": 0.1220703125, | |
| "rewards/rejected": -0.435546875, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.5232862375719518, | |
| "eval_logits/chosen": -6.28125, | |
| "eval_logits/rejected": -5.25, | |
| "eval_logps/chosen": -40.0, | |
| "eval_logps/rejected": -51.25, | |
| "eval_loss": 1975.6800537109375, | |
| "eval_rewards/accuracies": 0.6953125, | |
| "eval_rewards/chosen": -0.390625, | |
| "eval_rewards/margins": 0.111328125, | |
| "eval_rewards/rejected": -0.5, | |
| "eval_runtime": 46.6326, | |
| "eval_samples_per_second": 42.888, | |
| "eval_steps_per_second": 0.686, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.533751962323391, | |
| "grad_norm": 4077.684789792123, | |
| "learning_rate": 4.90825564889749e-05, | |
| "logits/chosen": -7.78125, | |
| "logits/rejected": -6.90625, | |
| "logps/chosen": -36.5, | |
| "logps/rejected": -51.0, | |
| "loss": 1976.4, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.35546875, | |
| "rewards/margins": 0.142578125, | |
| "rewards/rejected": -0.498046875, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.54421768707483, | |
| "grad_norm": 4358.732276897791, | |
| "learning_rate": 4.899899543831288e-05, | |
| "logits/chosen": -4.75, | |
| "logits/rejected": -4.03125, | |
| "logps/chosen": -29.625, | |
| "logps/rejected": -38.0, | |
| "loss": 2079.7, | |
| "rewards/accuracies": 0.643750011920929, | |
| "rewards/chosen": -0.28515625, | |
| "rewards/margins": 0.08154296875, | |
| "rewards/rejected": -0.3671875, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.554683411826269, | |
| "grad_norm": 3739.9337988801567, | |
| "learning_rate": 4.891187051966765e-05, | |
| "logits/chosen": -1.75, | |
| "logits/rejected": -0.8828125, | |
| "logps/chosen": -39.5, | |
| "logps/rejected": -48.75, | |
| "loss": 2026.75, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.384765625, | |
| "rewards/margins": 0.09228515625, | |
| "rewards/rejected": -0.4765625, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.565149136577708, | |
| "grad_norm": 3892.4229404778835, | |
| "learning_rate": 4.882119467115191e-05, | |
| "logits/chosen": -2.4375, | |
| "logits/rejected": -1.0, | |
| "logps/chosen": -30.0, | |
| "logps/rejected": -39.0, | |
| "loss": 1965.8, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": -0.2890625, | |
| "rewards/margins": 0.08837890625, | |
| "rewards/rejected": -0.376953125, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.5756148613291471, | |
| "grad_norm": 4349.921441684305, | |
| "learning_rate": 4.872698135819402e-05, | |
| "logits/chosen": -5.125, | |
| "logits/rejected": -3.84375, | |
| "logps/chosen": -34.25, | |
| "logps/rejected": -43.5, | |
| "loss": 2059.9, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -0.33203125, | |
| "rewards/margins": 0.0908203125, | |
| "rewards/rejected": -0.423828125, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.5860805860805861, | |
| "grad_norm": 4026.6692112230544, | |
| "learning_rate": 4.862924457153831e-05, | |
| "logits/chosen": -7.875, | |
| "logits/rejected": -6.90625, | |
| "logps/chosen": -29.75, | |
| "logps/rejected": -41.5, | |
| "loss": 2073.5, | |
| "rewards/accuracies": 0.768750011920929, | |
| "rewards/chosen": -0.28515625, | |
| "rewards/margins": 0.1162109375, | |
| "rewards/rejected": -0.40234375, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.5965463108320251, | |
| "grad_norm": 4101.184967902337, | |
| "learning_rate": 4.8527998825167484e-05, | |
| "logits/chosen": -7.15625, | |
| "logits/rejected": -6.46875, | |
| "logps/chosen": -34.5, | |
| "logps/rejected": -42.25, | |
| "loss": 1990.05, | |
| "rewards/accuracies": 0.6937500238418579, | |
| "rewards/chosen": -0.333984375, | |
| "rewards/margins": 0.0751953125, | |
| "rewards/rejected": -0.41015625, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.6070120355834642, | |
| "grad_norm": 4204.405428683835, | |
| "learning_rate": 4.842325915414728e-05, | |
| "logits/chosen": -8.25, | |
| "logits/rejected": -7.15625, | |
| "logps/chosen": -30.625, | |
| "logps/rejected": -41.75, | |
| "loss": 1977.65, | |
| "rewards/accuracies": 0.706250011920929, | |
| "rewards/chosen": -0.294921875, | |
| "rewards/margins": 0.10986328125, | |
| "rewards/rejected": -0.404296875, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.6174777603349032, | |
| "grad_norm": 4211.506376270018, | |
| "learning_rate": 4.831504111239373e-05, | |
| "logits/chosen": -8.5, | |
| "logits/rejected": -7.0625, | |
| "logps/chosen": -33.0, | |
| "logps/rejected": -42.25, | |
| "loss": 1970.7, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": -0.31640625, | |
| "rewards/margins": 0.09228515625, | |
| "rewards/rejected": -0.41015625, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.6279434850863422, | |
| "grad_norm": 3931.083121038388, | |
| "learning_rate": 4.8203360770363425e-05, | |
| "logits/chosen": -6.5625, | |
| "logits/rejected": -5.5, | |
| "logps/chosen": -39.25, | |
| "logps/rejected": -47.25, | |
| "loss": 1949.6, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.380859375, | |
| "rewards/margins": 0.0791015625, | |
| "rewards/rejected": -0.4609375, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.6279434850863422, | |
| "eval_logits/chosen": -6.34375, | |
| "eval_logits/rejected": -4.90625, | |
| "eval_logps/chosen": -34.5, | |
| "eval_logps/rejected": -44.0, | |
| "eval_loss": 1962.8800048828125, | |
| "eval_rewards/accuracies": 0.705078125, | |
| "eval_rewards/chosen": -0.333984375, | |
| "eval_rewards/margins": 0.09423828125, | |
| "eval_rewards/rejected": -0.427734375, | |
| "eval_runtime": 46.6265, | |
| "eval_samples_per_second": 42.894, | |
| "eval_steps_per_second": 0.686, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.6384092098377813, | |
| "grad_norm": 4391.606165253975, | |
| "learning_rate": 4.808823471266702e-05, | |
| "logits/chosen": -7.25, | |
| "logits/rejected": -5.40625, | |
| "logps/chosen": -35.0, | |
| "logps/rejected": -45.0, | |
| "loss": 2001.3, | |
| "rewards/accuracies": 0.706250011920929, | |
| "rewards/chosen": -0.337890625, | |
| "rewards/margins": 0.10107421875, | |
| "rewards/rejected": -0.4375, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.6488749345892203, | |
| "grad_norm": 3684.8667994750663, | |
| "learning_rate": 4.796968003560639e-05, | |
| "logits/chosen": -5.71875, | |
| "logits/rejected": -4.3125, | |
| "logps/chosen": -36.5, | |
| "logps/rejected": -48.0, | |
| "loss": 1885.2, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.353515625, | |
| "rewards/margins": 0.11376953125, | |
| "rewards/rejected": -0.46875, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.6593406593406593, | |
| "grad_norm": 6290.135269678066, | |
| "learning_rate": 4.7847714344635864e-05, | |
| "logits/chosen": -6.8125, | |
| "logits/rejected": -4.9375, | |
| "logps/chosen": -37.25, | |
| "logps/rejected": -50.0, | |
| "loss": 1819.05, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.361328125, | |
| "rewards/margins": 0.1279296875, | |
| "rewards/rejected": -0.48828125, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.6698063840920984, | |
| "grad_norm": 4914.712185265387, | |
| "learning_rate": 4.772235575174776e-05, | |
| "logits/chosen": -7.4375, | |
| "logits/rejected": -6.09375, | |
| "logps/chosen": -34.75, | |
| "logps/rejected": -46.25, | |
| "loss": 1973.6, | |
| "rewards/accuracies": 0.6812499761581421, | |
| "rewards/chosen": -0.3359375, | |
| "rewards/margins": 0.11328125, | |
| "rewards/rejected": -0.44921875, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.6802721088435374, | |
| "grad_norm": 5256.373215433523, | |
| "learning_rate": 4.7593622872782754e-05, | |
| "logits/chosen": -9.75, | |
| "logits/rejected": -8.375, | |
| "logps/chosen": -37.0, | |
| "logps/rejected": -46.75, | |
| "loss": 2042.55, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": -0.357421875, | |
| "rewards/margins": 0.09814453125, | |
| "rewards/rejected": -0.45703125, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.6907378335949764, | |
| "grad_norm": 5065.195474883797, | |
| "learning_rate": 4.746153482466542e-05, | |
| "logits/chosen": -10.3125, | |
| "logits/rejected": -8.375, | |
| "logps/chosen": -31.625, | |
| "logps/rejected": -43.0, | |
| "loss": 1953.8, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -0.3046875, | |
| "rewards/margins": 0.11328125, | |
| "rewards/rejected": -0.41796875, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.7012035583464155, | |
| "grad_norm": 5115.518646895519, | |
| "learning_rate": 4.732611122256537e-05, | |
| "logits/chosen": -10.375, | |
| "logits/rejected": -9.0625, | |
| "logps/chosen": -34.0, | |
| "logps/rejected": -44.75, | |
| "loss": 1888.45, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -0.330078125, | |
| "rewards/margins": 0.1064453125, | |
| "rewards/rejected": -0.435546875, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.7116692830978545, | |
| "grad_norm": 4269.686364758785, | |
| "learning_rate": 4.718737217698435e-05, | |
| "logits/chosen": -9.0625, | |
| "logits/rejected": -6.8125, | |
| "logps/chosen": -34.25, | |
| "logps/rejected": -47.25, | |
| "loss": 1987.0, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -0.33203125, | |
| "rewards/margins": 0.1279296875, | |
| "rewards/rejected": -0.4609375, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.7221350078492935, | |
| "grad_norm": 4183.4341708955535, | |
| "learning_rate": 4.704533829076982e-05, | |
| "logits/chosen": -7.625, | |
| "logits/rejected": -5.0625, | |
| "logps/chosen": -28.125, | |
| "logps/rejected": -40.75, | |
| "loss": 1849.1, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -0.26953125, | |
| "rewards/margins": 0.1240234375, | |
| "rewards/rejected": -0.39453125, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.7326007326007326, | |
| "grad_norm": 4787.269157504689, | |
| "learning_rate": 4.6900030656055455e-05, | |
| "logits/chosen": -7.125, | |
| "logits/rejected": -5.65625, | |
| "logps/chosen": -36.5, | |
| "logps/rejected": -45.5, | |
| "loss": 1951.75, | |
| "rewards/accuracies": 0.65625, | |
| "rewards/chosen": -0.353515625, | |
| "rewards/margins": 0.0888671875, | |
| "rewards/rejected": -0.44140625, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.7326007326007326, | |
| "eval_logits/chosen": -8.6875, | |
| "eval_logits/rejected": -7.0625, | |
| "eval_logps/chosen": -30.625, | |
| "eval_logps/rejected": -41.25, | |
| "eval_loss": 1956.095947265625, | |
| "eval_rewards/accuracies": 0.708984375, | |
| "eval_rewards/chosen": -0.294921875, | |
| "eval_rewards/margins": 0.10546875, | |
| "eval_rewards/rejected": -0.400390625, | |
| "eval_runtime": 46.6349, | |
| "eval_samples_per_second": 42.886, | |
| "eval_steps_per_second": 0.686, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.7430664573521716, | |
| "grad_norm": 5312.820427811685, | |
| "learning_rate": 4.6751470851128906e-05, | |
| "logits/chosen": -8.375, | |
| "logits/rejected": -6.96875, | |
| "logps/chosen": -34.5, | |
| "logps/rejected": -43.0, | |
| "loss": 2035.4, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": -0.33203125, | |
| "rewards/margins": 0.0849609375, | |
| "rewards/rejected": -0.41796875, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.7535321821036107, | |
| "grad_norm": 4109.790506971672, | |
| "learning_rate": 4.6599680937227454e-05, | |
| "logits/chosen": -9.0, | |
| "logits/rejected": -7.34375, | |
| "logps/chosen": -35.0, | |
| "logps/rejected": -45.5, | |
| "loss": 1875.55, | |
| "rewards/accuracies": 0.6937500238418579, | |
| "rewards/chosen": -0.337890625, | |
| "rewards/margins": 0.10546875, | |
| "rewards/rejected": -0.443359375, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.7639979068550498, | |
| "grad_norm": 4450.622153758956, | |
| "learning_rate": 4.644468345526183e-05, | |
| "logits/chosen": -7.90625, | |
| "logits/rejected": -6.78125, | |
| "logps/chosen": -42.75, | |
| "logps/rejected": -50.5, | |
| "loss": 1969.55, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": -0.416015625, | |
| "rewards/margins": 0.076171875, | |
| "rewards/rejected": -0.4921875, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.7744636316064888, | |
| "grad_norm": 5365.388943422686, | |
| "learning_rate": 4.628650142246894e-05, | |
| "logits/chosen": -7.53125, | |
| "logits/rejected": -5.9375, | |
| "logps/chosen": -36.5, | |
| "logps/rejected": -47.5, | |
| "loss": 1929.85, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.353515625, | |
| "rewards/margins": 0.1083984375, | |
| "rewards/rejected": -0.462890625, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.7849293563579278, | |
| "grad_norm": 3714.6576742177444, | |
| "learning_rate": 4.6125158328993765e-05, | |
| "logits/chosen": -9.25, | |
| "logits/rejected": -7.875, | |
| "logps/chosen": -33.5, | |
| "logps/rejected": -45.0, | |
| "loss": 1847.45, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -0.322265625, | |
| "rewards/margins": 0.11474609375, | |
| "rewards/rejected": -0.4375, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.7953950811093669, | |
| "grad_norm": 7303.030602648265, | |
| "learning_rate": 4.5960678134401064e-05, | |
| "logits/chosen": -9.1875, | |
| "logits/rejected": -7.6875, | |
| "logps/chosen": -33.5, | |
| "logps/rejected": -44.5, | |
| "loss": 1938.65, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.32421875, | |
| "rewards/margins": 0.10791015625, | |
| "rewards/rejected": -0.431640625, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.8058608058608059, | |
| "grad_norm": 3973.4821047327164, | |
| "learning_rate": 4.579308526411734e-05, | |
| "logits/chosen": -7.65625, | |
| "logits/rejected": -6.75, | |
| "logps/chosen": -43.0, | |
| "logps/rejected": -51.25, | |
| "loss": 1904.3, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": -0.41796875, | |
| "rewards/margins": 0.08349609375, | |
| "rewards/rejected": -0.5, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.8163265306122449, | |
| "grad_norm": 6193.009304380043, | |
| "learning_rate": 4.562240460580367e-05, | |
| "logits/chosen": -5.9375, | |
| "logits/rejected": -3.8125, | |
| "logps/chosen": -38.0, | |
| "logps/rejected": -50.5, | |
| "loss": 1953.7, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.369140625, | |
| "rewards/margins": 0.125, | |
| "rewards/rejected": -0.494140625, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.826792255363684, | |
| "grad_norm": 3829.537065411557, | |
| "learning_rate": 4.544866150565985e-05, | |
| "logits/chosen": -1.7421875, | |
| "logits/rejected": 0.1298828125, | |
| "logps/chosen": -37.75, | |
| "logps/rejected": -48.25, | |
| "loss": 1996.5, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -0.365234375, | |
| "rewards/margins": 0.10595703125, | |
| "rewards/rejected": -0.470703125, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.837257980115123, | |
| "grad_norm": 3261.5057492786596, | |
| "learning_rate": 4.527188176466051e-05, | |
| "logits/chosen": -0.69921875, | |
| "logits/rejected": 0.466796875, | |
| "logps/chosen": -36.5, | |
| "logps/rejected": -46.75, | |
| "loss": 1869.7, | |
| "rewards/accuracies": 0.706250011920929, | |
| "rewards/chosen": -0.353515625, | |
| "rewards/margins": 0.1015625, | |
