CharlesLi's picture
Model save
0fae834 verified
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 3.0,
"eval_steps": 100,
"global_step": 2865,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0010471204188481676,
"grad_norm": 5.383150394964546,
"learning_rate": 1.7421602787456448e-07,
"logits/chosen": -9.75,
"logits/rejected": -8.6875,
"logps/chosen": -760.0,
"logps/rejected": -1992.0,
"loss": 0.6914,
"rewards/accuracies": 0.0,
"rewards/chosen": 0.0,
"rewards/margins": 0.0,
"rewards/rejected": 0.0,
"step": 1
},
{
"epoch": 0.010471204188481676,
"grad_norm": 5.114507498453803,
"learning_rate": 1.7421602787456445e-06,
"logits/chosen": -10.8125,
"logits/rejected": -8.0,
"logps/chosen": -510.0,
"logps/rejected": -2080.0,
"loss": 0.6904,
"rewards/accuracies": 0.0833333358168602,
"rewards/chosen": -0.001861572265625,
"rewards/margins": -0.000347137451171875,
"rewards/rejected": -0.0015106201171875,
"step": 10
},
{
"epoch": 0.020942408376963352,
"grad_norm": 5.261864061781078,
"learning_rate": 3.484320557491289e-06,
"logits/chosen": -10.9375,
"logits/rejected": -8.375,
"logps/chosen": -486.0,
"logps/rejected": -2128.0,
"loss": 0.6306,
"rewards/accuracies": 0.699999988079071,
"rewards/chosen": -0.00732421875,
"rewards/margins": 0.1337890625,
"rewards/rejected": -0.1416015625,
"step": 20
},
{
"epoch": 0.031413612565445025,
"grad_norm": 1.703517034680173,
"learning_rate": 5.226480836236934e-06,
"logits/chosen": -10.875,
"logits/rejected": -8.1875,
"logps/chosen": -448.0,
"logps/rejected": -2272.0,
"loss": 0.3425,
"rewards/accuracies": 0.918749988079071,
"rewards/chosen": -0.1044921875,
"rewards/margins": 1.1640625,
"rewards/rejected": -1.265625,
"step": 30
},
{
"epoch": 0.041884816753926704,
"grad_norm": 0.5206726390816392,
"learning_rate": 6.968641114982578e-06,
"logits/chosen": -11.1875,
"logits/rejected": -8.9375,
"logps/chosen": -576.0,
"logps/rejected": -2544.0,
"loss": 0.2285,
"rewards/accuracies": 0.9125000238418579,
"rewards/chosen": -0.6796875,
"rewards/margins": 3.78125,
"rewards/rejected": -4.46875,
"step": 40
},
{
"epoch": 0.05235602094240838,
"grad_norm": 1.8734901520085283,
"learning_rate": 8.710801393728225e-06,
"logits/chosen": -11.9375,
"logits/rejected": -9.875,
"logps/chosen": -604.0,
"logps/rejected": -2432.0,
"loss": 0.2272,
"rewards/accuracies": 0.893750011920929,
"rewards/chosen": -0.7734375,
"rewards/margins": 3.765625,
"rewards/rejected": -4.53125,
"step": 50
},
{
"epoch": 0.06282722513089005,
"grad_norm": 1.612078607396359,
"learning_rate": 1.0452961672473868e-05,
"logits/chosen": -13.625,
"logits/rejected": -10.0625,
"logps/chosen": -516.0,
"logps/rejected": -2640.0,
"loss": 0.2082,
"rewards/accuracies": 0.9437500238418579,
"rewards/chosen": -0.55859375,
"rewards/margins": 3.734375,
"rewards/rejected": -4.3125,
"step": 60
},
{
"epoch": 0.07329842931937172,
"grad_norm": 2.7488417737574338,
"learning_rate": 1.2195121951219513e-05,
"logits/chosen": -14.4375,
"logits/rejected": -11.125,
"logps/chosen": -532.0,
"logps/rejected": -2432.0,
"loss": 0.2305,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -0.65625,
"rewards/margins": 3.203125,
"rewards/rejected": -3.859375,
"step": 70
},
{
"epoch": 0.08376963350785341,
"grad_norm": 1.4308700938878556,
"learning_rate": 1.3937282229965156e-05,
"logits/chosen": -14.4375,
"logits/rejected": -10.8125,
"logps/chosen": -516.0,
"logps/rejected": -2496.0,
"loss": 0.2194,
"rewards/accuracies": 0.918749988079071,
"rewards/chosen": -0.69921875,
"rewards/margins": 3.015625,
"rewards/rejected": -3.71875,
"step": 80
},
{
"epoch": 0.09424083769633508,
"grad_norm": 1.5045527997343389,
"learning_rate": 1.56794425087108e-05,
"logits/chosen": -14.5625,
"logits/rejected": -10.875,
"logps/chosen": -490.0,
"logps/rejected": -2688.0,
"loss": 0.181,
"rewards/accuracies": 0.956250011920929,
"rewards/chosen": -0.55078125,
"rewards/margins": 3.53125,
"rewards/rejected": -4.09375,
"step": 90
},
{
"epoch": 0.10471204188481675,
"grad_norm": 1.4856217419018194,
"learning_rate": 1.742160278745645e-05,
"logits/chosen": -14.1875,
"logits/rejected": -11.0,
"logps/chosen": -498.0,
"logps/rejected": -2640.0,
"loss": 0.1874,
"rewards/accuracies": 0.9624999761581421,
"rewards/chosen": -0.6796875,
"rewards/margins": 3.96875,
"rewards/rejected": -4.65625,
"step": 100
},
{
"epoch": 0.10471204188481675,
"eval_logits/chosen": -14.5625,
"eval_logits/rejected": -14.4375,
"eval_logps/chosen": -364.0,
"eval_logps/rejected": -346.0,
"eval_loss": 0.6750624775886536,
"eval_rewards/accuracies": 0.54296875,
"eval_rewards/chosen": -0.455078125,
"eval_rewards/margins": 0.115234375,
"eval_rewards/rejected": -0.5703125,
"eval_runtime": 48.587,
"eval_samples_per_second": 41.163,
"eval_steps_per_second": 0.659,
"step": 100
},
{
"epoch": 0.11518324607329843,
"grad_norm": 1.4313155908274708,
"learning_rate": 1.9163763066202093e-05,
"logits/chosen": -14.0,
"logits/rejected": -11.25,
"logps/chosen": -576.0,
"logps/rejected": -2480.0,
"loss": 0.2099,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -0.8125,
"rewards/margins": 3.078125,
"rewards/rejected": -3.890625,
"step": 110
},
{
"epoch": 0.1256544502617801,
"grad_norm": 1.1336185184558103,
"learning_rate": 2.0905923344947736e-05,
"logits/chosen": -12.1875,
"logits/rejected": -10.1875,
"logps/chosen": -556.0,
"logps/rejected": -2608.0,
"loss": 0.1613,
"rewards/accuracies": 0.90625,
"rewards/chosen": -0.8671875,
"rewards/margins": 4.5,
"rewards/rejected": -5.375,
"step": 120
},
{
"epoch": 0.13612565445026178,
"grad_norm": 1.2541733975975633,
"learning_rate": 2.264808362369338e-05,
"logits/chosen": -12.125,
"logits/rejected": -9.8125,
"logps/chosen": -512.0,
"logps/rejected": -2624.0,
"loss": 0.1661,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.6640625,
"rewards/margins": 4.25,
"rewards/rejected": -4.9375,
"step": 130
},
{
"epoch": 0.14659685863874344,
"grad_norm": 1.5518629051127435,
"learning_rate": 2.4390243902439026e-05,
"logits/chosen": -11.8125,
"logits/rejected": -9.4375,
"logps/chosen": -454.0,
"logps/rejected": -2560.0,
"loss": 0.1976,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.53125,
"rewards/margins": 3.84375,
"rewards/rejected": -4.375,
"step": 140
},
{
"epoch": 0.15706806282722513,
"grad_norm": 1.19572153801732,
"learning_rate": 2.6132404181184672e-05,
"logits/chosen": -11.8125,
"logits/rejected": -9.25,
"logps/chosen": -490.0,
"logps/rejected": -2528.0,
"loss": 0.1724,
"rewards/accuracies": 0.956250011920929,
"rewards/chosen": -0.671875,
"rewards/margins": 3.78125,
"rewards/rejected": -4.4375,
"step": 150
},
{
"epoch": 0.16753926701570682,
"grad_norm": 3.4260233181538906,
"learning_rate": 2.7874564459930312e-05,
"logits/chosen": -12.5,
"logits/rejected": -10.0625,
"logps/chosen": -512.0,
"logps/rejected": -2608.0,
"loss": 0.1893,
"rewards/accuracies": 0.956250011920929,
"rewards/chosen": -1.03125,
"rewards/margins": 4.1875,
"rewards/rejected": -5.21875,
"step": 160
},
{
"epoch": 0.17801047120418848,
"grad_norm": 1.2245404825349286,
"learning_rate": 2.9616724738675962e-05,
"logits/chosen": -12.6875,
"logits/rejected": -10.5625,
"logps/chosen": -636.0,
"logps/rejected": -2800.0,
"loss": 0.166,
"rewards/accuracies": 0.9312499761581421,
"rewards/chosen": -1.4375,
"rewards/margins": 4.40625,
"rewards/rejected": -5.875,
"step": 170
},
{
"epoch": 0.18848167539267016,
"grad_norm": 1.8520829315076202,
"learning_rate": 3.13588850174216e-05,
"logits/chosen": -12.75,
"logits/rejected": -10.125,
"logps/chosen": -548.0,
"logps/rejected": -2768.0,
"loss": 0.1018,
"rewards/accuracies": 0.96875,
"rewards/chosen": -1.0546875,
"rewards/margins": 4.59375,
"rewards/rejected": -5.65625,
"step": 180
},
{
"epoch": 0.19895287958115182,
"grad_norm": 5.627621274377512,
"learning_rate": 3.310104529616725e-05,
"logits/chosen": -13.5625,
"logits/rejected": -11.0,
"logps/chosen": -564.0,
"logps/rejected": -2752.0,
"loss": 0.2452,
"rewards/accuracies": 0.893750011920929,
"rewards/chosen": -1.2265625,
"rewards/margins": 4.84375,
"rewards/rejected": -6.0625,
"step": 190
},
{
"epoch": 0.2094240837696335,
"grad_norm": 1.5444075233572967,
"learning_rate": 3.48432055749129e-05,
"logits/chosen": -13.875,
"logits/rejected": -11.25,
"logps/chosen": -504.0,
"logps/rejected": -2656.0,
"loss": 0.1191,
"rewards/accuracies": 0.9437500238418579,
"rewards/chosen": -0.921875,
"rewards/margins": 4.59375,
"rewards/rejected": -5.5,
"step": 200
},
{
"epoch": 0.2094240837696335,
"eval_logits/chosen": -12.625,
"eval_logits/rejected": -12.4375,
"eval_logps/chosen": -392.0,
"eval_logps/rejected": -372.0,
"eval_loss": 0.7004374861717224,
"eval_rewards/accuracies": 0.521484375,
"eval_rewards/chosen": -0.73046875,
"eval_rewards/margins": 0.09619140625,
"eval_rewards/rejected": -0.82421875,
"eval_runtime": 46.5948,
"eval_samples_per_second": 42.923,
"eval_steps_per_second": 0.687,
"step": 200
},
{
"epoch": 0.2198952879581152,
"grad_norm": 1.6097474817305373,
"learning_rate": 3.6585365853658535e-05,
"logits/chosen": -12.0,
"logits/rejected": -9.875,
"logps/chosen": -512.0,
"logps/rejected": -2688.0,
"loss": 0.1345,
"rewards/accuracies": 0.9437500238418579,
"rewards/chosen": -0.921875,
"rewards/margins": 5.3125,
"rewards/rejected": -6.21875,
"step": 210
},
{
"epoch": 0.23036649214659685,
"grad_norm": 1.1282557298556626,
"learning_rate": 3.8327526132404185e-05,
"logits/chosen": -12.625,
"logits/rejected": -9.875,
"logps/chosen": -524.0,
"logps/rejected": -2560.0,
"loss": 0.173,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -0.91796875,
"rewards/margins": 4.28125,
"rewards/rejected": -5.1875,
"step": 220
},
{
"epoch": 0.24083769633507854,
"grad_norm": 2.3917439454970393,
"learning_rate": 4.006968641114983e-05,
"logits/chosen": -12.9375,
"logits/rejected": -10.1875,
"logps/chosen": -540.0,
"logps/rejected": -2976.0,
"loss": 0.1125,
"rewards/accuracies": 0.956250011920929,
"rewards/chosen": -1.2265625,
"rewards/margins": 5.4375,
"rewards/rejected": -6.6875,
"step": 230
},
{
"epoch": 0.2513089005235602,
"grad_norm": 0.8708214780821046,
"learning_rate": 4.181184668989547e-05,
"logits/chosen": -13.125,
"logits/rejected": -10.5,
"logps/chosen": -600.0,
"logps/rejected": -2800.0,
"loss": 0.1767,
"rewards/accuracies": 0.9624999761581421,
"rewards/chosen": -1.1484375,
"rewards/margins": 4.375,
"rewards/rejected": -5.53125,
"step": 240
},
{
"epoch": 0.2617801047120419,
"grad_norm": 2.063585523178906,
"learning_rate": 4.3554006968641115e-05,
"logits/chosen": -14.0,
"logits/rejected": -11.375,
"logps/chosen": -688.0,
"logps/rejected": -2672.0,
"loss": 0.1751,
"rewards/accuracies": 0.90625,
"rewards/chosen": -1.3046875,
"rewards/margins": 3.9375,
"rewards/rejected": -5.25,
"step": 250
},
{
"epoch": 0.27225130890052357,
"grad_norm": 2.400314222063967,
"learning_rate": 4.529616724738676e-05,
"logits/chosen": -13.1875,
"logits/rejected": -10.75,
"logps/chosen": -588.0,
"logps/rejected": -2752.0,
"loss": 0.1691,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -1.21875,
"rewards/margins": 4.3125,
"rewards/rejected": -5.53125,
"step": 260
},
{
"epoch": 0.28272251308900526,
"grad_norm": 1.5942017927553285,
"learning_rate": 4.703832752613241e-05,
"logits/chosen": -12.125,
"logits/rejected": -9.9375,
"logps/chosen": -548.0,
"logps/rejected": -2736.0,
"loss": 0.1519,
"rewards/accuracies": 0.9375,
"rewards/chosen": -1.0859375,
"rewards/margins": 4.78125,
"rewards/rejected": -5.875,
"step": 270
},
{
"epoch": 0.2931937172774869,
"grad_norm": 1.0340452700674418,
"learning_rate": 4.878048780487805e-05,
"logits/chosen": -12.4375,
"logits/rejected": -9.9375,
"logps/chosen": -584.0,
"logps/rejected": -2528.0,
"loss": 0.2433,
"rewards/accuracies": 0.918749988079071,
"rewards/chosen": -1.2109375,
"rewards/margins": 3.78125,
"rewards/rejected": -5.0,
"step": 280
},
{
"epoch": 0.3036649214659686,
"grad_norm": 1.0479833447400388,
"learning_rate": 4.999983293481418e-05,
"logits/chosen": -12.5625,
"logits/rejected": -9.875,
"logps/chosen": -560.0,
"logps/rejected": -2608.0,
"loss": 0.2085,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -1.03125,
"rewards/margins": 4.03125,
"rewards/rejected": -5.0625,
"step": 290
},
{
"epoch": 0.31413612565445026,
"grad_norm": 0.8428943855536154,
"learning_rate": 4.9996862949181595e-05,
"logits/chosen": -12.4375,
"logits/rejected": -9.8125,
"logps/chosen": -540.0,
"logps/rejected": -2832.0,
"loss": 0.1411,
"rewards/accuracies": 0.956250011920929,
"rewards/chosen": -1.109375,
"rewards/margins": 5.4375,
"rewards/rejected": -6.5625,
"step": 300
},
{
"epoch": 0.31413612565445026,
"eval_logits/chosen": -12.625,
"eval_logits/rejected": -12.25,
"eval_logps/chosen": -416.0,
"eval_logps/rejected": -394.0,
"eval_loss": 0.7418437600135803,
"eval_rewards/accuracies": 0.4765625,
"eval_rewards/chosen": -0.96875,
"eval_rewards/margins": 0.08935546875,
"eval_rewards/rejected": -1.0546875,
"eval_runtime": 46.6609,
"eval_samples_per_second": 42.862,
"eval_steps_per_second": 0.686,
"step": 300
},
{
"epoch": 0.32460732984293195,
"grad_norm": 1.6580722362095162,
"learning_rate": 4.9990180911526896e-05,
"logits/chosen": -12.875,
"logits/rejected": -10.375,
"logps/chosen": -580.0,
"logps/rejected": -2928.0,
"loss": 0.14,
"rewards/accuracies": 0.9437500238418579,
"rewards/chosen": -1.4140625,
"rewards/margins": 6.3125,
"rewards/rejected": -7.75,
"step": 310
},
{
"epoch": 0.33507853403141363,
"grad_norm": 2.16843393813157,
"learning_rate": 4.997978781413745e-05,
"logits/chosen": -13.125,
"logits/rejected": -10.375,
"logps/chosen": -624.0,
"logps/rejected": -2912.0,
"loss": 0.1882,
"rewards/accuracies": 0.9375,
"rewards/chosen": -1.78125,
"rewards/margins": 5.8125,
"rewards/rejected": -7.59375,
"step": 320
},
{
"epoch": 0.34554973821989526,
"grad_norm": 1.7067777576082523,
"learning_rate": 4.996568520039565e-05,
"logits/chosen": -12.25,
"logits/rejected": -9.6875,
"logps/chosen": -664.0,
"logps/rejected": -2816.0,
"loss": 0.1716,
"rewards/accuracies": 0.9437500238418579,
"rewards/chosen": -1.609375,
"rewards/margins": 4.9375,
"rewards/rejected": -6.5625,
"step": 330
},
{
"epoch": 0.35602094240837695,
"grad_norm": 0.6123468321085007,
"learning_rate": 4.994787516454973e-05,
"logits/chosen": -12.125,
"logits/rejected": -9.4375,
"logps/chosen": -632.0,
"logps/rejected": -2864.0,
"loss": 0.1528,
"rewards/accuracies": 0.956250011920929,
"rewards/chosen": -1.5625,
"rewards/margins": 5.75,
"rewards/rejected": -7.3125,
"step": 340
},
{
"epoch": 0.36649214659685864,
"grad_norm": 1.5157658657768138,
"learning_rate": 4.992636035140273e-05,
"logits/chosen": -11.8125,
"logits/rejected": -9.1875,