| "rewards/rejected": -0.455078125, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.837257980115123, | |
| "eval_logits/chosen": -1.203125, | |
| "eval_logits/rejected": 0.318359375, | |
| "eval_logps/chosen": -37.0, | |
| "eval_logps/rejected": -48.75, | |
| "eval_loss": 1889.72802734375, | |
| "eval_rewards/accuracies": 0.720703125, | |
| "eval_rewards/chosen": -0.359375, | |
| "eval_rewards/margins": 0.11474609375, | |
| "eval_rewards/rejected": -0.474609375, | |
| "eval_runtime": 46.6143, | |
| "eval_samples_per_second": 42.905, | |
| "eval_steps_per_second": 0.686, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.847723704866562, | |
| "grad_norm": 4187.625564328089, | |
| "learning_rate": 4.509209163472357e-05, | |
| "logits/chosen": -2.625, | |
| "logits/rejected": -0.984375, | |
| "logps/chosen": -40.5, | |
| "logps/rejected": -52.0, | |
| "loss": 1955.7, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.39453125, | |
| "rewards/margins": 0.1142578125, | |
| "rewards/rejected": -0.5078125, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.858189429618001, | |
| "grad_norm": 3650.4714165684713, | |
| "learning_rate": 4.4909317814811896e-05, | |
| "logits/chosen": -2.125, | |
| "logits/rejected": -1.0234375, | |
| "logps/chosen": -29.125, | |
| "logps/rejected": -39.5, | |
| "loss": 2008.05, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -0.279296875, | |
| "rewards/margins": 0.10400390625, | |
| "rewards/rejected": -0.3828125, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.8686551543694401, | |
| "grad_norm": 4072.0708858834196, | |
| "learning_rate": 4.472358744696845e-05, | |
| "logits/chosen": -0.8046875, | |
| "logits/rejected": -0.022216796875, | |
| "logps/chosen": -32.75, | |
| "logps/rejected": -43.75, | |
| "loss": 1983.65, | |
| "rewards/accuracies": 0.6812499761581421, | |
| "rewards/chosen": -0.31640625, | |
| "rewards/margins": 0.10791015625, | |
| "rewards/rejected": -0.423828125, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.8791208791208791, | |
| "grad_norm": 4794.376847332415, | |
| "learning_rate": 4.4534928112285645e-05, | |
| "logits/chosen": -1.2578125, | |
| "logits/rejected": 0.03662109375, | |
| "logps/chosen": -34.25, | |
| "logps/rejected": -46.0, | |
| "loss": 1856.45, | |
| "rewards/accuracies": 0.6937500238418579, | |
| "rewards/chosen": -0.330078125, | |
| "rewards/margins": 0.11767578125, | |
| "rewards/rejected": -0.44921875, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.8895866038723181, | |
| "grad_norm": 6057.883203642358, | |
| "learning_rate": 4.43433678268096e-05, | |
| "logits/chosen": -1.78125, | |
| "logits/rejected": -0.8046875, | |
| "logps/chosen": -39.5, | |
| "logps/rejected": -52.0, | |
| "loss": 1807.05, | |
| "rewards/accuracies": 0.706250011920929, | |
| "rewards/chosen": -0.3828125, | |
| "rewards/margins": 0.126953125, | |
| "rewards/rejected": -0.5078125, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.9000523286237572, | |
| "grad_norm": 6394.295301713368, | |
| "learning_rate": 4.414893503737968e-05, | |
| "logits/chosen": -3.140625, | |
| "logits/rejected": -2.046875, | |
| "logps/chosen": -48.5, | |
| "logps/rejected": -59.0, | |
| "loss": 1868.35, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -0.47265625, | |
| "rewards/margins": 0.10595703125, | |
| "rewards/rejected": -0.578125, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.9105180533751962, | |
| "grad_norm": 3885.696471424006, | |
| "learning_rate": 4.395165861740414e-05, | |
| "logits/chosen": -4.875, | |
| "logits/rejected": -3.546875, | |
| "logps/chosen": -39.25, | |
| "logps/rejected": -50.0, | |
| "loss": 1993.85, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": -0.380859375, | |
| "rewards/margins": 0.107421875, | |
| "rewards/rejected": -0.48828125, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.9209837781266352, | |
| "grad_norm": 4326.829158068277, | |
| "learning_rate": 4.375156786257241e-05, | |
| "logits/chosen": -5.28125, | |
| "logits/rejected": -4.0625, | |
| "logps/chosen": -31.875, | |
| "logps/rejected": -39.0, | |
| "loss": 1951.55, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.306640625, | |
| "rewards/margins": 0.07177734375, | |
| "rewards/rejected": -0.37890625, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.9314495028780743, | |
| "grad_norm": 4299.407454655135, | |
| "learning_rate": 4.354869248650466e-05, | |
| "logits/chosen": -7.65625, | |
| "logits/rejected": -6.28125, | |
| "logps/chosen": -39.0, | |
| "logps/rejected": -48.5, | |
| "loss": 1973.5, | |
| "rewards/accuracies": 0.6937500238418579, | |
| "rewards/chosen": -0.37890625, | |
| "rewards/margins": 0.09423828125, | |
| "rewards/rejected": -0.47265625, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.9419152276295133, | |
| "grad_norm": 4561.175753526089, | |
| "learning_rate": 4.334306261633931e-05, | |
| "logits/chosen": -7.625, | |
| "logits/rejected": -6.3125, | |
| "logps/chosen": -43.0, | |
| "logps/rejected": -54.5, | |
| "loss": 1905.45, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -0.41796875, | |
| "rewards/margins": 0.11669921875, | |
| "rewards/rejected": -0.53515625, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.9419152276295133, | |
| "eval_logits/chosen": -6.0625, | |
| "eval_logits/rejected": -4.21875, | |
| "eval_logps/chosen": -42.5, | |
| "eval_logps/rejected": -54.25, | |
| "eval_loss": 1903.8399658203125, | |
| "eval_rewards/accuracies": 0.70703125, | |
| "eval_rewards/chosen": -0.4140625, | |
| "eval_rewards/margins": 0.11669921875, | |
| "eval_rewards/rejected": -0.53125, | |
| "eval_runtime": 46.6216, | |
| "eval_samples_per_second": 42.899, | |
| "eval_steps_per_second": 0.686, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.9523809523809523, | |
| "grad_norm": 4477.18405503131, | |
| "learning_rate": 4.313470878825911e-05, | |
| "logits/chosen": -6.46875, | |
| "logits/rejected": -5.3125, | |
| "logps/chosen": -39.5, | |
| "logps/rejected": -48.5, | |
| "loss": 1987.15, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -0.3828125, | |
| "rewards/margins": 0.08837890625, | |
| "rewards/rejected": -0.47265625, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.9628466771323915, | |
| "grad_norm": 3537.953123756003, | |
| "learning_rate": 4.292366194295652e-05, | |
| "logits/chosen": -4.6875, | |
| "logits/rejected": -2.75, | |
| "logps/chosen": -44.25, | |
| "logps/rejected": -54.0, | |
| "loss": 1968.4, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.4296875, | |
| "rewards/margins": 0.0986328125, | |
| "rewards/rejected": -0.52734375, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.9733124018838305, | |
| "grad_norm": 4310.558147761052, | |
| "learning_rate": 4.2709953421038964e-05, | |
| "logits/chosen": -3.8125, | |
| "logits/rejected": -2.265625, | |
| "logps/chosen": -43.0, | |
| "logps/rejected": -52.5, | |
| "loss": 1890.1, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.416015625, | |
| "rewards/margins": 0.09619140625, | |
| "rewards/rejected": -0.51171875, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.9837781266352695, | |
| "grad_norm": 4464.873686345643, | |
| "learning_rate": 4.249361495837478e-05, | |
| "logits/chosen": -5.625, | |
| "logits/rejected": -3.84375, | |
| "logps/chosen": -30.375, | |
| "logps/rejected": -41.5, | |
| "loss": 1874.1, | |
| "rewards/accuracies": 0.6812499761581421, | |
| "rewards/chosen": -0.29296875, | |
| "rewards/margins": 0.1083984375, | |
| "rewards/rejected": -0.40234375, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.9942438513867086, | |
| "grad_norm": 4273.881496020123, | |
| "learning_rate": 4.227467868138035e-05, | |
| "logits/chosen": -3.859375, | |
| "logits/rejected": -2.75, | |
| "logps/chosen": -35.5, | |
| "logps/rejected": -45.25, | |
| "loss": 1970.95, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": -0.34375, | |
| "rewards/margins": 0.095703125, | |
| "rewards/rejected": -0.439453125, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.0047095761381475, | |
| "grad_norm": 3679.5937261212434, | |
| "learning_rate": 4.205317710224938e-05, | |
| "logits/chosen": -2.671875, | |
| "logits/rejected": -1.40625, | |
| "logps/chosen": -35.0, | |
| "logps/rejected": -49.75, | |
| "loss": 1739.95, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.33984375, | |
| "rewards/margins": 0.1474609375, | |
| "rewards/rejected": -0.48828125, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 1.0151753008895865, | |
| "grad_norm": 3950.037686563474, | |
| "learning_rate": 4.1829143114124733e-05, | |
| "logits/chosen": -2.3125, | |
| "logits/rejected": -0.734375, | |
| "logps/chosen": -40.5, | |
| "logps/rejected": -58.5, | |
| "loss": 1429.5, | |
| "rewards/accuracies": 0.84375, | |
| "rewards/chosen": -0.392578125, | |
| "rewards/margins": 0.1787109375, | |
| "rewards/rejected": -0.57421875, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 1.0256410256410255, | |
| "grad_norm": 6230.708696654123, | |
| "learning_rate": 4.160260998621387e-05, | |
| "logits/chosen": -2.03125, | |
| "logits/rejected": -0.26953125, | |
| "logps/chosen": -39.5, | |
| "logps/rejected": -60.5, | |
| "loss": 1382.6, | |
| "rewards/accuracies": 0.78125, | |
| "rewards/chosen": -0.3828125, | |
| "rewards/margins": 0.208984375, | |
| "rewards/rejected": -0.59375, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 1.0361067503924646, | |
| "grad_norm": 5880.076828887697, | |
| "learning_rate": 4.137361135884824e-05, | |
| "logits/chosen": -2.828125, | |
| "logits/rejected": -0.8359375, | |
| "logps/chosen": -33.25, | |
| "logps/rejected": -58.75, | |
| "loss": 1253.175, | |
| "rewards/accuracies": 0.862500011920929, | |
| "rewards/chosen": -0.3203125, | |
| "rewards/margins": 0.25390625, | |
| "rewards/rejected": -0.57421875, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 1.0465724751439036, | |
| "grad_norm": 7495.067230817937, | |
| "learning_rate": 4.114218123848777e-05, | |
| "logits/chosen": -2.03125, | |
| "logits/rejected": -0.71484375, | |
| "logps/chosen": -36.0, | |
| "logps/rejected": -58.5, | |
| "loss": 1301.1, | |
| "rewards/accuracies": 0.8062499761581421, | |
| "rewards/chosen": -0.34765625, | |
| "rewards/margins": 0.2255859375, | |
| "rewards/rejected": -0.57421875, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.0465724751439036, | |
| "eval_logits/chosen": -0.890625, | |
| "eval_logits/rejected": 0.2236328125, | |
| "eval_logps/chosen": -40.0, | |
| "eval_logps/rejected": -54.25, | |
| "eval_loss": 1946.8480224609375, | |
| "eval_rewards/accuracies": 0.7109375, | |
| "eval_rewards/chosen": -0.388671875, | |
| "eval_rewards/margins": 0.1416015625, | |
| "eval_rewards/rejected": -0.53125, | |
| "eval_runtime": 46.6351, | |
| "eval_samples_per_second": 42.886, | |
| "eval_steps_per_second": 0.686, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.0570381998953426, | |
| "grad_norm": 7835.697288777801, | |
| "learning_rate": 4.0908353992670824e-05, | |
| "logits/chosen": -1.6640625, | |
| "logits/rejected": 0.1650390625, | |
| "logps/chosen": -34.25, | |
| "logps/rejected": -58.5, | |
| "loss": 1235.95, | |
| "rewards/accuracies": 0.875, | |
| "rewards/chosen": -0.330078125, | |
| "rewards/margins": 0.240234375, | |
| "rewards/rejected": -0.5703125, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 1.0675039246467817, | |
| "grad_norm": 6181.4677608657885, | |
| "learning_rate": 4.067216434491059e-05, | |
| "logits/chosen": -1.015625, | |
| "logits/rejected": -0.0673828125, | |
| "logps/chosen": -42.0, | |
| "logps/rejected": -66.0, | |
| "loss": 1194.475, | |
| "rewards/accuracies": 0.8187500238418579, | |
| "rewards/chosen": -0.40625, | |
| "rewards/margins": 0.240234375, | |
| "rewards/rejected": -0.6484375, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 1.077969649398221, | |
| "grad_norm": 7190.947117620607, | |
| "learning_rate": 4.043364736953866e-05, | |
| "logits/chosen": -2.171875, | |
| "logits/rejected": -0.671875, | |
| "logps/chosen": -32.25, | |
| "logps/rejected": -57.5, | |
| "loss": 1192.325, | |
| "rewards/accuracies": 0.856249988079071, | |
| "rewards/chosen": -0.3125, | |
| "rewards/margins": 0.251953125, | |
| "rewards/rejected": -0.5625, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 1.08843537414966, | |
| "grad_norm": 7164.782648565083, | |
| "learning_rate": 4.019283848649643e-05, | |
| "logits/chosen": -1.09375, | |
| "logits/rejected": 0.494140625, | |
| "logps/chosen": -38.25, | |
| "logps/rejected": -60.75, | |
| "loss": 1268.525, | |
| "rewards/accuracies": 0.8374999761581421, | |
| "rewards/chosen": -0.37109375, | |
| "rewards/margins": 0.2236328125, | |
| "rewards/rejected": -0.59375, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 1.098901098901099, | |
| "grad_norm": 7380.219340334459, | |
| "learning_rate": 3.9949773456075215e-05, | |
| "logits/chosen": -2.359375, | |
| "logits/rejected": -1.234375, | |
| "logps/chosen": -35.5, | |
| "logps/rejected": -59.25, | |
| "loss": 1104.1, | |
| "rewards/accuracies": 0.862500011920929, | |
| "rewards/chosen": -0.34375, | |
| "rewards/margins": 0.236328125, | |
| "rewards/rejected": -0.58203125, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.109366823652538, | |
| "grad_norm": 8185.099824073727, | |
| "learning_rate": 3.970448837360585e-05, | |
| "logits/chosen": -2.0625, | |
| "logits/rejected": -0.40234375, | |
| "logps/chosen": -37.75, | |
| "logps/rejected": -63.0, | |
| "loss": 1202.1, | |
| "rewards/accuracies": 0.8500000238418579, | |
| "rewards/chosen": -0.3671875, | |
| "rewards/margins": 0.25, | |
| "rewards/rejected": -0.6171875, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 1.119832548403977, | |
| "grad_norm": 10839.48719985217, | |
| "learning_rate": 3.945701966409846e-05, | |
| "logits/chosen": -2.65625, | |
| "logits/rejected": -0.890625, | |
| "logps/chosen": -40.75, | |
| "logps/rejected": -65.5, | |
| "loss": 1217.7, | |
| "rewards/accuracies": 0.8687499761581421, | |
| "rewards/chosen": -0.396484375, | |
| "rewards/margins": 0.2451171875, | |
| "rewards/rejected": -0.640625, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 1.130298273155416, | |
| "grad_norm": 8780.088714527372, | |
| "learning_rate": 3.9207404076833374e-05, | |
| "logits/chosen": -2.109375, | |
| "logits/rejected": -1.4140625, | |
| "logps/chosen": -34.5, | |
| "logps/rejected": -65.5, | |
| "loss": 1189.975, | |
| "rewards/accuracies": 0.887499988079071, | |
| "rewards/chosen": -0.33203125, | |
| "rewards/margins": 0.3125, | |
| "rewards/rejected": -0.64453125, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 1.1407639979068551, | |
| "grad_norm": 6572.436093454694, | |
| "learning_rate": 3.895567867990379e-05, | |
| "logits/chosen": -2.828125, | |
| "logits/rejected": -1.53125, | |
| "logps/chosen": -37.5, | |
| "logps/rejected": -64.0, | |
| "loss": 1189.325, | |
| "rewards/accuracies": 0.8812500238418579, | |
| "rewards/chosen": -0.36328125, | |
| "rewards/margins": 0.265625, | |
| "rewards/rejected": -0.62890625, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 1.1512297226582942, | |
| "grad_norm": 7101.951636330339, | |
| "learning_rate": 3.870188085471114e-05, | |
| "logits/chosen": -2.03125, | |
| "logits/rejected": -1.140625, | |
| "logps/chosen": -36.75, | |