"logps/chosen": -620.0,
"logps/rejected": -2656.0,
"loss": 0.201,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -1.65625,
"rewards/margins": 3.921875,
"rewards/rejected": -5.5625,
"step": 350
},
{
"epoch": 0.3769633507853403,
"grad_norm": 2.005355889477844,
"learning_rate": 4.99011439559198e-05,
"logits/chosen": -12.1875,
"logits/rejected": -9.1875,
"logps/chosen": -620.0,
"logps/rejected": -3008.0,
"loss": 0.1689,
"rewards/accuracies": 0.96875,
"rewards/chosen": -1.6875,
"rewards/margins": 5.6875,
"rewards/rejected": -7.375,
"step": 360
},
{
"epoch": 0.387434554973822,
"grad_norm": 2.09292667541342,
"learning_rate": 4.987222972275369e-05,
"logits/chosen": -13.0,
"logits/rejected": -10.125,
"logps/chosen": -604.0,
"logps/rejected": -2672.0,
"loss": 0.1365,
"rewards/accuracies": 0.9437500238418579,
"rewards/chosen": -1.578125,
"rewards/margins": 4.6875,
"rewards/rejected": -6.28125,
"step": 370
},
{
"epoch": 0.39790575916230364,
"grad_norm": 1.3872205709392036,
"learning_rate": 4.983962194568872e-05,
"logits/chosen": -12.4375,
"logits/rejected": -10.0625,
"logps/chosen": -664.0,
"logps/rejected": -2720.0,
"loss": 0.1762,
"rewards/accuracies": 0.9125000238418579,
"rewards/chosen": -1.6953125,
"rewards/margins": 4.75,
"rewards/rejected": -6.4375,
"step": 380
},
{
"epoch": 0.4083769633507853,
"grad_norm": 1.0041746128923654,
"learning_rate": 4.9803325467003084e-05,
"logits/chosen": -11.9375,
"logits/rejected": -9.125,
"logps/chosen": -600.0,
"logps/rejected": -2752.0,
"loss": 0.1749,
"rewards/accuracies": 0.9312499761581421,
"rewards/chosen": -1.5,
"rewards/margins": 4.1875,
"rewards/rejected": -5.71875,
"step": 390
},
{
"epoch": 0.418848167539267,
"grad_norm": 1.4499848671151063,
"learning_rate": 4.9763345676749816e-05,
"logits/chosen": -11.9375,
"logits/rejected": -9.5,
"logps/chosen": -660.0,
"logps/rejected": -2784.0,
"loss": 0.1554,
"rewards/accuracies": 0.956250011920929,
"rewards/chosen": -1.7109375,
"rewards/margins": 4.71875,
"rewards/rejected": -6.4375,
"step": 400
},
{
"epoch": 0.418848167539267,
"eval_logits/chosen": -12.4375,
"eval_logits/rejected": -12.0625,
"eval_logps/chosen": -508.0,
"eval_logps/rejected": -484.0,
"eval_loss": 0.7756562232971191,
"eval_rewards/accuracies": 0.494140625,
"eval_rewards/chosen": -1.890625,
"eval_rewards/margins": 0.06884765625,
"eval_rewards/rejected": -1.9609375,
"eval_runtime": 46.6621,
"eval_samples_per_second": 42.861,
"eval_steps_per_second": 0.686,
"step": 400
},
{
"epoch": 0.4293193717277487,
"grad_norm": 0.9999172844726647,
"learning_rate": 4.9719688511956374e-05,
"logits/chosen": -12.625,
"logits/rejected": -9.8125,
"logps/chosen": -756.0,
"logps/rejected": -2896.0,
"loss": 0.2525,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -2.484375,
"rewards/margins": 4.5625,
"rewards/rejected": -7.03125,
"step": 410
},
{
"epoch": 0.4397905759162304,
"grad_norm": 1.0938663223211436,
"learning_rate": 4.967236045574294e-05,
"logits/chosen": -14.375,
"logits/rejected": -10.8125,
"logps/chosen": -616.0,
"logps/rejected": -2704.0,
"loss": 0.1363,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -1.9296875,
"rewards/margins": 3.953125,
"rewards/rejected": -5.875,
"step": 420
},
{
"epoch": 0.450261780104712,
"grad_norm": 3.5742944799105945,
"learning_rate": 4.96213685363597e-05,
"logits/chosen": -15.3125,
"logits/rejected": -11.625,
"logps/chosen": -624.0,
"logps/rejected": -2944.0,
"loss": 0.1818,
"rewards/accuracies": 0.9312499761581421,
"rewards/chosen": -2.03125,
"rewards/margins": 6.34375,
"rewards/rejected": -8.375,
"step": 430
},
{
"epoch": 0.4607329842931937,
"grad_norm": 0.9862192176921567,
"learning_rate": 4.956672032614315e-05,
"logits/chosen": -14.375,
"logits/rejected": -10.875,
"logps/chosen": -612.0,
"logps/rejected": -2896.0,
"loss": 0.1255,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -1.7421875,
"rewards/margins": 5.40625,
"rewards/rejected": -7.15625,
"step": 440
},
{
"epoch": 0.4712041884816754,
"grad_norm": 1.2720728098607696,
"learning_rate": 4.950842394039157e-05,
"logits/chosen": -13.75,
"logits/rejected": -10.6875,
"logps/chosen": -604.0,
"logps/rejected": -2736.0,
"loss": 0.1591,
"rewards/accuracies": 0.9437500238418579,
"rewards/chosen": -1.859375,
"rewards/margins": 4.625,
"rewards/rejected": -6.5,
"step": 450
},
{
"epoch": 0.4816753926701571,
"grad_norm": 1.1741368358140423,
"learning_rate": 4.944648803615994e-05,
"logits/chosen": -14.5625,
"logits/rejected": -10.8125,
"logps/chosen": -600.0,
"logps/rejected": -2992.0,
"loss": 0.123,
"rewards/accuracies": 0.96875,
"rewards/chosen": -1.796875,
"rewards/margins": 5.0,
"rewards/rejected": -6.78125,
"step": 460
},
{
"epoch": 0.49214659685863876,
"grad_norm": 1.0270724548446106,
"learning_rate": 4.938092181097433e-05,
"logits/chosen": -14.5,
"logits/rejected": -11.1875,
"logps/chosen": -672.0,
"logps/rejected": -2896.0,
"loss": 0.1099,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -2.109375,
"rewards/margins": 6.03125,
"rewards/rejected": -8.125,
"step": 470
},
{
"epoch": 0.5026178010471204,
"grad_norm": 0.9472936322210024,
"learning_rate": 4.931173500146607e-05,
"logits/chosen": -14.0625,
"logits/rejected": -10.625,
"logps/chosen": -696.0,
"logps/rejected": -3232.0,
"loss": 0.1348,
"rewards/accuracies": 0.96875,
"rewards/chosen": -2.390625,
"rewards/margins": 7.75,
"rewards/rejected": -10.125,
"step": 480
},
{
"epoch": 0.5130890052356021,
"grad_norm": 0.8788605376340061,
"learning_rate": 4.923893788192582e-05,
"logits/chosen": -13.75,
"logits/rejected": -10.125,
"logps/chosen": -720.0,
"logps/rejected": -3120.0,
"loss": 0.1589,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -2.5,
"rewards/margins": 5.75,
"rewards/rejected": -8.25,
"step": 490
},
{
"epoch": 0.5235602094240838,
"grad_norm": 1.4397812402574388,
"learning_rate": 4.9162541262777906e-05,
"logits/chosen": -13.9375,
"logits/rejected": -10.6875,
"logps/chosen": -696.0,
"logps/rejected": -2880.0,
"loss": 0.1334,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -2.65625,
"rewards/margins": 5.25,
"rewards/rejected": -7.90625,
"step": 500
},
{
"epoch": 0.5235602094240838,
"eval_logits/chosen": -15.0625,
"eval_logits/rejected": -14.75,
"eval_logps/chosen": -508.0,
"eval_logps/rejected": -482.0,
"eval_loss": 0.843999981880188,
"eval_rewards/accuracies": 0.48046875,
"eval_rewards/chosen": -1.890625,
"eval_rewards/margins": 0.04052734375,
"eval_rewards/rejected": -1.9296875,
"eval_runtime": 46.6493,
"eval_samples_per_second": 42.873,
"eval_steps_per_second": 0.686,
"step": 500
},
{
"epoch": 0.5340314136125655,
"grad_norm": 0.8201411873980885,
"learning_rate": 4.90825564889749e-05,
"logits/chosen": -14.8125,
"logits/rejected": -11.25,
"logps/chosen": -652.0,
"logps/rejected": -2880.0,
"loss": 0.1502,
"rewards/accuracies": 0.9375,
"rewards/chosen": -2.109375,
"rewards/margins": 5.34375,
"rewards/rejected": -7.4375,
"step": 510
},
{
"epoch": 0.5445026178010471,
"grad_norm": 1.7770677090321023,
"learning_rate": 4.899899543831288e-05,
"logits/chosen": -15.375,
"logits/rejected": -11.9375,
"logps/chosen": -668.0,
"logps/rejected": -2912.0,
"loss": 0.2062,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -1.9453125,
"rewards/margins": 5.0,
"rewards/rejected": -6.9375,
"step": 520
},
{
"epoch": 0.5549738219895288,
"grad_norm": 1.206280684623937,
"learning_rate": 4.891187051966765e-05,
"logits/chosen": -15.5,
"logits/rejected": -11.75,
"logps/chosen": -552.0,
"logps/rejected": -2672.0,
"loss": 0.1436,
"rewards/accuracies": 0.956250011920929,
"rewards/chosen": -1.421875,
"rewards/margins": 4.15625,
"rewards/rejected": -5.59375,
"step": 530
},
{
"epoch": 0.5654450261780105,
"grad_norm": 0.6891142422929127,
"learning_rate": 4.882119467115191e-05,
"logits/chosen": -15.4375,
"logits/rejected": -11.75,
"logps/chosen": -692.0,
"logps/rejected": -2768.0,
"loss": 0.1798,
"rewards/accuracies": 0.9437500238418579,
"rewards/chosen": -2.015625,
"rewards/margins": 4.46875,
"rewards/rejected": -6.46875,
"step": 540
},
{
"epoch": 0.5759162303664922,
"grad_norm": 16.22392122468538,
"learning_rate": 4.872698135819402e-05,
"logits/chosen": -14.9375,
"logits/rejected": -11.0625,
"logps/chosen": -600.0,
"logps/rejected": -2880.0,
"loss": 0.1436,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -1.9375,
"rewards/margins": 5.875,
"rewards/rejected": -7.8125,
"step": 550
},
{
"epoch": 0.5863874345549738,
"grad_norm": 0.8497144695012769,
"learning_rate": 4.862924457153831e-05,
"logits/chosen": -13.5625,
"logits/rejected": -10.125,
"logps/chosen": -632.0,
"logps/rejected": -2944.0,
"loss": 0.1325,
"rewards/accuracies": 0.9437500238418579,
"rewards/chosen": -2.078125,
"rewards/margins": 5.28125,
"rewards/rejected": -7.34375,
"step": 560
},
{
"epoch": 0.5968586387434555,
"grad_norm": 1.4563736140315904,
"learning_rate": 4.8527998825167484e-05,
"logits/chosen": -14.3125,
"logits/rejected": -10.5625,
"logps/chosen": -656.0,
"logps/rejected": -2992.0,
"loss": 0.1194,
"rewards/accuracies": 0.956250011920929,
"rewards/chosen": -1.96875,
"rewards/margins": 5.6875,
"rewards/rejected": -7.65625,
"step": 570
},
{
"epoch": 0.6073298429319371,
"grad_norm": 1.4139335168346614,
"learning_rate": 4.842325915414728e-05,
"logits/chosen": -14.1875,
"logits/rejected": -10.875,
"logps/chosen": -640.0,
"logps/rejected": -2944.0,
"loss": 0.1639,
"rewards/accuracies": 0.9624999761581421,
"rewards/chosen": -1.703125,
"rewards/margins": 5.75,
"rewards/rejected": -7.4375,
"step": 580
},
{
"epoch": 0.6178010471204188,
"grad_norm": 1.1746955773340486,
"learning_rate": 4.831504111239373e-05,
"logits/chosen": -14.75,
"logits/rejected": -11.0625,
"logps/chosen": -596.0,
"logps/rejected": -2880.0,
"loss": 0.1347,
"rewards/accuracies": 0.96875,
"rewards/chosen": -1.5625,
"rewards/margins": 4.59375,
"rewards/rejected": -6.125,
"step": 590
},
{
"epoch": 0.6282722513089005,
"grad_norm": 2.3560354618905595,
"learning_rate": 4.8203360770363425e-05,
"logits/chosen": -15.0,
"logits/rejected": -12.4375,
"logps/chosen": -672.0,
"logps/rejected": -2816.0,
"loss": 0.1429,
"rewards/accuracies": 0.9125000238418579,
"rewards/chosen": -2.109375,
"rewards/margins": 6.28125,
"rewards/rejected": -8.375,
"step": 600
},
{
"epoch": 0.6282722513089005,
"eval_logits/chosen": -14.875,
"eval_logits/rejected": -14.625,
"eval_logps/chosen": -482.0,
"eval_logps/rejected": -458.0,
"eval_loss": 0.817984402179718,
"eval_rewards/accuracies": 0.5078125,
"eval_rewards/chosen": -1.640625,
"eval_rewards/margins": 0.051513671875,
"eval_rewards/rejected": -1.6875,
"eval_runtime": 46.6448,
"eval_samples_per_second": 42.877,
"eval_steps_per_second": 0.686,
"step": 600
},
{
"epoch": 0.6387434554973822,
"grad_norm": 2.4829767453189406,
"learning_rate": 4.808823471266702e-05,
"logits/chosen": -14.3125,
"logits/rejected": -11.9375,
"logps/chosen": -720.0,
"logps/rejected": -2896.0,
"loss": 0.1839,
"rewards/accuracies": 0.9125000238418579,
"rewards/chosen": -2.046875,
"rewards/margins": 6.15625,
"rewards/rejected": -8.1875,
"step": 610
},
{
"epoch": 0.6492146596858639,
"grad_norm": 1.1162079736651396,
"learning_rate": 4.796968003560639e-05,
"logits/chosen": -14.0625,
"logits/rejected": -10.75,
"logps/chosen": -584.0,
"logps/rejected": -2832.0,
"loss": 0.1276,
"rewards/accuracies": 0.9624999761581421,
"rewards/chosen": -1.484375,
"rewards/margins": 4.40625,
"rewards/rejected": -5.90625,
"step": 620
},
{
"epoch": 0.6596858638743456,
"grad_norm": 0.8344561676739205,
"learning_rate": 4.7847714344635864e-05,
"logits/chosen": -13.875,
"logits/rejected": -10.5625,
"logps/chosen": -588.0,
"logps/rejected": -2912.0,
"loss": 0.1516,
"rewards/accuracies": 0.9375,
"rewards/chosen": -1.5390625,
"rewards/margins": 5.75,
"rewards/rejected": -7.28125,
"step": 630
},
{
"epoch": 0.6701570680628273,
"grad_norm": 0.8857409236852289,
"learning_rate": 4.772235575174776e-05,
"logits/chosen": -14.25,
"logits/rejected": -10.6875,
"logps/chosen": -600.0,
"logps/rejected": -2864.0,
"loss": 0.1633,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -1.46875,
"rewards/margins": 5.09375,
"rewards/rejected": -6.5625,
"step": 640
},
{
"epoch": 0.680628272251309,
"grad_norm": 0.6584223156093467,
"learning_rate": 4.7593622872782754e-05,
"logits/chosen": -15.375,
"logits/rejected": -11.6875,
"logps/chosen": -532.0,
"logps/rejected": -2832.0,
"loss": 0.1516,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -1.3203125,
"rewards/margins": 5.3125,
"rewards/rejected": -6.625,
"step": 650
},
{
"epoch": 0.6910994764397905,
"grad_norm": 0.7100195153241552,
"learning_rate": 4.746153482466542e-05,
"logits/chosen": -15.9375,
"logits/rejected": -11.9375,
"logps/chosen": -568.0,
"logps/rejected": -2896.0,
"loss": 0.1339,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -1.4765625,
"rewards/margins": 5.40625,
"rewards/rejected": -6.875,
"step": 660
},
{
"epoch": 0.7015706806282722,
"grad_norm": 0.9009807793234297,
"learning_rate": 4.732611122256537e-05,
"logits/chosen": -15.3125,
"logits/rejected": -12.3125,
"logps/chosen": -640.0,
"logps/rejected": -2736.0,
"loss": 0.1229,
"rewards/accuracies": 0.9375,
"rewards/chosen": -2.046875,
"rewards/margins": 4.8125,
"rewards/rejected": -6.84375,
"step": 670
},
{
"epoch": 0.7120418848167539,
"grad_norm": 2.44253283349393,
"learning_rate": 4.718737217698435e-05,
"logits/chosen": -15.125,
"logits/rejected": -11.75,
"logps/chosen": -696.0,
"logps/rejected": -2848.0,
"loss": 0.1285,
"rewards/accuracies": 0.9375,
"rewards/chosen": -2.1875,
"rewards/margins": 5.5625,
"rewards/rejected": -7.75,
"step": 680
},
{
"epoch": 0.7225130890052356,
"grad_norm": 0.8162247996732467,
"learning_rate": 4.704533829076982e-05,
"logits/chosen": -14.8125,
"logits/rejected": -11.125,
"logps/chosen": -664.0,
"logps/rejected": -2912.0,
"loss": 0.174,
"rewards/accuracies": 0.96875,
"rewards/chosen": -2.0625,
"rewards/margins": 6.25,
"rewards/rejected": -8.3125,
"step": 690
},
{
"epoch": 0.7329842931937173,
"grad_norm": 0.8467194346356433,
"learning_rate": 4.6900030656055455e-05,
"logits/chosen": -14.0625,
"logits/rejected": -10.125,
"logps/chosen": -668.0,
"logps/rejected": -2832.0,
"loss": 0.1228,
"rewards/accuracies": 0.9624999761581421,
"rewards/chosen": -1.75,
"rewards/margins": 4.65625,
"rewards/rejected": -6.375,
"step": 700
},
{
"epoch": 0.7329842931937173,
"eval_logits/chosen": -14.4375,
"eval_logits/rejected": -14.0625,
"eval_logps/chosen": -474.0,
"eval_logps/rejected": -460.0,
"eval_loss": 0.8036093711853027,
"eval_rewards/accuracies": 0.5,
"eval_rewards/chosen": -1.5625,
"eval_rewards/margins": 0.1611328125,