| "logps/rejected": -62.5, | |
| "loss": 1193.05, | |
| "rewards/accuracies": 0.831250011920929, | |
| "rewards/chosen": -0.35546875, | |
| "rewards/margins": 0.2578125, | |
| "rewards/rejected": -0.61328125, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 1.1512297226582942, | |
| "eval_logits/chosen": -1.609375, | |
| "eval_logits/rejected": -0.392578125, | |
| "eval_logps/chosen": -45.0, | |
| "eval_logps/rejected": -59.25, | |
| "eval_loss": 1939.2320556640625, | |
| "eval_rewards/accuracies": 0.703125, | |
| "eval_rewards/chosen": -0.439453125, | |
| "eval_rewards/margins": 0.140625, | |
| "eval_rewards/rejected": -0.578125, | |
| "eval_runtime": 46.6519, | |
| "eval_samples_per_second": 42.871, | |
| "eval_steps_per_second": 0.686, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 1.1616954474097332, | |
| "grad_norm": 6651.254087225532, | |
| "learning_rate": 3.844604829041395e-05, | |
| "logits/chosen": -2.59375, | |
| "logits/rejected": -0.88671875, | |
| "logps/chosen": -35.0, | |
| "logps/rejected": -61.75, | |
| "loss": 1090.7, | |
| "rewards/accuracies": 0.8812500238418579, | |
| "rewards/chosen": -0.33984375, | |
| "rewards/margins": 0.265625, | |
| "rewards/rejected": -0.60546875, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 1.1721611721611722, | |
| "grad_norm": 9748.046632952837, | |
| "learning_rate": 3.818821897833095e-05, | |
| "logits/chosen": -2.234375, | |
| "logits/rejected": -1.296875, | |
| "logps/chosen": -38.5, | |
| "logps/rejected": -63.0, | |
| "loss": 1266.675, | |
| "rewards/accuracies": 0.875, | |
| "rewards/chosen": -0.373046875, | |
| "rewards/margins": 0.24609375, | |
| "rewards/rejected": -0.6171875, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 1.1826268969126112, | |
| "grad_norm": 9025.401205694165, | |
| "learning_rate": 3.792843120629935e-05, | |
| "logits/chosen": -3.21875, | |
| "logits/rejected": -1.6796875, | |
| "logps/chosen": -37.0, | |
| "logps/rejected": -61.5, | |
| "loss": 1179.5, | |
| "rewards/accuracies": 0.8500000238418579, | |
| "rewards/chosen": -0.359375, | |
| "rewards/margins": 0.2421875, | |
| "rewards/rejected": -0.6015625, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 1.1930926216640503, | |
| "grad_norm": 7700.277198720241, | |
| "learning_rate": 3.766672355298907e-05, | |
| "logits/chosen": -2.484375, | |
| "logits/rejected": -1.5703125, | |
| "logps/chosen": -42.0, | |
| "logps/rejected": -67.5, | |
| "loss": 1097.85, | |
| "rewards/accuracies": 0.875, | |
| "rewards/chosen": -0.41015625, | |
| "rewards/margins": 0.25390625, | |
| "rewards/rejected": -0.6640625, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 1.2035583464154893, | |
| "grad_norm": 7406.88151590704, | |
| "learning_rate": 3.740313488217373e-05, | |
| "logits/chosen": -1.8671875, | |
| "logits/rejected": -1.125, | |
| "logps/chosen": -36.75, | |
| "logps/rejected": -60.75, | |
| "loss": 1066.9, | |
| "rewards/accuracies": 0.8500000238418579, | |
| "rewards/chosen": -0.357421875, | |
| "rewards/margins": 0.2392578125, | |
| "rewards/rejected": -0.59765625, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 1.2140240711669283, | |
| "grad_norm": 10062.87129041597, | |
| "learning_rate": 3.7137704336959464e-05, | |
| "logits/chosen": -2.34375, | |
| "logits/rejected": -0.42578125, | |
| "logps/chosen": -37.25, | |
| "logps/rejected": -65.5, | |
| "loss": 1143.6375, | |
| "rewards/accuracies": 0.8812500238418579, | |
| "rewards/chosen": -0.361328125, | |
| "rewards/margins": 0.283203125, | |
| "rewards/rejected": -0.64453125, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 1.2244897959183674, | |
| "grad_norm": 9593.472904181473, | |
| "learning_rate": 3.687047133397201e-05, | |
| "logits/chosen": -3.15625, | |
| "logits/rejected": -1.9375, | |
| "logps/chosen": -37.0, | |
| "logps/rejected": -63.5, | |
| "loss": 1189.675, | |
| "rewards/accuracies": 0.875, | |
| "rewards/chosen": -0.357421875, | |
| "rewards/margins": 0.263671875, | |
| "rewards/rejected": -0.62109375, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 1.2349555206698064, | |
| "grad_norm": 9630.370229592447, | |
| "learning_rate": 3.6601475557503415e-05, | |
| "logits/chosen": -3.5625, | |
| "logits/rejected": -2.578125, | |
| "logps/chosen": -38.25, | |
| "logps/rejected": -62.5, | |
| "loss": 1206.6, | |
| "rewards/accuracies": 0.824999988079071, | |
| "rewards/chosen": -0.37109375, | |
| "rewards/margins": 0.240234375, | |
| "rewards/rejected": -0.61328125, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 1.2454212454212454, | |
| "grad_norm": 9723.781700401587, | |
| "learning_rate": 3.633075695361882e-05, | |
| "logits/chosen": -3.578125, | |
| "logits/rejected": -1.7890625, | |
| "logps/chosen": -35.25, | |
| "logps/rejected": -59.0, | |
| "loss": 1144.6125, | |
| "rewards/accuracies": 0.8374999761581421, | |
| "rewards/chosen": -0.341796875, | |
| "rewards/margins": 0.236328125, | |
| "rewards/rejected": -0.578125, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 1.2558869701726845, | |
| "grad_norm": 7769.589050451526, | |
| "learning_rate": 3.6058355724224477e-05, | |
| "logits/chosen": -3.296875, | |
| "logits/rejected": -1.671875, | |
| "logps/chosen": -38.5, | |
| "logps/rejected": -62.0, | |
| "loss": 1162.575, | |
| "rewards/accuracies": 0.8687499761581421, | |
| "rewards/chosen": -0.373046875, | |
| "rewards/margins": 0.2353515625, | |
| "rewards/rejected": -0.609375, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 1.2558869701726845, | |
| "eval_logits/chosen": -2.09375, | |
| "eval_logits/rejected": -0.7109375, | |
| "eval_logps/chosen": -45.5, | |
| "eval_logps/rejected": -59.75, | |
| "eval_loss": 1908.47998046875, | |
| "eval_rewards/accuracies": 0.70703125, | |
| "eval_rewards/chosen": -0.443359375, | |
| "eval_rewards/margins": 0.140625, | |
| "eval_rewards/rejected": -0.5859375, | |
| "eval_runtime": 46.6645, | |
| "eval_samples_per_second": 42.859, | |
| "eval_steps_per_second": 0.686, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 1.2663526949241235, | |
| "grad_norm": 8576.285386312578, | |
| "learning_rate": 3.5784312321097744e-05, | |
| "logits/chosen": -3.796875, | |
| "logits/rejected": -1.3515625, | |
| "logps/chosen": -36.25, | |
| "logps/rejected": -62.5, | |
| "loss": 1206.75, | |
| "rewards/accuracies": 0.862500011920929, | |
| "rewards/chosen": -0.349609375, | |
| "rewards/margins": 0.263671875, | |
| "rewards/rejected": -0.61328125, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 1.2768184196755625, | |
| "grad_norm": 8649.361667137813, | |
| "learning_rate": 3.550866743987994e-05, | |
| "logits/chosen": -4.53125, | |
| "logits/rejected": -2.34375, | |
| "logps/chosen": -35.75, | |
| "logps/rejected": -60.0, | |
| "loss": 1275.575, | |
| "rewards/accuracies": 0.8374999761581421, | |
| "rewards/chosen": -0.345703125, | |
| "rewards/margins": 0.2431640625, | |
| "rewards/rejected": -0.58984375, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 1.2872841444270016, | |
| "grad_norm": 9415.034562036817, | |
| "learning_rate": 3.5231462014033024e-05, | |
| "logits/chosen": -4.21875, | |
| "logits/rejected": -2.234375, | |
| "logps/chosen": -39.0, | |
| "logps/rejected": -67.5, | |
| "loss": 979.1, | |
| "rewards/accuracies": 0.862500011920929, | |
| "rewards/chosen": -0.376953125, | |
| "rewards/margins": 0.287109375, | |
| "rewards/rejected": -0.6640625, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 1.2977498691784406, | |
| "grad_norm": 10041.89800952126, | |
| "learning_rate": 3.495273720876095e-05, | |
| "logits/chosen": -3.578125, | |
| "logits/rejected": -2.0, | |
| "logps/chosen": -34.75, | |
| "logps/rejected": -64.0, | |
| "loss": 1114.05, | |
| "rewards/accuracies": 0.862500011920929, | |
| "rewards/chosen": -0.3359375, | |
| "rewards/margins": 0.29296875, | |
| "rewards/rejected": -0.62890625, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 1.3082155939298796, | |
| "grad_norm": 9651.641063912584, | |
| "learning_rate": 3.4672534414896616e-05, | |
| "logits/chosen": -3.625, | |
| "logits/rejected": -1.203125, | |
| "logps/chosen": -33.0, | |
| "logps/rejected": -61.25, | |
| "loss": 1100.4625, | |
| "rewards/accuracies": 0.8812500238418579, | |
| "rewards/chosen": -0.318359375, | |
| "rewards/margins": 0.28125, | |
| "rewards/rejected": -0.59765625, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 1.3186813186813187, | |
| "grad_norm": 8447.898428030107, | |
| "learning_rate": 3.439089524275526e-05, | |
| "logits/chosen": -2.1875, | |
| "logits/rejected": -1.1328125, | |
| "logps/chosen": -39.25, | |
| "logps/rejected": -64.5, | |
| "loss": 1179.5, | |
| "rewards/accuracies": 0.8374999761581421, | |
| "rewards/chosen": -0.380859375, | |
| "rewards/margins": 0.25, | |
| "rewards/rejected": -0.6328125, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 1.3291470434327577, | |
| "grad_norm": 10410.825126425476, | |
| "learning_rate": 3.410786151595533e-05, | |
| "logits/chosen": -2.75, | |
| "logits/rejected": -0.82421875, | |
| "logps/chosen": -37.5, | |
| "logps/rejected": -61.0, | |
| "loss": 1168.6312, | |
| "rewards/accuracies": 0.856249988079071, | |
| "rewards/chosen": -0.36328125, | |
| "rewards/margins": 0.2333984375, | |
| "rewards/rejected": -0.59765625, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 1.3396127681841967, | |
| "grad_norm": 8129.07527591415, | |
| "learning_rate": 3.382347526520765e-05, | |
| "logits/chosen": -3.1875, | |
| "logits/rejected": -0.75, | |
| "logps/chosen": -39.5, | |
| "logps/rejected": -66.5, | |
| "loss": 1126.65, | |
| "rewards/accuracies": 0.8500000238418579, | |
| "rewards/chosen": -0.3828125, | |
| "rewards/margins": 0.26953125, | |
| "rewards/rejected": -0.65234375, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 1.3500784929356358, | |
| "grad_norm": 10143.760493390431, | |
| "learning_rate": 3.353777872207381e-05, | |
| "logits/chosen": -3.390625, | |
| "logits/rejected": -1.21875, | |
| "logps/chosen": -38.25, | |
| "logps/rejected": -66.0, | |
| "loss": 1167.7875, | |
| "rewards/accuracies": 0.875, | |
| "rewards/chosen": -0.37109375, | |
| "rewards/margins": 0.275390625, | |
| "rewards/rejected": -0.6484375, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 1.3605442176870748, | |
| "grad_norm": 7994.674300557725, | |
| "learning_rate": 3.325081431269475e-05, | |
| "logits/chosen": -3.65625, | |
| "logits/rejected": -1.78125, | |
| "logps/chosen": -35.0, | |
| "logps/rejected": -60.75, | |
| "loss": 1153.3, | |
| "rewards/accuracies": 0.84375, | |
| "rewards/chosen": -0.337890625, | |
| "rewards/margins": 0.2578125, | |
| "rewards/rejected": -0.59375, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 1.3605442176870748, | |
| "eval_logits/chosen": -2.828125, | |
| "eval_logits/rejected": -1.359375, | |
| "eval_logps/chosen": -41.25, | |
| "eval_logps/rejected": -54.75, | |
| "eval_loss": 1974.0799560546875, | |
| "eval_rewards/accuracies": 0.697265625, | |
| "eval_rewards/chosen": -0.400390625, | |
| "eval_rewards/margins": 0.1357421875, | |
| "eval_rewards/rejected": -0.53515625, | |
| "eval_runtime": 46.6845, | |
| "eval_samples_per_second": 42.841, | |
| "eval_steps_per_second": 0.685, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 1.3710099424385138, | |
| "grad_norm": 8697.592617990815, | |
| "learning_rate": 3.296262465149046e-05, | |
| "logits/chosen": -3.765625, | |
| "logits/rejected": -1.5234375, | |
| "logps/chosen": -36.0, | |
| "logps/rejected": -60.5, | |
| "loss": 1146.475, | |
| "rewards/accuracies": 0.8187500238418579, | |
| "rewards/chosen": -0.34765625, | |
| "rewards/margins": 0.24609375, | |
| "rewards/rejected": -0.59375, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 1.3814756671899528, | |
| "grad_norm": 9391.935748961541, | |
| "learning_rate": 3.2673252534831687e-05, | |
| "logits/chosen": -3.390625, | |
| "logits/rejected": -2.234375, | |
| "logps/chosen": -39.0, | |
| "logps/rejected": -66.5, | |
| "loss": 1115.05, | |
| "rewards/accuracies": 0.8687499761581421, | |
| "rewards/chosen": -0.37890625, | |
| "rewards/margins": 0.275390625, | |
| "rewards/rejected": -0.65625, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 1.3919413919413919, | |
| "grad_norm": 9405.955722946464, | |
| "learning_rate": 3.2382740934684696e-05, | |
| "logits/chosen": -3.078125, | |
| "logits/rejected": -2.390625, | |
| "logps/chosen": -40.5, | |
| "logps/rejected": -64.5, | |
| "loss": 1119.15, | |
| "rewards/accuracies": 0.831250011920929, | |
| "rewards/chosen": -0.39453125, | |
| "rewards/margins": 0.23828125, | |
| "rewards/rejected": -0.6328125, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 1.402407116692831, | |
| "grad_norm": 8995.929870129534, | |
| "learning_rate": 3.209113299222982e-05, | |
| "logits/chosen": -3.390625, | |
| "logits/rejected": -1.5234375, | |
| "logps/chosen": -39.5, | |
| "logps/rejected": -66.5, | |
| "loss": 1076.45, | |
| "rewards/accuracies": 0.8374999761581421, | |
| "rewards/chosen": -0.384765625, | |
| "rewards/margins": 0.26953125, | |
| "rewards/rejected": -0.65625, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 1.41287284144427, | |
| "grad_norm": 10594.90991667782, | |
| "learning_rate": 3.179847201145505e-05, | |
| "logits/chosen": -3.125, | |
| "logits/rejected": -1.515625, | |
| "logps/chosen": -39.0, | |
| "logps/rejected": -67.5, | |
| "loss": 1190.6, | |
| "rewards/accuracies": 0.8500000238418579, | |
| "rewards/chosen": -0.380859375, | |
| "rewards/margins": 0.28125, | |
| "rewards/rejected": -0.66015625, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 1.423338566195709, | |
| "grad_norm": 11652.429690802166, | |
| "learning_rate": 3.150480145272528e-05, | |
| "logits/chosen": -2.921875, | |
| "logits/rejected": -1.6640625, | |
| "logps/chosen": -41.75, | |
| "logps/rejected": -67.5, | |
| "loss": 1116.775, | |
| "rewards/accuracies": 0.856249988079071, | |
| "rewards/chosen": -0.40625, | |
| "rewards/margins": 0.255859375, | |
| "rewards/rejected": -0.6640625, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 1.433804290947148, | |
| "grad_norm": 11196.087566594675, | |
| "learning_rate": 3.121016492632848e-05, | |
| "logits/chosen": -3.171875, | |
| "logits/rejected": -1.2734375, | |
| "logps/chosen": -42.0, | |
| "logps/rejected": -66.0, | |
| "loss": 1212.525, | |
| "rewards/accuracies": 0.831250011920929, | |
| "rewards/chosen": -0.40625, | |
| "rewards/margins": 0.2421875, | |
| "rewards/rejected": -0.6484375, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 1.4442700156985873, | |
| "grad_norm": 10675.945011113197, | |
| "learning_rate": 3.091460618599951e-05, | |
| "logits/chosen": -3.59375, | |
| "logits/rejected": -1.03125, | |
| "logps/chosen": -33.5, | |
| "logps/rejected": -60.25, | |
| "loss": 1076.975, | |
| "rewards/accuracies": 0.8374999761581421, | |
| "rewards/chosen": -0.32421875, | |
| "rewards/margins": 0.265625, | |
| "rewards/rejected": -0.58984375, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 1.454735740450026, | |
| "grad_norm": 8879.030864305605, | |
| "learning_rate": 3.061816912242265e-05, | |
| "logits/chosen": -3.390625, | |
| "logits/rejected": -1.3984375, | |
| "logps/chosen": -43.75, | |
| "logps/rejected": -66.5, | |
| "loss": 1251.55, | |
| "rewards/accuracies": 0.8187500238418579, | |
| "rewards/chosen": -0.42578125, | |
| "rewards/margins": 0.2275390625, | |
| "rewards/rejected": -0.65234375, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 1.4652014652014653, | |
| "grad_norm": 10709.248753680347, | |
| "learning_rate": 3.0320897756713785e-05, | |
| "logits/chosen": -2.796875, | |
| "logits/rejected": -1.7578125, | |
| "logps/chosen": -40.75, | |
| "logps/rejected": -65.5, | |
| "loss": 1084.875, | |
| "rewards/accuracies": 0.8062499761581421, | |
| "rewards/chosen": -0.39453125, | |
| "rewards/margins": 0.2490234375, | |
| "rewards/rejected": -0.64453125, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 1.4652014652014653, | |
| "eval_logits/chosen": -1.5078125, | |
| "eval_logits/rejected": 0.0020751953125, | |
| "eval_logps/chosen": -48.0, | |
| "eval_logps/rejected": -61.5, | |
| "eval_loss": 1926.9439697265625, | |
| "eval_rewards/accuracies": 0.705078125, | |
| "eval_rewards/chosen": -0.46875, | |
| "eval_rewards/margins": 0.1337890625, | |
| "eval_rewards/rejected": -0.6015625, | |
| "eval_runtime": 46.6627, | |
| "eval_samples_per_second": 42.861, | |
| "eval_steps_per_second": 0.686, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 1.4756671899529041, | |
| "grad_norm": 9666.230125982887, | |
| "learning_rate": 3.002283623388332e-05, | |
| "logits/chosen": -2.375, | |
| "logits/rejected": -1.1015625, | |
| "logps/chosen": -42.5, | |
| "logps/rejected": -69.0, | |
| "loss": 1077.15, | |
| "rewards/accuracies": 0.831250011920929, | |
| "rewards/chosen": -0.412109375, | |
| "rewards/margins": 0.265625, | |
| "rewards/rejected": -0.6796875, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 1.4861329147043434, | |
| "grad_norm": 10303.308050384552, | |
| "learning_rate": 2.9724028816280507e-05, | |
| "logits/chosen": -1.4296875, | |
| "logits/rejected": -0.076171875, | |
| "logps/chosen": -40.5, | |
| "logps/rejected": -64.5, | |
| "loss": 1134.025, | |
| "rewards/accuracies": 0.8062499761581421, | |
| "rewards/chosen": -0.392578125, | |
| "rewards/margins": 0.240234375, | |
| "rewards/rejected": -0.6328125, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 1.4965986394557822, | |
| "grad_norm": 8111.223373735439, | |
| "learning_rate": 2.9424519877020524e-05, | |
| "logits/chosen": -2.1875, | |
| "logits/rejected": -0.59765625, | |
| "logps/chosen": -38.75, | |
| "logps/rejected": -69.0, | |
| "loss": 1053.025, | |
| "rewards/accuracies": 0.8687499761581421, | |
| "rewards/chosen": -0.375, | |
| "rewards/margins": 0.3046875, | |
| "rewards/rejected": -0.6796875, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 1.5070643642072215, | |
| "grad_norm": 9271.022934240376, | |
| "learning_rate": 2.912435389339504e-05, | |
| "logits/chosen": -1.8828125, | |
| "logits/rejected": -0.474609375, | |
| "logps/chosen": -41.25, | |
| "logps/rejected": -70.5, | |
| "loss": 1076.0, | |
| "rewards/accuracies": 0.875, | |
| "rewards/chosen": -0.400390625, | |
| "rewards/margins": 0.294921875, | |
| "rewards/rejected": -0.6953125, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 1.5175300889586603, | |
| "grad_norm": 10795.556968506937, | |
| "learning_rate": 2.8823575440267226e-05, | |
| "logits/chosen": -1.8671875, | |
| "logits/rejected": -0.359375, | |
| "logps/chosen": -39.0, | |
| "logps/rejected": -67.0, | |
| "loss": 1066.1125, | |
| "rewards/accuracies": 0.8500000238418579, | |
| "rewards/chosen": -0.37890625, | |
| "rewards/margins": 0.279296875, | |
| "rewards/rejected": -0.66015625, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 1.5279958137100995, | |
| "grad_norm": 13074.968160019085, | |
| "learning_rate": 2.8522229183452465e-05, | |
| "logits/chosen": -1.890625, | |
| "logits/rejected": -0.498046875, | |
| "logps/chosen": -37.0, | |
| "logps/rejected": -65.5, | |
| "loss": 1025.6, | |
| "rewards/accuracies": 0.8812500238418579, | |
| "rewards/chosen": -0.357421875, | |
| "rewards/margins": 0.28515625, | |
| "rewards/rejected": -0.64453125, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 1.5384615384615383, | |
| "grad_norm": 9864.762264491641, | |
| "learning_rate": 2.8220359873085373e-05, | |
| "logits/chosen": -2.65625, | |
| "logits/rejected": -1.8203125, | |
| "logps/chosen": -37.5, | |
| "logps/rejected": -65.0, | |
| "loss": 1193.8, | |
| "rewards/accuracies": 0.875, | |
| "rewards/chosen": -0.36328125, | |
| "rewards/margins": 0.27734375, | |
| "rewards/rejected": -0.640625, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 1.5489272632129776, | |
| "grad_norm": 7831.786878143211, | |
| "learning_rate": 2.7918012336974388e-05, | |
| "logits/chosen": -1.8515625, | |
| "logits/rejected": 0.08203125, | |
| "logps/chosen": -41.75, | |
| "logps/rejected": -68.0, | |
| "loss": 1150.0375, | |
| "rewards/accuracies": 0.8062499761581421, | |
| "rewards/chosen": -0.40625, | |
| "rewards/margins": 0.26171875, | |
| "rewards/rejected": -0.66796875, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 1.5593929879644164, | |
| "grad_norm": 8967.435998707904, | |
| "learning_rate": 2.7615231473944815e-05, | |
| "logits/chosen": -1.953125, | |
| "logits/rejected": 0.0595703125, | |
| "logps/chosen": -40.75, | |
| "logps/rejected": -65.5, | |
| "loss": 1070.975, | |
| "rewards/accuracies": 0.862500011920929, | |
| "rewards/chosen": -0.396484375, | |
| "rewards/margins": 0.24609375, | |
| "rewards/rejected": -0.640625, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 1.5698587127158556, | |
| "grad_norm": 6952.356226028971, | |
| "learning_rate": 2.7312062247171327e-05, | |
| "logits/chosen": -3.390625, | |
| "logits/rejected": -1.484375, | |
| "logps/chosen": -38.75, | |
| "logps/rejected": -65.5, | |
| "loss": 1031.2313, | |
| "rewards/accuracies": 0.893750011920929, | |
| "rewards/chosen": -0.375, | |
| "rewards/margins": 0.265625, | |
| "rewards/rejected": -0.640625, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.5698587127158556, | |
| "eval_logits/chosen": -1.6640625, | |
| "eval_logits/rejected": -0.1064453125, | |
| "eval_logps/chosen": -42.0, | |
| "eval_logps/rejected": -56.75, | |
| "eval_loss": 1931.583984375, | |
| "eval_rewards/accuracies": 0.703125, | |
| "eval_rewards/chosen": -0.408203125, | |
| "eval_rewards/margins": 0.146484375, | |
| "eval_rewards/rejected": -0.5546875, | |
| "eval_runtime": 46.6756, | |
| "eval_samples_per_second": 42.849, | |
| "eval_steps_per_second": 0.686, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.5803244374672945, | |
| "grad_norm": 9277.218612214656, | |
| "learning_rate": 2.7008549677500876e-05, | |
| "logits/chosen": -2.359375, | |
| "logits/rejected": -0.7421875, | |
| "logps/chosen": -37.75, | |
| "logps/rejected": -67.5, | |
| "loss": 1031.0563, | |
| "rewards/accuracies": 0.8999999761581421, | |
| "rewards/chosen": -0.365234375, | |
| "rewards/margins": 0.298828125, | |
| "rewards/rejected": -0.6640625, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 1.5907901622187337, | |
| "grad_norm": 10192.700017356783, | |
| "learning_rate": 2.670473883676709e-05, | |
| "logits/chosen": -2.546875, | |
| "logits/rejected": -0.84765625, | |
| "logps/chosen": -44.75, | |
| "logps/rejected": -70.5, | |
| "loss": 1109.45, | |
| "rewards/accuracies": 0.8812500238418579, | |
| "rewards/chosen": -0.4375, | |
| "rewards/margins": 0.255859375, | |
| "rewards/rejected": -0.69140625, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 1.6012558869701727, | |
| "grad_norm": 8462.581154368665, | |
| "learning_rate": 2.6400674841097074e-05, | |
| "logits/chosen": -3.15625, | |
| "logits/rejected": -0.87890625, | |
| "logps/chosen": -40.25, | |
| "logps/rejected": -68.0, | |
| "loss": 1058.2625, | |
| "rewards/accuracies": 0.84375, | |
| "rewards/chosen": -0.390625, | |
| "rewards/margins": 0.279296875, | |
| "rewards/rejected": -0.66796875, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 1.6117216117216118, | |
| "grad_norm": 7524.475468446226, | |
| "learning_rate": 2.609640284421158e-05, | |
| "logits/chosen": -2.71875, | |
| "logits/rejected": -0.90234375, | |
| "logps/chosen": -44.75, | |
| "logps/rejected": -71.0, | |
| "loss": 1165.125, | |
| "rewards/accuracies": 0.862500011920929, | |
| "rewards/chosen": -0.435546875, | |
| "rewards/margins": 0.259765625, | |
| "rewards/rejected": -0.6953125, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 1.6221873364730508, | |
| "grad_norm": 8871.088977019943, | |
| "learning_rate": 2.5791968030719726e-05, | |
| "logits/chosen": -2.515625, | |
| "logits/rejected": -1.09375, | |
| "logps/chosen": -42.75, | |
| "logps/rejected": -72.5, | |
| "loss": 1106.3625, | |
| "rewards/accuracies": 0.887499988079071, | |
| "rewards/chosen": -0.416015625, | |
| "rewards/margins": 0.294921875, | |
| "rewards/rejected": -0.7109375, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 1.6326530612244898, | |
| "grad_norm": 7471.0132589589575, | |
| "learning_rate": 2.548741560940902e-05, | |
| "logits/chosen": -1.9375, | |
| "logits/rejected": 0.013671875, | |
| "logps/chosen": -38.75, | |
| "logps/rejected": -70.5, | |
| "loss": 1148.125, | |
| "rewards/accuracies": 0.90625, | |
| "rewards/chosen": -0.375, | |
| "rewards/margins": 0.31640625, | |
| "rewards/rejected": -0.69140625, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 1.6431187859759289, | |
| "grad_norm": 8269.446489050108, | |
| "learning_rate": 2.5182790806531787e-05, | |
| "logits/chosen": -1.859375, | |
| "logits/rejected": 0.2890625, | |
| "logps/chosen": -39.0, | |
| "logps/rejected": -67.5, | |
| "loss": 1015.9875, | |
| "rewards/accuracies": 0.875, | |
| "rewards/chosen": -0.37890625, | |
| "rewards/margins": 0.28515625, | |
| "rewards/rejected": -0.6640625, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 1.653584510727368, | |
| "grad_norm": 7316.039226485421, | |
| "learning_rate": 2.4878138859089074e-05, | |
| "logits/chosen": -2.359375, | |
| "logits/rejected": -0.1376953125, | |
| "logps/chosen": -38.0, | |
| "logps/rejected": -67.0, | |
| "loss": 987.9125, | |
| "rewards/accuracies": 0.856249988079071, | |
| "rewards/chosen": -0.3671875, | |
| "rewards/margins": 0.291015625, | |
| "rewards/rejected": -0.66015625, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 1.664050235478807, | |
| "grad_norm": 9893.591542860187, | |
| "learning_rate": 2.4573505008112924e-05, | |
| "logits/chosen": -2.53125, | |
| "logits/rejected": -0.2275390625, | |
| "logps/chosen": -40.0, | |
| "logps/rejected": -68.0, | |
| "loss": 1048.725, | |
| "rewards/accuracies": 0.862500011920929, | |
| "rewards/chosen": -0.388671875, | |
| "rewards/margins": 0.279296875, | |
| "rewards/rejected": -0.66796875, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 1.674515960230246, | |
| "grad_norm": 9995.774821829142, | |
| "learning_rate": 2.4268934491948032e-05, | |
| "logits/chosen": -2.328125, | |
| "logits/rejected": -0.66796875, | |
| "logps/chosen": -37.75, | |
| "logps/rejected": -65.5, | |
| "loss": 1090.75, | |
| "rewards/accuracies": 0.856249988079071, | |
| "rewards/chosen": -0.365234375, | |
| "rewards/margins": 0.27734375, | |
| "rewards/rejected": -0.64453125, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 1.674515960230246, | |
| "eval_logits/chosen": -1.375, | |
| "eval_logits/rejected": 0.048583984375, | |
| "eval_logps/chosen": -44.25, | |
| "eval_logps/rejected": -58.25, | |
| "eval_loss": 1936.1280517578125, | |
| "eval_rewards/accuracies": 0.697265625, | |
| "eval_rewards/chosen": -0.431640625, | |
| "eval_rewards/margins": 0.1396484375, | |
| "eval_rewards/rejected": -0.5703125, | |
| "eval_runtime": 46.6439, | |
| "eval_samples_per_second": 42.878, | |
| "eval_steps_per_second": 0.686, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 1.684981684981685, | |
| "grad_norm": 7038.989244645223, | |
| "learning_rate": 2.396447253953385e-05, | |
| "logits/chosen": -2.546875, | |
| "logits/rejected": -0.9375, | |
| "logps/chosen": -41.75, | |
| "logps/rejected": -69.0, | |
| "loss": 1195.05, | |
| "rewards/accuracies": 0.856249988079071, | |
| "rewards/chosen": -0.40625, | |
| "rewards/margins": 0.271484375, | |
| "rewards/rejected": -0.67578125, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 1.695447409733124, | |
| "grad_norm": 8599.815631859326, | |
| "learning_rate": 2.3660164363687998e-05, | |
| "logits/chosen": -2.21875, | |
| "logits/rejected": -0.21484375, | |
| "logps/chosen": -37.25, | |
| "logps/rejected": -67.0, | |
| "loss": 1089.8875, | |
| "rewards/accuracies": 0.8999999761581421, | |
| "rewards/chosen": -0.361328125, | |
| "rewards/margins": 0.296875, | |
| "rewards/rejected": -0.65625, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 1.705913134484563, | |
| "grad_norm": 10626.30645492799, | |
| "learning_rate": 2.33560551543922e-05, | |
| "logits/chosen": -1.9921875, | |
| "logits/rejected": -0.91796875, | |
| "logps/chosen": -38.5, | |
| "logps/rejected": -62.5, | |
| "loss": 1061.85, | |
| "rewards/accuracies": 0.8812500238418579, | |
| "rewards/chosen": -0.373046875, | |
| "rewards/margins": 0.240234375, | |
| "rewards/rejected": -0.61328125, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 1.716378859236002, | |
| "grad_norm": 10763.633862520066, | |
| "learning_rate": 2.3052190072081493e-05, | |
| "logits/chosen": -2.96875, | |
| "logits/rejected": -1.5625, | |
| "logps/chosen": -43.25, | |
| "logps/rejected": -70.5, | |
| "loss": 1122.325, | |
| "rewards/accuracies": 0.8500000238418579, | |
| "rewards/chosen": -0.421875, | |
| "rewards/margins": 0.26953125, | |
| "rewards/rejected": -0.69140625, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 1.7268445839874411, | |
| "grad_norm": 8849.159213814466, | |
| "learning_rate": 2.2748614240937864e-05, | |
| "logits/chosen": -3.53125, | |
| "logits/rejected": -1.7578125, | |
| "logps/chosen": -38.75, | |
| "logps/rejected": -69.5, | |
| "loss": 1024.4, | |
| "rewards/accuracies": 0.8999999761581421, | |
| "rewards/chosen": -0.376953125, | |
| "rewards/margins": 0.306640625, | |
| "rewards/rejected": -0.68359375, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 1.7373103087388801, | |
| "grad_norm": 10079.09023346274, | |
| "learning_rate": 2.2445372742189334e-05, | |
| "logits/chosen": -2.59375, | |
| "logits/rejected": -1.3046875, | |
| "logps/chosen": -42.75, | |
| "logps/rejected": -71.0, | |
| "loss": 1100.4875, | |
| "rewards/accuracies": 0.8687499761581421, | |
| "rewards/chosen": -0.416015625, | |
| "rewards/margins": 0.283203125, | |
| "rewards/rejected": -0.69921875, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 1.7477760334903192, | |
| "grad_norm": 9121.97349017397, | |
| "learning_rate": 2.214251060741528e-05, | |
| "logits/chosen": -3.0625, | |
| "logits/rejected": -1.15625, | |
| "logps/chosen": -44.5, | |
| "logps/rejected": -71.0, | |
| "loss": 1091.0125, | |
| "rewards/accuracies": 0.8500000238418579, | |
| "rewards/chosen": -0.43359375, | |
| "rewards/margins": 0.265625, | |
| "rewards/rejected": -0.69921875, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 1.7582417582417582, | |
| "grad_norm": 9357.037647793195, | |
| "learning_rate": 2.1840072811859302e-05, | |
| "logits/chosen": -2.375, | |
| "logits/rejected": -0.75390625, | |
| "logps/chosen": -39.0, | |