"eval_rewards/rejected": -1.7265625,
"eval_runtime": 46.6813,
"eval_samples_per_second": 42.844,
"eval_steps_per_second": 0.685,
"step": 700
},
{
"epoch": 0.743455497382199,
"grad_norm": 0.9033330861109353,
"learning_rate": 4.6751470851128906e-05,
"logits/chosen": -14.5,
"logits/rejected": -10.75,
"logps/chosen": -624.0,
"logps/rejected": -3040.0,
"loss": 0.1066,
"rewards/accuracies": 0.9624999761581421,
"rewards/chosen": -1.8203125,
"rewards/margins": 6.625,
"rewards/rejected": -8.4375,
"step": 710
},
{
"epoch": 0.7539267015706806,
"grad_norm": 0.892572852718864,
"learning_rate": 4.6599680937227454e-05,
"logits/chosen": -14.5625,
"logits/rejected": -11.375,
"logps/chosen": -612.0,
"logps/rejected": -2800.0,
"loss": 0.1621,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -1.9296875,
"rewards/margins": 5.625,
"rewards/rejected": -7.53125,
"step": 720
},
{
"epoch": 0.7643979057591623,
"grad_norm": 1.7528419008060139,
"learning_rate": 4.644468345526183e-05,
"logits/chosen": -14.625,
"logits/rejected": -11.25,
"logps/chosen": -636.0,
"logps/rejected": -2784.0,
"loss": 0.1697,
"rewards/accuracies": 0.9437500238418579,
"rewards/chosen": -1.8984375,
"rewards/margins": 4.96875,
"rewards/rejected": -6.875,
"step": 730
},
{
"epoch": 0.774869109947644,
"grad_norm": 1.06748750642103,
"learning_rate": 4.628650142246894e-05,
"logits/chosen": -15.375,
"logits/rejected": -11.8125,
"logps/chosen": -600.0,
"logps/rejected": -2800.0,
"loss": 0.1503,
"rewards/accuracies": 0.9437500238418579,
"rewards/chosen": -1.7734375,
"rewards/margins": 4.78125,
"rewards/rejected": -6.5625,
"step": 740
},
{
"epoch": 0.7853403141361257,
"grad_norm": 1.8873391012169038,
"learning_rate": 4.6125158328993765e-05,
"logits/chosen": -15.875,
"logits/rejected": -12.625,
"logps/chosen": -752.0,
"logps/rejected": -2832.0,
"loss": 0.1457,
"rewards/accuracies": 0.9437500238418579,
"rewards/chosen": -2.328125,
"rewards/margins": 4.8125,
"rewards/rejected": -7.15625,
"step": 750
},
{
"epoch": 0.7958115183246073,
"grad_norm": 1.4062630391550905,
"learning_rate": 4.5960678134401064e-05,
"logits/chosen": -15.1875,
"logits/rejected": -11.75,
"logps/chosen": -672.0,
"logps/rejected": -3152.0,
"loss": 0.1307,
"rewards/accuracies": 0.956250011920929,
"rewards/chosen": -2.140625,
"rewards/margins": 7.0,
"rewards/rejected": -9.125,
"step": 760
},
{
"epoch": 0.806282722513089,
"grad_norm": 0.8650076923919571,
"learning_rate": 4.579308526411734e-05,
"logits/chosen": -14.5,
"logits/rejected": -11.125,
"logps/chosen": -556.0,
"logps/rejected": -2784.0,
"loss": 0.1539,
"rewards/accuracies": 0.981249988079071,
"rewards/chosen": -1.5390625,
"rewards/margins": 5.15625,
"rewards/rejected": -6.6875,
"step": 770
},
{
"epoch": 0.8167539267015707,
"grad_norm": 1.0409443273649337,
"learning_rate": 4.562240460580367e-05,
"logits/chosen": -14.0,
"logits/rejected": -10.9375,
"logps/chosen": -676.0,
"logps/rejected": -2688.0,
"loss": 0.1523,
"rewards/accuracies": 0.9125000238418579,
"rewards/chosen": -1.9453125,
"rewards/margins": 4.1875,
"rewards/rejected": -6.125,
"step": 780
},
{
"epoch": 0.8272251308900523,
"grad_norm": 1.1302794834047873,
"learning_rate": 4.544866150565985e-05,
"logits/chosen": -14.5,
"logits/rejected": -10.6875,
"logps/chosen": -600.0,
"logps/rejected": -2944.0,
"loss": 0.1236,
"rewards/accuracies": 0.96875,
"rewards/chosen": -1.765625,
"rewards/margins": 5.0625,
"rewards/rejected": -6.8125,
"step": 790
},
{
"epoch": 0.837696335078534,
"grad_norm": 0.9058381810609959,
"learning_rate": 4.527188176466051e-05,
"logits/chosen": -14.375,
"logits/rejected": -11.3125,
"logps/chosen": -604.0,
"logps/rejected": -2880.0,
"loss": 0.1182,
"rewards/accuracies": 0.9437500238418579,
"rewards/chosen": -1.96875,
"rewards/margins": 5.40625,
"rewards/rejected": -7.375,
"step": 800
},
{
"epoch": 0.837696335078534,
"eval_logits/chosen": -15.125,
"eval_logits/rejected": -14.875,
"eval_logps/chosen": -504.0,
"eval_logps/rejected": -480.0,
"eval_loss": 0.8425156474113464,
"eval_rewards/accuracies": 0.5078125,
"eval_rewards/chosen": -1.84375,
"eval_rewards/margins": 0.06640625,
"eval_rewards/rejected": -1.9140625,
"eval_runtime": 46.6693,
"eval_samples_per_second": 42.855,
"eval_steps_per_second": 0.686,
"step": 800
},
{
"epoch": 0.8481675392670157,
"grad_norm": 0.8731598694365104,
"learning_rate": 4.509209163472357e-05,
"logits/chosen": -14.6875,
"logits/rejected": -11.375,
"logps/chosen": -696.0,
"logps/rejected": -2880.0,
"loss": 0.1395,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -2.15625,
"rewards/margins": 5.3125,
"rewards/rejected": -7.46875,
"step": 810
},
{
"epoch": 0.8586387434554974,
"grad_norm": 1.1922079481298178,
"learning_rate": 4.4909317814811896e-05,
"logits/chosen": -15.3125,
"logits/rejected": -11.5,
"logps/chosen": -632.0,
"logps/rejected": -2976.0,
"loss": 0.1471,
"rewards/accuracies": 0.96875,
"rewards/chosen": -2.078125,
"rewards/margins": 5.6875,
"rewards/rejected": -7.78125,
"step": 820
},
{
"epoch": 0.8691099476439791,
"grad_norm": 1.2574184391827443,
"learning_rate": 4.472358744696845e-05,
"logits/chosen": -14.9375,
"logits/rejected": -11.375,
"logps/chosen": -636.0,
"logps/rejected": -2816.0,
"loss": 0.1532,
"rewards/accuracies": 0.9312499761581421,
"rewards/chosen": -1.984375,
"rewards/margins": 5.09375,
"rewards/rejected": -7.09375,
"step": 830
},
{
"epoch": 0.8795811518324608,
"grad_norm": 1.048769867012725,
"learning_rate": 4.4534928112285645e-05,
"logits/chosen": -14.875,
"logits/rejected": -11.0625,
"logps/chosen": -628.0,
"logps/rejected": -2720.0,
"loss": 0.1238,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -1.75,
"rewards/margins": 4.65625,
"rewards/rejected": -6.40625,
"step": 840
},
{
"epoch": 0.8900523560209425,
"grad_norm": 1.1070511362542446,
"learning_rate": 4.43433678268096e-05,
"logits/chosen": -14.6875,
"logits/rejected": -10.9375,
"logps/chosen": -648.0,
"logps/rejected": -2864.0,
"loss": 0.1421,
"rewards/accuracies": 0.9375,
"rewards/chosen": -1.984375,
"rewards/margins": 5.25,
"rewards/rejected": -7.21875,
"step": 850
},
{
"epoch": 0.900523560209424,
"grad_norm": 0.9664871476110467,
"learning_rate": 4.414893503737968e-05,
"logits/chosen": -14.8125,
"logits/rejected": -10.9375,
"logps/chosen": -604.0,
"logps/rejected": -2912.0,
"loss": 0.1264,
"rewards/accuracies": 0.96875,
"rewards/chosen": -1.9609375,
"rewards/margins": 5.5625,
"rewards/rejected": -7.5,
"step": 860
},
{
"epoch": 0.9109947643979057,
"grad_norm": 1.467464147079793,
"learning_rate": 4.395165861740414e-05,
"logits/chosen": -14.8125,
"logits/rejected": -11.4375,
"logps/chosen": -648.0,
"logps/rejected": -3040.0,
"loss": 0.104,
"rewards/accuracies": 0.9312499761581421,
"rewards/chosen": -2.25,
"rewards/margins": 6.625,
"rewards/rejected": -8.875,
"step": 870
},
{
"epoch": 0.9214659685863874,
"grad_norm": 1.2405036649763637,
"learning_rate": 4.375156786257241e-05,
"logits/chosen": -14.3125,
"logits/rejected": -10.9375,
"logps/chosen": -724.0,
"logps/rejected": -3168.0,
"loss": 0.1082,
"rewards/accuracies": 0.96875,
"rewards/chosen": -2.453125,
"rewards/margins": 6.875,
"rewards/rejected": -9.3125,
"step": 880
},
{
"epoch": 0.9319371727748691,
"grad_norm": 1.7565573776910457,
"learning_rate": 4.354869248650466e-05,
"logits/chosen": -14.4375,
"logits/rejected": -10.9375,
"logps/chosen": -748.0,
"logps/rejected": -2928.0,
"loss": 0.1455,
"rewards/accuracies": 0.90625,
"rewards/chosen": -2.6875,
"rewards/margins": 5.59375,
"rewards/rejected": -8.25,
"step": 890
},
{
"epoch": 0.9424083769633508,
"grad_norm": 0.6456546631026289,
"learning_rate": 4.334306261633931e-05,
"logits/chosen": -13.75,
"logits/rejected": -10.6875,
"logps/chosen": -676.0,
"logps/rejected": -2944.0,
"loss": 0.1471,
"rewards/accuracies": 0.9312499761581421,
"rewards/chosen": -2.546875,
"rewards/margins": 5.09375,
"rewards/rejected": -7.625,
"step": 900
},
{
"epoch": 0.9424083769633508,
"eval_logits/chosen": -13.9375,
"eval_logits/rejected": -13.625,
"eval_logps/chosen": -560.0,
"eval_logps/rejected": -540.0,
"eval_loss": 0.8238124847412109,
"eval_rewards/accuracies": 0.50390625,
"eval_rewards/chosen": -2.421875,
"eval_rewards/margins": 0.09912109375,
"eval_rewards/rejected": -2.53125,
"eval_runtime": 46.6755,
"eval_samples_per_second": 42.849,
"eval_steps_per_second": 0.686,
"step": 900
},
{
"epoch": 0.9528795811518325,
"grad_norm": 1.5156530888151523,
"learning_rate": 4.313470878825911e-05,
"logits/chosen": -13.875,
"logits/rejected": -11.0625,
"logps/chosen": -708.0,
"logps/rejected": -2880.0,
"loss": 0.1612,
"rewards/accuracies": 0.9312499761581421,
"rewards/chosen": -2.5,
"rewards/margins": 4.96875,
"rewards/rejected": -7.5,
"step": 910
},
{
"epoch": 0.9633507853403142,
"grad_norm": 2.1207386083227266,
"learning_rate": 4.292366194295652e-05,
"logits/chosen": -13.4375,
"logits/rejected": -10.0625,
"logps/chosen": -728.0,
"logps/rejected": -2960.0,
"loss": 0.1493,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -2.328125,
"rewards/margins": 5.3125,
"rewards/rejected": -7.625,
"step": 920
},
{
"epoch": 0.9738219895287958,
"grad_norm": 1.6517330796682208,
"learning_rate": 4.2709953421038964e-05,
"logits/chosen": -13.3125,
"logits/rejected": -9.875,
"logps/chosen": -612.0,
"logps/rejected": -2976.0,
"loss": 0.1224,
"rewards/accuracies": 0.956250011920929,
"rewards/chosen": -2.171875,
"rewards/margins": 5.21875,
"rewards/rejected": -7.375,
"step": 930
},
{
"epoch": 0.9842931937172775,
"grad_norm": 0.9291317173048833,
"learning_rate": 4.249361495837478e-05,
"logits/chosen": -13.625,
"logits/rejected": -9.875,
"logps/chosen": -636.0,
"logps/rejected": -3120.0,
"loss": 0.1223,
"rewards/accuracies": 0.96875,
"rewards/chosen": -2.328125,
"rewards/margins": 6.21875,
"rewards/rejected": -8.5625,
"step": 940
},
{
"epoch": 0.9947643979057592,
"grad_norm": 0.6015729919156609,
"learning_rate": 4.227467868138035e-05,
"logits/chosen": -14.0625,
"logits/rejected": -10.4375,
"logps/chosen": -672.0,
"logps/rejected": -3056.0,
"loss": 0.1331,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -2.125,
"rewards/margins": 5.8125,
"rewards/rejected": -7.9375,
"step": 950
},
{
"epoch": 1.0052356020942408,
"grad_norm": 1.1877047561484144,
"learning_rate": 4.205317710224938e-05,
"logits/chosen": -14.125,
"logits/rejected": -10.1875,
"logps/chosen": -580.0,
"logps/rejected": -2832.0,
"loss": 0.0921,
"rewards/accuracies": 0.96875,
"rewards/chosen": -1.6796875,
"rewards/margins": 5.53125,
"rewards/rejected": -7.1875,
"step": 960
},
{
"epoch": 1.0157068062827226,
"grad_norm": 0.7161458288379944,
"learning_rate": 4.1829143114124733e-05,
"logits/chosen": -13.9375,
"logits/rejected": -10.4375,
"logps/chosen": -624.0,
"logps/rejected": -3072.0,
"loss": 0.0532,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -1.859375,
"rewards/margins": 6.40625,
"rewards/rejected": -8.25,
"step": 970
},
{
"epoch": 1.0261780104712042,
"grad_norm": 0.23417374788952794,
"learning_rate": 4.160260998621387e-05,
"logits/chosen": -14.1875,
"logits/rejected": -10.5625,
"logps/chosen": -616.0,
"logps/rejected": -3104.0,
"loss": 0.0381,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -1.8359375,
"rewards/margins": 6.71875,
"rewards/rejected": -8.5625,
"step": 980
},
{
"epoch": 1.036649214659686,
"grad_norm": 0.2105775243248358,
"learning_rate": 4.137361135884824e-05,
"logits/chosen": -14.75,
"logits/rejected": -11.3125,
"logps/chosen": -628.0,
"logps/rejected": -3072.0,
"loss": 0.0513,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -1.9921875,
"rewards/margins": 7.125,
"rewards/rejected": -9.125,
"step": 990
},
{
"epoch": 1.0471204188481675,
"grad_norm": 0.8545296703197597,
"learning_rate": 4.114218123848777e-05,
"logits/chosen": -14.75,
"logits/rejected": -10.75,
"logps/chosen": -652.0,
"logps/rejected": -3312.0,
"loss": 0.0251,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -2.171875,
"rewards/margins": 8.0625,
"rewards/rejected": -10.1875,
"step": 1000
},
{
"epoch": 1.0471204188481675,
"eval_logits/chosen": -14.25,
"eval_logits/rejected": -14.0,
"eval_logps/chosen": -548.0,
"eval_logps/rejected": -532.0,
"eval_loss": 0.8727031350135803,
"eval_rewards/accuracies": 0.48046875,
"eval_rewards/chosen": -2.296875,
"eval_rewards/margins": 0.12890625,
"eval_rewards/rejected": -2.421875,
"eval_runtime": 46.6732,
"eval_samples_per_second": 42.851,
"eval_steps_per_second": 0.686,
"step": 1000
},
{
"epoch": 1.057591623036649,
"grad_norm": 0.48986300123951065,
"learning_rate": 4.0908353992670824e-05,
"logits/chosen": -14.3125,
"logits/rejected": -10.875,
"logps/chosen": -636.0,
"logps/rejected": -3296.0,
"loss": 0.0335,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -2.0625,
"rewards/margins": 8.625,
"rewards/rejected": -10.6875,
"step": 1010
},
{
"epoch": 1.068062827225131,
"grad_norm": 1.6918803585937856,
"learning_rate": 4.067216434491059e-05,
"logits/chosen": -14.1875,
"logits/rejected": -10.8125,
"logps/chosen": -660.0,
"logps/rejected": -3120.0,
"loss": 0.0527,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -2.171875,
"rewards/margins": 7.0,
"rewards/rejected": -9.1875,
"step": 1020
},
{
"epoch": 1.0785340314136125,
"grad_norm": 0.15373524739811048,
"learning_rate": 4.043364736953866e-05,
"logits/chosen": -13.375,
"logits/rejected": -10.1875,
"logps/chosen": -684.0,
"logps/rejected": -2944.0,
"loss": 0.0249,
"rewards/accuracies": 1.0,
"rewards/chosen": -2.125,
"rewards/margins": 6.8125,
"rewards/rejected": -8.9375,
"step": 1030
},
{
"epoch": 1.0890052356020943,
"grad_norm": 1.0061698664033307,
"learning_rate": 4.019283848649643e-05,
"logits/chosen": -14.25,
"logits/rejected": -11.375,
"logps/chosen": -684.0,
"logps/rejected": -3184.0,
"loss": 0.0394,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -2.28125,
"rewards/margins": 7.65625,
"rewards/rejected": -9.9375,
"step": 1040
},
{
"epoch": 1.0994764397905759,
"grad_norm": 0.6791257572978986,
"learning_rate": 3.9949773456075215e-05,
"logits/chosen": -13.9375,
"logits/rejected": -11.25,
"logps/chosen": -764.0,
"logps/rejected": -3088.0,
"loss": 0.0288,
"rewards/accuracies": 0.96875,
"rewards/chosen": -2.328125,
"rewards/margins": 6.9375,
"rewards/rejected": -9.25,
"step": 1050
},
{
"epoch": 1.1099476439790577,
"grad_norm": 0.5457170734133783,
"learning_rate": 3.970448837360585e-05,
"logits/chosen": -13.5625,
"logits/rejected": -11.1875,
"logps/chosen": -728.0,
"logps/rejected": -3104.0,
"loss": 0.0225,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -2.3125,
"rewards/margins": 7.5,
"rewards/rejected": -9.8125,
"step": 1060
},
{
"epoch": 1.1204188481675392,