| "logps/rejected": -67.0, | |
| "loss": 1107.35, | |
| "rewards/accuracies": 0.887499988079071, | |
| "rewards/chosen": -0.37890625, | |
| "rewards/margins": 0.283203125, | |
| "rewards/rejected": -0.66015625, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 1.7687074829931972, | |
| "grad_norm": 10279.44403350473, | |
| "learning_rate": 2.1538104267750287e-05, | |
| "logits/chosen": -3.390625, | |
| "logits/rejected": -0.84375, | |
| "logps/chosen": -43.0, | |
| "logps/rejected": -70.5, | |
| "loss": 1014.4, | |
| "rewards/accuracies": 0.8812500238418579, | |
| "rewards/chosen": -0.419921875, | |
| "rewards/margins": 0.2734375, | |
| "rewards/rejected": -0.6953125, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 1.7791732077446363, | |
| "grad_norm": 9264.175201513302, | |
| "learning_rate": 2.1236649817632952e-05, | |
| "logits/chosen": -3.15625, | |
| "logits/rejected": -1.6171875, | |
| "logps/chosen": -40.0, | |
| "logps/rejected": -66.5, | |
| "loss": 1097.5375, | |
| "rewards/accuracies": 0.8812500238418579, | |
| "rewards/chosen": -0.38671875, | |
| "rewards/margins": 0.267578125, | |
| "rewards/rejected": -0.65625, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 1.7791732077446363, | |
| "eval_logits/chosen": -2.234375, | |
| "eval_logits/rejected": -0.66015625, | |
| "eval_logps/chosen": -47.5, | |
| "eval_logps/rejected": -62.0, | |
| "eval_loss": 1975.2960205078125, | |
| "eval_rewards/accuracies": 0.70703125, | |
| "eval_rewards/chosen": -0.46484375, | |
| "eval_rewards/margins": 0.14453125, | |
| "eval_rewards/rejected": -0.609375, | |
| "eval_runtime": 46.6341, | |
| "eval_samples_per_second": 42.887, | |
| "eval_steps_per_second": 0.686, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 1.7896389324960753, | |
| "grad_norm": 8784.877449207876, | |
| "learning_rate": 2.093575422770872e-05, | |
| "logits/chosen": -3.125, | |
| "logits/rejected": -1.125, | |
| "logps/chosen": -45.0, | |
| "logps/rejected": -74.5, | |
| "loss": 1117.75, | |
| "rewards/accuracies": 0.8812500238418579, | |
| "rewards/chosen": -0.4375, | |
| "rewards/margins": 0.29296875, | |
| "rewards/rejected": -0.73046875, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 1.8001046572475143, | |
| "grad_norm": 10219.448663581172, | |
| "learning_rate": 2.0635462181187827e-05, | |
| "logits/chosen": -2.484375, | |
| "logits/rejected": -0.99609375, | |
| "logps/chosen": -45.0, | |
| "logps/rejected": -72.0, | |
| "loss": 1013.7125, | |
| "rewards/accuracies": 0.8687499761581421, | |
| "rewards/chosen": -0.4375, | |
| "rewards/margins": 0.267578125, | |
| "rewards/rejected": -0.70703125, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 1.8105703819989536, | |
| "grad_norm": 9720.261798588188, | |
| "learning_rate": 2.0335818271653917e-05, | |
| "logits/chosen": -2.71875, | |
| "logits/rejected": -0.68359375, | |
| "logps/chosen": -38.0, | |
| "logps/rejected": -67.5, | |
| "loss": 1072.375, | |
| "rewards/accuracies": 0.8812500238418579, | |
| "rewards/chosen": -0.369140625, | |
| "rewards/margins": 0.29296875, | |
| "rewards/rejected": -0.66015625, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 1.8210361067503924, | |
| "grad_norm": 9492.664346406444, | |
| "learning_rate": 2.0036866996441815e-05, | |
| "logits/chosen": -1.828125, | |
| "logits/rejected": -0.05517578125, | |
| "logps/chosen": -39.25, | |
| "logps/rejected": -67.5, | |
| "loss": 1050.4125, | |
| "rewards/accuracies": 0.893750011920929, | |
| "rewards/chosen": -0.37890625, | |
| "rewards/margins": 0.28515625, | |
| "rewards/rejected": -0.6640625, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 1.8315018315018317, | |
| "grad_norm": 11311.667425346242, | |
| "learning_rate": 1.9738652750029677e-05, | |
| "logits/chosen": -1.921875, | |
| "logits/rejected": 0.23828125, | |
| "logps/chosen": -40.5, | |
| "logps/rejected": -67.5, | |
| "loss": 1152.8375, | |
| "rewards/accuracies": 0.875, | |
| "rewards/chosen": -0.39453125, | |
| "rewards/margins": 0.271484375, | |
| "rewards/rejected": -0.6640625, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 1.8419675562532705, | |
| "grad_norm": 9088.398258506126, | |
| "learning_rate": 1.944121981744631e-05, | |
| "logits/chosen": -1.6484375, | |
| "logits/rejected": -0.416015625, | |
| "logps/chosen": -42.25, | |
| "logps/rejected": -68.5, | |
| "loss": 1055.575, | |
| "rewards/accuracies": 0.8500000238418579, | |
| "rewards/chosen": -0.412109375, | |
| "rewards/margins": 0.26171875, | |
| "rewards/rejected": -0.67578125, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 1.8524332810047097, | |
| "grad_norm": 7761.853592937917, | |
| "learning_rate": 1.9144612367694887e-05, | |
| "logits/chosen": -1.9375, | |
| "logits/rejected": -0.74609375, | |
| "logps/chosen": -37.5, | |
| "logps/rejected": -66.5, | |
| "loss": 1022.4875, | |
| "rewards/accuracies": 0.8687499761581421, | |
| "rewards/chosen": -0.36328125, | |
| "rewards/margins": 0.291015625, | |
| "rewards/rejected": -0.65234375, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 1.8628990057561485, | |
| "grad_norm": 9759.990942068207, | |
| "learning_rate": 1.8848874447193803e-05, | |
| "logits/chosen": -2.359375, | |
| "logits/rejected": -1.1015625, | |
| "logps/chosen": -41.0, | |
| "logps/rejected": -71.0, | |
| "loss": 1096.6, | |
| "rewards/accuracies": 0.8999999761581421, | |
| "rewards/chosen": -0.3984375, | |
| "rewards/margins": 0.296875, | |
| "rewards/rejected": -0.6953125, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 1.8733647305075878, | |
| "grad_norm": 9243.508889928567, | |
| "learning_rate": 1.855404997323571e-05, | |
| "logits/chosen": -1.796875, | |
| "logits/rejected": -0.10693359375, | |
| "logps/chosen": -38.75, | |
| "logps/rejected": -66.0, | |
| "loss": 1058.8312, | |
| "rewards/accuracies": 0.831250011920929, | |
| "rewards/chosen": -0.376953125, | |
| "rewards/margins": 0.271484375, | |
| "rewards/rejected": -0.6484375, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 1.8838304552590266, | |
| "grad_norm": 8579.526765575463, | |
| "learning_rate": 1.82601827274658e-05, | |
| "logits/chosen": -0.81640625, | |
| "logits/rejected": 0.271484375, | |
| "logps/chosen": -39.5, | |
| "logps/rejected": -70.0, | |
| "loss": 1031.15, | |
| "rewards/accuracies": 0.875, | |
| "rewards/chosen": -0.384765625, | |
| "rewards/margins": 0.302734375, | |
| "rewards/rejected": -0.6875, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 1.8838304552590266, | |
| "eval_logits/chosen": -0.8125, | |
| "eval_logits/rejected": 0.451171875, | |
| "eval_logps/chosen": -48.0, | |
| "eval_logps/rejected": -62.25, | |
| "eval_loss": 1964.511962890625, | |
| "eval_rewards/accuracies": 0.708984375, | |
| "eval_rewards/chosen": -0.466796875, | |
| "eval_rewards/margins": 0.1416015625, | |
| "eval_rewards/rejected": -0.609375, | |
| "eval_runtime": 46.6426, | |
| "eval_samples_per_second": 42.879, | |
| "eval_steps_per_second": 0.686, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 1.8942961800104658, | |
| "grad_norm": 8135.3261433073, | |
| "learning_rate": 1.7967316349380213e-05, | |
| "logits/chosen": -1.578125, | |
| "logits/rejected": -0.6953125, | |
| "logps/chosen": -41.5, | |
| "logps/rejected": -72.0, | |
| "loss": 1102.7875, | |
| "rewards/accuracies": 0.893750011920929, | |
| "rewards/chosen": -0.404296875, | |
| "rewards/margins": 0.30078125, | |
| "rewards/rejected": -0.70703125, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 1.9047619047619047, | |
| "grad_norm": 7825.056542443668, | |
| "learning_rate": 1.7675494329845516e-05, | |
| "logits/chosen": -1.296875, | |
| "logits/rejected": -0.125, | |
| "logps/chosen": -41.25, | |
| "logps/rejected": -70.0, | |
| "loss": 1035.3375, | |
| "rewards/accuracies": 0.862500011920929, | |
| "rewards/chosen": -0.40234375, | |
| "rewards/margins": 0.28515625, | |
| "rewards/rejected": -0.6875, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 1.915227629513344, | |
| "grad_norm": 11461.051859237983, | |
| "learning_rate": 1.738476000464026e-05, | |
| "logits/chosen": -1.3125, | |
| "logits/rejected": -0.875, | |
| "logps/chosen": -40.25, | |
| "logps/rejected": -67.5, | |
| "loss": 1084.275, | |
| "rewards/accuracies": 0.8374999761581421, | |
| "rewards/chosen": -0.390625, | |
| "rewards/margins": 0.26953125, | |
| "rewards/rejected": -0.66015625, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 1.9256933542647827, | |
| "grad_norm": 12433.213268186713, | |
| "learning_rate": 1.7095156548019646e-05, | |
| "logits/chosen": -2.28125, | |
| "logits/rejected": -0.412109375, | |
| "logps/chosen": -41.0, | |
| "logps/rejected": -71.0, | |
| "loss": 1066.6875, | |
| "rewards/accuracies": 0.862500011920929, | |
| "rewards/chosen": -0.3984375, | |
| "rewards/margins": 0.30078125, | |
| "rewards/rejected": -0.69921875, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 1.936159079016222, | |
| "grad_norm": 8306.651614769393, | |
| "learning_rate": 1.6806726966304064e-05, | |
| "logits/chosen": -1.9609375, | |
| "logits/rejected": -0.53515625, | |
| "logps/chosen": -40.25, | |
| "logps/rejected": -68.0, | |
| "loss": 1081.3125, | |
| "rewards/accuracies": 0.887499988079071, | |
| "rewards/chosen": -0.392578125, | |
| "rewards/margins": 0.275390625, | |
| "rewards/rejected": -0.66796875, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 1.9466248037676608, | |
| "grad_norm": 9222.62135037718, | |
| "learning_rate": 1.6519514091492625e-05, | |
| "logits/chosen": -1.7421875, | |
| "logits/rejected": 0.11279296875, | |
| "logps/chosen": -40.0, | |
| "logps/rejected": -67.5, | |
| "loss": 1005.5125, | |
| "rewards/accuracies": 0.875, | |
| "rewards/chosen": -0.388671875, | |
| "rewards/margins": 0.2734375, | |
| "rewards/rejected": -0.6640625, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 1.9570905285191, | |
| "grad_norm": 7198.0091554735, | |
| "learning_rate": 1.6233560574902627e-05, | |
| "logits/chosen": -2.15625, | |
| "logits/rejected": 0.25, | |
| "logps/chosen": -34.5, | |
| "logps/rejected": -65.5, | |
| "loss": 1023.975, | |
| "rewards/accuracies": 0.9125000238418579, | |
| "rewards/chosen": -0.333984375, | |
| "rewards/margins": 0.30859375, | |
| "rewards/rejected": -0.64453125, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 1.9675562532705388, | |
| "grad_norm": 9333.908724930976, | |
| "learning_rate": 1.5948908880835753e-05, | |
| "logits/chosen": -2.859375, | |
| "logits/rejected": -0.8671875, | |
| "logps/chosen": -42.0, | |
| "logps/rejected": -69.0, | |
| "loss": 1058.6625, | |
| "rewards/accuracies": 0.8500000238418579, | |
| "rewards/chosen": -0.408203125, | |
| "rewards/margins": 0.267578125, | |
| "rewards/rejected": -0.67578125, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 1.978021978021978, | |
| "grad_norm": 10625.75203106232, | |
| "learning_rate": 1.5665601280272124e-05, | |
| "logits/chosen": -2.234375, | |
| "logits/rejected": -0.60546875, | |
| "logps/chosen": -42.5, | |
| "logps/rejected": -69.0, | |
| "loss": 1106.7125, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": -0.416015625, | |
| "rewards/margins": 0.263671875, | |
| "rewards/rejected": -0.6796875, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 1.988487702773417, | |
| "grad_norm": 8536.343422928834, | |
| "learning_rate": 1.538367984459298e-05, | |
| "logits/chosen": -1.7890625, | |
| "logits/rejected": -0.671875, | |
| "logps/chosen": -41.0, | |
| "logps/rejected": -71.5, | |
| "loss": 1012.0125, | |
| "rewards/accuracies": 0.8999999761581421, | |
| "rewards/chosen": -0.3984375, | |
| "rewards/margins": 0.306640625, | |
| "rewards/rejected": -0.703125, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 1.988487702773417, | |
| "eval_logits/chosen": -0.7578125, | |
| "eval_logits/rejected": 0.61328125, | |
| "eval_logps/chosen": -46.25, | |
| "eval_logps/rejected": -60.25, | |
| "eval_loss": 1937.0240478515625, | |
| "eval_rewards/accuracies": 0.703125, | |
| "eval_rewards/chosen": -0.451171875, | |
| "eval_rewards/margins": 0.140625, | |
| "eval_rewards/rejected": -0.58984375, | |
| "eval_runtime": 46.6417, | |
| "eval_samples_per_second": 42.88, | |
| "eval_steps_per_second": 0.686, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 1.9989534275248562, | |
| "grad_norm": 10292.385522443983, | |
| "learning_rate": 1.5103186439333133e-05, | |
| "logits/chosen": -0.84375, | |
| "logits/rejected": 0.98828125, | |
| "logps/chosen": -40.5, | |
| "logps/rejected": -69.5, | |
| "loss": 1054.25, | |
| "rewards/accuracies": 0.875, | |
| "rewards/chosen": -0.392578125, | |
| "rewards/margins": 0.2890625, | |
| "rewards/rejected": -0.68359375, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 2.009419152276295, | |
| "grad_norm": 3404.3900099848256, | |
| "learning_rate": 1.4824162717963829e-05, | |
| "logits/chosen": -2.484375, | |
| "logits/rejected": -0.09130859375, | |
| "logps/chosen": -35.25, | |
| "logps/rejected": -80.0, | |
| "loss": 334.6734, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -0.33984375, | |
| "rewards/margins": 0.4453125, | |
| "rewards/rejected": -0.78515625, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 2.0198848770277342, | |
| "grad_norm": 3042.697635714066, | |
| "learning_rate": 1.4546650115707192e-05, | |
| "logits/chosen": -0.81640625, | |
| "logits/rejected": 0.77734375, | |
| "logps/chosen": -34.5, | |
| "logps/rejected": -79.5, | |
| "loss": 295.2094, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -0.333984375, | |
| "rewards/margins": 0.44921875, | |
| "rewards/rejected": -0.78515625, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 2.030350601779173, | |
| "grad_norm": 3688.6942973537653, | |
| "learning_rate": 1.4270689843383111e-05, | |
| "logits/chosen": -1.2421875, | |
| "logits/rejected": 0.423828125, | |
| "logps/chosen": -37.0, | |
| "logps/rejected": -79.0, | |
| "loss": 293.6031, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -0.359375, | |
| "rewards/margins": 0.419921875, | |
| "rewards/rejected": -0.77734375, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 2.0408163265306123, | |
| "grad_norm": 5299.3148495287605, | |
| "learning_rate": 1.3996322881289347e-05, | |
| "logits/chosen": -1.953125, | |
| "logits/rejected": 0.640625, | |
| "logps/chosen": -34.5, | |
| "logps/rejected": -77.5, | |
| "loss": 296.2625, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -0.333984375, | |
| "rewards/margins": 0.4296875, | |
| "rewards/rejected": -0.76171875, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 2.051282051282051, | |
| "grad_norm": 3786.1928929717064, | |
| "learning_rate": 1.3723589973115963e-05, | |
| "logits/chosen": -1.890625, | |
| "logits/rejected": -1.25, | |
| "logps/chosen": -37.5, | |
| "logps/rejected": -80.5, | |
| "loss": 284.2031, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -0.36328125, | |
| "rewards/margins": 0.431640625, | |
| "rewards/rejected": -0.796875, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 2.0617477760334904, | |
| "grad_norm": 4923.235483322859, | |
| "learning_rate": 1.3452531619894837e-05, | |
| "logits/chosen": -2.515625, | |
| "logits/rejected": -0.28515625, | |
| "logps/chosen": -32.0, | |
| "logps/rejected": -78.5, | |
| "loss": 298.7875, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -0.30859375, | |