"grad_norm": 1.0828616058439433,
"learning_rate": 3.945701966409846e-05,
"logits/chosen": -13.625,
"logits/rejected": -10.4375,
"logps/chosen": -704.0,
"logps/rejected": -3184.0,
"loss": 0.0274,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -2.34375,
"rewards/margins": 8.0625,
"rewards/rejected": -10.375,
"step": 1070
},
{
"epoch": 1.130890052356021,
"grad_norm": 0.1680477906744099,
"learning_rate": 3.9207404076833374e-05,
"logits/chosen": -13.375,
"logits/rejected": -10.375,
"logps/chosen": -644.0,
"logps/rejected": -3280.0,
"loss": 0.0345,
"rewards/accuracies": 1.0,
"rewards/chosen": -2.390625,
"rewards/margins": 8.1875,
"rewards/rejected": -10.5625,
"step": 1080
},
{
"epoch": 1.1413612565445026,
"grad_norm": 0.3730345032803325,
"learning_rate": 3.895567867990379e-05,
"logits/chosen": -14.25,
"logits/rejected": -10.875,
"logps/chosen": -636.0,
"logps/rejected": -3152.0,
"loss": 0.0231,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -2.140625,
"rewards/margins": 7.28125,
"rewards/rejected": -9.375,
"step": 1090
},
{
"epoch": 1.1518324607329844,
"grad_norm": 1.1651313536885508,
"learning_rate": 3.870188085471114e-05,
"logits/chosen": -13.4375,
"logits/rejected": -10.5,
"logps/chosen": -744.0,
"logps/rejected": -3152.0,
"loss": 0.033,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -2.328125,
"rewards/margins": 7.375,
"rewards/rejected": -9.6875,
"step": 1100
},
{
"epoch": 1.1518324607329844,
"eval_logits/chosen": -14.125,
"eval_logits/rejected": -13.75,
"eval_logps/chosen": -524.0,
"eval_logps/rejected": -500.0,
"eval_loss": 0.8287187218666077,
"eval_rewards/accuracies": 0.5078125,
"eval_rewards/chosen": -2.0625,
"eval_rewards/margins": 0.05029296875,
"eval_rewards/rejected": -2.109375,
"eval_runtime": 46.6805,
"eval_samples_per_second": 42.844,
"eval_steps_per_second": 0.686,
"step": 1100
},
{
"epoch": 1.162303664921466,
"grad_norm": 0.17645294546217305,
"learning_rate": 3.844604829041395e-05,
"logits/chosen": -14.1875,
"logits/rejected": -11.625,
"logps/chosen": -668.0,
"logps/rejected": -3056.0,
"loss": 0.0319,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -2.234375,
"rewards/margins": 7.53125,
"rewards/rejected": -9.75,
"step": 1110
},
{
"epoch": 1.1727748691099475,
"grad_norm": 0.23126830068369286,
"learning_rate": 3.818821897833095e-05,
"logits/chosen": -14.0,
"logits/rejected": -10.9375,
"logps/chosen": -740.0,
"logps/rejected": -3136.0,
"loss": 0.0234,
"rewards/accuracies": 0.981249988079071,
"rewards/chosen": -2.359375,
"rewards/margins": 7.0,
"rewards/rejected": -9.3125,
"step": 1120
},
{
"epoch": 1.1832460732984293,
"grad_norm": 0.5650854225333866,
"learning_rate": 3.792843120629935e-05,
"logits/chosen": -13.6875,
"logits/rejected": -11.0,
"logps/chosen": -600.0,
"logps/rejected": -3072.0,
"loss": 0.0402,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -2.125,
"rewards/margins": 7.5,
"rewards/rejected": -9.625,
"step": 1130
},
{
"epoch": 1.193717277486911,
"grad_norm": 0.0798648848931708,
"learning_rate": 3.766672355298907e-05,
"logits/chosen": -12.875,
"logits/rejected": -10.4375,
"logps/chosen": -616.0,
"logps/rejected": -3200.0,
"loss": 0.0472,
"rewards/accuracies": 1.0,
"rewards/chosen": -2.171875,
"rewards/margins": 7.875,
"rewards/rejected": -10.0625,
"step": 1140
},
{
"epoch": 1.2041884816753927,
"grad_norm": 0.3554084702213716,
"learning_rate": 3.740313488217373e-05,
"logits/chosen": -11.625,
"logits/rejected": -10.0,
"logps/chosen": -716.0,
"logps/rejected": -3072.0,
"loss": 0.0276,
"rewards/accuracies": 0.981249988079071,
"rewards/chosen": -2.453125,
"rewards/margins": 7.03125,
"rewards/rejected": -9.5,
"step": 1150
},
{
"epoch": 1.2146596858638743,
"grad_norm": 0.29037304424736843,
"learning_rate": 3.7137704336959464e-05,
"logits/chosen": -12.1875,
"logits/rejected": -10.5,
"logps/chosen": -700.0,
"logps/rejected": -3088.0,
"loss": 0.0138,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -2.34375,
"rewards/margins": 7.0625,
"rewards/rejected": -9.375,
"step": 1160
},
{
"epoch": 1.225130890052356,
"grad_norm": 0.08425843067421462,
"learning_rate": 3.687047133397201e-05,
"logits/chosen": -10.375,
"logits/rejected": -9.6875,
"logps/chosen": -684.0,
"logps/rejected": -3104.0,
"loss": 0.0215,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -2.578125,
"rewards/margins": 7.625,
"rewards/rejected": -10.1875,
"step": 1170
},
{
"epoch": 1.2356020942408377,
"grad_norm": 0.12197324846540991,
"learning_rate": 3.6601475557503415e-05,
"logits/chosen": -10.875,
"logits/rejected": -9.4375,
"logps/chosen": -704.0,
"logps/rejected": -3328.0,
"loss": 0.0273,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -2.46875,
"rewards/margins": 8.3125,
"rewards/rejected": -10.75,
"step": 1180
},
{
"epoch": 1.2460732984293195,
"grad_norm": 0.16577602672607503,
"learning_rate": 3.633075695361882e-05,
"logits/chosen": -11.625,
"logits/rejected": -9.5625,
"logps/chosen": -664.0,
"logps/rejected": -3056.0,
"loss": 0.0288,
"rewards/accuracies": 0.981249988079071,
"rewards/chosen": -2.171875,
"rewards/margins": 7.34375,
"rewards/rejected": -9.5,
"step": 1190
},
{
"epoch": 1.256544502617801,
"grad_norm": 0.05987519009810323,
"learning_rate": 3.6058355724224477e-05,
"logits/chosen": -11.6875,
"logits/rejected": -9.625,
"logps/chosen": -632.0,
"logps/rejected": -3232.0,
"loss": 0.0204,
"rewards/accuracies": 1.0,
"rewards/chosen": -2.234375,
"rewards/margins": 7.53125,
"rewards/rejected": -9.75,
"step": 1200
},
{
"epoch": 1.256544502617801,
"eval_logits/chosen": -11.9375,
"eval_logits/rejected": -11.125,
"eval_logps/chosen": -552.0,
"eval_logps/rejected": -536.0,
"eval_loss": 0.8518593907356262,
"eval_rewards/accuracies": 0.53125,
"eval_rewards/chosen": -2.328125,
"eval_rewards/margins": 0.1376953125,
"eval_rewards/rejected": -2.46875,
"eval_runtime": 46.6848,
"eval_samples_per_second": 42.84,
"eval_steps_per_second": 0.685,
"step": 1200
},
{
"epoch": 1.2670157068062826,
"grad_norm": 0.1627098148506589,
"learning_rate": 3.5784312321097744e-05,
"logits/chosen": -12.875,
"logits/rejected": -10.0,
"logps/chosen": -752.0,
"logps/rejected": -3296.0,
"loss": 0.0235,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -2.625,
"rewards/margins": 7.875,
"rewards/rejected": -10.5,
"step": 1210
},
{
"epoch": 1.2774869109947644,
"grad_norm": 0.13464006590081817,
"learning_rate": 3.550866743987994e-05,
"logits/chosen": -11.9375,
"logits/rejected": -9.875,
"logps/chosen": -780.0,
"logps/rejected": -3232.0,
"loss": 0.0166,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.03125,
"rewards/margins": 8.625,
"rewards/rejected": -11.6875,
"step": 1220
},
{
"epoch": 1.2879581151832462,
"grad_norm": 0.9084253071757399,
"learning_rate": 3.5231462014033024e-05,
"logits/chosen": -11.5,
"logits/rejected": -9.625,
"logps/chosen": -804.0,
"logps/rejected": -3072.0,
"loss": 0.0525,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -3.0625,
"rewards/margins": 7.59375,
"rewards/rejected": -10.625,
"step": 1230
},
{
"epoch": 1.2984293193717278,
"grad_norm": 0.2784576348074477,
"learning_rate": 3.495273720876095e-05,
"logits/chosen": -10.625,
"logits/rejected": -8.4375,
"logps/chosen": -708.0,
"logps/rejected": -3152.0,
"loss": 0.0436,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -2.75,
"rewards/margins": 7.09375,
"rewards/rejected": -9.8125,
"step": 1240
},
{
"epoch": 1.3089005235602094,
"grad_norm": 1.2236219172229694,
"learning_rate": 3.4672534414896616e-05,
"logits/chosen": -12.1875,
"logits/rejected": -9.75,
"logps/chosen": -736.0,
"logps/rejected": -3072.0,
"loss": 0.021,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -2.546875,
"rewards/margins": 7.46875,
"rewards/rejected": -10.0,
"step": 1250
},
{
"epoch": 1.3193717277486912,
"grad_norm": 0.18858202438326535,
"learning_rate": 3.439089524275526e-05,
"logits/chosen": -13.5625,
"logits/rejected": -10.3125,
"logps/chosen": -688.0,
"logps/rejected": -3152.0,
"loss": 0.0237,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -2.390625,
"rewards/margins": 7.96875,
"rewards/rejected": -10.375,
"step": 1260
},
{
"epoch": 1.3298429319371727,
"grad_norm": 0.6065842713253152,
"learning_rate": 3.410786151595533e-05,
"logits/chosen": -13.375,
"logits/rejected": -9.625,
"logps/chosen": -724.0,
"logps/rejected": -3136.0,
"loss": 0.0308,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -2.53125,
"rewards/margins": 8.0625,
"rewards/rejected": -10.5625,
"step": 1270
},
{
"epoch": 1.3403141361256545,
"grad_norm": 0.45088832762971004,
"learning_rate": 3.382347526520765e-05,
"logits/chosen": -11.125,
"logits/rejected": -9.125,
"logps/chosen": -728.0,
"logps/rejected": -3280.0,
"loss": 0.0189,
"rewards/accuracies": 1.0,
"rewards/chosen": -2.71875,
"rewards/margins": 8.3125,
"rewards/rejected": -11.0,
"step": 1280
},
{
"epoch": 1.350785340314136,
"grad_norm": 0.8292060636688661,
"learning_rate": 3.353777872207381e-05,
"logits/chosen": -10.875,
"logits/rejected": -8.8125,
"logps/chosen": -804.0,
"logps/rejected": -3296.0,
"loss": 0.0262,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -3.015625,
"rewards/margins": 8.0625,
"rewards/rejected": -11.0625,
"step": 1290
},
{
"epoch": 1.3612565445026177,
"grad_norm": 0.31659313702185066,
"learning_rate": 3.325081431269475e-05,
"logits/chosen": -10.1875,
"logits/rejected": -8.6875,
"logps/chosen": -716.0,
"logps/rejected": -3328.0,
"loss": 0.0168,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -2.71875,
"rewards/margins": 8.5625,
"rewards/rejected": -11.25,
"step": 1300
},
{
"epoch": 1.3612565445026177,
"eval_logits/chosen": -9.8125,
"eval_logits/rejected": -8.9375,
"eval_logps/chosen": -608.0,
"eval_logps/rejected": -596.0,
"eval_loss": 0.8706718683242798,
"eval_rewards/accuracies": 0.509765625,
"eval_rewards/chosen": -2.890625,
"eval_rewards/margins": 0.1748046875,
"eval_rewards/rejected": -3.0625,
"eval_runtime": 46.6761,
"eval_samples_per_second": 42.849,
"eval_steps_per_second": 0.686,
"step": 1300
},
{
"epoch": 1.3717277486910995,
"grad_norm": 0.6076938704078351,
"learning_rate": 3.296262465149046e-05,
"logits/chosen": -10.5,
"logits/rejected": -8.5625,
"logps/chosen": -768.0,
"logps/rejected": -3232.0,
"loss": 0.0203,
"rewards/accuracies": 1.0,
"rewards/chosen": -2.703125,
"rewards/margins": 8.5625,
"rewards/rejected": -11.25,
"step": 1310
},
{
"epoch": 1.3821989528795813,
"grad_norm": 0.846082019549806,
"learning_rate": 3.2673252534831687e-05,
"logits/chosen": -11.5,
"logits/rejected": -8.8125,
"logps/chosen": -700.0,
"logps/rejected": -3200.0,
"loss": 0.028,
"rewards/accuracies": 1.0,
"rewards/chosen": -2.671875,
"rewards/margins": 7.875,
"rewards/rejected": -10.5625,
"step": 1320
},
{
"epoch": 1.3926701570680629,
"grad_norm": 0.16687099170949768,
"learning_rate": 3.2382740934684696e-05,
"logits/chosen": -12.25,
"logits/rejected": -9.6875,
"logps/chosen": -684.0,
"logps/rejected": -3216.0,
"loss": 0.0225,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -2.671875,
"rewards/margins": 7.96875,
"rewards/rejected": -10.625,
"step": 1330
},
{
"epoch": 1.4031413612565444,
"grad_norm": 0.8832943677406042,
"learning_rate": 3.209113299222982e-05,
"logits/chosen": -11.9375,
"logits/rejected": -9.4375,
"logps/chosen": -764.0,
"logps/rejected": -3184.0,
"loss": 0.031,
"rewards/accuracies": 0.981249988079071,
"rewards/chosen": -2.90625,
"rewards/margins": 8.125,
"rewards/rejected": -11.0625,
"step": 1340
},
{
"epoch": 1.4136125654450262,
"grad_norm": 0.7417873572970306,
"learning_rate": 3.179847201145505e-05,
"logits/chosen": -11.9375,
"logits/rejected": -9.0,
"logps/chosen": -728.0,
"logps/rejected": -3200.0,
"loss": 0.0392,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -2.71875,
"rewards/margins": 7.6875,
"rewards/rejected": -10.375,
"step": 1350
},
{
"epoch": 1.4240837696335078,
"grad_norm": 0.055517908366391734,
"learning_rate": 3.150480145272528e-05,
"logits/chosen": -10.875,
"logits/rejected": -9.3125,
"logps/chosen": -772.0,
"logps/rejected": -3248.0,
"loss": 0.0178,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -2.71875,
"rewards/margins": 7.0,
"rewards/rejected": -9.75,
"step": 1360
},
{
"epoch": 1.4345549738219896,
"grad_norm": 0.20553880448041817,
"learning_rate": 3.121016492632848e-05,
"logits/chosen": -11.25,
"logits/rejected": -9.1875,
"logps/chosen": -724.0,
"logps/rejected": -3152.0,
"loss": 0.0154,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -2.390625,
"rewards/margins": 7.8125,
"rewards/rejected": -10.1875,
"step": 1370
},
{
"epoch": 1.4450261780104712,
"grad_norm": 0.12069218938775955,
"learning_rate": 3.091460618599951e-05,
"logits/chosen": -11.6875,
"logits/rejected": -8.875,
"logps/chosen": -696.0,
"logps/rejected": -3360.0,
"loss": 0.0183,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -2.609375,
"rewards/margins": 8.625,
"rewards/rejected": -11.25,
"step": 1380
},
{
"epoch": 1.4554973821989527,
"grad_norm": 0.4217138174114351,
"learning_rate": 3.061816912242265e-05,
"logits/chosen": -11.4375,
"logits/rejected": -9.5,
"logps/chosen": -712.0,
"logps/rejected": -3456.0,
"loss": 0.0225,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -2.53125,
"rewards/margins": 9.5,
"rewards/rejected": -12.0,
"step": 1390
},
{
"epoch": 1.4659685863874345,
"grad_norm": 1.428839471308891,
"learning_rate": 3.0320897756713785e-05,
"logits/chosen": -12.4375,
"logits/rejected": -10.1875,
"logps/chosen": -684.0,
"logps/rejected": -3328.0,
"loss": 0.0183,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -2.578125,
"rewards/margins": 9.0,
"rewards/rejected": -11.5625,
"step": 1400
},
{
"epoch": 1.4659685863874345,
"eval_logits/chosen": -12.125,
"eval_logits/rejected": -11.5625,
"eval_logps/chosen": -580.0,
"eval_logps/rejected": -564.0,
"eval_loss": 0.9054843783378601,
"eval_rewards/accuracies": 0.5,
"eval_rewards/chosen": -2.625,
"eval_rewards/margins": 0.1171875,
"eval_rewards/rejected": -2.734375,
"eval_runtime": 46.6867,
"eval_samples_per_second": 42.839,
"eval_steps_per_second": 0.685,
"step": 1400
},
{
"epoch": 1.4764397905759163,
"grad_norm": 0.5124237302194357,
"learning_rate": 3.002283623388332e-05,
"logits/chosen": -12.1875,
"logits/rejected": -9.375,
"logps/chosen": -712.0,
"logps/rejected": -3264.0,
"loss": 0.0209,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -2.515625,
"rewards/margins": 8.625,
"rewards/rejected": -11.125,
"step": 1410
},
{
"epoch": 1.486910994764398,
"grad_norm": 0.08316278936626349,
"learning_rate": 2.9724028816280507e-05,
"logits/chosen": -11.4375,
"logits/rejected": -9.3125,
"logps/chosen": -728.0,
"logps/rejected": -3376.0,
"loss": 0.0232,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -2.8125,
"rewards/margins": 9.0,
"rewards/rejected": -11.8125,
"step": 1420
},
{
"epoch": 1.4973821989528795,
"grad_norm": 0.5782257689219309,
"learning_rate": 2.9424519877020524e-05,
"logits/chosen": -12.125,
"logits/rejected": -9.4375,
"logps/chosen": -680.0,
"logps/rejected": -3168.0,
"loss": 0.033,
"rewards/accuracies": 0.981249988079071,
"rewards/chosen": -2.703125,
"rewards/margins": 8.1875,
"rewards/rejected": -10.875,
"step": 1430
},
{
"epoch": 1.5078534031413613,
"grad_norm": 1.4321153446872121,
"learning_rate": 2.912435389339504e-05,
"logits/chosen": -11.6875,
"logits/rejected": -9.5,
"logps/chosen": -656.0,
"logps/rejected": -3280.0,
"loss": 0.0278,
"rewards/accuracies": 1.0,
"rewards/chosen": -2.75,
"rewards/margins": 8.1875,
"rewards/rejected": -10.9375,
"step": 1440
},
{
"epoch": 1.518324607329843,
"grad_norm": 0.15245151905720722,
"learning_rate": 2.8823575440267226e-05,
"logits/chosen": -12.0625,
"logits/rejected": -9.375,
"logps/chosen": -704.0,
"logps/rejected": -3216.0,
"loss": 0.0148,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -2.5625,
"rewards/margins": 8.0625,
"rewards/rejected": -10.625,
"step": 1450
},
{
"epoch": 1.5287958115183247,
"grad_norm": 0.14986527427133123,
"learning_rate": 2.8522229183452465e-05,
"logits/chosen": -12.1875,
"logits/rejected": -9.1875,
"logps/chosen": -684.0,
"logps/rejected": -3456.0,
"loss": 0.0175,
"rewards/accuracies": 1.0,
"rewards/chosen": -2.4375,
"rewards/margins": 9.125,
"rewards/rejected": -11.5,
"step": 1460
},
{
"epoch": 1.5392670157068062,
"grad_norm": 0.7180708388013607,
"learning_rate": 2.8220359873085373e-05,
"logits/chosen": -12.0625,
"logits/rejected": -9.5625,
"logps/chosen": -760.0,
"logps/rejected": -3280.0,
"loss": 0.0367,
"rewards/accuracies": 0.981249988079071,
"rewards/chosen": -2.84375,
"rewards/margins": 8.625,
"rewards/rejected": -11.5,
"step": 1470
},
{
"epoch": 1.5497382198952878,
"grad_norm": 0.3617305537324189,
"learning_rate": 2.7918012336974388e-05,
"logits/chosen": -11.8125,
"logits/rejected": -9.3125,
"logps/chosen": -808.0,
"logps/rejected": -3280.0,
"loss": 0.0291,
"rewards/accuracies": 1.0,
"rewards/chosen": -2.671875,
"rewards/margins": 7.75,
"rewards/rejected": -10.4375,
"step": 1480
},
{
"epoch": 1.5602094240837696,
"grad_norm": 0.5038483189400726,
"learning_rate": 2.7615231473944815e-05,
"logits/chosen": -12.125,
"logits/rejected": -9.375,
"logps/chosen": -684.0,
"logps/rejected": -3248.0,
"loss": 0.0279,
"rewards/accuracies": 1.0,
"rewards/chosen": -2.46875,
"rewards/margins": 7.625,
"rewards/rejected": -10.0625,
"step": 1490
},
{
"epoch": 1.5706806282722514,
"grad_norm": 1.3513590249662928,
"learning_rate": 2.7312062247171327e-05,
"logits/chosen": -12.4375,
"logits/rejected": -9.3125,
"logps/chosen": -724.0,
"logps/rejected": -3296.0,
"loss": 0.0258,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -2.65625,
"rewards/margins": 7.3125,
"rewards/rejected": -9.9375,
"step": 1500
},
{
"epoch": 1.5706806282722514,
"eval_logits/chosen": -12.625,
"eval_logits/rejected": -12.1875,
"eval_logps/chosen": -556.0,
"eval_logps/rejected": -536.0,
"eval_loss": 0.879687488079071,
"eval_rewards/accuracies": 0.509765625,
"eval_rewards/chosen": -2.390625,
"eval_rewards/margins": 0.06298828125,
"eval_rewards/rejected": -2.453125,
"eval_runtime": 46.674,
"eval_samples_per_second": 42.85,
"eval_steps_per_second": 0.686,
"step": 1500
},
{
"epoch": 1.581151832460733,
"grad_norm": 0.3041968940714293,
"learning_rate": 2.7008549677500876e-05,
"logits/chosen": -12.9375,
"logits/rejected": -9.4375,
"logps/chosen": -740.0,
"logps/rejected": -3248.0,
"loss": 0.0152,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -2.65625,
"rewards/margins": 7.375,
"rewards/rejected": -10.0625,
"step": 1510
},
{
"epoch": 1.5916230366492146,
"grad_norm": 0.5469357880233887,
"learning_rate": 2.670473883676709e-05,
"logits/chosen": -12.6875,
"logits/rejected": -9.6875,
"logps/chosen": -692.0,
"logps/rejected": -3232.0,
"loss": 0.0274,
"rewards/accuracies": 1.0,
"rewards/chosen": -2.5625,
"rewards/margins": 8.0625,
"rewards/rejected": -10.625,
"step": 1520
},
{
"epoch": 1.6020942408376964,
"grad_norm": 0.2653801856297482,
"learning_rate": 2.6400674841097074e-05,
"logits/chosen": -12.25,
"logits/rejected": -9.3125,
"logps/chosen": -724.0,
"logps/rejected": -3296.0,
"loss": 0.0227,
"rewards/accuracies": 0.981249988079071,
"rewards/chosen": -2.65625,
"rewards/margins": 8.3125,
"rewards/rejected": -11.0,
"step": 1530
},
{
"epoch": 1.6125654450261782,
"grad_norm": 0.18280914580132127,
"learning_rate": 2.609640284421158e-05,
"logits/chosen": -11.6875,
"logits/rejected": -8.8125,
"logps/chosen": -740.0,
"logps/rejected": -3280.0,
"loss": 0.018,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -3.03125,
"rewards/margins": 7.96875,
"rewards/rejected": -11.0,
"step": 1540
},
{
"epoch": 1.6230366492146597,
"grad_norm": 0.3818923768163967,
"learning_rate": 2.5791968030719726e-05,
"logits/chosen": -11.125,
"logits/rejected": -8.8125,
"logps/chosen": -760.0,
"logps/rejected": -3232.0,
"loss": 0.0251,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -3.09375,
"rewards/margins": 8.375,
"rewards/rejected": -11.5,
"step": 1550
},
{
"epoch": 1.6335078534031413,
"grad_norm": 0.16991647640622323,
"learning_rate": 2.548741560940902e-05,
"logits/chosen": -11.4375,
"logits/rejected": -9.3125,
"logps/chosen": -844.0,
"logps/rejected": -3248.0,
"loss": 0.0277,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -3.109375,
"rewards/margins": 8.3125,
"rewards/rejected": -11.4375,
"step": 1560
},
{
"epoch": 1.6439790575916229,
"grad_norm": 1.1702222667758566,
"learning_rate": 2.5182790806531787e-05,
"logits/chosen": -11.75,
"logits/rejected": -8.5,
"logps/chosen": -768.0,
"logps/rejected": -3360.0,
"loss": 0.0129,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.1875,
"rewards/margins": 8.25,
"rewards/rejected": -11.4375,
"step": 1570
},
{
"epoch": 1.6544502617801047,
"grad_norm": 0.4242136823142166,
"learning_rate": 2.4878138859089074e-05,
"logits/chosen": -11.25,
"logits/rejected": -8.5625,
"logps/chosen": -732.0,
"logps/rejected": -3280.0,
"loss": 0.0381,
"rewards/accuracies": 0.981249988079071,
"rewards/chosen": -3.0625,
"rewards/margins": 8.375,
"rewards/rejected": -11.4375,
"step": 1580
},
{
"epoch": 1.6649214659685865,
"grad_norm": 0.3430643188529773,
"learning_rate": 2.4573505008112924e-05,
"logits/chosen": -10.875,
"logits/rejected": -8.0,
"logps/chosen": -784.0,
"logps/rejected": -3488.0,
"loss": 0.0229,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -3.25,
"rewards/margins": 9.0625,
"rewards/rejected": -12.3125,
"step": 1590
},
{
"epoch": 1.675392670157068,
"grad_norm": 0.4254208893779586,
"learning_rate": 2.4268934491948032e-05,
"logits/chosen": -10.5,
"logits/rejected": -8.125,
"logps/chosen": -796.0,
"logps/rejected": -3152.0,
"loss": 0.0168,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -3.09375,
"rewards/margins": 8.3125,
"rewards/rejected": -11.375,
"step": 1600
},
{
"epoch": 1.675392670157068,
"eval_logits/chosen": -11.0,
"eval_logits/rejected": -10.4375,
"eval_logps/chosen": -616.0,
"eval_logps/rejected": -600.0,
"eval_loss": 0.9114062786102295,
"eval_rewards/accuracies": 0.505859375,
"eval_rewards/chosen": -2.984375,
"eval_rewards/margins": 0.1337890625,
"eval_rewards/rejected": -3.125,
"eval_runtime": 46.6669,
"eval_samples_per_second": 42.857,
"eval_steps_per_second": 0.686,
"step": 1600
},
{
"epoch": 1.6858638743455496,
"grad_norm": 0.20711363961601834,
"learning_rate": 2.396447253953385e-05,
"logits/chosen": -11.5625,
"logits/rejected": -8.9375,
"logps/chosen": -736.0,
"logps/rejected": -3296.0,
"loss": 0.0117,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -3.015625,
"rewards/margins": 8.25,
"rewards/rejected": -11.25,
"step": 1610
},
{
"epoch": 1.6963350785340314,
"grad_norm": 0.06324536553229601,
"learning_rate": 2.3660164363687998e-05,
"logits/chosen": -11.9375,
"logits/rejected": -8.8125,
"logps/chosen": -828.0,
"logps/rejected": -3216.0,
"loss": 0.0743,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.109375,
"rewards/margins": 8.375,
"rewards/rejected": -11.5,
"step": 1620
},
{
"epoch": 1.7068062827225132,
"grad_norm": 1.1135053775770383,
"learning_rate": 2.33560551543922e-05,
"logits/chosen": -11.8125,
"logits/rejected": -9.375,
"logps/chosen": -740.0,
"logps/rejected": -3360.0,
"loss": 0.1195,
"rewards/accuracies": 0.981249988079071,
"rewards/chosen": -3.390625,
"rewards/margins": 8.25,
"rewards/rejected": -11.6875,
"step": 1630
},
{
"epoch": 1.7172774869109948,
"grad_norm": 1.750893471166311,
"learning_rate": 2.3052190072081493e-05,
"logits/chosen": -12.0,
"logits/rejected": -9.0,
"logps/chosen": -792.0,
"logps/rejected": -3264.0,
"loss": 0.0348,
"rewards/accuracies": 0.981249988079071,
"rewards/chosen": -3.109375,
"rewards/margins": 7.84375,
"rewards/rejected": -10.9375,
"step": 1640
},
{
"epoch": 1.7277486910994764,
"grad_norm": 1.4041590017761676,
"learning_rate": 2.2748614240937864e-05,
"logits/chosen": -11.75,
"logits/rejected": -9.4375,
"logps/chosen": -768.0,
"logps/rejected": -3104.0,
"loss": 0.0297,
"rewards/accuracies": 1.0,
"rewards/chosen": -2.890625,
"rewards/margins": 7.625,
"rewards/rejected": -10.5,
"step": 1650
},
{
"epoch": 1.738219895287958,
"grad_norm": 0.3409463719358081,
"learning_rate": 2.2445372742189334e-05,
"logits/chosen": -10.0625,
"logits/rejected": -8.375,
"logps/chosen": -744.0,
"logps/rejected": -3232.0,
"loss": 0.0204,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -2.890625,
"rewards/margins": 8.0625,
"rewards/rejected": -10.9375,
"step": 1660
},
{
"epoch": 1.7486910994764397,
"grad_norm": 0.22584350677842213,
"learning_rate": 2.214251060741528e-05,
"logits/chosen": -9.0,
"logits/rejected": -8.1875,
"logps/chosen": -748.0,
"logps/rejected": -3328.0,
"loss": 0.0505,
"rewards/accuracies": 0.981249988079071,
"rewards/chosen": -3.09375,
"rewards/margins": 7.9375,
"rewards/rejected": -11.0,
"step": 1670
},
{
"epoch": 1.7591623036649215,
"grad_norm": 0.3166308674994499,
"learning_rate": 2.1840072811859302e-05,
"logits/chosen": -9.5,
"logits/rejected": -8.1875,
"logps/chosen": -788.0,
"logps/rejected": -3088.0,
"loss": 0.0245,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -2.90625,
"rewards/margins": 7.0,
"rewards/rejected": -9.9375,
"step": 1680
},
{
"epoch": 1.7696335078534031,
"grad_norm": 0.09110494775536958,
"learning_rate": 2.1538104267750287e-05,
"logits/chosen": -9.25,
"logits/rejected": -8.125,
"logps/chosen": -684.0,
"logps/rejected": -3376.0,
"loss": 0.0104,
"rewards/accuracies": 1.0,
"rewards/chosen": -2.765625,
"rewards/margins": 8.0625,
"rewards/rejected": -10.875,
"step": 1690
},
{
"epoch": 1.7801047120418847,
"grad_norm": 0.2632723364629577,
"learning_rate": 2.1236649817632952e-05,
"logits/chosen": -9.8125,
"logits/rejected": -8.4375,
"logps/chosen": -712.0,
"logps/rejected": -3328.0,
"loss": 0.0313,
"rewards/accuracies": 1.0,
"rewards/chosen": -2.875,
"rewards/margins": 8.375,
"rewards/rejected": -11.25,
"step": 1700
},
{
"epoch": 1.7801047120418847,
"eval_logits/chosen": -9.25,
"eval_logits/rejected": -8.5625,
"eval_logps/chosen": -584.0,
"eval_logps/rejected": -564.0,
"eval_loss": 0.9135624766349792,
"eval_rewards/accuracies": 0.501953125,
"eval_rewards/chosen": -2.65625,
"eval_rewards/margins": 0.078125,
"eval_rewards/rejected": -2.734375,
"eval_runtime": 46.6787,
"eval_samples_per_second": 42.846,
"eval_steps_per_second": 0.686,
"step": 1700
},
{
"epoch": 1.7905759162303665,
"grad_norm": 0.27009047046079143,
"learning_rate": 2.093575422770872e-05,
"logits/chosen": -9.9375,
"logits/rejected": -8.5,
"logps/chosen": -772.0,
"logps/rejected": -3200.0,
"loss": 0.0203,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -2.828125,
"rewards/margins": 7.9375,
"rewards/rejected": -10.75,
"step": 1710
},
{
"epoch": 1.8010471204188483,
"grad_norm": 0.5098621727638959,
"learning_rate": 2.0635462181187827e-05,
"logits/chosen": -9.625,
"logits/rejected": -8.5,
"logps/chosen": -744.0,
"logps/rejected": -3376.0,
"loss": 0.0143,
"rewards/accuracies": 1.0,
"rewards/chosen": -2.765625,
"rewards/margins": 8.5625,
"rewards/rejected": -11.375,
"step": 1720
},
{
"epoch": 1.8115183246073299,
"grad_norm": 0.25924435447205474,
"learning_rate": 2.0335818271653917e-05,
"logits/chosen": -9.5,
"logits/rejected": -8.5,
"logps/chosen": -664.0,
"logps/rejected": -3344.0,
"loss": 0.03,
"rewards/accuracies": 1.0,
"rewards/chosen": -2.640625,
"rewards/margins": 9.0,
"rewards/rejected": -11.625,
"step": 1730
},
{
"epoch": 1.8219895287958114,
"grad_norm": 0.3792615916407521,
"learning_rate": 2.0036866996441815e-05,
"logits/chosen": -9.0,
"logits/rejected": -8.4375,
"logps/chosen": -720.0,
"logps/rejected": -3280.0,
"loss": 0.0218,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -2.765625,
"rewards/margins": 8.3125,
"rewards/rejected": -11.125,
"step": 1740
},
{
"epoch": 1.8324607329842932,
"grad_norm": 0.2020890106171226,
"learning_rate": 1.9738652750029677e-05,
"logits/chosen": -9.8125,
"logits/rejected": -8.5,
"logps/chosen": -780.0,
"logps/rejected": -3168.0,
"loss": 0.0177,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -3.0,
"rewards/margins": 7.78125,
"rewards/rejected": -10.75,
"step": 1750
},
{
"epoch": 1.8429319371727748,
"grad_norm": 0.10541884603523646,
"learning_rate": 1.944121981744631e-05,
"logits/chosen": -9.75,
"logits/rejected": -8.8125,
"logps/chosen": -728.0,
"logps/rejected": -3264.0,
"loss": 0.0205,
"rewards/accuracies": 1.0,
"rewards/chosen": -2.859375,
"rewards/margins": 8.8125,
"rewards/rejected": -11.6875,
"step": 1760
},
{
"epoch": 1.8534031413612566,
"grad_norm": 0.2308210879343531,
"learning_rate": 1.9144612367694887e-05,
"logits/chosen": -10.4375,
"logits/rejected": -9.0625,
"logps/chosen": -736.0,
"logps/rejected": -3136.0,
"loss": 0.0237,
"rewards/accuracies": 0.981249988079071,
"rewards/chosen": -2.9375,
"rewards/margins": 7.96875,
"rewards/rejected": -10.9375,
"step": 1770
},
{
"epoch": 1.8638743455497382,
"grad_norm": 0.8653260872030909,
"learning_rate": 1.8848874447193803e-05,
"logits/chosen": -11.25,
"logits/rejected": -8.875,
"logps/chosen": -744.0,
"logps/rejected": -3280.0,
"loss": 0.0114,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.0,
"rewards/margins": 8.625,
"rewards/rejected": -11.625,
"step": 1780
},
{
"epoch": 1.8743455497382198,
"grad_norm": 0.7519984693440278,
"learning_rate": 1.855404997323571e-05,
"logits/chosen": -11.5625,
"logits/rejected": -9.4375,
"logps/chosen": -784.0,