| "rewards/margins": 0.46484375, | |
| "rewards/rejected": -0.7734375, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 2.072213500784929, | |
| "grad_norm": 3097.9047416637304, | |
| "learning_rate": 1.3183188073985245e-05, | |
| "logits/chosen": -2.078125, | |
| "logits/rejected": -0.94921875, | |
| "logps/chosen": -37.0, | |
| "logps/rejected": -81.5, | |
| "loss": 283.2094, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -0.359375, | |
| "rewards/margins": 0.447265625, | |
| "rewards/rejected": -0.8046875, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 2.0826792255363684, | |
| "grad_norm": 3785.553091279569, | |
| "learning_rate": 1.2915599333096351e-05, | |
| "logits/chosen": -1.8125, | |
| "logits/rejected": -0.1298828125, | |
| "logps/chosen": -32.75, | |
| "logps/rejected": -76.5, | |
| "loss": 275.4453, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": -0.31640625, | |
| "rewards/margins": 0.4375, | |
| "rewards/rejected": -0.75, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 2.0931449502878072, | |
| "grad_norm": 3344.601291463624, | |
| "learning_rate": 1.264980513434752e-05, | |
| "logits/chosen": -1.1953125, | |
| "logits/rejected": 0.77734375, | |
| "logps/chosen": -36.25, | |
| "logps/rejected": -81.5, | |
| "loss": 262.0437, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -0.3515625, | |
| "rewards/margins": 0.44921875, | |
| "rewards/rejected": -0.80078125, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 2.0931449502878072, | |
| "eval_logits/chosen": -0.875, | |
| "eval_logits/rejected": 0.54296875, | |
| "eval_logps/chosen": -47.75, | |
| "eval_logps/rejected": -60.75, | |
| "eval_loss": 1950.9439697265625, | |
| "eval_rewards/accuracies": 0.689453125, | |
| "eval_rewards/chosen": -0.466796875, | |
| "eval_rewards/margins": 0.130859375, | |
| "eval_rewards/rejected": -0.59765625, | |
| "eval_runtime": 46.6397, | |
| "eval_samples_per_second": 42.882, | |
| "eval_steps_per_second": 0.686, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 2.1036106750392465, | |
| "grad_norm": 3830.2668567613723, | |
| "learning_rate": 1.2385844948367324e-05, | |
| "logits/chosen": -2.484375, | |
| "logits/rejected": -0.54296875, | |
| "logps/chosen": -33.25, | |
| "logps/rejected": -78.0, | |
| "loss": 297.3906, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.3203125, | |
| "rewards/margins": 0.447265625, | |
| "rewards/rejected": -0.76953125, | |
| "step": 2010 | |
| }, | |
| { | |
| "epoch": 2.1140763997906853, | |
| "grad_norm": 3642.7432810562595, | |
| "learning_rate": 1.2123757973432114e-05, | |
| "logits/chosen": -2.140625, | |
| "logits/rejected": -0.34375, | |
| "logps/chosen": -33.25, | |
| "logps/rejected": -80.5, | |
| "loss": 241.9187, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.322265625, | |
| "rewards/margins": 0.46875, | |
| "rewards/rejected": -0.7890625, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 2.1245421245421245, | |
| "grad_norm": 3167.084834423103, | |
| "learning_rate": 1.1863583129645106e-05, | |
| "logits/chosen": -2.109375, | |
| "logits/rejected": 0.2216796875, | |
| "logps/chosen": -32.25, | |
| "logps/rejected": -78.5, | |
| "loss": 274.7688, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -0.310546875, | |
| "rewards/margins": 0.46484375, | |
| "rewards/rejected": -0.7734375, | |
| "step": 2030 | |
| }, | |
| { | |
| "epoch": 2.1350078492935634, | |
| "grad_norm": 4181.52618149087, | |
| "learning_rate": 1.1605359053156607e-05, | |
| "logits/chosen": -2.015625, | |
| "logits/rejected": -1.3125, | |
| "logps/chosen": -36.75, | |
| "logps/rejected": -81.5, | |
| "loss": 289.4187, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -0.35546875, | |
| "rewards/margins": 0.447265625, | |
| "rewards/rejected": -0.8046875, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 2.1454735740450026, | |
| "grad_norm": 4098.998758672205, | |
| "learning_rate": 1.1349124090426644e-05, | |
| "logits/chosen": -1.6796875, | |
| "logits/rejected": 0.060546875, | |
| "logps/chosen": -35.25, | |
| "logps/rejected": -78.5, | |
| "loss": 286.8375, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -0.341796875, | |
| "rewards/margins": 0.43359375, | |
| "rewards/rejected": -0.7734375, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 2.155939298796442, | |
| "grad_norm": 3885.840013335884, | |
| "learning_rate": 1.1094916292530406e-05, | |
| "logits/chosen": -1.9609375, | |
| "logits/rejected": -0.326171875, | |
| "logps/chosen": -36.0, | |
| "logps/rejected": -79.5, | |
| "loss": 292.1344, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -0.34765625, | |
| "rewards/margins": 0.435546875, | |
| "rewards/rejected": -0.78515625, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 2.1664050235478807, | |
| "grad_norm": 3039.6946116565987, | |
| "learning_rate": 1.0842773409507623e-05, | |
| "logits/chosen": -1.5234375, | |
| "logits/rejected": -0.2109375, | |
| "logps/chosen": -37.5, | |
| "logps/rejected": -82.5, | |
| "loss": 253.5188, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -0.36328125, | |
| "rewards/margins": 0.4453125, | |
| "rewards/rejected": -0.8125, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 2.17687074829932, | |
| "grad_norm": 3464.812862285144, | |
| "learning_rate": 1.0592732884756754e-05, | |
| "logits/chosen": -1.9609375, | |
| "logits/rejected": -0.058837890625, | |
| "logps/chosen": -36.75, | |
| "logps/rejected": -81.5, | |
| "loss": 244.6594, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -0.357421875, | |
| "rewards/margins": 0.4453125, | |
| "rewards/rejected": -0.8046875, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 2.1873364730507587, | |
| "grad_norm": 3760.6082335100064, | |
| "learning_rate": 1.0344831849474507e-05, | |
| "logits/chosen": -2.234375, | |
| "logits/rejected": -0.0966796875, | |
| "logps/chosen": -37.5, | |
| "logps/rejected": -81.0, | |
| "loss": 266.0391, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -0.361328125, | |
| "rewards/margins": 0.43359375, | |
| "rewards/rejected": -0.796875, | |
| "step": 2090 | |
| }, | |
| { | |
| "epoch": 2.197802197802198, | |
| "grad_norm": 3932.41173164906, | |
| "learning_rate": 1.0099107117141879e-05, | |
| "logits/chosen": -2.03125, | |
| "logits/rejected": 0.12109375, | |
| "logps/chosen": -31.75, | |
| "logps/rejected": -79.0, | |
| "loss": 266.8375, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -0.306640625, | |
| "rewards/margins": 0.470703125, | |
| "rewards/rejected": -0.77734375, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 2.197802197802198, | |
| "eval_logits/chosen": -1.25, | |
| "eval_logits/rejected": 0.220703125, | |
| "eval_logps/chosen": -47.25, | |
| "eval_logps/rejected": -60.25, | |
| "eval_loss": 1943.8719482421875, | |
| "eval_rewards/accuracies": 0.708984375, | |
| "eval_rewards/chosen": -0.4609375, | |
| "eval_rewards/margins": 0.1279296875, | |
| "eval_rewards/rejected": -0.58984375, | |
| "eval_runtime": 46.6437, | |
| "eval_samples_per_second": 42.878, | |
| "eval_steps_per_second": 0.686, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 2.208267922553637, | |
| "grad_norm": 3337.652337760505, | |
| "learning_rate": 9.855595178057334e-06, | |
| "logits/chosen": -3.03125, | |
| "logits/rejected": -0.70703125, | |
| "logps/chosen": -35.0, | |
| "logps/rejected": -79.5, | |
| "loss": 274.5297, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": -0.337890625, | |
| "rewards/margins": 0.447265625, | |
| "rewards/rejected": -0.78515625, | |
| "step": 2110 | |
| }, | |
| { | |
| "epoch": 2.218733647305076, | |
| "grad_norm": 4470.906968618023, | |
| "learning_rate": 9.614332193917949e-06, | |
| "logits/chosen": -2.859375, | |
| "logits/rejected": -1.0234375, | |
| "logps/chosen": -35.0, | |
| "logps/rejected": -80.0, | |
| "loss": 275.0906, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -0.337890625, | |
| "rewards/margins": 0.451171875, | |
| "rewards/rejected": -0.7890625, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 2.229199372056515, | |
| "grad_norm": 2872.101613491522, | |
| "learning_rate": 9.375353992449385e-06, | |
| "logits/chosen": -2.703125, | |
| "logits/rejected": -0.71875, | |
| "logps/chosen": -32.25, | |
| "logps/rejected": -75.5, | |
| "loss": 279.3453, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -0.310546875, | |
| "rewards/margins": 0.4296875, | |
| "rewards/rejected": -0.7421875, | |
| "step": 2130 | |
| }, | |
| { | |
| "epoch": 2.239665096807954, | |
| "grad_norm": 3881.7549504165245, | |
| "learning_rate": 9.138696062085442e-06, | |
| "logits/chosen": -2.625, | |
| "logits/rejected": -0.859375, | |
| "logps/chosen": -36.0, | |
| "logps/rejected": -81.5, | |
| "loss": 245.9406, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -0.34765625, | |
| "rewards/margins": 0.45703125, | |
| "rewards/rejected": -0.8046875, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 2.250130821559393, | |
| "grad_norm": 3580.0301370150332, | |
| "learning_rate": 8.904393546698007e-06, | |
| "logits/chosen": -3.40625, | |
| "logits/rejected": -1.0234375, | |
| "logps/chosen": -32.5, | |
| "logps/rejected": -79.0, | |
| "loss": 260.1594, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.314453125, | |
| "rewards/margins": 0.46484375, | |
| "rewards/rejected": -0.78125, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 2.260596546310832, | |
| "grad_norm": 3704.930148331641, | |
| "learning_rate": 8.672481240378142e-06, | |
| "logits/chosen": -2.390625, | |
| "logits/rejected": -0.6796875, | |
| "logps/chosen": -31.875, | |
| "logps/rejected": -78.0, | |
| "loss": 282.3813, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -0.306640625, | |
| "rewards/margins": 0.458984375, | |
| "rewards/rejected": -0.765625, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 2.271062271062271, | |
| "grad_norm": 5125.6639928215, | |
| "learning_rate": 8.44299358226919e-06, | |
| "logits/chosen": -2.859375, | |
| "logits/rejected": -0.421875, | |
| "logps/chosen": -35.0, | |
| "logps/rejected": -79.5, | |
| "loss": 284.0906, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.3359375, | |
| "rewards/margins": 0.443359375, | |
| "rewards/rejected": -0.78125, | |
| "step": 2170 | |
| }, | |
| { | |
| "epoch": 2.2815279958137102, | |
| "grad_norm": 3967.456324463655, | |
| "learning_rate": 8.215964651452457e-06, | |
| "logits/chosen": -2.09375, | |
| "logits/rejected": -0.0771484375, | |
| "logps/chosen": -35.75, | |
| "logps/rejected": -80.0, | |
| "loss": 260.3852, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -0.345703125, | |
| "rewards/margins": 0.4453125, | |
| "rewards/rejected": -0.7890625, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 2.291993720565149, | |
| "grad_norm": 4171.682811610293, | |
| "learning_rate": 7.991428161886502e-06, | |
| "logits/chosen": -1.4140625, | |
| "logits/rejected": -0.373046875, | |
| "logps/chosen": -35.0, | |
| "logps/rejected": -79.0, | |
| "loss": 240.6531, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -0.337890625, | |
| "rewards/margins": 0.44140625, | |
| "rewards/rejected": -0.78125, | |
| "step": 2190 | |
| }, | |
| { | |
| "epoch": 2.3024594453165883, | |
| "grad_norm": 4579.135990984435, | |
| "learning_rate": 7.76941745740061e-06, | |
| "logits/chosen": -1.4375, | |
| "logits/rejected": 0.494140625, | |
| "logps/chosen": -34.25, | |
| "logps/rejected": -77.0, | |
| "loss": 284.8125, | |
| "rewards/accuracies": 0.956250011920929, | |
| "rewards/chosen": -0.33203125, | |
| "rewards/margins": 0.4296875, | |
| "rewards/rejected": -0.76171875, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 2.3024594453165883, | |
| "eval_logits/chosen": -0.55078125, | |
| "eval_logits/rejected": 0.81640625, | |
| "eval_logps/chosen": -49.75, | |
| "eval_logps/rejected": -62.75, | |
| "eval_loss": 1946.751953125, | |
| "eval_rewards/accuracies": 0.693359375, | |
| "eval_rewards/chosen": -0.48828125, | |
| "eval_rewards/margins": 0.12890625, | |
| "eval_rewards/rejected": -0.6171875, | |
| "eval_runtime": 46.6324, | |
| "eval_samples_per_second": 42.889, | |
| "eval_steps_per_second": 0.686, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 2.312925170068027, | |
| "grad_norm": 4283.740192370733, | |
| "learning_rate": 7.549965506743123e-06, | |
| "logits/chosen": -1.484375, | |
| "logits/rejected": 0.8671875, | |
| "logps/chosen": -36.75, | |
| "logps/rejected": -80.0, | |
| "loss": 265.1969, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -0.35546875, | |
| "rewards/margins": 0.43359375, | |
| "rewards/rejected": -0.7890625, | |
| "step": 2210 | |
| }, | |
| { | |
| "epoch": 2.3233908948194664, | |
| "grad_norm": 3741.4558632073504, | |
| "learning_rate": 7.333104898685631e-06, | |
| "logits/chosen": -1.8359375, | |
| "logits/rejected": -0.279296875, | |
| "logps/chosen": -35.75, | |
| "logps/rejected": -80.0, | |
| "loss": 273.525, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -0.345703125, | |
| "rewards/margins": 0.44140625, | |
| "rewards/rejected": -0.78515625, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 2.333856619570905, | |
| "grad_norm": 4373.998315282991, | |
| "learning_rate": 7.118867837183518e-06, | |
| "logits/chosen": -1.7265625, | |
| "logits/rejected": 0.035400390625, | |
| "logps/chosen": -33.5, | |
| "logps/rejected": -76.0, | |
| "loss": 276.8969, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -0.322265625, | |
| "rewards/margins": 0.427734375, | |
| "rewards/rejected": -0.75, | |
| "step": 2230 | |
| }, | |
| { | |
| "epoch": 2.3443223443223444, | |
| "grad_norm": 3628.679876275693, | |
| "learning_rate": 6.9072861365936045e-06, | |
| "logits/chosen": -1.4921875, | |
| "logits/rejected": 0.000244140625, | |
| "logps/chosen": -34.0, | |
| "logps/rejected": -79.0, | |
| "loss": 218.9469, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -0.330078125, | |
| "rewards/margins": 0.44921875, | |
| "rewards/rejected": -0.77734375, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 2.3547880690737832, | |
| "grad_norm": 4214.892357889748, | |
| "learning_rate": 6.6983912169497005e-06, | |
| "logits/chosen": -2.21875, | |
| "logits/rejected": 0.2373046875, | |
| "logps/chosen": -35.0, | |
| "logps/rejected": -80.0, | |
| "loss": 256.275, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.33984375, | |
| "rewards/margins": 0.44921875, | |
| "rewards/rejected": -0.7890625, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 2.3652537938252225, | |
| "grad_norm": 3802.5032756305554, | |
| "learning_rate": 6.49221409929677e-06, | |
| "logits/chosen": -2.71875, | |
| "logits/rejected": -0.60546875, | |
| "logps/chosen": -34.5, | |
| "logps/rejected": -78.0, | |
| "loss": 286.425, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -0.333984375, | |
| "rewards/margins": 0.43359375, | |
| "rewards/rejected": -0.76953125, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 2.3757195185766613, | |
| "grad_norm": 5391.821239507111, | |
| "learning_rate": 6.2887854010842075e-06, | |
| "logits/chosen": -1.2578125, | |
| "logits/rejected": 0.73046875, | |
| "logps/chosen": -36.25, | |
| "logps/rejected": -80.0, | |
| "loss": 290.5078, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -0.3515625, | |
| "rewards/margins": 0.439453125, | |
| "rewards/rejected": -0.7890625, | |
| "step": 2270 | |
| }, | |
| { | |
| "epoch": 2.3861852433281006, | |
| "grad_norm": 5966.69628837423, | |
| "learning_rate": 6.0881353316191385e-06, | |
| "logits/chosen": -1.8671875, | |
| "logits/rejected": -0.1494140625, | |
| "logps/chosen": -32.75, | |
| "logps/rejected": -80.0, | |
| "loss": 249.5656, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -0.314453125, | |
| "rewards/margins": 0.470703125, | |
| "rewards/rejected": -0.78515625, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 2.3966509680795394, | |
| "grad_norm": 4197.930242179745, | |
| "learning_rate": 5.890293687580381e-06, | |
| "logits/chosen": -1.609375, | |
| "logits/rejected": -0.337890625, | |
| "logps/chosen": -33.75, | |
| "logps/rejected": -79.0, | |
| "loss": 285.875, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -0.326171875, | |
| "rewards/margins": 0.44921875, | |
| "rewards/rejected": -0.77734375, | |
| "step": 2290 | |
| }, | |
| { | |
| "epoch": 2.4071166928309786, | |
| "grad_norm": 3369.3755944689015, | |
| "learning_rate": 5.695289848593532e-06, | |
| "logits/chosen": -1.3125, | |
| "logits/rejected": 0.0267333984375, | |
| "logps/chosen": -34.0, | |
| "logps/rejected": -80.0, | |
| "loss": 303.8625, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -0.328125, | |
| "rewards/margins": 0.45703125, | |
| "rewards/rejected": -0.78515625, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 2.4071166928309786, | |
| "eval_logits/chosen": -0.408203125, | |
| "eval_logits/rejected": 0.9296875, | |
| "eval_logps/chosen": -50.25, | |
| "eval_logps/rejected": -63.0, | |
| "eval_loss": 1945.9840087890625, | |
| "eval_rewards/accuracies": 0.697265625, | |
| "eval_rewards/chosen": -0.490234375, | |
| "eval_rewards/margins": 0.1279296875, | |
| "eval_rewards/rejected": -0.6171875, | |
| "eval_runtime": 46.6407, | |
| "eval_samples_per_second": 42.881, | |
| "eval_steps_per_second": 0.686, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 2.4175824175824174, | |
| "grad_norm": 4204.7981902465735, | |
| "learning_rate": 5.50315277286812e-06, | |
| "logits/chosen": -1.5859375, | |
| "logits/rejected": -0.2216796875, | |
| "logps/chosen": -36.5, | |
| "logps/rejected": -81.5, | |
| "loss": 270.3938, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -0.35546875, | |
| "rewards/margins": 0.44921875, | |
| "rewards/rejected": -0.8046875, | |
| "step": 2310 | |
| }, | |
| { | |
| "epoch": 2.4280481423338567, | |
| "grad_norm": 4292.971305020472, | |
| "learning_rate": 5.313910992897281e-06, | |
| "logits/chosen": -1.5234375, | |
| "logits/rejected": 0.22265625, | |
| "logps/chosen": -32.5, | |
| "logps/rejected": -79.5, | |
| "loss": 277.925, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -0.314453125, | |
| "rewards/margins": 0.46875, | |
| "rewards/rejected": -0.78515625, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 2.4385138670852955, | |
| "grad_norm": 3910.815399541148, | |
| "learning_rate": 5.127592611220658e-06, | |
| "logits/chosen": -1.765625, | |
| "logits/rejected": 0.154296875, | |
| "logps/chosen": -34.0, | |
| "logps/rejected": -79.5, | |
| "loss": 267.6719, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -0.330078125, | |
| "rewards/margins": 0.455078125, | |
| "rewards/rejected": -0.78515625, | |
| "step": 2330 | |
| }, | |
| { | |
| "epoch": 2.4489795918367347, | |
| "grad_norm": 4948.556784449565, | |
| "learning_rate": 4.944225296251159e-06, | |
| "logits/chosen": -2.65625, | |
| "logits/rejected": -0.55859375, | |
| "logps/chosen": -35.0, | |
| "logps/rejected": -79.5, | |
| "loss": 263.3422, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -0.337890625, | |
| "rewards/margins": 0.447265625, | |
| "rewards/rejected": -0.78515625, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 2.4594453165881736, | |
| "grad_norm": 4404.439378979044, | |
| "learning_rate": 4.763836278166172e-06, | |
| "logits/chosen": -1.5, | |
| "logits/rejected": 0.427734375, | |
| "logps/chosen": -35.0, | |
| "logps/rejected": -80.5, | |
| "loss": 237.7969, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -0.337890625, | |
| "rewards/margins": 0.45703125, | |
| "rewards/rejected": -0.79296875, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 2.469911041339613, | |
| "grad_norm": 4450.484801842572, | |
| "learning_rate": 4.586452344863876e-06, | |
| "logits/chosen": -2.265625, | |
| "logits/rejected": -0.11962890625, | |
| "logps/chosen": -33.5, | |
| "logps/rejected": -80.5, | |
| "loss": 266.0813, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -0.32421875, | |
| "rewards/margins": 0.46875, | |
| "rewards/rejected": -0.79296875, | |
| "step": 2360 | |
| }, | |
| { | |
| "epoch": 2.4803767660910516, | |
| "grad_norm": 4956.687754875603, | |
| "learning_rate": 4.412099837985192e-06, | |
| "logits/chosen": -1.8515625, | |
| "logits/rejected": 0.025634765625, | |
| "logps/chosen": -34.25, | |
| "logps/rejected": -77.5, | |
| "loss": 251.275, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -0.33203125, | |
| "rewards/margins": 0.4296875, | |
| "rewards/rejected": -0.76171875, | |
| "step": 2370 | |
| }, | |
| { | |
| "epoch": 2.490842490842491, | |
| "grad_norm": 4005.4194802668694, | |
| "learning_rate": 4.240804649002089e-06, | |
| "logits/chosen": -1.8046875, | |
| "logits/rejected": -0.158203125, | |
| "logps/chosen": -35.75, | |
| "logps/rejected": -82.0, | |
| "loss": 242.5094, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -0.345703125, | |
| "rewards/margins": 0.4609375, | |
| "rewards/rejected": -0.8046875, | |
| "step": 2380 | |
| }, | |
| { | |
| "epoch": 2.50130821559393, | |
| "grad_norm": 3554.8449593153937, | |
| "learning_rate": 4.072592215372598e-06, | |
| "logits/chosen": -1.3359375, | |
| "logits/rejected": 0.65234375, | |
| "logps/chosen": -34.25, | |
| "logps/rejected": -78.0, | |
| "loss": 247.2594, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -0.330078125, | |
| "rewards/margins": 0.4375, | |
| "rewards/rejected": -0.76953125, | |
| "step": 2390 | |
| }, | |
| { | |
| "epoch": 2.511773940345369, | |
| "grad_norm": 4180.3983116237, | |
| "learning_rate": 3.907487516763389e-06, | |
| "logits/chosen": -2.265625, | |
| "logits/rejected": 0.26953125, | |
| "logps/chosen": -33.25, | |
| "logps/rejected": -80.0, | |
| "loss": 266.5266, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -0.3203125, | |
| "rewards/margins": 0.46484375, | |
| "rewards/rejected": -0.78515625, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 2.511773940345369, | |
| "eval_logits/chosen": -0.66015625, | |
| "eval_logits/rejected": 0.7578125, | |
| "eval_logps/chosen": -49.25, | |
| "eval_logps/rejected": -62.25, | |
| "eval_loss": 1952.06396484375, | |
| "eval_rewards/accuracies": 0.69140625, | |
| "eval_rewards/chosen": -0.48046875, | |
| "eval_rewards/margins": 0.12890625, | |
| "eval_rewards/rejected": -0.609375, | |
| "eval_runtime": 46.6415, | |
| "eval_samples_per_second": 42.88, | |
| "eval_steps_per_second": 0.686, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 2.5222396650968077, | |
| "grad_norm": 3826.4953305908193, | |
| "learning_rate": 3.745515071340286e-06, | |
| "logits/chosen": -2.28125, | |
| "logits/rejected": -0.39453125, | |
| "logps/chosen": -35.5, | |
| "logps/rejected": -80.5, | |
| "loss": 252.3469, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -0.34375, | |
| "rewards/margins": 0.44921875, | |
| "rewards/rejected": -0.79296875, | |
| "step": 2410 | |
| }, | |
| { | |
| "epoch": 2.532705389848247, | |
| "grad_norm": 4764.813189644713, | |
| "learning_rate": 3.586698932127236e-06, | |
| "logits/chosen": -2.234375, | |
| "logits/rejected": -0.404296875, | |
| "logps/chosen": -34.5, | |
| "logps/rejected": -78.0, | |
| "loss": 263.8766, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": -0.333984375, | |
| "rewards/margins": 0.43359375, | |
| "rewards/rejected": -0.765625, | |
| "step": 2420 | |
| }, | |
| { | |
| "epoch": 2.5431711145996863, | |
| "grad_norm": 4171.107449760552, | |
| "learning_rate": 3.4310626834344744e-06, | |
| "logits/chosen": -1.546875, | |
| "logits/rejected": 0.13671875, | |
| "logps/chosen": -36.75, | |
| "logps/rejected": -83.0, | |
| "loss": 224.9984, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -0.35546875, | |
| "rewards/margins": 0.462890625, | |
| "rewards/rejected": -0.81640625, | |
| "step": 2430 | |
| }, | |
| { | |
| "epoch": 2.553636839351125, | |
| "grad_norm": 4102.3909768368685, | |
| "learning_rate": 3.2786294373562343e-06, | |
| "logits/chosen": -0.74609375, | |
| "logits/rejected": 1.25, | |
| "logps/chosen": -38.75, | |
| "logps/rejected": -83.0, | |
| "loss": 262.3375, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -0.376953125, | |
| "rewards/margins": 0.44140625, | |
| "rewards/rejected": -0.81640625, | |
| "step": 2440 | |
| }, | |
| { | |
| "epoch": 2.564102564102564, | |
| "grad_norm": 3893.4175221603296, | |
| "learning_rate": 3.1294218303385526e-06, | |
| "logits/chosen": -2.171875, | |
| "logits/rejected": -0.1962890625, | |
| "logps/chosen": -35.0, | |
| "logps/rejected": -78.0, | |
| "loss": 237.5687, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -0.33984375, | |
| "rewards/margins": 0.4296875, | |
| "rewards/rejected": -0.76953125, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 2.574568288854003, | |
| "grad_norm": 4111.362788925979, | |
| "learning_rate": 2.9834620198177483e-06, | |
| "logits/chosen": -1.7421875, | |
| "logits/rejected": 0.498046875, | |
| "logps/chosen": -32.0, | |
| "logps/rejected": -78.5, | |
| "loss": 275.925, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -0.310546875, | |
| "rewards/margins": 0.462890625, | |
| "rewards/rejected": -0.7734375, | |
| "step": 2460 | |
| }, | |
| { | |
| "epoch": 2.5850340136054424, | |
| "grad_norm": 3414.175847679286, | |
| "learning_rate": 2.840771680930068e-06, | |
| "logits/chosen": -2.34375, | |
| "logits/rejected": -0.765625, | |
| "logps/chosen": -35.75, | |
| "logps/rejected": -80.5, | |
| "loss": 285.8687, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -0.34765625, | |
| "rewards/margins": 0.44921875, | |
| "rewards/rejected": -0.796875, | |
| "step": 2470 | |
| }, | |
| { | |
| "epoch": 2.595499738356881, | |
| "grad_norm": 3854.650308007856, | |
| "learning_rate": 2.701372003292868e-06, | |
| "logits/chosen": -1.6328125, | |
| "logits/rejected": -0.34765625, | |
| "logps/chosen": -36.5, | |
| "logps/rejected": -82.5, | |
| "loss": 269.2109, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -0.353515625, | |
| "rewards/margins": 0.45703125, | |
| "rewards/rejected": -0.8125, | |
| "step": 2480 | |
| }, | |
| { | |
| "epoch": 2.60596546310832, | |
| "grad_norm": 3563.238727553535, | |
| "learning_rate": 2.56528368785795e-06, | |
| "logits/chosen": -1.875, | |
| "logits/rejected": -0.439453125, | |
| "logps/chosen": -36.25, | |
| "logps/rejected": -81.0, | |
| "loss": 236.7375, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -0.349609375, | |
| "rewards/margins": 0.447265625, | |
| "rewards/rejected": -0.796875, | |
| "step": 2490 | |
| }, | |
| { | |
| "epoch": 2.6164311878597593, | |
| "grad_norm": 3735.7861634533815, | |
| "learning_rate": 2.432526943837471e-06, | |
| "logits/chosen": -2.0625, | |
| "logits/rejected": -0.1103515625, | |
| "logps/chosen": -34.0, | |
| "logps/rejected": -81.5, | |
| "loss": 220.4344, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -0.328125, | |
| "rewards/margins": 0.474609375, | |
| "rewards/rejected": -0.80078125, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 2.6164311878597593, | |
| "eval_logits/chosen": -0.5625, | |
| "eval_logits/rejected": 0.8671875, | |
| "eval_logps/chosen": -49.25, | |
| "eval_logps/rejected": -62.25, | |
| "eval_loss": 1944.1280517578125, | |
| "eval_rewards/accuracies": 0.697265625, | |
| "eval_rewards/chosen": -0.48046875, | |
| "eval_rewards/margins": 0.130859375, | |
| "eval_rewards/rejected": -0.609375, | |
| "eval_runtime": 46.6315, | |
| "eval_samples_per_second": 42.889, | |
| "eval_steps_per_second": 0.686, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 2.6268969126111985, | |
| "grad_norm": 4242.2425362307295, | |
| "learning_rate": 2.3031214857028446e-06, | |
| "logits/chosen": -2.53125, | |
| "logits/rejected": -0.1328125, | |
| "logps/chosen": -34.75, | |
| "logps/rejected": -79.5, | |
| "loss": 251.6406, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -0.3359375, | |
| "rewards/margins": 0.447265625, | |
| "rewards/rejected": -0.78125, | |
| "step": 2510 | |
| }, | |
| { | |
| "epoch": 2.6373626373626373, | |
| "grad_norm": 4687.909539511884, | |
| "learning_rate": 2.177086530257139e-06, | |
| "logits/chosen": -1.375, | |
| "logits/rejected": 0.73828125, | |
| "logps/chosen": -34.25, | |
| "logps/rejected": -80.5, | |
| "loss": 248.1813, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -0.330078125, | |
| "rewards/margins": 0.458984375, | |
| "rewards/rejected": -0.79296875, | |
| "step": 2520 | |
| }, | |
| { | |
| "epoch": 2.647828362114076, | |
| "grad_norm": 3160.237413504292, | |
| "learning_rate": 2.054440793781354e-06, | |
| "logits/chosen": -2.234375, | |
| "logits/rejected": -0.478515625, | |
| "logps/chosen": -34.0, | |
| "logps/rejected": -77.5, | |
| "loss": 274.0031, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -0.328125, | |
| "rewards/margins": 0.43359375, | |
| "rewards/rejected": -0.76171875, | |
| "step": 2530 | |
| }, | |
| { | |
| "epoch": 2.6582940868655154, | |
| "grad_norm": 3034.7024963741933, | |
| "learning_rate": 1.9352024892550587e-06, | |
| "logits/chosen": -1.9140625, | |
| "logits/rejected": 0.384765625, | |
| "logps/chosen": -38.25, | |
| "logps/rejected": -80.5, | |
| "loss": 293.4234, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -0.37109375, | |
| "rewards/margins": 0.423828125, | |
| "rewards/rejected": -0.796875, | |
| "step": 2540 | |
| }, | |
| { | |
| "epoch": 2.6687598116169546, | |
| "grad_norm": 4232.045922036638, | |
| "learning_rate": 1.8193893236517379e-06, | |
| "logits/chosen": -1.453125, | |
| "logits/rejected": 0.58203125, | |
| "logps/chosen": -34.75, | |
| "logps/rejected": -80.0, | |
| "loss": 253.5469, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -0.337890625, | |
| "rewards/margins": 0.44921875, | |
| "rewards/rejected": -0.78515625, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 2.6792255363683934, | |
| "grad_norm": 4539.125793989451, | |
| "learning_rate": 1.7070184953092956e-06, | |
| "logits/chosen": -1.890625, | |
| "logits/rejected": 0.083984375, | |
| "logps/chosen": -36.0, | |
| "logps/rejected": -81.0, | |
| "loss": 265.0047, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -0.349609375, | |
| "rewards/margins": 0.447265625, | |
| "rewards/rejected": -0.796875, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 2.6896912611198327, | |
| "grad_norm": 4893.350885913185, | |
| "learning_rate": 1.5981066913760918e-06, | |
| "logits/chosen": -1.671875, | |
| "logits/rejected": 0.08740234375, | |
| "logps/chosen": -33.75, | |
| "logps/rejected": -78.5, | |
| "loss": 248.0687, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -0.326171875, | |
| "rewards/margins": 0.44921875, | |