"logps/rejected": -3312.0,
"loss": 0.0188,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -3.265625,
"rewards/margins": 8.875,
"rewards/rejected": -12.125,
"step": 1790
},
{
"epoch": 1.8848167539267016,
"grad_norm": 0.5830137533632717,
"learning_rate": 1.82601827274658e-05,
"logits/chosen": -10.3125,
"logits/rejected": -8.6875,
"logps/chosen": -756.0,
"logps/rejected": -3456.0,
"loss": 0.0207,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -3.234375,
"rewards/margins": 8.1875,
"rewards/rejected": -11.4375,
"step": 1800
},
{
"epoch": 1.8848167539267016,
"eval_logits/chosen": -9.5,
"eval_logits/rejected": -8.6875,
"eval_logps/chosen": -628.0,
"eval_logps/rejected": -612.0,
"eval_loss": 0.9313593506813049,
"eval_rewards/accuracies": 0.505859375,
"eval_rewards/chosen": -3.078125,
"eval_rewards/margins": 0.12890625,
"eval_rewards/rejected": -3.21875,
"eval_runtime": 46.6579,
"eval_samples_per_second": 42.865,
"eval_steps_per_second": 0.686,
"step": 1800
},
{
"epoch": 1.8952879581151834,
"grad_norm": 0.2745819183478167,
"learning_rate": 1.7967316349380213e-05,
"logits/chosen": -10.5625,
"logits/rejected": -8.625,
"logps/chosen": -784.0,
"logps/rejected": -3168.0,
"loss": 0.023,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -3.203125,
"rewards/margins": 7.75,
"rewards/rejected": -10.9375,
"step": 1810
},
{
"epoch": 1.905759162303665,
"grad_norm": 0.14176972322114517,
"learning_rate": 1.7675494329845516e-05,
"logits/chosen": -9.6875,
"logits/rejected": -8.3125,
"logps/chosen": -752.0,
"logps/rejected": -3408.0,
"loss": 0.0317,
"rewards/accuracies": 0.981249988079071,
"rewards/chosen": -3.328125,
"rewards/margins": 8.625,
"rewards/rejected": -11.9375,
"step": 1820
},
{
"epoch": 1.9162303664921465,
"grad_norm": 0.10386858276948377,
"learning_rate": 1.738476000464026e-05,
"logits/chosen": -8.9375,
"logits/rejected": -8.375,
"logps/chosen": -768.0,
"logps/rejected": -3392.0,
"loss": 0.0459,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -3.296875,
"rewards/margins": 8.4375,
"rewards/rejected": -11.75,
"step": 1830
},
{
"epoch": 1.9267015706806283,
"grad_norm": 0.7925230265817749,
"learning_rate": 1.7095156548019646e-05,
"logits/chosen": -9.75,
"logits/rejected": -8.875,
"logps/chosen": -788.0,
"logps/rejected": -3328.0,
"loss": 0.0298,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.109375,
"rewards/margins": 8.375,
"rewards/rejected": -11.5,
"step": 1840
},
{
"epoch": 1.93717277486911,
"grad_norm": 0.5948701627622711,
"learning_rate": 1.6806726966304064e-05,
"logits/chosen": -9.125,
"logits/rejected": -8.5625,
"logps/chosen": -732.0,
"logps/rejected": -3232.0,
"loss": 0.0455,
"rewards/accuracies": 1.0,
"rewards/chosen": -2.90625,
"rewards/margins": 7.875,
"rewards/rejected": -10.75,
"step": 1850
},
{
"epoch": 1.9476439790575917,
"grad_norm": 1.2260863023470558,
"learning_rate": 1.6519514091492625e-05,
"logits/chosen": -8.125,
"logits/rejected": -8.5,
"logps/chosen": -756.0,
"logps/rejected": -3360.0,
"loss": 0.0237,
"rewards/accuracies": 1.0,
"rewards/chosen": -2.8125,
"rewards/margins": 7.71875,
"rewards/rejected": -10.5625,
"step": 1860
},
{
"epoch": 1.9581151832460733,
"grad_norm": 0.5710670914911604,
"learning_rate": 1.6233560574902627e-05,
"logits/chosen": -9.3125,
"logits/rejected": -8.0625,
"logps/chosen": -780.0,
"logps/rejected": -3056.0,
"loss": 0.0221,
"rewards/accuracies": 1.0,
"rewards/chosen": -2.90625,
"rewards/margins": 7.75,
"rewards/rejected": -10.6875,
"step": 1870
},
{
"epoch": 1.9685863874345548,
"grad_norm": 0.702991920422923,
"learning_rate": 1.5948908880835753e-05,
"logits/chosen": -8.6875,
"logits/rejected": -8.125,
"logps/chosen": -736.0,
"logps/rejected": -3440.0,
"loss": 0.0359,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -3.140625,
"rewards/margins": 8.5,
"rewards/rejected": -11.625,
"step": 1880
},
{
"epoch": 1.9790575916230366,
"grad_norm": 0.06556838387047441,
"learning_rate": 1.5665601280272124e-05,
"logits/chosen": -8.0625,
"logits/rejected": -7.78125,
"logps/chosen": -752.0,
"logps/rejected": -3600.0,
"loss": 0.011,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -3.234375,
"rewards/margins": 9.1875,
"rewards/rejected": -12.375,
"step": 1890
},
{
"epoch": 1.9895287958115184,
"grad_norm": 0.06552112666165379,
"learning_rate": 1.538367984459298e-05,
"logits/chosen": -7.5625,
"logits/rejected": -6.875,
"logps/chosen": -788.0,
"logps/rejected": -3152.0,
"loss": 0.0155,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -3.421875,
"rewards/margins": 8.625,
"rewards/rejected": -12.0,
"step": 1900
},
{
"epoch": 1.9895287958115184,
"eval_logits/chosen": -6.65625,
"eval_logits/rejected": -5.75,
"eval_logps/chosen": -640.0,
"eval_logps/rejected": -628.0,
"eval_loss": 0.9222187399864197,
"eval_rewards/accuracies": 0.505859375,
"eval_rewards/chosen": -3.234375,
"eval_rewards/margins": 0.1416015625,
"eval_rewards/rejected": -3.375,
"eval_runtime": 46.6828,
"eval_samples_per_second": 42.842,
"eval_steps_per_second": 0.685,
"step": 1900
},
{
"epoch": 2.0,
"grad_norm": 0.3268371320207235,
"learning_rate": 1.5103186439333133e-05,
"logits/chosen": -7.46875,
"logits/rejected": -6.6875,
"logps/chosen": -724.0,
"logps/rejected": -3488.0,
"loss": 0.0119,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.03125,
"rewards/margins": 10.125,
"rewards/rejected": -13.125,
"step": 1910
},
{
"epoch": 2.0104712041884816,
"grad_norm": 0.02182437835886761,
"learning_rate": 1.4824162717963829e-05,
"logits/chosen": -8.6875,
"logits/rejected": -7.28125,
"logps/chosen": -820.0,
"logps/rejected": -3344.0,
"loss": 0.0028,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.265625,
"rewards/margins": 9.75,
"rewards/rejected": -13.0,
"step": 1920
},
{
"epoch": 2.020942408376963,
"grad_norm": 0.023306464376482382,
"learning_rate": 1.4546650115707192e-05,
"logits/chosen": -8.125,
"logits/rejected": -8.0,
"logps/chosen": -788.0,
"logps/rejected": -3536.0,
"loss": 0.0022,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.140625,
"rewards/margins": 10.8125,
"rewards/rejected": -13.9375,
"step": 1930
},
{
"epoch": 2.031413612565445,
"grad_norm": 0.031002994861580514,
"learning_rate": 1.4270689843383111e-05,
"logits/chosen": -8.625,
"logits/rejected": -8.1875,
"logps/chosen": -824.0,
"logps/rejected": -3648.0,
"loss": 0.0026,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.328125,
"rewards/margins": 10.9375,
"rewards/rejected": -14.25,
"step": 1940
},
{
"epoch": 2.0418848167539267,
"grad_norm": 0.021937600032229335,
"learning_rate": 1.3996322881289347e-05,
"logits/chosen": -9.0625,
"logits/rejected": -7.875,
"logps/chosen": -804.0,
"logps/rejected": -3504.0,
"loss": 0.001,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.390625,
"rewards/margins": 10.625,
"rewards/rejected": -14.0,
"step": 1950
},
{
"epoch": 2.0523560209424083,
"grad_norm": 0.05411474884908041,
"learning_rate": 1.3723589973115963e-05,
"logits/chosen": -7.6875,
"logits/rejected": -7.8125,
"logps/chosen": -792.0,
"logps/rejected": -3632.0,
"loss": 0.0011,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.515625,
"rewards/margins": 11.125,
"rewards/rejected": -14.625,
"step": 1960
},
{
"epoch": 2.06282722513089,
"grad_norm": 0.043554948629682254,
"learning_rate": 1.3452531619894837e-05,
"logits/chosen": -8.0625,
"logits/rejected": -7.90625,
"logps/chosen": -808.0,
"logps/rejected": -3552.0,
"loss": 0.0013,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.53125,
"rewards/margins": 10.9375,
"rewards/rejected": -14.5,
"step": 1970
},
{
"epoch": 2.073298429319372,
"grad_norm": 0.09897469175915552,
"learning_rate": 1.3183188073985245e-05,
"logits/chosen": -8.0625,
"logits/rejected": -7.375,
"logps/chosen": -852.0,
"logps/rejected": -3616.0,
"loss": 0.0007,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.21875,
"rewards/margins": 11.5625,
"rewards/rejected": -14.8125,
"step": 1980
},
{
"epoch": 2.0837696335078535,
"grad_norm": 0.039372365254048386,
"learning_rate": 1.2915599333096351e-05,
"logits/chosen": -9.125,
"logits/rejected": -7.5,
"logps/chosen": -772.0,
"logps/rejected": -3488.0,
"loss": 0.001,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.390625,
"rewards/margins": 10.9375,
"rewards/rejected": -14.375,
"step": 1990
},
{
"epoch": 2.094240837696335,
"grad_norm": 0.009130169550997198,
"learning_rate": 1.264980513434752e-05,
"logits/chosen": -8.6875,
"logits/rejected": -7.8125,
"logps/chosen": -772.0,
"logps/rejected": -3648.0,
"loss": 0.0013,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.53125,
"rewards/margins": 11.5,
"rewards/rejected": -15.0625,
"step": 2000
},
{
"epoch": 2.094240837696335,
"eval_logits/chosen": -7.53125,
"eval_logits/rejected": -6.625,
"eval_logps/chosen": -668.0,
"eval_logps/rejected": -656.0,
"eval_loss": 0.9954218864440918,
"eval_rewards/accuracies": 0.501953125,
"eval_rewards/chosen": -3.484375,
"eval_rewards/margins": 0.1884765625,
"eval_rewards/rejected": -3.671875,
"eval_runtime": 46.6887,
"eval_samples_per_second": 42.837,
"eval_steps_per_second": 0.685,
"step": 2000
},
{
"epoch": 2.1047120418848166,
"grad_norm": 0.03298589424505654,
"learning_rate": 1.2385844948367324e-05,
"logits/chosen": -9.25,
"logits/rejected": -7.96875,
"logps/chosen": -780.0,
"logps/rejected": -3744.0,
"loss": 0.0018,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.390625,
"rewards/margins": 11.875,
"rewards/rejected": -15.25,
"step": 2010
},
{
"epoch": 2.115183246073298,
"grad_norm": 0.017880300499474658,
"learning_rate": 1.2123757973432114e-05,
"logits/chosen": -8.875,
"logits/rejected": -8.125,
"logps/chosen": -780.0,
"logps/rejected": -3552.0,
"loss": 0.0035,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.578125,
"rewards/margins": 11.3125,
"rewards/rejected": -14.9375,
"step": 2020
},
{
"epoch": 2.1256544502617802,
"grad_norm": 1.6005294212732664,
"learning_rate": 1.1863583129645106e-05,
"logits/chosen": -8.125,
"logits/rejected": -7.28125,
"logps/chosen": -880.0,
"logps/rejected": -3792.0,
"loss": 0.0089,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.890625,
"rewards/margins": 12.0,
"rewards/rejected": -15.875,
"step": 2030
},
{
"epoch": 2.136125654450262,
"grad_norm": 0.011910210453604731,
"learning_rate": 1.1605359053156607e-05,
"logits/chosen": -9.625,
"logits/rejected": -7.5625,
"logps/chosen": -772.0,
"logps/rejected": -3520.0,
"loss": 0.0012,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.390625,
"rewards/margins": 11.75,
"rewards/rejected": -15.125,
"step": 2040
},
{
"epoch": 2.1465968586387434,
"grad_norm": 0.04169291932656863,
"learning_rate": 1.1349124090426644e-05,
"logits/chosen": -8.875,
"logits/rejected": -7.65625,
"logps/chosen": -880.0,
"logps/rejected": -3872.0,
"loss": 0.002,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.640625,
"rewards/margins": 12.25,
"rewards/rejected": -15.875,
"step": 2050
},
{
"epoch": 2.157068062827225,
"grad_norm": 0.022259546669978078,
"learning_rate": 1.1094916292530406e-05,
"logits/chosen": -8.6875,
"logits/rejected": -7.5,
"logps/chosen": -772.0,
"logps/rejected": -3824.0,
"loss": 0.0013,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.40625,
"rewards/margins": 12.25,
"rewards/rejected": -15.625,
"step": 2060
},
{
"epoch": 2.167539267015707,
"grad_norm": 0.009122866823369599,
"learning_rate": 1.0842773409507623e-05,
"logits/chosen": -8.375,
"logits/rejected": -8.0,
"logps/chosen": -760.0,
"logps/rejected": -3632.0,
"loss": 0.001,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.234375,
"rewards/margins": 12.5625,
"rewards/rejected": -15.75,
"step": 2070
},
{
"epoch": 2.1780104712041886,
"grad_norm": 0.03562808106889683,
"learning_rate": 1.0592732884756754e-05,
"logits/chosen": -9.0,
"logits/rejected": -7.65625,
"logps/chosen": -800.0,
"logps/rejected": -3808.0,
"loss": 0.0021,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -3.578125,
"rewards/margins": 11.875,
"rewards/rejected": -15.4375,
"step": 2080
},
{
"epoch": 2.18848167539267,
"grad_norm": 0.04979002339490782,
"learning_rate": 1.0344831849474507e-05,
"logits/chosen": -8.375,
"logits/rejected": -7.5,
"logps/chosen": -804.0,
"logps/rejected": -3664.0,
"loss": 0.0009,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.640625,
"rewards/margins": 11.75,
"rewards/rejected": -15.375,
"step": 2090
},
{
"epoch": 2.1989528795811517,
"grad_norm": 0.06530394342391863,
"learning_rate": 1.0099107117141879e-05,
"logits/chosen": -8.9375,
"logits/rejected": -7.625,
"logps/chosen": -848.0,
"logps/rejected": -3648.0,
"loss": 0.0018,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.765625,
"rewards/margins": 11.875,
"rewards/rejected": -15.625,
"step": 2100
},
{
"epoch": 2.1989528795811517,
"eval_logits/chosen": -7.59375,
"eval_logits/rejected": -6.75,
"eval_logps/chosen": -684.0,
"eval_logps/rejected": -676.0,
"eval_loss": 1.0399062633514404,
"eval_rewards/accuracies": 0.498046875,
"eval_rewards/chosen": -3.65625,
"eval_rewards/margins": 0.2119140625,
"eval_rewards/rejected": -3.875,
"eval_runtime": 46.6847,
"eval_samples_per_second": 42.841,
"eval_steps_per_second": 0.685,
"step": 2100
},
{
"epoch": 2.2094240837696333,
"grad_norm": 0.02163541774016879,
"learning_rate": 9.855595178057334e-06,
"logits/chosen": -9.125,
"logits/rejected": -7.28125,
"logps/chosen": -904.0,
"logps/rejected": -3664.0,
"loss": 0.006,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -3.671875,
"rewards/margins": 11.75,
"rewards/rejected": -15.375,
"step": 2110
},
{
"epoch": 2.2198952879581153,
"grad_norm": 0.07350503818478554,
"learning_rate": 9.614332193917949e-06,
"logits/chosen": -8.75,
"logits/rejected": -7.84375,
"logps/chosen": -884.0,
"logps/rejected": -3632.0,
"loss": 0.0012,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.765625,
"rewards/margins": 11.5,
"rewards/rejected": -15.25,
"step": 2120
},
{
"epoch": 2.230366492146597,
"grad_norm": 0.01889128250233415,
"learning_rate": 9.375353992449385e-06,
"logits/chosen": -8.625,
"logits/rejected": -7.8125,
"logps/chosen": -764.0,
"logps/rejected": -3632.0,
"loss": 0.0009,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.5,
"rewards/margins": 11.5625,
"rewards/rejected": -15.0625,
"step": 2130
},
{
"epoch": 2.2408376963350785,
"grad_norm": 0.04071837909560905,
"learning_rate": 9.138696062085442e-06,
"logits/chosen": -8.5,
"logits/rejected": -7.78125,
"logps/chosen": -792.0,
"logps/rejected": -3728.0,
"loss": 0.0017,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.546875,
"rewards/margins": 12.125,
"rewards/rejected": -15.6875,
"step": 2140
},
{
"epoch": 2.25130890052356,
"grad_norm": 0.021469389564482435,
"learning_rate": 8.904393546698007e-06,
"logits/chosen": -8.8125,
"logits/rejected": -7.75,
"logps/chosen": -860.0,
"logps/rejected": -3712.0,
"loss": 0.0007,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.703125,
"rewards/margins": 12.0,
"rewards/rejected": -15.75,
"step": 2150
},
{
"epoch": 2.261780104712042,
"grad_norm": 0.020402906943641254,
"learning_rate": 8.672481240378142e-06,
"logits/chosen": -9.0625,
"logits/rejected": -7.90625,
"logps/chosen": -900.0,
"logps/rejected": -3632.0,
"loss": 0.0016,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.75,
"rewards/margins": 11.75,
"rewards/rejected": -15.5,
"step": 2160
},
{
"epoch": 2.2722513089005236,
"grad_norm": 0.022142962972623925,
"learning_rate": 8.44299358226919e-06,
"logits/chosen": -9.375,
"logits/rejected": -7.65625,
"logps/chosen": -888.0,
"logps/rejected": -3856.0,
"loss": 0.0021,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.890625,
"rewards/margins": 12.3125,
"rewards/rejected": -16.25,
"step": 2170
},
{
"epoch": 2.282722513089005,
"grad_norm": 0.03300282470334173,
"learning_rate": 8.215964651452457e-06,
"logits/chosen": -9.375,
"logits/rejected": -7.96875,
"logps/chosen": -876.0,
"logps/rejected": -3728.0,
"loss": 0.0011,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.84375,
"rewards/margins": 12.3125,
"rewards/rejected": -16.25,
"step": 2180
},
{
"epoch": 2.2931937172774868,
"grad_norm": 0.01194477775545041,
"learning_rate": 7.991428161886502e-06,
"logits/chosen": -8.75,
"logits/rejected": -8.3125,
"logps/chosen": -876.0,
"logps/rejected": -3840.0,
"loss": 0.0051,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.890625,
"rewards/margins": 12.3125,
"rewards/rejected": -16.25,
"step": 2190
},
{
"epoch": 2.303664921465969,
"grad_norm": 0.017861975092218173,
"learning_rate": 7.76941745740061e-06,
"logits/chosen": -8.1875,
"logits/rejected": -7.4375,
"logps/chosen": -764.0,
"logps/rejected": -3888.0,
"loss": 0.0012,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.59375,
"rewards/margins": 12.5625,
"rewards/rejected": -16.125,
"step": 2200
},
{
"epoch": 2.303664921465969,
"eval_logits/chosen": -8.0625,
"eval_logits/rejected": -7.25,
"eval_logps/chosen": -700.0,
"eval_logps/rejected": -692.0,
"eval_loss": 1.0474375486373901,
"eval_rewards/accuracies": 0.5,
"eval_rewards/chosen": -3.8125,
"eval_rewards/margins": 0.236328125,
"eval_rewards/rejected": -4.03125,
"eval_runtime": 46.6842,
"eval_samples_per_second": 42.841,
"eval_steps_per_second": 0.685,
"step": 2200
},
{
"epoch": 2.3141361256544504,
"grad_norm": 0.11896214244265217,
"learning_rate": 7.549965506743123e-06,
"logits/chosen": -8.5,
"logits/rejected": -7.75,
"logps/chosen": -752.0,
"logps/rejected": -3696.0,
"loss": 0.0017,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.671875,
"rewards/margins": 12.5,
"rewards/rejected": -16.125,
"step": 2210
},
{
"epoch": 2.324607329842932,
"grad_norm": 0.019423890288326043,
"learning_rate": 7.333104898685631e-06,
"logits/chosen": -8.375,
"logits/rejected": -8.1875,
"logps/chosen": -780.0,
"logps/rejected": -3648.0,
"loss": 0.0011,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.59375,
"rewards/margins": 12.25,
"rewards/rejected": -15.875,
"step": 2220
},
{
"epoch": 2.3350785340314135,
"grad_norm": 0.02394615444689975,
"learning_rate": 7.118867837183518e-06,
"logits/chosen": -8.6875,
"logits/rejected": -7.46875,
"logps/chosen": -752.0,
"logps/rejected": -3872.0,
"loss": 0.0009,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.5625,
"rewards/margins": 13.3125,
"rewards/rejected": -16.875,
"step": 2230
},
{
"epoch": 2.345549738219895,
"grad_norm": 0.019734030502550154,
"learning_rate": 6.9072861365936045e-06,
"logits/chosen": -8.875,
"logits/rejected": -7.84375,
"logps/chosen": -840.0,
"logps/rejected": -3888.0,
"loss": 0.0007,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.796875,
"rewards/margins": 12.375,
"rewards/rejected": -16.125,
"step": 2240
},
{
"epoch": 2.356020942408377,
"grad_norm": 0.01768539026521244,
"learning_rate": 6.6983912169497005e-06,
"logits/chosen": -8.9375,
"logits/rejected": -8.25,
"logps/chosen": -856.0,
"logps/rejected": -3872.0,
"loss": 0.0008,
"rewards/accuracies": 1.0,
"rewards/chosen": -4.09375,
"rewards/margins": 12.4375,
"rewards/rejected": -16.5,
"step": 2250
},
{
"epoch": 2.3664921465968587,
"grad_norm": 0.011464514421830191,
"learning_rate": 6.49221409929677e-06,
"logits/chosen": -9.1875,
"logits/rejected": -7.5625,
"logps/chosen": -872.0,
"logps/rejected": -3888.0,
"loss": 0.0013,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.890625,
"rewards/margins": 12.75,
"rewards/rejected": -16.625,
"step": 2260
},
{
"epoch": 2.3769633507853403,
"grad_norm": 0.010222790937320952,
"learning_rate": 6.2887854010842075e-06,
"logits/chosen": -8.5,
"logits/rejected": -8.1875,
"logps/chosen": -808.0,
"logps/rejected": -3856.0,
"loss": 0.0015,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.84375,
"rewards/margins": 13.3125,
"rewards/rejected": -17.125,
"step": 2270
},
{
"epoch": 2.387434554973822,
"grad_norm": 0.010206334269551233,
"learning_rate": 6.0881353316191385e-06,
"logits/chosen": -8.8125,
"logits/rejected": -7.8125,
"logps/chosen": -856.0,
"logps/rejected": -3840.0,
"loss": 0.0006,
"rewards/accuracies": 1.0,
"rewards/chosen": -4.125,
"rewards/margins": 12.4375,
"rewards/rejected": -16.625,
"step": 2280
},
{
"epoch": 2.3979057591623034,
"grad_norm": 0.009622631120446971,
"learning_rate": 5.890293687580381e-06,
"logits/chosen": -8.8125,
"logits/rejected": -8.125,
"logps/chosen": -804.0,
"logps/rejected": -3904.0,
"loss": 0.0007,
"rewards/accuracies": 1.0,
"rewards/chosen": -4.0625,
"rewards/margins": 12.4375,
"rewards/rejected": -16.5,
"step": 2290
},
{
"epoch": 2.4083769633507854,
"grad_norm": 0.030647640756532212,
"learning_rate": 5.695289848593532e-06,
"logits/chosen": -8.75,
"logits/rejected": -7.46875,
"logps/chosen": -872.0,
"logps/rejected": -3680.0,
"loss": 0.0012,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.96875,
"rewards/margins": 11.875,
"rewards/rejected": -15.875,
"step": 2300
},
{
"epoch": 2.4083769633507854,
"eval_logits/chosen": -7.96875,
"eval_logits/rejected": -7.125,
"eval_logps/chosen": -712.0,
"eval_logps/rejected": -708.0,
"eval_loss": 1.0702812671661377,
"eval_rewards/accuracies": 0.4921875,
"eval_rewards/chosen": -3.953125,
"eval_rewards/margins": 0.2451171875,
"eval_rewards/rejected": -4.1875,
"eval_runtime": 46.6799,
"eval_samples_per_second": 42.845,
"eval_steps_per_second": 0.686,
"step": 2300
},
{
"epoch": 2.418848167539267,
"grad_norm": 0.008362554383941635,
"learning_rate": 5.50315277286812e-06,
"logits/chosen": -8.5,
"logits/rejected": -7.65625,
"logps/chosen": -832.0,
"logps/rejected": -3856.0,
"loss": 0.0007,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.9375,
"rewards/margins": 12.625,
"rewards/rejected": -16.625,
"step": 2310
},
{
"epoch": 2.4293193717277486,
"grad_norm": 0.06802426279148979,
"learning_rate": 5.313910992897281e-06,
"logits/chosen": -9.125,
"logits/rejected": -7.875,
"logps/chosen": -812.0,
"logps/rejected": -3744.0,
"loss": 0.002,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.84375,
"rewards/margins": 12.25,
"rewards/rejected": -16.125,
"step": 2320
},
{
"epoch": 2.4397905759162306,
"grad_norm": 0.018358968493733585,
"learning_rate": 5.127592611220658e-06,
"logits/chosen": -8.5,
"logits/rejected": -7.8125,
"logps/chosen": -844.0,
"logps/rejected": -3968.0,
"loss": 0.0005,
"rewards/accuracies": 1.0,
"rewards/chosen": -4.0625,
"rewards/margins": 12.4375,
"rewards/rejected": -16.5,
"step": 2330
},
{
"epoch": 2.450261780104712,
"grad_norm": 0.01214231761251931,
"learning_rate": 4.944225296251159e-06,
"logits/chosen": -9.0,
"logits/rejected": -7.25,
"logps/chosen": -860.0,
"logps/rejected": -3760.0,
"loss": 0.0006,
"rewards/accuracies": 1.0,
"rewards/chosen": -4.0,
"rewards/margins": 12.6875,
"rewards/rejected": -16.625,
"step": 2340
},
{
"epoch": 2.4607329842931938,
"grad_norm": 0.11751543802006448,
"learning_rate": 4.763836278166172e-06,
"logits/chosen": -8.5,
"logits/rejected": -7.6875,
"logps/chosen": -864.0,
"logps/rejected": -3856.0,
"loss": 0.0015,
"rewards/accuracies": 1.0,
"rewards/chosen": -4.125,
"rewards/margins": 12.9375,
"rewards/rejected": -17.125,
"step": 2350
},
{
"epoch": 2.4712041884816753,
"grad_norm": 0.023447273775509123,
"learning_rate": 4.586452344863876e-06,
"logits/chosen": -9.375,
"logits/rejected": -8.1875,
"logps/chosen": -856.0,
"logps/rejected": -3808.0,
"loss": 0.0013,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.921875,
"rewards/margins": 13.125,
"rewards/rejected": -17.0,
"step": 2360
},
{
"epoch": 2.481675392670157,
"grad_norm": 0.03208818774985124,
"learning_rate": 4.412099837985192e-06,
"logits/chosen": -8.375,
"logits/rejected": -7.3125,
"logps/chosen": -780.0,
"logps/rejected": -3792.0,
"loss": 0.001,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.765625,
"rewards/margins": 13.0625,
"rewards/rejected": -16.75,
"step": 2370
},
{
"epoch": 2.492146596858639,
"grad_norm": 0.0181958527105786,
"learning_rate": 4.240804649002089e-06,
"logits/chosen": -8.4375,
"logits/rejected": -7.34375,
"logps/chosen": -800.0,
"logps/rejected": -3968.0,
"loss": 0.0008,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.921875,
"rewards/margins": 13.25,
"rewards/rejected": -17.25,
"step": 2380
},
{
"epoch": 2.5026178010471205,
"grad_norm": 0.015887955967711368,
"learning_rate": 4.072592215372598e-06,
"logits/chosen": -9.75,
"logits/rejected": -8.0625,
"logps/chosen": -864.0,
"logps/rejected": -3728.0,
"loss": 0.0008,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.859375,
"rewards/margins": 12.5625,
"rewards/rejected": -16.375,
"step": 2390
},
{
"epoch": 2.513089005235602,
"grad_norm": 0.004684993804307615,
"learning_rate": 3.907487516763389e-06,
"logits/chosen": -8.4375,
"logits/rejected": -7.625,
"logps/chosen": -864.0,
"logps/rejected": -3888.0,
"loss": 0.0014,
"rewards/accuracies": 1.0,
"rewards/chosen": -4.0,
"rewards/margins": 13.25,
"rewards/rejected": -17.25,
"step": 2400
},
{
"epoch": 2.513089005235602,
"eval_logits/chosen": -8.0,
"eval_logits/rejected": -7.125,
"eval_logps/chosen": -724.0,
"eval_logps/rejected": -720.0,
"eval_loss": 1.0871562957763672,
"eval_rewards/accuracies": 0.498046875,
"eval_rewards/chosen": -4.03125,
"eval_rewards/margins": 0.259765625,
"eval_rewards/rejected": -4.3125,
"eval_runtime": 46.6914,
"eval_samples_per_second": 42.834,
"eval_steps_per_second": 0.685,
"step": 2400
},
{
"epoch": 2.5235602094240837,
"grad_norm": 0.08711724610615093,
"learning_rate": 3.745515071340286e-06,
"logits/chosen": -9.25,
"logits/rejected": -8.4375,
"logps/chosen": -924.0,
"logps/rejected": -3760.0,
"loss": 0.001,
"rewards/accuracies": 1.0,
"rewards/chosen": -4.03125,
"rewards/margins": 12.875,
"rewards/rejected": -16.875,
"step": 2410
},
{
"epoch": 2.5340314136125652,
"grad_norm": 0.01169919594059341,
"learning_rate": 3.586698932127236e-06,
"logits/chosen": -9.0,
"logits/rejected": -7.25,
"logps/chosen": -876.0,
"logps/rejected": -4000.0,
"loss": 0.0007,
"rewards/accuracies": 1.0,
"rewards/chosen": -4.15625,
"rewards/margins": 13.125,
"rewards/rejected": -17.25,
"step": 2420
},
{
"epoch": 2.5445026178010473,
"grad_norm": 0.015264520650282581,
"learning_rate": 3.4310626834344744e-06,
"logits/chosen": -8.5625,
"logits/rejected": -7.21875,
"logps/chosen": -932.0,
"logps/rejected": -3696.0,
"loss": 0.0085,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -3.96875,
"rewards/margins": 12.375,
"rewards/rejected": -16.375,
"step": 2430
},
{
"epoch": 2.554973821989529,
"grad_norm": 0.05058738993381474,
"learning_rate": 3.2786294373562343e-06,
"logits/chosen": -8.875,
"logits/rejected": -7.5625,
"logps/chosen": -816.0,
"logps/rejected": -3824.0,
"loss": 0.0009,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.921875,
"rewards/margins": 12.9375,
"rewards/rejected": -16.875,
"step": 2440
},
{
"epoch": 2.5654450261780104,
"grad_norm": 0.03479353006827816,
"learning_rate": 3.1294218303385526e-06,
"logits/chosen": -8.8125,
"logits/rejected": -7.5625,
"logps/chosen": -848.0,
"logps/rejected": -3824.0,
"loss": 0.0009,
"rewards/accuracies": 1.0,
"rewards/chosen": -4.0,
"rewards/margins": 12.5,
"rewards/rejected": -16.5,
"step": 2450
},
{
"epoch": 2.5759162303664924,
"grad_norm": 0.003868268032536823,
"learning_rate": 2.9834620198177483e-06,
"logits/chosen": -8.5625,
"logits/rejected": -7.28125,
"logps/chosen": -852.0,
"logps/rejected": -3840.0,
"loss": 0.001,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.96875,
"rewards/margins": 12.6875,
"rewards/rejected": -16.625,
"step": 2460
},
{
"epoch": 2.5863874345549736,
"grad_norm": 0.03862081964356321,
"learning_rate": 2.840771680930068e-06,
"logits/chosen": -8.3125,
"logits/rejected": -7.3125,
"logps/chosen": -860.0,
"logps/rejected": -3632.0,
"loss": 0.0007,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.90625,
"rewards/margins": 12.5625,
"rewards/rejected": -16.5,
"step": 2470
},
{
"epoch": 2.5968586387434556,
"grad_norm": 0.015396324685458427,
"learning_rate": 2.701372003292868e-06,
"logits/chosen": -9.25,
"logits/rejected": -7.625,
"logps/chosen": -864.0,
"logps/rejected": -3888.0,
"loss": 0.0011,
"rewards/accuracies": 1.0,
"rewards/chosen": -4.09375,
"rewards/margins": 12.6875,
"rewards/rejected": -16.75,
"step": 2480
},
{
"epoch": 2.607329842931937,
"grad_norm": 0.03375290620843567,
"learning_rate": 2.56528368785795e-06,
"logits/chosen": -8.1875,
"logits/rejected": -7.34375,
"logps/chosen": -876.0,
"logps/rejected": -3920.0,
"loss": 0.0016,
"rewards/accuracies": 1.0,
"rewards/chosen": -4.0625,
"rewards/margins": 12.75,
"rewards/rejected": -16.875,
"step": 2490
},
{
"epoch": 2.6178010471204187,
"grad_norm": 0.030348494854253092,
"learning_rate": 2.432526943837471e-06,
"logits/chosen": -8.0625,
"logits/rejected": -8.0,
"logps/chosen": -768.0,
"logps/rejected": -4016.0,
"loss": 0.0013,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.75,
"rewards/margins": 13.625,
"rewards/rejected": -17.375,
"step": 2500
},
{
"epoch": 2.6178010471204187,
"eval_logits/chosen": -7.8125,
"eval_logits/rejected": -6.9375,
"eval_logps/chosen": -716.0,
"eval_logps/rejected": -712.0,
"eval_loss": 1.07826566696167,
"eval_rewards/accuracies": 0.501953125,
"eval_rewards/chosen": -3.96875,
"eval_rewards/margins": 0.2490234375,
"eval_rewards/rejected": -4.21875,
"eval_runtime": 46.6867,
"eval_samples_per_second": 42.839,
"eval_steps_per_second": 0.685,
"step": 2500
},
{
"epoch": 2.6282722513089007,
"grad_norm": 0.006129277312494226,
"learning_rate": 2.3031214857028446e-06,
"logits/chosen": -8.6875,
"logits/rejected": -7.65625,
"logps/chosen": -832.0,
"logps/rejected": -3808.0,
"loss": 0.0011,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.984375,
"rewards/margins": 12.5,
"rewards/rejected": -16.5,
"step": 2510
},
{
"epoch": 2.6387434554973823,
"grad_norm": 0.006911450636428596,
"learning_rate": 2.177086530257139e-06,
"logits/chosen": -8.875,
"logits/rejected": -7.3125,
"logps/chosen": -804.0,
"logps/rejected": -3808.0,
"loss": 0.0007,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.828125,
"rewards/margins": 12.75,
"rewards/rejected": -16.5,
"step": 2520
},
{
"epoch": 2.649214659685864,
"grad_norm": 0.04862872764055925,
"learning_rate": 2.054440793781354e-06,
"logits/chosen": -8.875,
"logits/rejected": -7.75,
"logps/chosen": -880.0,
"logps/rejected": -3872.0,
"loss": 0.0008,
"rewards/accuracies": 1.0,
"rewards/chosen": -4.15625,
"rewards/margins": 12.875,
"rewards/rejected": -17.0,
"step": 2530
},
{
"epoch": 2.6596858638743455,
"grad_norm": 0.005007668355864154,
"learning_rate": 1.9352024892550587e-06,
"logits/chosen": -9.0,
"logits/rejected": -7.71875,
"logps/chosen": -876.0,
"logps/rejected": -3968.0,
"loss": 0.0005,
"rewards/accuracies": 1.0,
"rewards/chosen": -4.125,
"rewards/margins": 13.0,
"rewards/rejected": -17.125,
"step": 2540
},
{
"epoch": 2.670157068062827,
"grad_norm": 0.2408050010574279,
"learning_rate": 1.8193893236517379e-06,
"logits/chosen": -8.0,
"logits/rejected": -7.375,
"logps/chosen": -836.0,
"logps/rejected": -3744.0,
"loss": 0.0012,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.984375,
"rewards/margins": 12.5625,
"rewards/rejected": -16.625,
"step": 2550
},
{
"epoch": 2.680628272251309,
"grad_norm": 0.013213353488241575,
"learning_rate": 1.7070184953092956e-06,
"logits/chosen": -8.5625,
"logits/rejected": -7.5625,
"logps/chosen": -884.0,
"logps/rejected": -3744.0,
"loss": 0.0008,
"rewards/accuracies": 1.0,
"rewards/chosen": -4.0,
"rewards/margins": 12.5625,
"rewards/rejected": -16.5,
"step": 2560
},
{
"epoch": 2.6910994764397906,
"grad_norm": 0.05457368229109858,
"learning_rate": 1.5981066913760918e-06,
"logits/chosen": -8.1875,
"logits/rejected": -7.0625,
"logps/chosen": -852.0,
"logps/rejected": -3680.0,
"loss": 0.0006,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.921875,
"rewards/margins": 12.125,
"rewards/rejected": -16.0,
"step": 2570
},
{
"epoch": 2.701570680628272,
"grad_norm": 0.01251571811368801,
"learning_rate": 1.492670085332884e-06,
"logits/chosen": -8.75,
"logits/rejected": -7.28125,
"logps/chosen": -844.0,
"logps/rejected": -3760.0,
"loss": 0.0007,
"rewards/accuracies": 1.0,
"rewards/chosen": -4.0,
"rewards/margins": 12.5,
"rewards/rejected": -16.5,
"step": 2580
},
{
"epoch": 2.712041884816754,
"grad_norm": 0.034401445021969033,
"learning_rate": 1.3907243345910787e-06,
"logits/chosen": -8.25,
"logits/rejected": -7.8125,
"logps/chosen": -852.0,
"logps/rejected": -3824.0,
"loss": 0.001,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.875,
"rewards/margins": 12.9375,
"rewards/rejected": -16.75,
"step": 2590
},
{
"epoch": 2.7225130890052354,
"grad_norm": 0.011273006495123644,
"learning_rate": 1.292284578167538e-06,
"logits/chosen": -8.5625,
"logits/rejected": -7.875,
"logps/chosen": -788.0,
"logps/rejected": -3872.0,
"loss": 0.001,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.90625,
"rewards/margins": 12.75,
"rewards/rejected": -16.625,
"step": 2600
},
{
"epoch": 2.7225130890052354,
"eval_logits/chosen": -7.84375,
"eval_logits/rejected": -6.96875,
"eval_logps/chosen": -720.0,
"eval_logps/rejected": -712.0,
"eval_loss": 1.0849374532699585,
"eval_rewards/accuracies": 0.501953125,
"eval_rewards/chosen": -4.0,
"eval_rewards/margins": 0.251953125,
"eval_rewards/rejected": -4.25,
"eval_runtime": 46.5427,
"eval_samples_per_second": 42.971,
"eval_steps_per_second": 0.688,
"step": 2600
},
{
"epoch": 2.7329842931937174,
"grad_norm": 0.03159285372319596,
"learning_rate": 1.1973654344364925e-06,
"logits/chosen": -8.875,
"logits/rejected": -7.5,
"logps/chosen": -848.0,
"logps/rejected": -3776.0,
"loss": 0.001,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.84375,
"rewards/margins": 12.4375,
"rewards/rejected": -16.25,
"step": 2610
},
{
"epoch": 2.743455497382199,
"grad_norm": 0.012429183603076915,
"learning_rate": 1.1059809989586583e-06,
"logits/chosen": -8.5625,
"logits/rejected": -7.84375,
"logps/chosen": -792.0,
"logps/rejected": -3872.0,
"loss": 0.0009,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.78125,
"rewards/margins": 13.0625,
"rewards/rejected": -16.875,
"step": 2620
},
{
"epoch": 2.7539267015706805,
"grad_norm": 0.02094386822942291,
"learning_rate": 1.018144842388044e-06,
"logits/chosen": -8.375,
"logits/rejected": -7.8125,
"logps/chosen": -812.0,
"logps/rejected": -3856.0,
"loss": 0.0004,
"rewards/accuracies": 1.0,
"rewards/chosen": -4.0,
"rewards/margins": 12.75,
"rewards/rejected": -16.75,
"step": 2630
},
{
"epoch": 2.7643979057591626,
"grad_norm": 0.0158444216088217,
"learning_rate": 9.338700084567109e-07,
"logits/chosen": -8.375,
"logits/rejected": -7.78125,
"logps/chosen": -868.0,
"logps/rejected": -3904.0,
"loss": 0.0011,
"rewards/accuracies": 1.0,
"rewards/chosen": -4.09375,
"rewards/margins": 12.875,
"rewards/rejected": -17.0,
"step": 2640
},
{
"epoch": 2.774869109947644,
"grad_norm": 0.049501589253906776,
"learning_rate": 8.531690120377606e-07,
"logits/chosen": -8.9375,
"logits/rejected": -7.25,
"logps/chosen": -860.0,
"logps/rejected": -3744.0,
"loss": 0.0008,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.625,
"rewards/margins": 13.25,
"rewards/rejected": -16.875,
"step": 2650
},
{
"epoch": 2.7853403141361257,
"grad_norm": 0.04080167254594831,
"learning_rate": 7.760538372868636e-07,
"logits/chosen": -8.1875,
"logits/rejected": -7.6875,
"logps/chosen": -776.0,
"logps/rejected": -3936.0,
"loss": 0.0008,
"rewards/accuracies": 1.0,
"rewards/chosen": -4.0,
"rewards/margins": 13.3125,
"rewards/rejected": -17.25,
"step": 2660
},
{
"epoch": 2.7958115183246073,
"grad_norm": 0.05882021967860267,
"learning_rate": 7.025359358626166e-07,
"logits/chosen": -8.75,
"logits/rejected": -8.0625,
"logps/chosen": -836.0,
"logps/rejected": -3808.0,
"loss": 0.0007,
"rewards/accuracies": 1.0,
"rewards/chosen": -4.03125,
"rewards/margins": 12.8125,
"rewards/rejected": -16.75,
"step": 2670
},
{
"epoch": 2.806282722513089,
"grad_norm": 0.017859368186850088,
"learning_rate": 6.326262252259463e-07,
"logits/chosen": -9.375,
"logits/rejected": -8.1875,
"logps/chosen": -892.0,
"logps/rejected": -3792.0,
"loss": 0.0008,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.9375,
"rewards/margins": 13.0,
"rewards/rejected": -16.875,
"step": 2680
},
{
"epoch": 2.816753926701571,
"grad_norm": 0.023097358474203793,
"learning_rate": 5.66335087018871e-07,
"logits/chosen": -8.375,
"logits/rejected": -7.3125,
"logps/chosen": -904.0,
"logps/rejected": -3888.0,
"loss": 0.0009,
"rewards/accuracies": 1.0,
"rewards/chosen": -4.0625,
"rewards/margins": 12.9375,
"rewards/rejected": -17.0,
"step": 2690
},
{
"epoch": 2.8272251308900525,
"grad_norm": 0.017302407810572212,
"learning_rate": 5.036723655228226e-07,
"logits/chosen": -8.875,
"logits/rejected": -7.3125,
"logps/chosen": -864.0,
"logps/rejected": -3696.0,
"loss": 0.0008,
"rewards/accuracies": 1.0,
"rewards/chosen": -4.0625,
"rewards/margins": 12.0,
"rewards/rejected": -16.125,
"step": 2700
},
{
"epoch": 2.8272251308900525,
"eval_logits/chosen": -7.875,
"eval_logits/rejected": -7.0,
"eval_logps/chosen": -716.0,
"eval_logps/rejected": -712.0,
"eval_loss": 1.082421898841858,
"eval_rewards/accuracies": 0.498046875,
"eval_rewards/chosen": -3.984375,
"eval_rewards/margins": 0.251953125,
"eval_rewards/rejected": -4.25,
"eval_runtime": 46.6118,
"eval_samples_per_second": 42.908,
"eval_steps_per_second": 0.687,
"step": 2700
},
{
"epoch": 2.837696335078534,
"grad_norm": 0.014940979281921807,
"learning_rate": 4.446473661967432e-07,
"logits/chosen": -9.0,
"logits/rejected": -7.78125,
"logps/chosen": -872.0,
"logps/rejected": -3840.0,
"loss": 0.0006,
"rewards/accuracies": 1.0,
"rewards/chosen": -4.03125,
"rewards/margins": 12.5,
"rewards/rejected": -16.5,
"step": 2710
},
{
"epoch": 2.8481675392670156,
"grad_norm": 0.01640919947378091,
"learning_rate": 3.892688542952411e-07,
"logits/chosen": -9.3125,
"logits/rejected": -7.5625,
"logps/chosen": -904.0,
"logps/rejected": -3840.0,
"loss": 0.001,
"rewards/accuracies": 1.0,
"rewards/chosen": -4.03125,
"rewards/margins": 13.0625,
"rewards/rejected": -17.125,
"step": 2720
},
{
"epoch": 2.858638743455497,
"grad_norm": 0.28412281326001143,
"learning_rate": 3.3754505356694e-07,
"logits/chosen": -8.5,
"logits/rejected": -7.8125,
"logps/chosen": -844.0,
"logps/rejected": -3728.0,
"loss": 0.0009,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.828125,
"rewards/margins": 12.75,
"rewards/rejected": -16.625,
"step": 2730
},
{
"epoch": 2.869109947643979,
"grad_norm": 0.04247549144423903,
"learning_rate": 2.8948364503322276e-07,
"logits/chosen": -9.375,
"logits/rejected": -7.625,
"logps/chosen": -920.0,
"logps/rejected": -3952.0,
"loss": 0.0008,
"rewards/accuracies": 1.0,
"rewards/chosen": -4.09375,
"rewards/margins": 12.5625,
"rewards/rejected": -16.625,
"step": 2740
},
{
"epoch": 2.8795811518324608,
"grad_norm": 0.01505103940038022,
"learning_rate": 2.4509176584762706e-07,
"logits/chosen": -8.8125,
"logits/rejected": -7.65625,
"logps/chosen": -812.0,
"logps/rejected": -3760.0,
"loss": 0.0023,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -3.84375,
"rewards/margins": 12.25,
"rewards/rejected": -16.125,
"step": 2750
},
{
"epoch": 2.8900523560209423,
"grad_norm": 0.017570848428039923,
"learning_rate": 2.0437600823595694e-07,
"logits/chosen": -9.0,
"logits/rejected": -7.6875,
"logps/chosen": -864.0,
"logps/rejected": -4032.0,
"loss": 0.0012,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.984375,
"rewards/margins": 13.4375,
"rewards/rejected": -17.375,
"step": 2760
},
{
"epoch": 2.900523560209424,
"grad_norm": 0.028669907369084778,
"learning_rate": 1.6734241851733e-07,
"logits/chosen": -8.9375,
"logits/rejected": -7.875,
"logps/chosen": -844.0,
"logps/rejected": -3776.0,
"loss": 0.0006,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.96875,
"rewards/margins": 12.875,
"rewards/rejected": -16.875,
"step": 2770
},
{
"epoch": 2.9109947643979055,
"grad_norm": 0.02137053040173603,
"learning_rate": 1.339964962062984e-07,
"logits/chosen": -8.8125,
"logits/rejected": -7.78125,
"logps/chosen": -892.0,
"logps/rejected": -3936.0,
"loss": 0.0066,
"rewards/accuracies": 1.0,
"rewards/chosen": -4.15625,
"rewards/margins": 13.25,
"rewards/rejected": -17.375,
"step": 2780
},
{
"epoch": 2.9214659685863875,
"grad_norm": 0.01618173739957357,
"learning_rate": 1.0434319319617136e-07,
"logits/chosen": -9.4375,
"logits/rejected": -7.75,
"logps/chosen": -836.0,
"logps/rejected": -3808.0,
"loss": 0.0007,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.9375,
"rewards/margins": 12.75,
"rewards/rejected": -16.75,
"step": 2790
},
{
"epoch": 2.931937172774869,
"grad_norm": 0.004542141523474725,
"learning_rate": 7.838691302365363e-08,
"logits/chosen": -8.75,
"logits/rejected": -7.9375,
"logps/chosen": -792.0,
"logps/rejected": -3840.0,
"loss": 0.0007,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.921875,
"rewards/margins": 13.0625,
"rewards/rejected": -17.0,
"step": 2800
},
{
"epoch": 2.931937172774869,
"eval_logits/chosen": -7.875,
"eval_logits/rejected": -7.0,
"eval_logps/chosen": -716.0,
"eval_logps/rejected": -712.0,
"eval_loss": 1.0827969312667847,
"eval_rewards/accuracies": 0.5,
"eval_rewards/chosen": -3.984375,
"eval_rewards/margins": 0.25,
"eval_rewards/rejected": -4.25,
"eval_runtime": 46.7005,
"eval_samples_per_second": 42.826,
"eval_steps_per_second": 0.685,
"step": 2800
},
{
"epoch": 2.9424083769633507,
"grad_norm": 0.012706448563563316,
"learning_rate": 5.613151021490459e-08,
"logits/chosen": -9.1875,
"logits/rejected": -7.875,
"logps/chosen": -840.0,
"logps/rejected": -3808.0,
"loss": 0.0022,
"rewards/accuracies": 1.0,
"rewards/chosen": -4.03125,
"rewards/margins": 12.25,
"rewards/rejected": -16.25,
"step": 2810
},
{
"epoch": 2.9528795811518327,
"grad_norm": 0.012881189468920841,
"learning_rate": 3.7580289713159965e-08,
"logits/chosen": -8.875,
"logits/rejected": -7.40625,
"logps/chosen": -792.0,
"logps/rejected": -3872.0,
"loss": 0.0009,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.734375,
"rewards/margins": 12.9375,
"rewards/rejected": -16.625,
"step": 2820
},
{
"epoch": 2.9633507853403143,
"grad_norm": 1.282803529345366,
"learning_rate": 2.2736006387927277e-08,
"logits/chosen": -8.375,
"logits/rejected": -7.8125,
"logps/chosen": -812.0,
"logps/rejected": -3664.0,
"loss": 0.0036,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.875,
"rewards/margins": 12.125,
"rewards/rejected": -16.0,
"step": 2830
},
{
"epoch": 2.973821989528796,
"grad_norm": 0.08296661587181926,
"learning_rate": 1.1600864625893604e-08,
"logits/chosen": -9.75,
"logits/rejected": -7.71875,
"logps/chosen": -912.0,
"logps/rejected": -3840.0,
"loss": 0.001,
"rewards/accuracies": 1.0,
"rewards/chosen": -4.0625,
"rewards/margins": 11.9375,
"rewards/rejected": -16.0,
"step": 2840
},
{
"epoch": 2.9842931937172774,
"grad_norm": 0.020709192670580245,
"learning_rate": 4.176518003573549e-09,
"logits/chosen": -9.375,
"logits/rejected": -7.9375,
"logps/chosen": -852.0,
"logps/rejected": -3792.0,
"loss": 0.0007,
"rewards/accuracies": 1.0,
"rewards/chosen": -4.0625,
"rewards/margins": 12.4375,
"rewards/rejected": -16.5,
"step": 2850
},
{
"epoch": 2.994764397905759,
"grad_norm": 0.022535033752319897,
"learning_rate": 4.640690417528903e-10,
"logits/chosen": -9.0625,
"logits/rejected": -8.5625,
"logps/chosen": -792.0,
"logps/rejected": -3648.0,
"loss": 0.0013,
"rewards/accuracies": 1.0,
"rewards/chosen": -3.625,
"rewards/margins": 12.4375,
"rewards/rejected": -16.0,
"step": 2860
},
{
"epoch": 3.0,
"step": 2865,
"total_flos": 0.0,
"train_loss": 0.06649821210216997,
"train_runtime": 12540.8656,
"train_samples_per_second": 14.621,
"train_steps_per_second": 0.228
}
],
"logging_steps": 10,
"max_steps": 2865,
"num_input_tokens_seen": 0,
"num_train_epochs": 3,
"save_steps": 500,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": false,
"should_training_stop": false
},
"attributes": {}
}
},
"total_flos": 0.0,
"train_batch_size": 8,
"trial_name": null,
"trial_params": null
}