| "rewards/rejected": -0.77734375, | |
| "step": 2570 | |
| }, | |
| { | |
| "epoch": 2.7001569858712715, | |
| "grad_norm": 3351.6310952210156, | |
| "learning_rate": 1.492670085332884e-06, | |
| "logits/chosen": -2.328125, | |
| "logits/rejected": -0.1708984375, | |
| "logps/chosen": -35.75, | |
| "logps/rejected": -79.5, | |
| "loss": 274.0656, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -0.345703125, | |
| "rewards/margins": 0.439453125, | |
| "rewards/rejected": -0.78515625, | |
| "step": 2580 | |
| }, | |
| { | |
| "epoch": 2.7106227106227108, | |
| "grad_norm": 5428.20446354956, | |
| "learning_rate": 1.3907243345910787e-06, | |
| "logits/chosen": -0.953125, | |
| "logits/rejected": 0.72265625, | |
| "logps/chosen": -35.75, | |
| "logps/rejected": -80.5, | |
| "loss": 258.9125, | |
| "rewards/accuracies": 0.949999988079071, | |
| "rewards/chosen": -0.345703125, | |
| "rewards/margins": 0.447265625, | |
| "rewards/rejected": -0.79296875, | |
| "step": 2590 | |
| }, | |
| { | |
| "epoch": 2.7210884353741496, | |
| "grad_norm": 3048.3926423024454, | |
| "learning_rate": 1.292284578167538e-06, | |
| "logits/chosen": -2.0, | |
| "logits/rejected": -0.26171875, | |
| "logps/chosen": -34.25, | |
| "logps/rejected": -79.0, | |
| "loss": 253.4812, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -0.33203125, | |
| "rewards/margins": 0.447265625, | |
| "rewards/rejected": -0.77734375, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 2.7210884353741496, | |
| "eval_logits/chosen": -0.546875, | |
| "eval_logits/rejected": 0.87890625, | |
| "eval_logps/chosen": -50.0, | |
| "eval_logps/rejected": -63.0, | |
| "eval_loss": 1938.112060546875, | |
| "eval_rewards/accuracies": 0.69140625, | |
| "eval_rewards/chosen": -0.48828125, | |
| "eval_rewards/margins": 0.1298828125, | |
| "eval_rewards/rejected": -0.6171875, | |
| "eval_runtime": 46.6307, | |
| "eval_samples_per_second": 42.89, | |
| "eval_steps_per_second": 0.686, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 2.731554160125589, | |
| "grad_norm": 3753.4915926874996, | |
| "learning_rate": 1.1973654344364925e-06, | |
| "logits/chosen": -1.78125, | |
| "logits/rejected": -0.034912109375, | |
| "logps/chosen": -33.0, | |
| "logps/rejected": -77.5, | |
| "loss": 295.4578, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -0.318359375, | |
| "rewards/margins": 0.447265625, | |
| "rewards/rejected": -0.765625, | |
| "step": 2610 | |
| }, | |
| { | |
| "epoch": 2.7420198848770276, | |
| "grad_norm": 4719.267541608869, | |
| "learning_rate": 1.1059809989586583e-06, | |
| "logits/chosen": -1.5625, | |
| "logits/rejected": 0.67578125, | |
| "logps/chosen": -36.5, | |
| "logps/rejected": -79.0, | |
| "loss": 273.5328, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.3515625, | |
| "rewards/margins": 0.427734375, | |
| "rewards/rejected": -0.77734375, | |
| "step": 2620 | |
| }, | |
| { | |
| "epoch": 2.752485609628467, | |
| "grad_norm": 4595.85212684217, | |
| "learning_rate": 1.018144842388044e-06, | |
| "logits/chosen": -1.125, | |
| "logits/rejected": 0.26171875, | |
| "logps/chosen": -34.75, | |
| "logps/rejected": -79.0, | |
| "loss": 283.9656, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -0.3359375, | |
| "rewards/margins": 0.4453125, | |
| "rewards/rejected": -0.78125, | |
| "step": 2630 | |
| }, | |
| { | |
| "epoch": 2.7629513343799057, | |
| "grad_norm": 4827.667356282864, | |
| "learning_rate": 9.338700084567109e-07, | |
| "logits/chosen": -1.6328125, | |
| "logits/rejected": 0.98046875, | |
| "logps/chosen": -35.0, | |
| "logps/rejected": -81.0, | |
| "loss": 243.4219, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -0.337890625, | |
| "rewards/margins": 0.4609375, | |
| "rewards/rejected": -0.80078125, | |
| "step": 2640 | |
| }, | |
| { | |
| "epoch": 2.773417059131345, | |
| "grad_norm": 4340.477074199542, | |
| "learning_rate": 8.531690120377606e-07, | |
| "logits/chosen": -1.640625, | |
| "logits/rejected": 0.25390625, | |
| "logps/chosen": -34.75, | |
| "logps/rejected": -79.0, | |
| "loss": 251.1687, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -0.337890625, | |
| "rewards/margins": 0.439453125, | |
| "rewards/rejected": -0.77734375, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 2.7838827838827838, | |
| "grad_norm": 4693.846523907441, | |
| "learning_rate": 7.760538372868636e-07, | |
| "logits/chosen": -2.796875, | |
| "logits/rejected": -0.2080078125, | |
| "logps/chosen": -34.75, | |
| "logps/rejected": -78.0, | |
| "loss": 280.85, | |
| "rewards/accuracies": 0.956250011920929, | |
| "rewards/chosen": -0.3359375, | |
| "rewards/margins": 0.431640625, | |
| "rewards/rejected": -0.76953125, | |
| "step": 2660 | |
| }, | |
| { | |
| "epoch": 2.794348508634223, | |
| "grad_norm": 4496.841146534937, | |
| "learning_rate": 7.025359358626166e-07, | |
| "logits/chosen": -1.84375, | |
| "logits/rejected": 0.78125, | |
| "logps/chosen": -36.25, | |
| "logps/rejected": -79.0, | |
| "loss": 303.5328, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -0.349609375, | |
| "rewards/margins": 0.42578125, | |
| "rewards/rejected": -0.77734375, | |
| "step": 2670 | |
| }, | |
| { | |
| "epoch": 2.804814233385662, | |
| "grad_norm": 4003.651277944437, | |
| "learning_rate": 6.326262252259463e-07, | |
| "logits/chosen": -1.9765625, | |
| "logits/rejected": 0.0002193450927734375, | |
| "logps/chosen": -34.75, | |
| "logps/rejected": -79.0, | |
| "loss": 269.95, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -0.3359375, | |
| "rewards/margins": 0.443359375, | |
| "rewards/rejected": -0.77734375, | |
| "step": 2680 | |
| }, | |
| { | |
| "epoch": 2.815279958137101, | |
| "grad_norm": 5395.243953617292, | |
| "learning_rate": 5.66335087018871e-07, | |
| "logits/chosen": -1.0703125, | |
| "logits/rejected": 0.59375, | |
| "logps/chosen": -35.0, | |
| "logps/rejected": -78.5, | |
| "loss": 253.0734, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -0.337890625, | |
| "rewards/margins": 0.435546875, | |
| "rewards/rejected": -0.7734375, | |
| "step": 2690 | |
| }, | |
| { | |
| "epoch": 2.82574568288854, | |
| "grad_norm": 4644.550937579759, | |
| "learning_rate": 5.036723655228226e-07, | |
| "logits/chosen": -1.6328125, | |
| "logits/rejected": 0.86328125, | |
| "logps/chosen": -35.5, | |
| "logps/rejected": -79.5, | |
| "loss": 271.3984, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -0.341796875, | |
| "rewards/margins": 0.439453125, | |
| "rewards/rejected": -0.78125, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 2.82574568288854, | |
| "eval_logits/chosen": -0.6328125, | |
| "eval_logits/rejected": 0.8046875, | |
| "eval_logps/chosen": -49.75, | |
| "eval_logps/rejected": -63.0, | |
| "eval_loss": 1943.8719482421875, | |
| "eval_rewards/accuracies": 0.6953125, | |
| "eval_rewards/chosen": -0.486328125, | |
| "eval_rewards/margins": 0.1298828125, | |
| "eval_rewards/rejected": -0.6171875, | |
| "eval_runtime": 46.6413, | |
| "eval_samples_per_second": 42.88, | |
| "eval_steps_per_second": 0.686, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 2.836211407639979, | |
| "grad_norm": 4168.0524961807905, | |
| "learning_rate": 4.446473661967432e-07, | |
| "logits/chosen": -1.9453125, | |
| "logits/rejected": -0.15234375, | |
| "logps/chosen": -33.5, | |
| "logps/rejected": -78.5, | |
| "loss": 255.0938, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -0.322265625, | |
| "rewards/margins": 0.44921875, | |
| "rewards/rejected": -0.7734375, | |
| "step": 2710 | |
| }, | |
| { | |
| "epoch": 2.846677132391418, | |
| "grad_norm": 4731.848967595873, | |
| "learning_rate": 3.892688542952411e-07, | |
| "logits/chosen": -2.03125, | |
| "logits/rejected": 0.259765625, | |
| "logps/chosen": -34.5, | |
| "logps/rejected": -81.0, | |
| "loss": 249.5801, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -0.333984375, | |
| "rewards/margins": 0.462890625, | |
| "rewards/rejected": -0.796875, | |
| "step": 2720 | |
| }, | |
| { | |
| "epoch": 2.857142857142857, | |
| "grad_norm": 4175.562951556529, | |
| "learning_rate": 3.3754505356694e-07, | |
| "logits/chosen": -1.5390625, | |
| "logits/rejected": -0.06640625, | |
| "logps/chosen": -37.0, | |
| "logps/rejected": -81.5, | |
| "loss": 277.4219, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -0.359375, | |
| "rewards/margins": 0.4453125, | |
| "rewards/rejected": -0.8046875, | |
| "step": 2730 | |
| }, | |
| { | |
| "epoch": 2.867608581894296, | |
| "grad_norm": 3583.2889682548266, | |
| "learning_rate": 2.8948364503322276e-07, | |
| "logits/chosen": -1.8671875, | |
| "logits/rejected": -0.48828125, | |
| "logps/chosen": -35.5, | |
| "logps/rejected": -80.0, | |
| "loss": 250.1406, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -0.341796875, | |
| "rewards/margins": 0.4453125, | |
| "rewards/rejected": -0.7890625, | |
| "step": 2740 | |
| }, | |
| { | |
| "epoch": 2.8780743066457353, | |
| "grad_norm": 2980.5880563421206, | |
| "learning_rate": 2.4509176584762706e-07, | |
| "logits/chosen": -1.7421875, | |
| "logits/rejected": 0.5390625, | |
| "logps/chosen": -36.5, | |
| "logps/rejected": -81.0, | |
| "loss": 244.3047, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -0.353515625, | |
| "rewards/margins": 0.44140625, | |
| "rewards/rejected": -0.796875, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 2.8885400313971745, | |
| "grad_norm": 5403.186749117958, | |
| "learning_rate": 2.0437600823595694e-07, | |
| "logits/chosen": -1.7109375, | |
| "logits/rejected": 0.4765625, | |
| "logps/chosen": -35.5, | |
| "logps/rejected": -80.0, | |
| "loss": 262.7922, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -0.34375, | |
| "rewards/margins": 0.4453125, | |
| "rewards/rejected": -0.7890625, | |
| "step": 2760 | |
| }, | |
| { | |
| "epoch": 2.8990057561486133, | |
| "grad_norm": 4394.019455000847, | |
| "learning_rate": 1.6734241851733e-07, | |
| "logits/chosen": -2.15625, | |
| "logits/rejected": 0.443359375, | |
| "logps/chosen": -34.75, | |
| "logps/rejected": -78.0, | |
| "loss": 240.3281, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -0.3359375, | |
| "rewards/margins": 0.431640625, | |
| "rewards/rejected": -0.765625, | |
| "step": 2770 | |
| }, | |
| { | |
| "epoch": 2.909471480900052, | |
| "grad_norm": 3107.885413754039, | |
| "learning_rate": 1.339964962062984e-07, | |
| "logits/chosen": -2.484375, | |
| "logits/rejected": 0.11767578125, | |
| "logps/chosen": -34.0, | |
| "logps/rejected": -79.5, | |
| "loss": 285.7141, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -0.330078125, | |
| "rewards/margins": 0.451171875, | |
| "rewards/rejected": -0.78125, | |
| "step": 2780 | |
| }, | |
| { | |
| "epoch": 2.9199372056514914, | |
| "grad_norm": 4323.489854055578, | |
| "learning_rate": 1.0434319319617136e-07, | |
| "logits/chosen": -1.3671875, | |
| "logits/rejected": 0.73828125, | |
| "logps/chosen": -32.75, | |
| "logps/rejected": -76.0, | |
| "loss": 269.3797, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -0.318359375, | |
| "rewards/margins": 0.431640625, | |
| "rewards/rejected": -0.75, | |
| "step": 2790 | |
| }, | |
| { | |
| "epoch": 2.9304029304029307, | |
| "grad_norm": 4086.3821813016075, | |
| "learning_rate": 7.838691302365363e-08, | |
| "logits/chosen": -1.8203125, | |
| "logits/rejected": 0.4453125, | |
| "logps/chosen": -37.0, | |
| "logps/rejected": -79.0, | |
| "loss": 292.8133, | |
| "rewards/accuracies": 0.96875, | |
| "rewards/chosen": -0.357421875, | |
| "rewards/margins": 0.419921875, | |
| "rewards/rejected": -0.77734375, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 2.9304029304029307, | |
| "eval_logits/chosen": -0.63671875, | |
| "eval_logits/rejected": 0.80078125, | |
| "eval_logps/chosen": -49.75, | |
| "eval_logps/rejected": -62.75, | |
| "eval_loss": 1943.3599853515625, | |
| "eval_rewards/accuracies": 0.6953125, | |
| "eval_rewards/chosen": -0.486328125, | |
| "eval_rewards/margins": 0.130859375, | |
| "eval_rewards/rejected": -0.6171875, | |
| "eval_runtime": 46.6375, | |
| "eval_samples_per_second": 42.884, | |
| "eval_steps_per_second": 0.686, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 2.9408686551543695, | |
| "grad_norm": 3913.1495796453646, | |
| "learning_rate": 5.613151021490459e-08, | |
| "logits/chosen": -1.75, | |
| "logits/rejected": 0.208984375, | |
| "logps/chosen": -35.0, | |
| "logps/rejected": -80.5, | |
| "loss": 243.7562, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -0.337890625, | |
| "rewards/margins": 0.455078125, | |
| "rewards/rejected": -0.79296875, | |
| "step": 2810 | |
| }, | |
| { | |
| "epoch": 2.9513343799058083, | |
| "grad_norm": 4461.982780780038, | |
| "learning_rate": 3.7580289713159965e-08, | |
| "logits/chosen": -1.9609375, | |
| "logits/rejected": 0.1962890625, | |
| "logps/chosen": -35.75, | |
| "logps/rejected": -80.5, | |
| "loss": 254.1328, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -0.34765625, | |
| "rewards/margins": 0.447265625, | |
| "rewards/rejected": -0.79296875, | |
| "step": 2820 | |
| }, | |
| { | |
| "epoch": 2.9618001046572475, | |
| "grad_norm": 3840.914716988201, | |
| "learning_rate": 2.2736006387927277e-08, | |
| "logits/chosen": -2.328125, | |
| "logits/rejected": -0.1630859375, | |
| "logps/chosen": -33.75, | |
| "logps/rejected": -78.0, | |
| "loss": 303.0531, | |
| "rewards/accuracies": 0.96875, | |
| "rewards/chosen": -0.326171875, | |
| "rewards/margins": 0.44140625, | |
| "rewards/rejected": -0.76953125, | |
| "step": 2830 | |
| }, | |
| { | |
| "epoch": 2.9722658294086868, | |
| "grad_norm": 4497.708441219362, | |
| "learning_rate": 1.1600864625893604e-08, | |
| "logits/chosen": -1.6796875, | |
| "logits/rejected": 0.92578125, | |
| "logps/chosen": -34.0, | |
| "logps/rejected": -81.0, | |
| "loss": 283.6375, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -0.328125, | |
| "rewards/margins": 0.46875, | |
| "rewards/rejected": -0.796875, | |
| "step": 2840 | |
| }, | |
| { | |
| "epoch": 2.9827315541601256, | |
| "grad_norm": 3919.123386328894, | |
| "learning_rate": 4.176518003573549e-09, | |
| "logits/chosen": -1.7421875, | |
| "logits/rejected": -0.396484375, | |
| "logps/chosen": -34.25, | |
| "logps/rejected": -78.5, | |
| "loss": 255.6219, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -0.330078125, | |
| "rewards/margins": 0.4453125, | |
| "rewards/rejected": -0.77734375, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 2.9931972789115644, | |
| "grad_norm": 5148.738766884023, | |
| "learning_rate": 4.640690417528903e-10, | |
| "logits/chosen": -1.125, | |
| "logits/rejected": -0.034423828125, | |
| "logps/chosen": -35.25, | |
| "logps/rejected": -80.5, | |
| "loss": 220.5062, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -0.33984375, | |
| "rewards/margins": 0.44921875, | |
| "rewards/rejected": -0.7890625, | |
| "step": 2860 | |
| }, | |
| { | |
| "epoch": 2.998430141287284, | |
| "step": 2865, | |
| "total_flos": 0.0, | |
| "train_loss": 0.0, | |
| "train_runtime": 0.0514, | |
| "train_samples_per_second": 3570504.364, | |
| "train_steps_per_second": 55776.355 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 2865, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 0.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |