CharlesLi's picture
Model save
3ee5d97 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": 1.8063274317033295,
"learning_rate": 1.7421602787456448e-07,
"logits/chosen": -11.625,
"logits/rejected": -11.75,
"logps/chosen": -466.0,
"logps/rejected": -764.0,
"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": 1.7540902935294356,
"learning_rate": 1.7421602787456445e-06,
"logits/chosen": -11.125,
"logits/rejected": -10.5,
"logps/chosen": -428.0,
"logps/rejected": -756.0,
"loss": 0.6917,
"rewards/accuracies": 0.1875,
"rewards/chosen": 1.6689300537109375e-05,
"rewards/margins": 0.00183868408203125,
"rewards/rejected": -0.00182342529296875,
"step": 10
},
{
"epoch": 0.020942408376963352,
"grad_norm": 1.6463329190564504,
"learning_rate": 3.484320557491289e-06,
"logits/chosen": -11.0,
"logits/rejected": -10.8125,
"logps/chosen": -402.0,
"logps/rejected": -576.0,
"loss": 0.6889,
"rewards/accuracies": 0.3687500059604645,
"rewards/chosen": 0.0004425048828125,
"rewards/margins": 0.01068115234375,
"rewards/rejected": -0.01025390625,
"step": 20
},
{
"epoch": 0.031413612565445025,
"grad_norm": 1.6756892071812357,
"learning_rate": 5.226480836236934e-06,
"logits/chosen": -11.625,
"logits/rejected": -11.375,
"logps/chosen": -430.0,
"logps/rejected": -532.0,
"loss": 0.6761,
"rewards/accuracies": 0.5874999761581421,
"rewards/chosen": -0.0255126953125,
"rewards/margins": 0.02685546875,
"rewards/rejected": -0.052490234375,
"step": 30
},
{
"epoch": 0.041884816753926704,
"grad_norm": 1.9641900886044579,
"learning_rate": 6.968641114982578e-06,
"logits/chosen": -12.4375,
"logits/rejected": -12.0,
"logps/chosen": -448.0,
"logps/rejected": -616.0,
"loss": 0.6367,
"rewards/accuracies": 0.6312500238418579,
"rewards/chosen": -0.1884765625,
"rewards/margins": 0.12255859375,
"rewards/rejected": -0.3125,
"step": 40
},
{
"epoch": 0.05235602094240838,
"grad_norm": 2.0191787395657532,
"learning_rate": 8.710801393728225e-06,
"logits/chosen": -12.1875,
"logits/rejected": -12.0,
"logps/chosen": -462.0,
"logps/rejected": -700.0,
"loss": 0.5846,
"rewards/accuracies": 0.675000011920929,
"rewards/chosen": -0.51171875,
"rewards/margins": 0.421875,
"rewards/rejected": -0.9296875,
"step": 50
},
{
"epoch": 0.06282722513089005,
"grad_norm": 1.8488023940540366,
"learning_rate": 1.0452961672473868e-05,
"logits/chosen": -11.0,
"logits/rejected": -10.9375,
"logps/chosen": -398.0,
"logps/rejected": -692.0,
"loss": 0.5395,
"rewards/accuracies": 0.6812499761581421,
"rewards/chosen": -0.50390625,
"rewards/margins": 0.76171875,
"rewards/rejected": -1.265625,
"step": 60
},
{
"epoch": 0.07329842931937172,
"grad_norm": 2.023635176453907,
"learning_rate": 1.2195121951219513e-05,
"logits/chosen": -12.0,
"logits/rejected": -11.6875,
"logps/chosen": -516.0,
"logps/rejected": -848.0,
"loss": 0.5344,
"rewards/accuracies": 0.706250011920929,
"rewards/chosen": -0.81640625,
"rewards/margins": 1.0,
"rewards/rejected": -1.8125,
"step": 70
},
{
"epoch": 0.08376963350785341,
"grad_norm": 2.12542473241284,
"learning_rate": 1.3937282229965156e-05,
"logits/chosen": -12.3125,
"logits/rejected": -12.1875,
"logps/chosen": -506.0,
"logps/rejected": -924.0,
"loss": 0.4981,
"rewards/accuracies": 0.768750011920929,
"rewards/chosen": -1.0859375,
"rewards/margins": 1.53125,
"rewards/rejected": -2.625,
"step": 80
},
{
"epoch": 0.09424083769633508,
"grad_norm": 3.3651617907270945,
"learning_rate": 1.56794425087108e-05,
"logits/chosen": -12.125,
"logits/rejected": -11.875,
"logps/chosen": -506.0,
"logps/rejected": -720.0,
"loss": 0.5293,
"rewards/accuracies": 0.706250011920929,
"rewards/chosen": -1.4609375,
"rewards/margins": 0.9921875,
"rewards/rejected": -2.453125,
"step": 90
},
{
"epoch": 0.10471204188481675,
"grad_norm": 1.973205604013508,
"learning_rate": 1.742160278745645e-05,
"logits/chosen": -11.75,
"logits/rejected": -11.5625,
"logps/chosen": -656.0,
"logps/rejected": -968.0,
"loss": 0.5077,
"rewards/accuracies": 0.731249988079071,
"rewards/chosen": -1.796875,
"rewards/margins": 1.234375,
"rewards/rejected": -3.03125,
"step": 100
},
{
"epoch": 0.10471204188481675,
"eval_logits/chosen": -10.8125,
"eval_logits/rejected": -10.4375,
"eval_logps/chosen": -442.0,
"eval_logps/rejected": -438.0,
"eval_loss": 0.6684062480926514,
"eval_rewards/accuracies": 0.619140625,
"eval_rewards/chosen": -1.2421875,
"eval_rewards/margins": 0.2490234375,
"eval_rewards/rejected": -1.4921875,
"eval_runtime": 59.8417,
"eval_samples_per_second": 33.422,
"eval_steps_per_second": 0.535,
"step": 100
},
{
"epoch": 0.11518324607329843,
"grad_norm": 2.5970473202527726,
"learning_rate": 1.9163763066202093e-05,
"logits/chosen": -10.4375,
"logits/rejected": -10.1875,
"logps/chosen": -552.0,
"logps/rejected": -880.0,
"loss": 0.4552,
"rewards/accuracies": 0.762499988079071,
"rewards/chosen": -1.3671875,
"rewards/margins": 1.4296875,
"rewards/rejected": -2.796875,
"step": 110
},
{
"epoch": 0.1256544502617801,
"grad_norm": 2.854077453487409,
"learning_rate": 2.0905923344947736e-05,
"logits/chosen": -10.375,
"logits/rejected": -10.3125,
"logps/chosen": -564.0,
"logps/rejected": -864.0,
"loss": 0.4893,
"rewards/accuracies": 0.78125,
"rewards/chosen": -2.015625,
"rewards/margins": 1.375,
"rewards/rejected": -3.390625,
"step": 120
},
{
"epoch": 0.13612565445026178,
"grad_norm": 2.7737113586040296,
"learning_rate": 2.264808362369338e-05,
"logits/chosen": -9.9375,
"logits/rejected": -10.0,
"logps/chosen": -584.0,
"logps/rejected": -880.0,
"loss": 0.4747,
"rewards/accuracies": 0.75,
"rewards/chosen": -1.7734375,
"rewards/margins": 1.125,
"rewards/rejected": -2.90625,
"step": 130
},
{
"epoch": 0.14659685863874344,
"grad_norm": 3.0608627183503203,
"learning_rate": 2.4390243902439026e-05,
"logits/chosen": -12.625,
"logits/rejected": -12.375,
"logps/chosen": -680.0,
"logps/rejected": -1032.0,
"loss": 0.4602,
"rewards/accuracies": 0.78125,
"rewards/chosen": -2.34375,
"rewards/margins": 1.421875,
"rewards/rejected": -3.765625,
"step": 140
},
{
"epoch": 0.15706806282722513,
"grad_norm": 2.484418385070813,
"learning_rate": 2.6132404181184672e-05,
"logits/chosen": -12.5625,
"logits/rejected": -12.25,
"logps/chosen": -636.0,
"logps/rejected": -876.0,
"loss": 0.4866,
"rewards/accuracies": 0.793749988079071,
"rewards/chosen": -2.125,
"rewards/margins": 1.09375,
"rewards/rejected": -3.21875,
"step": 150
},
{
"epoch": 0.16753926701570682,
"grad_norm": 2.0226295334690176,
"learning_rate": 2.7874564459930312e-05,
"logits/chosen": -10.6875,
"logits/rejected": -10.125,
"logps/chosen": -640.0,
"logps/rejected": -916.0,
"loss": 0.4459,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -2.484375,
"rewards/margins": 1.28125,
"rewards/rejected": -3.765625,
"step": 160
},
{
"epoch": 0.17801047120418848,
"grad_norm": 2.0895897190431674,
"learning_rate": 2.9616724738675962e-05,
"logits/chosen": -9.875,
"logits/rejected": -9.4375,
"logps/chosen": -684.0,
"logps/rejected": -1104.0,
"loss": 0.4326,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": -2.59375,
"rewards/margins": 1.96875,
"rewards/rejected": -4.5625,
"step": 170
},
{
"epoch": 0.18848167539267016,
"grad_norm": 3.29537870269347,
"learning_rate": 3.13588850174216e-05,
"logits/chosen": -9.8125,
"logits/rejected": -9.5625,
"logps/chosen": -712.0,
"logps/rejected": -1120.0,
"loss": 0.4356,
"rewards/accuracies": 0.8187500238418579,
"rewards/chosen": -2.75,
"rewards/margins": 2.125,
"rewards/rejected": -4.875,
"step": 180
},
{
"epoch": 0.19895287958115182,
"grad_norm": 2.849134635344997,
"learning_rate": 3.310104529616725e-05,
"logits/chosen": -9.25,
"logits/rejected": -9.3125,
"logps/chosen": -760.0,
"logps/rejected": -1160.0,
"loss": 0.4261,
"rewards/accuracies": 0.7875000238418579,
"rewards/chosen": -3.640625,
"rewards/margins": 1.9140625,
"rewards/rejected": -5.5625,
"step": 190
},
{
"epoch": 0.2094240837696335,
"grad_norm": 2.429664636550645,
"learning_rate": 3.48432055749129e-05,
"logits/chosen": -9.625,
"logits/rejected": -9.0625,
"logps/chosen": -760.0,
"logps/rejected": -1096.0,
"loss": 0.436,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -3.09375,
"rewards/margins": 1.6171875,
"rewards/rejected": -4.71875,
"step": 200
},
{
"epoch": 0.2094240837696335,
"eval_logits/chosen": -10.1875,
"eval_logits/rejected": -9.6875,
"eval_logps/chosen": -608.0,
"eval_logps/rejected": -620.0,
"eval_loss": 0.7755937576293945,
"eval_rewards/accuracies": 0.619140625,
"eval_rewards/chosen": -2.921875,
"eval_rewards/margins": 0.4140625,
"eval_rewards/rejected": -3.328125,
"eval_runtime": 58.0065,
"eval_samples_per_second": 34.479,
"eval_steps_per_second": 0.552,
"step": 200
},
{
"epoch": 0.2198952879581152,
"grad_norm": 2.4936763250468115,
"learning_rate": 3.6585365853658535e-05,
"logits/chosen": -9.6875,
"logits/rejected": -9.5,
"logps/chosen": -668.0,
"logps/rejected": -916.0,
"loss": 0.4676,
"rewards/accuracies": 0.768750011920929,
"rewards/chosen": -3.03125,
"rewards/margins": 1.109375,
"rewards/rejected": -4.125,
"step": 210
},
{
"epoch": 0.23036649214659685,
"grad_norm": 1.8405521756648462,
"learning_rate": 3.8327526132404185e-05,
"logits/chosen": -10.875,
"logits/rejected": -10.25,
"logps/chosen": -628.0,
"logps/rejected": -980.0,
"loss": 0.422,
"rewards/accuracies": 0.84375,
"rewards/chosen": -2.0625,
"rewards/margins": 1.484375,
"rewards/rejected": -3.546875,
"step": 220
},
{
"epoch": 0.24083769633507854,
"grad_norm": 2.1750713523389993,
"learning_rate": 4.006968641114983e-05,
"logits/chosen": -10.375,
"logits/rejected": -10.25,
"logps/chosen": -732.0,
"logps/rejected": -988.0,
"loss": 0.4533,
"rewards/accuracies": 0.8187500238418579,
"rewards/chosen": -3.390625,
"rewards/margins": 1.5625,
"rewards/rejected": -4.96875,
"step": 230
},
{
"epoch": 0.2513089005235602,
"grad_norm": 3.0094548174605706,
"learning_rate": 4.181184668989547e-05,
"logits/chosen": -9.5625,
"logits/rejected": -8.875,
"logps/chosen": -768.0,
"logps/rejected": -1096.0,
"loss": 0.4516,
"rewards/accuracies": 0.793749988079071,
"rewards/chosen": -3.96875,
"rewards/margins": 1.75,
"rewards/rejected": -5.71875,
"step": 240
},
{
"epoch": 0.2617801047120419,
"grad_norm": 4.574481724454628,
"learning_rate": 4.3554006968641115e-05,
"logits/chosen": -9.625,
"logits/rejected": -9.0625,
"logps/chosen": -808.0,
"logps/rejected": -1144.0,
"loss": 0.4399,
"rewards/accuracies": 0.7875000238418579,
"rewards/chosen": -3.546875,
"rewards/margins": 1.890625,
"rewards/rejected": -5.4375,
"step": 250
},
{
"epoch": 0.27225130890052357,
"grad_norm": 1.6260781248514558,
"learning_rate": 4.529616724738676e-05,
"logits/chosen": -11.875,
"logits/rejected": -11.0,
"logps/chosen": -664.0,
"logps/rejected": -1080.0,
"loss": 0.4231,
"rewards/accuracies": 0.8062499761581421,
"rewards/chosen": -2.875,
"rewards/margins": 2.234375,
"rewards/rejected": -5.125,
"step": 260
},
{
"epoch": 0.28272251308900526,
"grad_norm": 2.2558451392028247,
"learning_rate": 4.703832752613241e-05,
"logits/chosen": -13.6875,
"logits/rejected": -12.875,
"logps/chosen": -812.0,
"logps/rejected": -1264.0,
"loss": 0.4575,
"rewards/accuracies": 0.7875000238418579,
"rewards/chosen": -4.375,
"rewards/margins": 2.46875,
"rewards/rejected": -6.84375,
"step": 270
},
{
"epoch": 0.2931937172774869,
"grad_norm": 2.3484177998566267,
"learning_rate": 4.878048780487805e-05,
"logits/chosen": -10.125,
"logits/rejected": -9.5625,
"logps/chosen": -708.0,
"logps/rejected": -1040.0,
"loss": 0.4801,
"rewards/accuracies": 0.8062499761581421,
"rewards/chosen": -3.484375,
"rewards/margins": 1.609375,
"rewards/rejected": -5.09375,
"step": 280
},
{
"epoch": 0.3036649214659686,
"grad_norm": 2.7489572217900498,
"learning_rate": 4.999983293481418e-05,
"logits/chosen": -11.1875,
"logits/rejected": -10.625,
"logps/chosen": -696.0,
"logps/rejected": -1056.0,
"loss": 0.4238,
"rewards/accuracies": 0.762499988079071,
"rewards/chosen": -3.125,
"rewards/margins": 1.515625,
"rewards/rejected": -4.65625,
"step": 290
},
{
"epoch": 0.31413612565445026,
"grad_norm": 2.789909810863701,
"learning_rate": 4.9996862949181595e-05,
"logits/chosen": -9.3125,
"logits/rejected": -8.5625,
"logps/chosen": -872.0,
"logps/rejected": -1272.0,
"loss": 0.4375,
"rewards/accuracies": 0.8187500238418579,
"rewards/chosen": -4.59375,
"rewards/margins": 2.34375,
"rewards/rejected": -6.9375,
"step": 300
},
{
"epoch": 0.31413612565445026,
"eval_logits/chosen": -9.9375,
"eval_logits/rejected": -8.9375,
"eval_logps/chosen": -720.0,
"eval_logps/rejected": -752.0,
"eval_loss": 0.7544218897819519,
"eval_rewards/accuracies": 0.642578125,
"eval_rewards/chosen": -4.0,
"eval_rewards/margins": 0.6328125,
"eval_rewards/rejected": -4.625,
"eval_runtime": 58.0391,
"eval_samples_per_second": 34.46,
"eval_steps_per_second": 0.551,
"step": 300
},
{
"epoch": 0.32460732984293195,
"grad_norm": 2.139399440257054,
"learning_rate": 4.9990180911526896e-05,
"logits/chosen": -11.0625,
"logits/rejected": -10.375,
"logps/chosen": -736.0,
"logps/rejected": -1064.0,
"loss": 0.4142,
"rewards/accuracies": 0.8187500238418579,
"rewards/chosen": -3.640625,
"rewards/margins": 1.59375,
"rewards/rejected": -5.21875,
"step": 310
},
{
"epoch": 0.33507853403141363,
"grad_norm": 2.087886814297133,
"learning_rate": 4.997978781413745e-05,
"logits/chosen": -11.75,
"logits/rejected": -11.0625,
"logps/chosen": -864.0,
"logps/rejected": -1176.0,
"loss": 0.4401,
"rewards/accuracies": 0.8187500238418579,
"rewards/chosen": -4.1875,
"rewards/margins": 2.0,
"rewards/rejected": -6.1875,
"step": 320
},
{
"epoch": 0.34554973821989526,
"grad_norm": 2.9860809907477717,
"learning_rate": 4.996568520039565e-05,
"logits/chosen": -10.8125,
"logits/rejected": -10.0625,
"logps/chosen": -752.0,
"logps/rejected": -1160.0,
"loss": 0.4124,
"rewards/accuracies": 0.793749988079071,
"rewards/chosen": -3.359375,
"rewards/margins": 2.203125,
"rewards/rejected": -5.5625,
"step": 330
},
{
"epoch": 0.35602094240837695,
"grad_norm": 3.3123350838882004,
"learning_rate": 4.994787516454973e-05,
"logits/chosen": -8.9375,
"logits/rejected": -8.3125,
"logps/chosen": -736.0,
"logps/rejected": -1032.0,
"loss": 0.4521,
"rewards/accuracies": 0.8187500238418579,
"rewards/chosen": -3.40625,
"rewards/margins": 1.7578125,
"rewards/rejected": -5.15625,
"step": 340
},
{
"epoch": 0.36649214659685864,
"grad_norm": 3.6220493645258722,
"learning_rate": 4.992636035140273e-05,
"logits/chosen": -11.1875,
"logits/rejected": -10.6875,
"logps/chosen": -672.0,
"logps/rejected": -1104.0,
"loss": 0.453,
"rewards/accuracies": 0.8187500238418579,
"rewards/chosen": -3.0625,
"rewards/margins": 1.9921875,
"rewards/rejected": -5.0625,
"step": 350
},
{
"epoch": 0.3769633507853403,
"grad_norm": 3.024998451027464,
"learning_rate": 4.99011439559198e-05,
"logits/chosen": -10.4375,
"logits/rejected": -9.5,
"logps/chosen": -832.0,
"logps/rejected": -1168.0,
"loss": 0.4398,
"rewards/accuracies": 0.768750011920929,
"rewards/chosen": -4.28125,
"rewards/margins": 1.9453125,
"rewards/rejected": -6.21875,
"step": 360
},
{
"epoch": 0.387434554973822,
"grad_norm": 2.944100623023382,
"learning_rate": 4.987222972275369e-05,
"logits/chosen": -9.125,
"logits/rejected": -8.0625,
"logps/chosen": -900.0,
"logps/rejected": -1280.0,
"loss": 0.3701,
"rewards/accuracies": 0.862500011920929,
"rewards/chosen": -5.0625,
"rewards/margins": 2.203125,
"rewards/rejected": -7.25,
"step": 370
},
{
"epoch": 0.39790575916230364,
"grad_norm": 2.1733594797976847,
"learning_rate": 4.983962194568872e-05,
"logits/chosen": -9.3125,
"logits/rejected": -8.375,
"logps/chosen": -856.0,
"logps/rejected": -1416.0,
"loss": 0.4489,
"rewards/accuracies": 0.8125,
"rewards/chosen": -4.46875,
"rewards/margins": 2.40625,
"rewards/rejected": -6.875,
"step": 380
},
{
"epoch": 0.4083769633507853,
"grad_norm": 2.764564967841632,
"learning_rate": 4.9803325467003084e-05,
"logits/chosen": -10.5,
"logits/rejected": -9.625,
"logps/chosen": -832.0,
"logps/rejected": -1312.0,
"loss": 0.4218,
"rewards/accuracies": 0.78125,
"rewards/chosen": -4.15625,
"rewards/margins": 2.25,
"rewards/rejected": -6.40625,
"step": 390
},
{
"epoch": 0.418848167539267,
"grad_norm": 2.878701058404269,
"learning_rate": 4.9763345676749816e-05,
"logits/chosen": -9.125,
"logits/rejected": -8.9375,
"logps/chosen": -832.0,
"logps/rejected": -1136.0,
"loss": 0.4641,
"rewards/accuracies": 0.768750011920929,
"rewards/chosen": -4.4375,
"rewards/margins": 1.59375,
"rewards/rejected": -6.0,
"step": 400
},
{
"epoch": 0.418848167539267,
"eval_logits/chosen": -10.6875,
"eval_logits/rejected": -9.8125,
"eval_logps/chosen": -680.0,
"eval_logps/rejected": -708.0,
"eval_loss": 0.7597812414169312,
"eval_rewards/accuracies": 0.626953125,
"eval_rewards/chosen": -3.59375,
"eval_rewards/margins": 0.609375,
"eval_rewards/rejected": -4.21875,
"eval_runtime": 58.0275,
"eval_samples_per_second": 34.466,
"eval_steps_per_second": 0.551,
"step": 400
},
{
"epoch": 0.4293193717277487,
"grad_norm": 2.7467231444747,
"learning_rate": 4.9719688511956374e-05,
"logits/chosen": -12.3125,
"logits/rejected": -11.4375,
"logps/chosen": -740.0,
"logps/rejected": -1120.0,
"loss": 0.4219,
"rewards/accuracies": 0.831250011920929,
"rewards/chosen": -3.15625,
"rewards/margins": 1.9453125,
"rewards/rejected": -5.09375,
"step": 410
},
{
"epoch": 0.4397905759162304,
"grad_norm": 2.6877513768315153,
"learning_rate": 4.967236045574294e-05,
"logits/chosen": -10.25,
"logits/rejected": -9.4375,
"logps/chosen": -904.0,
"logps/rejected": -1360.0,
"loss": 0.4096,
"rewards/accuracies": 0.831250011920929,
"rewards/chosen": -5.0,
"rewards/margins": 2.515625,
"rewards/rejected": -7.5,
"step": 420
},
{
"epoch": 0.450261780104712,
"grad_norm": 2.4541402615722943,
"learning_rate": 4.96213685363597e-05,
"logits/chosen": -10.0625,
"logits/rejected": -9.4375,
"logps/chosen": -828.0,
"logps/rejected": -1232.0,
"loss": 0.4047,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": -4.34375,
"rewards/margins": 2.046875,
"rewards/rejected": -6.375,
"step": 430
},
{
"epoch": 0.4607329842931937,
"grad_norm": 8.480851620621909,
"learning_rate": 4.956672032614315e-05,
"logits/chosen": -11.125,
"logits/rejected": -10.125,
"logps/chosen": -868.0,
"logps/rejected": -1328.0,
"loss": 0.4771,
"rewards/accuracies": 0.856249988079071,
"rewards/chosen": -4.53125,
"rewards/margins": 2.296875,
"rewards/rejected": -6.84375,
"step": 440
},
{
"epoch": 0.4712041884816754,
"grad_norm": 3.3429782910739307,
"learning_rate": 4.950842394039157e-05,
"logits/chosen": -10.9375,
"logits/rejected": -9.875,
"logps/chosen": -944.0,
"logps/rejected": -1432.0,
"loss": 0.384,
"rewards/accuracies": 0.793749988079071,
"rewards/chosen": -5.5625,
"rewards/margins": 2.859375,
"rewards/rejected": -8.4375,
"step": 450
},
{
"epoch": 0.4816753926701571,
"grad_norm": 2.577537832835451,
"learning_rate": 4.944648803615994e-05,
"logits/chosen": -13.6875,
"logits/rejected": -12.5625,
"logps/chosen": -800.0,
"logps/rejected": -1344.0,
"loss": 0.373,
"rewards/accuracies": 0.8687499761581421,
"rewards/chosen": -3.921875,
"rewards/margins": 2.78125,
"rewards/rejected": -6.71875,
"step": 460
},
{
"epoch": 0.49214659685863876,
"grad_norm": 2.285069442856761,
"learning_rate": 4.938092181097433e-05,
"logits/chosen": -12.125,
"logits/rejected": -11.1875,
"logps/chosen": -836.0,
"logps/rejected": -1320.0,
"loss": 0.3602,
"rewards/accuracies": 0.856249988079071,
"rewards/chosen": -4.3125,
"rewards/margins": 3.0,
"rewards/rejected": -7.3125,
"step": 470
},
{
"epoch": 0.5026178010471204,
"grad_norm": 3.744947560739054,
"learning_rate": 4.931173500146607e-05,
"logits/chosen": -10.0,
"logits/rejected": -9.125,
"logps/chosen": -972.0,
"logps/rejected": -1648.0,
"loss": 0.3695,
"rewards/accuracies": 0.84375,
"rewards/chosen": -5.9375,
"rewards/margins": 4.3125,
"rewards/rejected": -10.25,
"step": 480
},
{
"epoch": 0.5130890052356021,
"grad_norm": 2.4031196763833784,
"learning_rate": 4.923893788192582e-05,
"logits/chosen": -11.9375,
"logits/rejected": -10.6875,
"logps/chosen": -772.0,
"logps/rejected": -1248.0,
"loss": 0.4325,
"rewards/accuracies": 0.7562500238418579,
"rewards/chosen": -3.703125,
"rewards/margins": 2.828125,
"rewards/rejected": -6.53125,
"step": 490
},
{
"epoch": 0.5235602094240838,
"grad_norm": 2.592865431371261,
"learning_rate": 4.9162541262777906e-05,
"logits/chosen": -10.9375,
"logits/rejected": -9.875,
"logps/chosen": -800.0,
"logps/rejected": -1184.0,
"loss": 0.3819,
"rewards/accuracies": 0.8125,
"rewards/chosen": -3.8125,
"rewards/margins": 2.25,
"rewards/rejected": -6.0625,
"step": 500
},
{
"epoch": 0.5235602094240838,
"eval_logits/chosen": -9.125,
"eval_logits/rejected": -7.84375,
"eval_logps/chosen": -828.0,
"eval_logps/rejected": -872.0,
"eval_loss": 0.8648124933242798,
"eval_rewards/accuracies": 0.607421875,
"eval_rewards/chosen": -5.09375,
"eval_rewards/margins": 0.73828125,
"eval_rewards/rejected": -5.84375,
"eval_runtime": 58.1009,
"eval_samples_per_second": 34.423,
"eval_steps_per_second": 0.551,
"step": 500
},
{
"epoch": 0.5340314136125655,
"grad_norm": 3.0110293202441127,
"learning_rate": 4.90825564889749e-05,
"logits/chosen": -11.25,
"logits/rejected": -10.5625,
"logps/chosen": -956.0,
"logps/rejected": -1304.0,
"loss": 0.4049,
"rewards/accuracies": 0.78125,
"rewards/chosen": -5.15625,
"rewards/margins": 2.4375,
"rewards/rejected": -7.59375,
"step": 510
},
{
"epoch": 0.5445026178010471,
"grad_norm": 2.0340669281367276,
"learning_rate": 4.899899543831288e-05,
"logits/chosen": -13.75,
"logits/rejected": -12.8125,
"logps/chosen": -944.0,
"logps/rejected": -1336.0,
"loss": 0.3994,
"rewards/accuracies": 0.7875000238418579,
"rewards/chosen": -5.09375,
"rewards/margins": 2.8125,
"rewards/rejected": -7.90625,
"step": 520
},
{
"epoch": 0.5549738219895288,
"grad_norm": 2.7953601723688077,
"learning_rate": 4.891187051966765e-05,
"logits/chosen": -14.25,
"logits/rejected": -13.375,
"logps/chosen": -904.0,
"logps/rejected": -1256.0,
"loss": 0.4259,
"rewards/accuracies": 0.84375,
"rewards/chosen": -4.75,
"rewards/margins": 2.234375,
"rewards/rejected": -7.0,
"step": 530
},
{
"epoch": 0.5654450261780105,
"grad_norm": 2.5496327215392003,
"learning_rate": 4.882119467115191e-05,
"logits/chosen": -12.0,
"logits/rejected": -11.125,
"logps/chosen": -896.0,
"logps/rejected": -1312.0,
"loss": 0.3684,
"rewards/accuracies": 0.824999988079071,
"rewards/chosen": -4.9375,
"rewards/margins": 2.453125,
"rewards/rejected": -7.375,
"step": 540
},
{
"epoch": 0.5759162303664922,
"grad_norm": 2.7273105698314297,
"learning_rate": 4.872698135819402e-05,
"logits/chosen": -10.8125,
"logits/rejected": -9.5,
"logps/chosen": -904.0,
"logps/rejected": -1368.0,
"loss": 0.4103,
"rewards/accuracies": 0.8125,
"rewards/chosen": -5.09375,
"rewards/margins": 2.265625,
"rewards/rejected": -7.34375,
"step": 550
},
{
"epoch": 0.5863874345549738,
"grad_norm": 3.4110420331727327,
"learning_rate": 4.862924457153831e-05,
"logits/chosen": -10.375,
"logits/rejected": -8.75,
"logps/chosen": -772.0,
"logps/rejected": -1264.0,
"loss": 0.4044,
"rewards/accuracies": 0.824999988079071,
"rewards/chosen": -4.34375,
"rewards/margins": 2.640625,
"rewards/rejected": -7.0,
"step": 560
},
{
"epoch": 0.5968586387434555,
"grad_norm": 2.2212758447762853,
"learning_rate": 4.8527998825167484e-05,
"logits/chosen": -9.0625,
"logits/rejected": -8.375,
"logps/chosen": -932.0,
"logps/rejected": -1440.0,
"loss": 0.3597,
"rewards/accuracies": 0.824999988079071,
"rewards/chosen": -5.5625,
"rewards/margins": 2.96875,
"rewards/rejected": -8.5,
"step": 570
},
{
"epoch": 0.6073298429319371,
"grad_norm": 2.2008674200730027,
"learning_rate": 4.842325915414728e-05,
"logits/chosen": -11.5,
"logits/rejected": -10.4375,
"logps/chosen": -764.0,
"logps/rejected": -1128.0,
"loss": 0.3619,
"rewards/accuracies": 0.862500011920929,
"rewards/chosen": -4.03125,
"rewards/margins": 2.21875,
"rewards/rejected": -6.28125,
"step": 580
},
{
"epoch": 0.6178010471204188,
"grad_norm": 3.4441444529122323,
"learning_rate": 4.831504111239373e-05,
"logits/chosen": -9.3125,
"logits/rejected": -8.4375,
"logps/chosen": -996.0,
"logps/rejected": -1624.0,
"loss": 0.3718,
"rewards/accuracies": 0.875,
"rewards/chosen": -5.78125,
"rewards/margins": 3.5625,
"rewards/rejected": -9.375,
"step": 590
},
{
"epoch": 0.6282722513089005,
"grad_norm": 3.806302960457914,
"learning_rate": 4.8203360770363425e-05,
"logits/chosen": -11.0625,
"logits/rejected": -9.8125,
"logps/chosen": -916.0,
"logps/rejected": -1392.0,
"loss": 0.4052,
"rewards/accuracies": 0.8187500238418579,
"rewards/chosen": -5.09375,
"rewards/margins": 2.9375,
"rewards/rejected": -8.0625,
"step": 600
},
{
"epoch": 0.6282722513089005,
"eval_logits/chosen": -10.625,
"eval_logits/rejected": -9.3125,
"eval_logps/chosen": -836.0,
"eval_logps/rejected": -880.0,
"eval_loss": 0.8810781240463257,
"eval_rewards/accuracies": 0.6015625,
"eval_rewards/chosen": -5.1875,
"eval_rewards/margins": 0.74609375,
"eval_rewards/rejected": -5.9375,
"eval_runtime": 58.2046,
"eval_samples_per_second": 34.362,
"eval_steps_per_second": 0.55,
"step": 600
},
{
"epoch": 0.6387434554973822,
"grad_norm": 3.2140734398889665,
"learning_rate": 4.808823471266702e-05,
"logits/chosen": -10.75,
"logits/rejected": -9.5,
"logps/chosen": -928.0,
"logps/rejected": -1440.0,
"loss": 0.3796,
"rewards/accuracies": 0.8187500238418579,
"rewards/chosen": -5.34375,
"rewards/margins": 3.015625,
"rewards/rejected": -8.375,
"step": 610
},
{
"epoch": 0.6492146596858639,
"grad_norm": 2.7471351853583386,
"learning_rate": 4.796968003560639e-05,
"logits/chosen": -11.75,
"logits/rejected": -10.6875,
"logps/chosen": -908.0,
"logps/rejected": -1352.0,
"loss": 0.4275,
"rewards/accuracies": 0.831250011920929,
"rewards/chosen": -5.03125,
"rewards/margins": 2.328125,
"rewards/rejected": -7.375,
"step": 620
},
{
"epoch": 0.6596858638743456,
"grad_norm": 2.0471606564299822,
"learning_rate": 4.7847714344635864e-05,
"logits/chosen": -12.0625,
"logits/rejected": -10.8125,
"logps/chosen": -836.0,
"logps/rejected": -1192.0,
"loss": 0.4019,
"rewards/accuracies": 0.78125,
"rewards/chosen": -4.5,
"rewards/margins": 2.171875,
"rewards/rejected": -6.6875,
"step": 630
},
{
"epoch": 0.6701570680628273,
"grad_norm": 2.978874738269952,
"learning_rate": 4.772235575174776e-05,
"logits/chosen": -9.375,
"logits/rejected": -8.1875,
"logps/chosen": -976.0,
"logps/rejected": -1376.0,
"loss": 0.3906,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -5.4375,
"rewards/margins": 2.515625,
"rewards/rejected": -7.96875,
"step": 640
},
{
"epoch": 0.680628272251309,
"grad_norm": 2.301538130098716,
"learning_rate": 4.7593622872782754e-05,
"logits/chosen": -9.0,
"logits/rejected": -8.125,
"logps/chosen": -804.0,
"logps/rejected": -1160.0,
"loss": 0.4023,
"rewards/accuracies": 0.7875000238418579,
"rewards/chosen": -4.1875,
"rewards/margins": 2.0625,
"rewards/rejected": -6.25,
"step": 650
},
{
"epoch": 0.6910994764397905,
"grad_norm": 2.1972619524514827,
"learning_rate": 4.746153482466542e-05,
"logits/chosen": -10.5,
"logits/rejected": -9.1875,
"logps/chosen": -804.0,
"logps/rejected": -1152.0,
"loss": 0.3982,
"rewards/accuracies": 0.8187500238418579,
"rewards/chosen": -4.1875,
"rewards/margins": 2.0,
"rewards/rejected": -6.21875,
"step": 660
},
{
"epoch": 0.7015706806282722,
"grad_norm": 4.230181847181701,
"learning_rate": 4.732611122256537e-05,
"logits/chosen": -8.875,
"logits/rejected": -7.71875,
"logps/chosen": -1048.0,
"logps/rejected": -1424.0,
"loss": 0.3639,
"rewards/accuracies": 0.8187500238418579,
"rewards/chosen": -6.09375,
"rewards/margins": 2.625,
"rewards/rejected": -8.75,
"step": 670
},
{
"epoch": 0.7120418848167539,
"grad_norm": 2.538746728233518,
"learning_rate": 4.718737217698435e-05,
"logits/chosen": -8.5625,
"logits/rejected": -7.65625,
"logps/chosen": -868.0,
"logps/rejected": -1344.0,
"loss": 0.4231,
"rewards/accuracies": 0.8374999761581421,
"rewards/chosen": -4.875,
"rewards/margins": 2.59375,
"rewards/rejected": -7.4375,
"step": 680
},
{
"epoch": 0.7225130890052356,
"grad_norm": 2.5146784410118865,
"learning_rate": 4.704533829076982e-05,
"logits/chosen": -8.8125,
"logits/rejected": -7.96875,
"logps/chosen": -868.0,
"logps/rejected": -1296.0,
"loss": 0.3832,
"rewards/accuracies": 0.8062499761581421,
"rewards/chosen": -4.71875,
"rewards/margins": 2.15625,
"rewards/rejected": -6.875,
"step": 690
},
{
"epoch": 0.7329842931937173,
"grad_norm": 2.8090381226600427,
"learning_rate": 4.6900030656055455e-05,
"logits/chosen": -8.75,
"logits/rejected": -7.375,
"logps/chosen": -888.0,
"logps/rejected": -1256.0,
"loss": 0.397,
"rewards/accuracies": 0.8125,
"rewards/chosen": -4.78125,
"rewards/margins": 2.265625,
"rewards/rejected": -7.03125,
"step": 700
},
{
"epoch": 0.7329842931937173,
"eval_logits/chosen": -9.125,
"eval_logits/rejected": -7.5,
"eval_logps/chosen": -780.0,
"eval_logps/rejected": -824.0,
"eval_loss": 0.7825937271118164,
"eval_rewards/accuracies": 0.64453125,
"eval_rewards/chosen": -4.59375,
"eval_rewards/margins": 0.7578125,
"eval_rewards/rejected": -5.34375,
"eval_runtime": 58.0834,
"eval_samples_per_second": 34.433,
"eval_steps_per_second": 0.551,
"step": 700
},
{
"epoch": 0.743455497382199,
"grad_norm": 2.3024394544522653,
"learning_rate": 4.6751470851128906e-05,
"logits/chosen": -10.5625,
"logits/rejected": -8.8125,
"logps/chosen": -852.0,
"logps/rejected": -1296.0,
"loss": 0.3865,
"rewards/accuracies": 0.8125,
"rewards/chosen": -4.1875,
"rewards/margins": 1.921875,
"rewards/rejected": -6.09375,
"step": 710
},
{
"epoch": 0.7539267015706806,
"grad_norm": 2.721268194844064,
"learning_rate": 4.6599680937227454e-05,
"logits/chosen": -8.8125,
"logits/rejected": -7.25,
"logps/chosen": -944.0,
"logps/rejected": -1432.0,
"loss": 0.3884,
"rewards/accuracies": 0.7875000238418579,
"rewards/chosen": -5.1875,
"rewards/margins": 2.234375,
"rewards/rejected": -7.4375,
"step": 720
},
{
"epoch": 0.7643979057591623,
"grad_norm": 2.55939754842322,
"learning_rate": 4.644468345526183e-05,
"logits/chosen": -10.6875,
"logits/rejected": -9.3125,
"logps/chosen": -832.0,
"logps/rejected": -1344.0,
"loss": 0.3805,
"rewards/accuracies": 0.831250011920929,
"rewards/chosen": -4.1875,
"rewards/margins": 2.71875,
"rewards/rejected": -6.90625,
"step": 730
},
{
"epoch": 0.774869109947644,
"grad_norm": 2.449408063388354,
"learning_rate": 4.628650142246894e-05,
"logits/chosen": -9.625,
"logits/rejected": -7.84375,
"logps/chosen": -872.0,
"logps/rejected": -1464.0,
"loss": 0.3282,
"rewards/accuracies": 0.8374999761581421,
"rewards/chosen": -4.96875,
"rewards/margins": 2.96875,
"rewards/rejected": -7.9375,
"step": 740
},
{
"epoch": 0.7853403141361257,
"grad_norm": 2.3801209913418444,
"learning_rate": 4.6125158328993765e-05,
"logits/chosen": -10.4375,
"logits/rejected": -8.9375,
"logps/chosen": -836.0,
"logps/rejected": -1272.0,
"loss": 0.3625,
"rewards/accuracies": 0.84375,
"rewards/chosen": -4.5,
"rewards/margins": 2.203125,
"rewards/rejected": -6.6875,
"step": 750
},
{
"epoch": 0.7958115183246073,
"grad_norm": 4.0396363478970025,
"learning_rate": 4.5960678134401064e-05,
"logits/chosen": -11.0,
"logits/rejected": -9.875,
"logps/chosen": -912.0,
"logps/rejected": -1336.0,
"loss": 0.3628,
"rewards/accuracies": 0.831250011920929,
"rewards/chosen": -5.09375,
"rewards/margins": 2.34375,
"rewards/rejected": -7.4375,
"step": 760
},
{
"epoch": 0.806282722513089,
"grad_norm": 3.432009297437936,
"learning_rate": 4.579308526411734e-05,
"logits/chosen": -10.25,
"logits/rejected": -9.25,
"logps/chosen": -944.0,
"logps/rejected": -1624.0,
"loss": 0.3766,
"rewards/accuracies": 0.8374999761581421,
"rewards/chosen": -5.53125,
"rewards/margins": 3.5,
"rewards/rejected": -9.0625,
"step": 770
},
{
"epoch": 0.8167539267015707,
"grad_norm": 4.625077395305937,
"learning_rate": 4.562240460580367e-05,
"logits/chosen": -10.75,
"logits/rejected": -9.75,
"logps/chosen": -868.0,
"logps/rejected": -1400.0,
"loss": 0.3583,
"rewards/accuracies": 0.918749988079071,
"rewards/chosen": -4.78125,
"rewards/margins": 3.015625,
"rewards/rejected": -7.8125,
"step": 780
},
{
"epoch": 0.8272251308900523,
"grad_norm": 2.1421911255049575,
"learning_rate": 4.544866150565985e-05,
"logits/chosen": -10.5625,
"logits/rejected": -9.0625,
"logps/chosen": -836.0,
"logps/rejected": -1328.0,
"loss": 0.3671,
"rewards/accuracies": 0.824999988079071,
"rewards/chosen": -4.40625,
"rewards/margins": 2.625,
"rewards/rejected": -7.03125,
"step": 790
},
{
"epoch": 0.837696335078534,
"grad_norm": 2.2520192600655484,
"learning_rate": 4.527188176466051e-05,
"logits/chosen": -8.5,
"logits/rejected": -7.53125,
"logps/chosen": -924.0,
"logps/rejected": -1232.0,
"loss": 0.3853,
"rewards/accuracies": 0.831250011920929,
"rewards/chosen": -4.96875,
"rewards/margins": 1.9296875,
"rewards/rejected": -6.875,
"step": 800
},
{
"epoch": 0.837696335078534,
"eval_logits/chosen": -7.3125,
"eval_logits/rejected": -5.96875,
"eval_logps/chosen": -904.0,
"eval_logps/rejected": -948.0,
"eval_loss": 0.8263437747955322,
"eval_rewards/accuracies": 0.6328125,
"eval_rewards/chosen": -5.84375,
"eval_rewards/margins": 0.74609375,
"eval_rewards/rejected": -6.59375,
"eval_runtime": 58.0963,
"eval_samples_per_second": 34.426,
"eval_steps_per_second": 0.551,
"step": 800
},
{
"epoch": 0.8481675392670157,
"grad_norm": 2.449013732161559,
"learning_rate": 4.509209163472357e-05,
"logits/chosen": -10.8125,
"logits/rejected": -9.6875,
"logps/chosen": -940.0,
"logps/rejected": -1376.0,
"loss": 0.364,
"rewards/accuracies": 0.8062499761581421,
"rewards/chosen": -5.25,
"rewards/margins": 2.53125,
"rewards/rejected": -7.78125,
"step": 810
},
{
"epoch": 0.8586387434554974,
"grad_norm": 1.5768361401709878,
"learning_rate": 4.4909317814811896e-05,
"logits/chosen": -13.3125,
"logits/rejected": -12.1875,
"logps/chosen": -788.0,
"logps/rejected": -1256.0,
"loss": 0.3818,
"rewards/accuracies": 0.8062499761581421,
"rewards/chosen": -3.90625,
"rewards/margins": 2.390625,
"rewards/rejected": -6.28125,
"step": 820
},
{
"epoch": 0.8691099476439791,
"grad_norm": 2.328485074111117,
"learning_rate": 4.472358744696845e-05,
"logits/chosen": -12.125,
"logits/rejected": -10.25,
"logps/chosen": -864.0,
"logps/rejected": -1336.0,
"loss": 0.3405,
"rewards/accuracies": 0.8687499761581421,
"rewards/chosen": -4.625,
"rewards/margins": 3.015625,
"rewards/rejected": -7.65625,
"step": 830
},
{
"epoch": 0.8795811518324608,
"grad_norm": 5.261036597624319,
"learning_rate": 4.4534928112285645e-05,
"logits/chosen": -8.375,
"logits/rejected": -6.96875,
"logps/chosen": -1072.0,
"logps/rejected": -1640.0,
"loss": 0.3824,
"rewards/accuracies": 0.831250011920929,
"rewards/chosen": -6.59375,
"rewards/margins": 3.515625,
"rewards/rejected": -10.125,
"step": 840
},
{
"epoch": 0.8900523560209425,
"grad_norm": 2.634118752370496,
"learning_rate": 4.43433678268096e-05,
"logits/chosen": -9.625,
"logits/rejected": -7.75,
"logps/chosen": -908.0,
"logps/rejected": -1416.0,
"loss": 0.3547,
"rewards/accuracies": 0.84375,
"rewards/chosen": -5.09375,
"rewards/margins": 2.9375,
"rewards/rejected": -8.0,
"step": 850
},
{
"epoch": 0.900523560209424,
"grad_norm": 2.8331048417442766,
"learning_rate": 4.414893503737968e-05,
"logits/chosen": -9.75,
"logits/rejected": -7.59375,
"logps/chosen": -864.0,
"logps/rejected": -1352.0,
"loss": 0.3458,
"rewards/accuracies": 0.8187500238418579,
"rewards/chosen": -4.9375,
"rewards/margins": 2.5625,
"rewards/rejected": -7.5,
"step": 860
},
{
"epoch": 0.9109947643979057,
"grad_norm": 3.8398249605822112,
"learning_rate": 4.395165861740414e-05,
"logits/chosen": -8.25,
"logits/rejected": -7.28125,
"logps/chosen": -984.0,
"logps/rejected": -1456.0,
"loss": 0.3513,
"rewards/accuracies": 0.793749988079071,
"rewards/chosen": -6.09375,
"rewards/margins": 2.734375,
"rewards/rejected": -8.8125,
"step": 870
},
{
"epoch": 0.9214659685863874,
"grad_norm": 2.180316849305192,
"learning_rate": 4.375156786257241e-05,
"logits/chosen": -9.8125,
"logits/rejected": -8.625,
"logps/chosen": -1056.0,
"logps/rejected": -1448.0,
"loss": 0.4113,
"rewards/accuracies": 0.824999988079071,
"rewards/chosen": -6.1875,
"rewards/margins": 2.359375,
"rewards/rejected": -8.5625,
"step": 880
},
{
"epoch": 0.9319371727748691,
"grad_norm": 2.7467798027876125,
"learning_rate": 4.354869248650466e-05,
"logits/chosen": -10.1875,
"logits/rejected": -8.6875,
"logps/chosen": -892.0,
"logps/rejected": -1296.0,
"loss": 0.3689,
"rewards/accuracies": 0.856249988079071,
"rewards/chosen": -4.84375,
"rewards/margins": 1.9609375,
"rewards/rejected": -6.8125,
"step": 890
},
{
"epoch": 0.9424083769633508,
"grad_norm": 4.563086230264344,
"learning_rate": 4.334306261633931e-05,
"logits/chosen": -6.125,
"logits/rejected": -4.65625,
"logps/chosen": -1080.0,
"logps/rejected": -1624.0,
"loss": 0.3438,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": -6.5,
"rewards/margins": 2.921875,
"rewards/rejected": -9.4375,
"step": 900
},
{
"epoch": 0.9424083769633508,
"eval_logits/chosen": -4.21875,
"eval_logits/rejected": -2.5,
"eval_logps/chosen": -1080.0,
"eval_logps/rejected": -1168.0,
"eval_loss": 1.0278280973434448,
"eval_rewards/accuracies": 0.623046875,
"eval_rewards/chosen": -7.59375,
"eval_rewards/margins": 1.234375,
"eval_rewards/rejected": -8.8125,
"eval_runtime": 58.0713,
"eval_samples_per_second": 34.44,
"eval_steps_per_second": 0.551,
"step": 900
},
{
"epoch": 0.9528795811518325,
"grad_norm": 2.9890514343476804,
"learning_rate": 4.313470878825911e-05,
"logits/chosen": -7.71875,
"logits/rejected": -5.96875,
"logps/chosen": -984.0,
"logps/rejected": -1496.0,
"loss": 0.3439,
"rewards/accuracies": 0.875,
"rewards/chosen": -5.78125,
"rewards/margins": 3.3125,
"rewards/rejected": -9.0625,
"step": 910
},
{
"epoch": 0.9633507853403142,
"grad_norm": 2.3424395250048717,
"learning_rate": 4.292366194295652e-05,
"logits/chosen": -7.71875,
"logits/rejected": -6.71875,
"logps/chosen": -864.0,
"logps/rejected": -1360.0,
"loss": 0.3476,
"rewards/accuracies": 0.856249988079071,
"rewards/chosen": -4.59375,
"rewards/margins": 2.65625,
"rewards/rejected": -7.25,
"step": 920
},
{
"epoch": 0.9738219895287958,
"grad_norm": 2.4625005168986287,
"learning_rate": 4.2709953421038964e-05,
"logits/chosen": -6.03125,
"logits/rejected": -5.1875,
"logps/chosen": -888.0,
"logps/rejected": -1328.0,
"loss": 0.3615,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -5.0625,
"rewards/margins": 2.34375,
"rewards/rejected": -7.375,
"step": 930
},
{
"epoch": 0.9842931937172775,
"grad_norm": 2.8075957960853977,
"learning_rate": 4.249361495837478e-05,
"logits/chosen": -7.40625,
"logits/rejected": -6.125,
"logps/chosen": -884.0,
"logps/rejected": -1328.0,
"loss": 0.3992,
"rewards/accuracies": 0.7875000238418579,
"rewards/chosen": -4.90625,
"rewards/margins": 2.4375,
"rewards/rejected": -7.3125,
"step": 940
},
{
"epoch": 0.9947643979057592,
"grad_norm": 2.112027052383725,
"learning_rate": 4.227467868138035e-05,
"logits/chosen": -9.1875,
"logits/rejected": -7.71875,
"logps/chosen": -824.0,
"logps/rejected": -1336.0,
"loss": 0.3351,
"rewards/accuracies": 0.8187500238418579,
"rewards/chosen": -4.1875,
"rewards/margins": 2.671875,
"rewards/rejected": -6.875,
"step": 950
},
{
"epoch": 1.0052356020942408,
"grad_norm": 1.7974965044496571,
"learning_rate": 4.205317710224938e-05,
"logits/chosen": -9.4375,
"logits/rejected": -8.0,
"logps/chosen": -868.0,
"logps/rejected": -1384.0,
"loss": 0.2713,
"rewards/accuracies": 0.8812500238418579,
"rewards/chosen": -4.5,
"rewards/margins": 3.3125,
"rewards/rejected": -7.84375,
"step": 960
},
{
"epoch": 1.0157068062827226,
"grad_norm": 1.0480920801900242,
"learning_rate": 4.1829143114124733e-05,
"logits/chosen": -8.375,
"logits/rejected": -6.78125,
"logps/chosen": -1016.0,
"logps/rejected": -1552.0,
"loss": 0.1308,
"rewards/accuracies": 0.9437500238418579,
"rewards/chosen": -6.0,
"rewards/margins": 4.3125,
"rewards/rejected": -10.3125,
"step": 970
},
{
"epoch": 1.0261780104712042,
"grad_norm": 2.8021821583883155,
"learning_rate": 4.160260998621387e-05,
"logits/chosen": -9.8125,
"logits/rejected": -7.40625,
"logps/chosen": -1104.0,
"logps/rejected": -1768.0,
"loss": 0.1152,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -6.5625,
"rewards/margins": 5.3125,
"rewards/rejected": -11.875,
"step": 980
},
{
"epoch": 1.036649214659686,
"grad_norm": 1.3473810935358794,
"learning_rate": 4.137361135884824e-05,
"logits/chosen": -8.375,
"logits/rejected": -6.75,
"logps/chosen": -1128.0,
"logps/rejected": -1928.0,
"loss": 0.0984,
"rewards/accuracies": 0.9437500238418579,
"rewards/chosen": -7.15625,
"rewards/margins": 5.71875,
"rewards/rejected": -12.875,
"step": 990
},
{
"epoch": 1.0471204188481675,
"grad_norm": 0.9868378244907727,
"learning_rate": 4.114218123848777e-05,
"logits/chosen": -8.6875,
"logits/rejected": -6.78125,
"logps/chosen": -1200.0,
"logps/rejected": -2008.0,
"loss": 0.0879,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -8.125,
"rewards/margins": 6.125,
"rewards/rejected": -14.25,
"step": 1000
},
{
"epoch": 1.0471204188481675,
"eval_logits/chosen": -8.5,
"eval_logits/rejected": -6.625,
"eval_logps/chosen": -1256.0,
"eval_logps/rejected": -1368.0,
"eval_loss": 1.2818750143051147,
"eval_rewards/accuracies": 0.60546875,
"eval_rewards/chosen": -9.375,
"eval_rewards/margins": 1.4375,
"eval_rewards/rejected": -10.8125,
"eval_runtime": 58.5401,
"eval_samples_per_second": 34.165,
"eval_steps_per_second": 0.547,
"step": 1000
},
{
"epoch": 1.057591623036649,
"grad_norm": 2.3045762509797156,
"learning_rate": 4.0908353992670824e-05,
"logits/chosen": -10.3125,
"logits/rejected": -7.71875,
"logps/chosen": -1096.0,
"logps/rejected": -1784.0,
"loss": 0.1203,
"rewards/accuracies": 0.96875,
"rewards/chosen": -7.03125,
"rewards/margins": 5.71875,
"rewards/rejected": -12.75,
"step": 1010
},
{
"epoch": 1.068062827225131,
"grad_norm": 1.9733278289270342,
"learning_rate": 4.067216434491059e-05,
"logits/chosen": -9.3125,
"logits/rejected": -7.59375,
"logps/chosen": -1104.0,
"logps/rejected": -1808.0,
"loss": 0.1101,
"rewards/accuracies": 0.9375,
"rewards/chosen": -7.03125,
"rewards/margins": 5.28125,
"rewards/rejected": -12.3125,
"step": 1020
},
{
"epoch": 1.0785340314136125,
"grad_norm": 1.7918292234449644,
"learning_rate": 4.043364736953866e-05,
"logits/chosen": -8.5625,
"logits/rejected": -6.25,
"logps/chosen": -1104.0,
"logps/rejected": -1808.0,
"loss": 0.1243,
"rewards/accuracies": 0.981249988079071,
"rewards/chosen": -6.59375,
"rewards/margins": 5.09375,
"rewards/rejected": -11.6875,
"step": 1030
},
{
"epoch": 1.0890052356020943,
"grad_norm": 2.61677417351855,
"learning_rate": 4.019283848649643e-05,
"logits/chosen": -8.625,
"logits/rejected": -6.46875,
"logps/chosen": -1088.0,
"logps/rejected": -1776.0,
"loss": 0.1363,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -6.78125,
"rewards/margins": 5.125,
"rewards/rejected": -11.875,
"step": 1040
},
{
"epoch": 1.0994764397905759,
"grad_norm": 1.2876619119555233,
"learning_rate": 3.9949773456075215e-05,
"logits/chosen": -7.28125,
"logits/rejected": -5.96875,
"logps/chosen": -1112.0,
"logps/rejected": -1776.0,
"loss": 0.0901,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -7.4375,
"rewards/margins": 4.5625,
"rewards/rejected": -12.0,
"step": 1050
},
{
"epoch": 1.1099476439790577,
"grad_norm": 1.3248585502931964,
"learning_rate": 3.970448837360585e-05,
"logits/chosen": -9.8125,
"logits/rejected": -7.9375,
"logps/chosen": -1184.0,
"logps/rejected": -1968.0,
"loss": 0.1022,
"rewards/accuracies": 0.9375,
"rewards/chosen": -7.34375,
"rewards/margins": 5.53125,
"rewards/rejected": -12.875,
"step": 1060
},
{
"epoch": 1.1204188481675392,
"grad_norm": 0.5581899433229645,
"learning_rate": 3.945701966409846e-05,
"logits/chosen": -8.25,
"logits/rejected": -6.75,
"logps/chosen": -1216.0,
"logps/rejected": -2208.0,
"loss": 0.1038,
"rewards/accuracies": 0.96875,
"rewards/chosen": -8.125,
"rewards/margins": 7.4375,
"rewards/rejected": -15.5625,
"step": 1070
},
{
"epoch": 1.130890052356021,
"grad_norm": 2.0502778396139356,
"learning_rate": 3.9207404076833374e-05,
"logits/chosen": -7.8125,
"logits/rejected": -5.8125,
"logps/chosen": -1232.0,
"logps/rejected": -2144.0,
"loss": 0.1158,
"rewards/accuracies": 0.96875,
"rewards/chosen": -7.875,
"rewards/margins": 6.875,
"rewards/rejected": -14.75,
"step": 1080
},
{
"epoch": 1.1413612565445026,
"grad_norm": 0.5813371202927899,
"learning_rate": 3.895567867990379e-05,
"logits/chosen": -8.1875,
"logits/rejected": -6.28125,
"logps/chosen": -1240.0,
"logps/rejected": -2096.0,
"loss": 0.0938,
"rewards/accuracies": 0.96875,
"rewards/chosen": -7.71875,
"rewards/margins": 6.4375,
"rewards/rejected": -14.1875,
"step": 1090
},
{
"epoch": 1.1518324607329844,
"grad_norm": 0.7690198357163421,
"learning_rate": 3.870188085471114e-05,
"logits/chosen": -8.5,
"logits/rejected": -6.59375,
"logps/chosen": -1152.0,
"logps/rejected": -1944.0,
"loss": 0.0875,
"rewards/accuracies": 0.96875,
"rewards/chosen": -7.40625,
"rewards/margins": 6.3125,
"rewards/rejected": -13.6875,
"step": 1100
},
{
"epoch": 1.1518324607329844,
"eval_logits/chosen": -5.25,
"eval_logits/rejected": -3.640625,
"eval_logps/chosen": -1352.0,
"eval_logps/rejected": -1464.0,
"eval_loss": 1.2598906755447388,
"eval_rewards/accuracies": 0.615234375,
"eval_rewards/chosen": -10.3125,
"eval_rewards/margins": 1.4609375,
"eval_rewards/rejected": -11.75,
"eval_runtime": 58.1396,
"eval_samples_per_second": 34.4,
"eval_steps_per_second": 0.55,
"step": 1100
},
{
"epoch": 1.162303664921466,
"grad_norm": 2.4538765791904598,
"learning_rate": 3.844604829041395e-05,
"logits/chosen": -6.4375,
"logits/rejected": -4.875,
"logps/chosen": -1256.0,
"logps/rejected": -2064.0,
"loss": 0.0946,
"rewards/accuracies": 0.956250011920929,
"rewards/chosen": -8.8125,
"rewards/margins": 6.71875,
"rewards/rejected": -15.5,
"step": 1110
},
{
"epoch": 1.1727748691099475,
"grad_norm": 2.516263352193615,
"learning_rate": 3.818821897833095e-05,
"logits/chosen": -6.15625,
"logits/rejected": -4.71875,
"logps/chosen": -1296.0,
"logps/rejected": -2064.0,
"loss": 0.108,
"rewards/accuracies": 0.9624999761581421,
"rewards/chosen": -8.75,
"rewards/margins": 6.09375,
"rewards/rejected": -14.8125,
"step": 1120
},
{
"epoch": 1.1832460732984293,
"grad_norm": 2.166374644331338,
"learning_rate": 3.792843120629935e-05,
"logits/chosen": -7.28125,
"logits/rejected": -5.15625,
"logps/chosen": -1184.0,
"logps/rejected": -2032.0,
"loss": 0.1058,
"rewards/accuracies": 0.981249988079071,
"rewards/chosen": -7.90625,
"rewards/margins": 6.15625,
"rewards/rejected": -14.0625,
"step": 1130
},
{
"epoch": 1.193717277486911,
"grad_norm": 3.2997291398627313,
"learning_rate": 3.766672355298907e-05,
"logits/chosen": -9.125,
"logits/rejected": -6.84375,
"logps/chosen": -1168.0,
"logps/rejected": -1992.0,
"loss": 0.089,
"rewards/accuracies": 0.956250011920929,
"rewards/chosen": -8.0625,
"rewards/margins": 6.34375,
"rewards/rejected": -14.4375,
"step": 1140
},
{
"epoch": 1.2041884816753927,
"grad_norm": 3.400189195399067,
"learning_rate": 3.740313488217373e-05,
"logits/chosen": -7.96875,
"logits/rejected": -5.71875,
"logps/chosen": -1344.0,
"logps/rejected": -2272.0,
"loss": 0.0931,
"rewards/accuracies": 0.956250011920929,
"rewards/chosen": -9.625,
"rewards/margins": 7.71875,
"rewards/rejected": -17.375,
"step": 1150
},
{
"epoch": 1.2146596858638743,
"grad_norm": 2.268851645331209,
"learning_rate": 3.7137704336959464e-05,
"logits/chosen": -7.0625,
"logits/rejected": -4.90625,
"logps/chosen": -1248.0,
"logps/rejected": -2144.0,
"loss": 0.1057,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -8.6875,
"rewards/margins": 7.21875,
"rewards/rejected": -15.9375,
"step": 1160
},
{
"epoch": 1.225130890052356,
"grad_norm": 2.564836954188546,
"learning_rate": 3.687047133397201e-05,
"logits/chosen": -8.5,
"logits/rejected": -6.3125,
"logps/chosen": -1112.0,
"logps/rejected": -1936.0,
"loss": 0.1028,
"rewards/accuracies": 0.981249988079071,
"rewards/chosen": -7.21875,
"rewards/margins": 5.90625,
"rewards/rejected": -13.125,
"step": 1170
},
{
"epoch": 1.2356020942408377,
"grad_norm": 2.002756526108138,
"learning_rate": 3.6601475557503415e-05,
"logits/chosen": -8.5,
"logits/rejected": -6.0,
"logps/chosen": -1112.0,
"logps/rejected": -1776.0,
"loss": 0.109,
"rewards/accuracies": 0.9375,
"rewards/chosen": -6.875,
"rewards/margins": 5.46875,
"rewards/rejected": -12.375,
"step": 1180
},
{
"epoch": 1.2460732984293195,
"grad_norm": 1.1157757472563545,
"learning_rate": 3.633075695361882e-05,
"logits/chosen": -7.5625,
"logits/rejected": -5.53125,
"logps/chosen": -1104.0,
"logps/rejected": -1880.0,
"loss": 0.1016,
"rewards/accuracies": 0.96875,
"rewards/chosen": -7.3125,
"rewards/margins": 5.5625,
"rewards/rejected": -12.875,
"step": 1190
},
{
"epoch": 1.256544502617801,
"grad_norm": 2.2654887415135736,
"learning_rate": 3.6058355724224477e-05,
"logits/chosen": -8.0625,
"logits/rejected": -6.1875,
"logps/chosen": -1080.0,
"logps/rejected": -1776.0,
"loss": 0.1119,
"rewards/accuracies": 0.96875,
"rewards/chosen": -6.84375,
"rewards/margins": 5.28125,
"rewards/rejected": -12.125,
"step": 1200
},
{
"epoch": 1.256544502617801,
"eval_logits/chosen": -6.21875,
"eval_logits/rejected": -4.375,
"eval_logps/chosen": -1112.0,
"eval_logps/rejected": -1200.0,
"eval_loss": 1.0712811946868896,
"eval_rewards/accuracies": 0.623046875,
"eval_rewards/chosen": -7.96875,
"eval_rewards/margins": 1.15625,
"eval_rewards/rejected": -9.125,
"eval_runtime": 58.1175,
"eval_samples_per_second": 34.413,
"eval_steps_per_second": 0.551,
"step": 1200
},
{
"epoch": 1.2670157068062826,
"grad_norm": 3.1995736024811086,
"learning_rate": 3.5784312321097744e-05,
"logits/chosen": -7.21875,
"logits/rejected": -5.78125,
"logps/chosen": -1200.0,
"logps/rejected": -1952.0,
"loss": 0.1094,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -7.84375,
"rewards/margins": 5.90625,
"rewards/rejected": -13.75,
"step": 1210
},
{
"epoch": 1.2774869109947644,
"grad_norm": 2.3880207640125812,
"learning_rate": 3.550866743987994e-05,
"logits/chosen": -7.5625,
"logits/rejected": -5.9375,
"logps/chosen": -1240.0,
"logps/rejected": -2008.0,
"loss": 0.1163,
"rewards/accuracies": 0.956250011920929,
"rewards/chosen": -8.125,
"rewards/margins": 6.1875,
"rewards/rejected": -14.3125,
"step": 1220
},
{
"epoch": 1.2879581151832462,
"grad_norm": 1.2877672608272677,
"learning_rate": 3.5231462014033024e-05,
"logits/chosen": -9.6875,
"logits/rejected": -8.25,
"logps/chosen": -1136.0,
"logps/rejected": -1976.0,
"loss": 0.0663,
"rewards/accuracies": 0.96875,
"rewards/chosen": -7.5,
"rewards/margins": 6.15625,
"rewards/rejected": -13.6875,
"step": 1230
},
{
"epoch": 1.2984293193717278,
"grad_norm": 2.9006891802521664,
"learning_rate": 3.495273720876095e-05,
"logits/chosen": -8.0,
"logits/rejected": -6.3125,
"logps/chosen": -1304.0,
"logps/rejected": -2128.0,
"loss": 0.0787,
"rewards/accuracies": 0.9624999761581421,
"rewards/chosen": -8.9375,
"rewards/margins": 6.84375,
"rewards/rejected": -15.8125,
"step": 1240
},
{
"epoch": 1.3089005235602094,
"grad_norm": 3.577493094473363,
"learning_rate": 3.4672534414896616e-05,
"logits/chosen": -8.0625,
"logits/rejected": -5.96875,
"logps/chosen": -1360.0,
"logps/rejected": -2288.0,
"loss": 0.1059,
"rewards/accuracies": 0.9375,
"rewards/chosen": -9.3125,
"rewards/margins": 7.28125,
"rewards/rejected": -16.625,
"step": 1250
},
{
"epoch": 1.3193717277486912,
"grad_norm": 3.0384989047846234,
"learning_rate": 3.439089524275526e-05,
"logits/chosen": -8.25,
"logits/rejected": -6.15625,
"logps/chosen": -1144.0,
"logps/rejected": -1896.0,
"loss": 0.1086,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -7.625,
"rewards/margins": 5.625,
"rewards/rejected": -13.25,
"step": 1260
},
{
"epoch": 1.3298429319371727,
"grad_norm": 1.757767807830736,
"learning_rate": 3.410786151595533e-05,
"logits/chosen": -8.8125,
"logits/rejected": -6.5625,
"logps/chosen": -1080.0,
"logps/rejected": -1656.0,
"loss": 0.103,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -6.96875,
"rewards/margins": 4.90625,
"rewards/rejected": -11.875,
"step": 1270
},
{
"epoch": 1.3403141361256545,
"grad_norm": 2.3487958569696237,
"learning_rate": 3.382347526520765e-05,
"logits/chosen": -8.875,
"logits/rejected": -6.1875,
"logps/chosen": -1184.0,
"logps/rejected": -1872.0,
"loss": 0.0921,
"rewards/accuracies": 0.9624999761581421,
"rewards/chosen": -7.75,
"rewards/margins": 5.53125,
"rewards/rejected": -13.25,
"step": 1280
},
{
"epoch": 1.350785340314136,
"grad_norm": 3.0105813208622387,
"learning_rate": 3.353777872207381e-05,
"logits/chosen": -7.3125,
"logits/rejected": -5.125,
"logps/chosen": -1296.0,
"logps/rejected": -2096.0,
"loss": 0.0977,
"rewards/accuracies": 0.9437500238418579,
"rewards/chosen": -8.875,
"rewards/margins": 6.0,
"rewards/rejected": -14.875,
"step": 1290
},
{
"epoch": 1.3612565445026177,
"grad_norm": 2.594136210328335,
"learning_rate": 3.325081431269475e-05,
"logits/chosen": -5.6875,
"logits/rejected": -4.0625,
"logps/chosen": -1280.0,
"logps/rejected": -2112.0,
"loss": 0.1083,
"rewards/accuracies": 0.9437500238418579,
"rewards/chosen": -9.0625,
"rewards/margins": 6.1875,
"rewards/rejected": -15.25,
"step": 1300
},
{
"epoch": 1.3612565445026177,
"eval_logits/chosen": -5.4375,
"eval_logits/rejected": -3.71875,
"eval_logps/chosen": -1336.0,
"eval_logps/rejected": -1440.0,
"eval_loss": 1.1731406450271606,
"eval_rewards/accuracies": 0.591796875,
"eval_rewards/chosen": -10.1875,
"eval_rewards/margins": 1.296875,
"eval_rewards/rejected": -11.5,
"eval_runtime": 58.157,
"eval_samples_per_second": 34.39,
"eval_steps_per_second": 0.55,
"step": 1300
},
{
"epoch": 1.3717277486910995,
"grad_norm": 2.02241876013486,
"learning_rate": 3.296262465149046e-05,
"logits/chosen": -7.65625,
"logits/rejected": -4.96875,
"logps/chosen": -1256.0,
"logps/rejected": -2048.0,
"loss": 0.0884,
"rewards/accuracies": 0.981249988079071,
"rewards/chosen": -8.3125,
"rewards/margins": 6.3125,
"rewards/rejected": -14.625,
"step": 1310
},
{
"epoch": 1.3821989528795813,
"grad_norm": 0.4926870546495341,
"learning_rate": 3.2673252534831687e-05,
"logits/chosen": -6.46875,
"logits/rejected": -4.0,
"logps/chosen": -1256.0,
"logps/rejected": -2080.0,
"loss": 0.0847,
"rewards/accuracies": 0.9624999761581421,
"rewards/chosen": -8.6875,
"rewards/margins": 6.40625,
"rewards/rejected": -15.0625,
"step": 1320
},
{
"epoch": 1.3926701570680629,
"grad_norm": 2.558741771265552,
"learning_rate": 3.2382740934684696e-05,
"logits/chosen": -5.84375,
"logits/rejected": -3.703125,
"logps/chosen": -1304.0,
"logps/rejected": -2208.0,
"loss": 0.1145,
"rewards/accuracies": 0.9375,
"rewards/chosen": -9.0625,
"rewards/margins": 6.375,
"rewards/rejected": -15.375,
"step": 1330
},
{
"epoch": 1.4031413612565444,
"grad_norm": 1.552014552924826,
"learning_rate": 3.209113299222982e-05,
"logits/chosen": -6.34375,
"logits/rejected": -5.0,
"logps/chosen": -1112.0,
"logps/rejected": -1888.0,
"loss": 0.0984,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -7.71875,
"rewards/margins": 5.46875,
"rewards/rejected": -13.1875,
"step": 1340
},
{
"epoch": 1.4136125654450262,
"grad_norm": 2.124398956704995,
"learning_rate": 3.179847201145505e-05,
"logits/chosen": -6.5625,
"logits/rejected": -4.59375,
"logps/chosen": -1184.0,
"logps/rejected": -2032.0,
"loss": 0.098,
"rewards/accuracies": 0.956250011920929,
"rewards/chosen": -7.65625,
"rewards/margins": 5.9375,
"rewards/rejected": -13.625,
"step": 1350
},
{
"epoch": 1.4240837696335078,
"grad_norm": 1.08454606303427,
"learning_rate": 3.150480145272528e-05,
"logits/chosen": -5.375,
"logits/rejected": -4.0625,
"logps/chosen": -1352.0,
"logps/rejected": -2144.0,
"loss": 0.0861,
"rewards/accuracies": 0.96875,
"rewards/chosen": -9.5,
"rewards/margins": 6.25,
"rewards/rejected": -15.75,
"step": 1360
},
{
"epoch": 1.4345549738219896,
"grad_norm": 1.9978997005808714,
"learning_rate": 3.121016492632848e-05,
"logits/chosen": -6.25,
"logits/rejected": -4.0,
"logps/chosen": -1336.0,
"logps/rejected": -2208.0,
"loss": 0.1025,
"rewards/accuracies": 0.96875,
"rewards/chosen": -9.3125,
"rewards/margins": 6.9375,
"rewards/rejected": -16.25,
"step": 1370
},
{
"epoch": 1.4450261780104712,
"grad_norm": 1.9735697507357144,
"learning_rate": 3.091460618599951e-05,
"logits/chosen": -5.8125,
"logits/rejected": -3.671875,
"logps/chosen": -1264.0,
"logps/rejected": -2208.0,
"loss": 0.1081,
"rewards/accuracies": 0.918749988079071,
"rewards/chosen": -8.6875,
"rewards/margins": 7.375,
"rewards/rejected": -16.125,
"step": 1380
},
{
"epoch": 1.4554973821989527,
"grad_norm": 1.5191892493409722,
"learning_rate": 3.061816912242265e-05,
"logits/chosen": -6.09375,
"logits/rejected": -4.46875,
"logps/chosen": -1256.0,
"logps/rejected": -2096.0,
"loss": 0.0908,
"rewards/accuracies": 0.956250011920929,
"rewards/chosen": -8.375,
"rewards/margins": 6.09375,
"rewards/rejected": -14.5,
"step": 1390
},
{
"epoch": 1.4659685863874345,
"grad_norm": 0.5614126710138799,
"learning_rate": 3.0320897756713785e-05,
"logits/chosen": -6.3125,
"logits/rejected": -5.03125,
"logps/chosen": -1200.0,
"logps/rejected": -2160.0,
"loss": 0.0827,
"rewards/accuracies": 0.96875,
"rewards/chosen": -7.8125,
"rewards/margins": 6.5625,
"rewards/rejected": -14.375,
"step": 1400
},
{
"epoch": 1.4659685863874345,
"eval_logits/chosen": -4.5,
"eval_logits/rejected": -2.609375,
"eval_logps/chosen": -1240.0,
"eval_logps/rejected": -1336.0,
"eval_loss": 1.0476875305175781,
"eval_rewards/accuracies": 0.615234375,
"eval_rewards/chosen": -9.25,
"eval_rewards/margins": 1.25,
"eval_rewards/rejected": -10.5,
"eval_runtime": 58.1464,
"eval_samples_per_second": 34.396,
"eval_steps_per_second": 0.55,
"step": 1400
},
{
"epoch": 1.4764397905759163,
"grad_norm": 2.745200814396308,
"learning_rate": 3.002283623388332e-05,
"logits/chosen": -5.75,
"logits/rejected": -3.84375,
"logps/chosen": -1176.0,
"logps/rejected": -1928.0,
"loss": 0.1027,
"rewards/accuracies": 0.9375,
"rewards/chosen": -8.125,
"rewards/margins": 5.625,
"rewards/rejected": -13.75,
"step": 1410
},
{
"epoch": 1.486910994764398,
"grad_norm": 2.555598207824252,
"learning_rate": 2.9724028816280507e-05,
"logits/chosen": -6.09375,
"logits/rejected": -4.46875,
"logps/chosen": -1200.0,
"logps/rejected": -1872.0,
"loss": 0.1278,
"rewards/accuracies": 0.9375,
"rewards/chosen": -8.1875,
"rewards/margins": 5.25,
"rewards/rejected": -13.4375,
"step": 1420
},
{
"epoch": 1.4973821989528795,
"grad_norm": 0.7131456666237508,
"learning_rate": 2.9424519877020524e-05,
"logits/chosen": -7.875,
"logits/rejected": -5.8125,
"logps/chosen": -1224.0,
"logps/rejected": -2096.0,
"loss": 0.0963,
"rewards/accuracies": 0.956250011920929,
"rewards/chosen": -8.0,
"rewards/margins": 6.0625,
"rewards/rejected": -14.0625,
"step": 1430
},
{
"epoch": 1.5078534031413613,
"grad_norm": 1.707600481216615,
"learning_rate": 2.912435389339504e-05,
"logits/chosen": -5.15625,
"logits/rejected": -2.921875,
"logps/chosen": -1272.0,
"logps/rejected": -2008.0,
"loss": 0.0961,
"rewards/accuracies": 0.956250011920929,
"rewards/chosen": -8.8125,
"rewards/margins": 5.6875,
"rewards/rejected": -14.5,
"step": 1440
},
{
"epoch": 1.518324607329843,
"grad_norm": 2.4378652164158736,
"learning_rate": 2.8823575440267226e-05,
"logits/chosen": -6.09375,
"logits/rejected": -4.875,
"logps/chosen": -1232.0,
"logps/rejected": -2040.0,
"loss": 0.1101,
"rewards/accuracies": 0.9312499761581421,
"rewards/chosen": -8.375,
"rewards/margins": 5.625,
"rewards/rejected": -14.0,
"step": 1450
},
{
"epoch": 1.5287958115183247,
"grad_norm": 2.591287103510408,
"learning_rate": 2.8522229183452465e-05,
"logits/chosen": -6.375,
"logits/rejected": -4.15625,
"logps/chosen": -1168.0,
"logps/rejected": -1880.0,
"loss": 0.1224,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -7.53125,
"rewards/margins": 5.84375,
"rewards/rejected": -13.375,
"step": 1460
},
{
"epoch": 1.5392670157068062,
"grad_norm": 1.4355497204707246,
"learning_rate": 2.8220359873085373e-05,
"logits/chosen": -6.53125,
"logits/rejected": -4.5625,
"logps/chosen": -1176.0,
"logps/rejected": -1960.0,
"loss": 0.091,
"rewards/accuracies": 0.981249988079071,
"rewards/chosen": -7.6875,
"rewards/margins": 5.78125,
"rewards/rejected": -13.5,
"step": 1470
},
{
"epoch": 1.5497382198952878,
"grad_norm": 2.3342297966674295,
"learning_rate": 2.7918012336974388e-05,
"logits/chosen": -6.3125,
"logits/rejected": -4.625,
"logps/chosen": -1240.0,
"logps/rejected": -2008.0,
"loss": 0.093,
"rewards/accuracies": 0.956250011920929,
"rewards/chosen": -8.375,
"rewards/margins": 6.125,
"rewards/rejected": -14.5,
"step": 1480
},
{
"epoch": 1.5602094240837696,
"grad_norm": 2.76625565400065,
"learning_rate": 2.7615231473944815e-05,
"logits/chosen": -5.53125,
"logits/rejected": -3.296875,
"logps/chosen": -1272.0,
"logps/rejected": -2040.0,
"loss": 0.0996,
"rewards/accuracies": 0.9624999761581421,
"rewards/chosen": -8.875,
"rewards/margins": 6.125,
"rewards/rejected": -15.0,
"step": 1490
},
{
"epoch": 1.5706806282722514,
"grad_norm": 2.7831456968564243,
"learning_rate": 2.7312062247171327e-05,
"logits/chosen": -6.8125,
"logits/rejected": -5.25,
"logps/chosen": -1168.0,
"logps/rejected": -1960.0,
"loss": 0.0913,
"rewards/accuracies": 0.956250011920929,
"rewards/chosen": -7.9375,
"rewards/margins": 5.5,
"rewards/rejected": -13.4375,
"step": 1500
},
{
"epoch": 1.5706806282722514,
"eval_logits/chosen": -4.78125,
"eval_logits/rejected": -2.96875,
"eval_logps/chosen": -1248.0,
"eval_logps/rejected": -1352.0,
"eval_loss": 1.0556719303131104,
"eval_rewards/accuracies": 0.626953125,
"eval_rewards/chosen": -9.25,
"eval_rewards/margins": 1.3828125,
"eval_rewards/rejected": -10.625,
"eval_runtime": 58.1134,
"eval_samples_per_second": 34.415,
"eval_steps_per_second": 0.551,
"step": 1500
},
{
"epoch": 1.581151832460733,
"grad_norm": 0.5286441607651609,
"learning_rate": 2.7008549677500876e-05,
"logits/chosen": -7.59375,
"logits/rejected": -5.6875,
"logps/chosen": -1208.0,
"logps/rejected": -1936.0,
"loss": 0.0858,
"rewards/accuracies": 0.981249988079071,
"rewards/chosen": -7.875,
"rewards/margins": 5.625,
"rewards/rejected": -13.5,
"step": 1510
},
{
"epoch": 1.5916230366492146,
"grad_norm": 1.8537005497385242,
"learning_rate": 2.670473883676709e-05,
"logits/chosen": -6.1875,
"logits/rejected": -4.625,
"logps/chosen": -1240.0,
"logps/rejected": -2192.0,
"loss": 0.082,
"rewards/accuracies": 0.9624999761581421,
"rewards/chosen": -8.25,
"rewards/margins": 6.4375,
"rewards/rejected": -14.6875,
"step": 1520
},
{
"epoch": 1.6020942408376964,
"grad_norm": 2.029074339280519,
"learning_rate": 2.6400674841097074e-05,
"logits/chosen": -6.3125,
"logits/rejected": -4.59375,
"logps/chosen": -1368.0,
"logps/rejected": -2112.0,
"loss": 0.0734,
"rewards/accuracies": 0.956250011920929,
"rewards/chosen": -9.625,
"rewards/margins": 6.125,
"rewards/rejected": -15.75,
"step": 1530
},
{
"epoch": 1.6125654450261782,
"grad_norm": 1.291872971705726,
"learning_rate": 2.609640284421158e-05,
"logits/chosen": -6.25,
"logits/rejected": -4.5,
"logps/chosen": -1392.0,
"logps/rejected": -2176.0,
"loss": 0.1275,
"rewards/accuracies": 0.9624999761581421,
"rewards/chosen": -10.0,
"rewards/margins": 6.3125,
"rewards/rejected": -16.25,
"step": 1540
},
{
"epoch": 1.6230366492146597,
"grad_norm": 2.096335474327163,
"learning_rate": 2.5791968030719726e-05,
"logits/chosen": -8.0625,
"logits/rejected": -5.9375,
"logps/chosen": -1304.0,
"logps/rejected": -2144.0,
"loss": 0.1051,
"rewards/accuracies": 0.96875,
"rewards/chosen": -8.625,
"rewards/margins": 6.4375,
"rewards/rejected": -15.0625,
"step": 1550
},
{
"epoch": 1.6335078534031413,
"grad_norm": 2.76680288050578,
"learning_rate": 2.548741560940902e-05,
"logits/chosen": -7.8125,
"logits/rejected": -6.09375,
"logps/chosen": -1192.0,
"logps/rejected": -2096.0,
"loss": 0.1054,
"rewards/accuracies": 0.9624999761581421,
"rewards/chosen": -7.96875,
"rewards/margins": 6.5,
"rewards/rejected": -14.5,
"step": 1560
},
{
"epoch": 1.6439790575916229,
"grad_norm": 1.272948320009889,
"learning_rate": 2.5182790806531787e-05,
"logits/chosen": -6.59375,
"logits/rejected": -5.46875,
"logps/chosen": -1296.0,
"logps/rejected": -2192.0,
"loss": 0.1061,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -8.75,
"rewards/margins": 7.0,
"rewards/rejected": -15.75,
"step": 1570
},
{
"epoch": 1.6544502617801047,
"grad_norm": 1.522501665293866,
"learning_rate": 2.4878138859089074e-05,
"logits/chosen": -6.4375,
"logits/rejected": -4.71875,
"logps/chosen": -1280.0,
"logps/rejected": -2208.0,
"loss": 0.0886,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -8.75,
"rewards/margins": 7.4375,
"rewards/rejected": -16.25,
"step": 1580
},
{
"epoch": 1.6649214659685865,
"grad_norm": 1.1308622718690846,
"learning_rate": 2.4573505008112924e-05,
"logits/chosen": -5.96875,
"logits/rejected": -4.125,
"logps/chosen": -1392.0,
"logps/rejected": -2160.0,
"loss": 0.0748,
"rewards/accuracies": 0.981249988079071,
"rewards/chosen": -9.9375,
"rewards/margins": 6.5,
"rewards/rejected": -16.375,
"step": 1590
},
{
"epoch": 1.675392670157068,
"grad_norm": 1.3684315880507028,
"learning_rate": 2.4268934491948032e-05,
"logits/chosen": -5.46875,
"logits/rejected": -4.28125,
"logps/chosen": -1376.0,
"logps/rejected": -2096.0,
"loss": 0.0813,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -9.6875,
"rewards/margins": 5.8125,
"rewards/rejected": -15.5,
"step": 1600
},
{
"epoch": 1.675392670157068,
"eval_logits/chosen": -2.78125,
"eval_logits/rejected": -1.015625,
"eval_logps/chosen": -1456.0,
"eval_logps/rejected": -1584.0,
"eval_loss": 1.2080937623977661,
"eval_rewards/accuracies": 0.623046875,
"eval_rewards/chosen": -11.4375,
"eval_rewards/margins": 1.5625,
"eval_rewards/rejected": -13.0,
"eval_runtime": 58.12,
"eval_samples_per_second": 34.412,
"eval_steps_per_second": 0.551,
"step": 1600
},
{
"epoch": 1.6858638743455496,
"grad_norm": 0.7301763980956629,
"learning_rate": 2.396447253953385e-05,
"logits/chosen": -5.78125,
"logits/rejected": -4.25,
"logps/chosen": -1344.0,
"logps/rejected": -2192.0,
"loss": 0.0717,
"rewards/accuracies": 0.981249988079071,
"rewards/chosen": -9.125,
"rewards/margins": 6.84375,
"rewards/rejected": -16.0,
"step": 1610
},
{
"epoch": 1.6963350785340314,
"grad_norm": 1.260461173695342,
"learning_rate": 2.3660164363687998e-05,
"logits/chosen": -6.03125,
"logits/rejected": -4.125,
"logps/chosen": -1304.0,
"logps/rejected": -2080.0,
"loss": 0.0889,
"rewards/accuracies": 0.96875,
"rewards/chosen": -8.5,
"rewards/margins": 6.15625,
"rewards/rejected": -14.6875,
"step": 1620
},
{
"epoch": 1.7068062827225132,
"grad_norm": 2.263220266861775,
"learning_rate": 2.33560551543922e-05,
"logits/chosen": -6.59375,
"logits/rejected": -4.46875,
"logps/chosen": -1208.0,
"logps/rejected": -1952.0,
"loss": 0.0901,
"rewards/accuracies": 0.9624999761581421,
"rewards/chosen": -7.78125,
"rewards/margins": 5.65625,
"rewards/rejected": -13.4375,
"step": 1630
},
{
"epoch": 1.7172774869109948,
"grad_norm": 2.9351955647980823,
"learning_rate": 2.3052190072081493e-05,
"logits/chosen": -5.71875,
"logits/rejected": -3.546875,
"logps/chosen": -1288.0,
"logps/rejected": -2032.0,
"loss": 0.0914,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -8.75,
"rewards/margins": 6.15625,
"rewards/rejected": -14.875,
"step": 1640
},
{
"epoch": 1.7277486910994764,
"grad_norm": 1.3216707885098384,
"learning_rate": 2.2748614240937864e-05,
"logits/chosen": -4.6875,
"logits/rejected": -3.4375,
"logps/chosen": -1304.0,
"logps/rejected": -2224.0,
"loss": 0.0579,
"rewards/accuracies": 0.96875,
"rewards/chosen": -9.125,
"rewards/margins": 6.8125,
"rewards/rejected": -15.9375,
"step": 1650
},
{
"epoch": 1.738219895287958,
"grad_norm": 1.9595015833294163,
"learning_rate": 2.2445372742189334e-05,
"logits/chosen": -4.6875,
"logits/rejected": -2.9375,
"logps/chosen": -1448.0,
"logps/rejected": -2368.0,
"loss": 0.0904,
"rewards/accuracies": 0.9624999761581421,
"rewards/chosen": -10.3125,
"rewards/margins": 6.90625,
"rewards/rejected": -17.25,
"step": 1660
},
{
"epoch": 1.7486910994764397,
"grad_norm": 2.514553191980556,
"learning_rate": 2.214251060741528e-05,
"logits/chosen": -5.375,
"logits/rejected": -3.859375,
"logps/chosen": -1456.0,
"logps/rejected": -2352.0,
"loss": 0.0781,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -10.375,
"rewards/margins": 7.0,
"rewards/rejected": -17.375,
"step": 1670
},
{
"epoch": 1.7591623036649215,
"grad_norm": 1.9304685422522236,
"learning_rate": 2.1840072811859302e-05,
"logits/chosen": -5.78125,
"logits/rejected": -3.953125,
"logps/chosen": -1440.0,
"logps/rejected": -2288.0,
"loss": 0.0867,
"rewards/accuracies": 0.956250011920929,
"rewards/chosen": -9.9375,
"rewards/margins": 6.78125,
"rewards/rejected": -16.75,
"step": 1680
},
{
"epoch": 1.7696335078534031,
"grad_norm": 2.328033848544687,
"learning_rate": 2.1538104267750287e-05,
"logits/chosen": -6.21875,
"logits/rejected": -4.59375,
"logps/chosen": -1280.0,
"logps/rejected": -2000.0,
"loss": 0.1053,
"rewards/accuracies": 0.9624999761581421,
"rewards/chosen": -8.875,
"rewards/margins": 5.5,
"rewards/rejected": -14.375,
"step": 1690
},
{
"epoch": 1.7801047120418847,
"grad_norm": 1.2305326304884159,
"learning_rate": 2.1236649817632952e-05,
"logits/chosen": -6.71875,
"logits/rejected": -5.375,
"logps/chosen": -1272.0,
"logps/rejected": -2000.0,
"loss": 0.0882,
"rewards/accuracies": 0.9437500238418579,
"rewards/chosen": -9.0625,
"rewards/margins": 5.375,
"rewards/rejected": -14.4375,
"step": 1700
},
{
"epoch": 1.7801047120418847,
"eval_logits/chosen": -4.78125,
"eval_logits/rejected": -3.046875,
"eval_logps/chosen": -1472.0,
"eval_logps/rejected": -1592.0,
"eval_loss": 1.1652499437332153,
"eval_rewards/accuracies": 0.634765625,
"eval_rewards/chosen": -11.5625,
"eval_rewards/margins": 1.453125,
"eval_rewards/rejected": -13.0,
"eval_runtime": 58.1451,
"eval_samples_per_second": 34.397,
"eval_steps_per_second": 0.55,
"step": 1700
},
{
"epoch": 1.7905759162303665,
"grad_norm": 3.687265844143873,
"learning_rate": 2.093575422770872e-05,
"logits/chosen": -5.71875,
"logits/rejected": -4.1875,
"logps/chosen": -1376.0,
"logps/rejected": -2144.0,
"loss": 0.0781,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -10.0,
"rewards/margins": 6.0625,
"rewards/rejected": -16.0,
"step": 1710
},
{
"epoch": 1.8010471204188483,
"grad_norm": 4.0018653273641105,
"learning_rate": 2.0635462181187827e-05,
"logits/chosen": -6.0,
"logits/rejected": -4.59375,
"logps/chosen": -1536.0,
"logps/rejected": -2528.0,
"loss": 0.0915,
"rewards/accuracies": 0.956250011920929,
"rewards/chosen": -11.3125,
"rewards/margins": 7.625,
"rewards/rejected": -18.875,
"step": 1720
},
{
"epoch": 1.8115183246073299,
"grad_norm": 2.029239865827198,
"learning_rate": 2.0335818271653917e-05,
"logits/chosen": -6.0625,
"logits/rejected": -4.59375,
"logps/chosen": -1456.0,
"logps/rejected": -2368.0,
"loss": 0.0728,
"rewards/accuracies": 0.96875,
"rewards/chosen": -10.5,
"rewards/margins": 6.90625,
"rewards/rejected": -17.5,
"step": 1730
},
{
"epoch": 1.8219895287958114,
"grad_norm": 3.042431324232482,
"learning_rate": 2.0036866996441815e-05,
"logits/chosen": -6.125,
"logits/rejected": -3.828125,
"logps/chosen": -1448.0,
"logps/rejected": -2304.0,
"loss": 0.0949,
"rewards/accuracies": 0.956250011920929,
"rewards/chosen": -10.5,
"rewards/margins": 6.71875,
"rewards/rejected": -17.25,
"step": 1740
},
{
"epoch": 1.8324607329842932,
"grad_norm": 3.5822601753058083,
"learning_rate": 1.9738652750029677e-05,
"logits/chosen": -5.4375,
"logits/rejected": -3.890625,
"logps/chosen": -1416.0,
"logps/rejected": -2304.0,
"loss": 0.0836,
"rewards/accuracies": 0.981249988079071,
"rewards/chosen": -9.875,
"rewards/margins": 7.1875,
"rewards/rejected": -17.125,
"step": 1750
},
{
"epoch": 1.8429319371727748,
"grad_norm": 1.8450887458195167,
"learning_rate": 1.944121981744631e-05,
"logits/chosen": -6.625,
"logits/rejected": -4.59375,
"logps/chosen": -1400.0,
"logps/rejected": -2336.0,
"loss": 0.0786,
"rewards/accuracies": 0.9624999761581421,
"rewards/chosen": -9.5,
"rewards/margins": 7.03125,
"rewards/rejected": -16.5,
"step": 1760
},
{
"epoch": 1.8534031413612566,
"grad_norm": 1.3004757675656184,
"learning_rate": 1.9144612367694887e-05,
"logits/chosen": -6.09375,
"logits/rejected": -3.875,
"logps/chosen": -1320.0,
"logps/rejected": -2112.0,
"loss": 0.0747,
"rewards/accuracies": 0.96875,
"rewards/chosen": -9.1875,
"rewards/margins": 6.53125,
"rewards/rejected": -15.6875,
"step": 1770
},
{
"epoch": 1.8638743455497382,
"grad_norm": 3.2717270160702627,
"learning_rate": 1.8848874447193803e-05,
"logits/chosen": -4.40625,
"logits/rejected": -2.59375,
"logps/chosen": -1376.0,
"logps/rejected": -2192.0,
"loss": 0.0875,
"rewards/accuracies": 0.956250011920929,
"rewards/chosen": -9.875,
"rewards/margins": 6.8125,
"rewards/rejected": -16.75,
"step": 1780
},
{
"epoch": 1.8743455497382198,
"grad_norm": 2.2738872292672188,
"learning_rate": 1.855404997323571e-05,
"logits/chosen": -5.59375,
"logits/rejected": -4.03125,
"logps/chosen": -1344.0,
"logps/rejected": -2112.0,
"loss": 0.1089,
"rewards/accuracies": 0.96875,
"rewards/chosen": -9.375,
"rewards/margins": 6.25,
"rewards/rejected": -15.5625,
"step": 1790
},
{
"epoch": 1.8848167539267016,
"grad_norm": 1.5164525663792208,
"learning_rate": 1.82601827274658e-05,
"logits/chosen": -5.15625,
"logits/rejected": -3.3125,
"logps/chosen": -1224.0,
"logps/rejected": -1944.0,
"loss": 0.0991,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -8.4375,
"rewards/margins": 5.84375,
"rewards/rejected": -14.3125,
"step": 1800
},
{
"epoch": 1.8848167539267016,
"eval_logits/chosen": -2.046875,
"eval_logits/rejected": -0.27734375,
"eval_logps/chosen": -1288.0,
"eval_logps/rejected": -1392.0,
"eval_loss": 1.0545624494552612,
"eval_rewards/accuracies": 0.62109375,
"eval_rewards/chosen": -9.6875,
"eval_rewards/margins": 1.3203125,
"eval_rewards/rejected": -11.0,
"eval_runtime": 58.1384,
"eval_samples_per_second": 34.401,
"eval_steps_per_second": 0.55,
"step": 1800
},
{
"epoch": 1.8952879581151834,
"grad_norm": 2.9694814485736423,
"learning_rate": 1.7967316349380213e-05,
"logits/chosen": -3.578125,
"logits/rejected": -2.640625,
"logps/chosen": -1208.0,
"logps/rejected": -1912.0,
"loss": 0.1131,
"rewards/accuracies": 0.9437500238418579,
"rewards/chosen": -8.375,
"rewards/margins": 5.59375,
"rewards/rejected": -13.9375,
"step": 1810
},
{
"epoch": 1.905759162303665,
"grad_norm": 1.5749081433945913,
"learning_rate": 1.7675494329845516e-05,
"logits/chosen": -4.0625,
"logits/rejected": -2.6875,
"logps/chosen": -1248.0,
"logps/rejected": -2008.0,
"loss": 0.0693,
"rewards/accuracies": 0.96875,
"rewards/chosen": -8.5625,
"rewards/margins": 5.8125,
"rewards/rejected": -14.375,
"step": 1820
},
{
"epoch": 1.9162303664921465,
"grad_norm": 0.6698994167162394,
"learning_rate": 1.738476000464026e-05,
"logits/chosen": -5.09375,
"logits/rejected": -3.171875,
"logps/chosen": -1176.0,
"logps/rejected": -1976.0,
"loss": 0.0809,
"rewards/accuracies": 0.96875,
"rewards/chosen": -7.875,
"rewards/margins": 6.40625,
"rewards/rejected": -14.25,
"step": 1830
},
{
"epoch": 1.9267015706806283,
"grad_norm": 1.4497176229531499,
"learning_rate": 1.7095156548019646e-05,
"logits/chosen": -6.4375,
"logits/rejected": -4.59375,
"logps/chosen": -1192.0,
"logps/rejected": -2064.0,
"loss": 0.0838,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -8.125,
"rewards/margins": 6.625,
"rewards/rejected": -14.75,
"step": 1840
},
{
"epoch": 1.93717277486911,
"grad_norm": 3.5982046514604686,
"learning_rate": 1.6806726966304064e-05,
"logits/chosen": -6.53125,
"logits/rejected": -4.5,
"logps/chosen": -1296.0,
"logps/rejected": -2112.0,
"loss": 0.1102,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -8.6875,
"rewards/margins": 6.46875,
"rewards/rejected": -15.1875,
"step": 1850
},
{
"epoch": 1.9476439790575917,
"grad_norm": 2.6408858202400953,
"learning_rate": 1.6519514091492625e-05,
"logits/chosen": -6.5,
"logits/rejected": -4.84375,
"logps/chosen": -1200.0,
"logps/rejected": -2000.0,
"loss": 0.0886,
"rewards/accuracies": 0.956250011920929,
"rewards/chosen": -8.375,
"rewards/margins": 6.5,
"rewards/rejected": -14.875,
"step": 1860
},
{
"epoch": 1.9581151832460733,
"grad_norm": 2.7605664353424575,
"learning_rate": 1.6233560574902627e-05,
"logits/chosen": -6.375,
"logits/rejected": -4.53125,
"logps/chosen": -1160.0,
"logps/rejected": -2008.0,
"loss": 0.1043,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -8.0625,
"rewards/margins": 6.1875,
"rewards/rejected": -14.25,
"step": 1870
},
{
"epoch": 1.9685863874345548,
"grad_norm": 1.4845822817168464,
"learning_rate": 1.5948908880835753e-05,
"logits/chosen": -6.4375,
"logits/rejected": -4.25,
"logps/chosen": -1208.0,
"logps/rejected": -1952.0,
"loss": 0.0814,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -8.125,
"rewards/margins": 6.09375,
"rewards/rejected": -14.25,
"step": 1880
},
{
"epoch": 1.9790575916230366,
"grad_norm": 0.9774032826819946,
"learning_rate": 1.5665601280272124e-05,
"logits/chosen": -6.71875,
"logits/rejected": -5.0625,
"logps/chosen": -1232.0,
"logps/rejected": -2032.0,
"loss": 0.0867,
"rewards/accuracies": 0.9375,
"rewards/chosen": -8.5,
"rewards/margins": 5.8125,
"rewards/rejected": -14.3125,
"step": 1890
},
{
"epoch": 1.9895287958115184,
"grad_norm": 1.8044319867937242,
"learning_rate": 1.538367984459298e-05,
"logits/chosen": -6.25,
"logits/rejected": -4.28125,
"logps/chosen": -1208.0,
"logps/rejected": -2032.0,
"loss": 0.0663,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -8.4375,
"rewards/margins": 6.15625,
"rewards/rejected": -14.5625,
"step": 1900
},
{
"epoch": 1.9895287958115184,
"eval_logits/chosen": -3.734375,
"eval_logits/rejected": -1.9765625,
"eval_logps/chosen": -1424.0,
"eval_logps/rejected": -1552.0,
"eval_loss": 1.1601874828338623,
"eval_rewards/accuracies": 0.634765625,
"eval_rewards/chosen": -11.0625,
"eval_rewards/margins": 1.53125,
"eval_rewards/rejected": -12.625,
"eval_runtime": 58.1124,
"eval_samples_per_second": 34.416,
"eval_steps_per_second": 0.551,
"step": 1900
},
{
"epoch": 2.0,
"grad_norm": 1.4053488815203414,
"learning_rate": 1.5103186439333133e-05,
"logits/chosen": -5.21875,
"logits/rejected": -4.03125,
"logps/chosen": -1272.0,
"logps/rejected": -2160.0,
"loss": 0.0996,
"rewards/accuracies": 0.9624999761581421,
"rewards/chosen": -8.9375,
"rewards/margins": 6.9375,
"rewards/rejected": -15.875,
"step": 1910
},
{
"epoch": 2.0104712041884816,
"grad_norm": 0.2750995325531386,
"learning_rate": 1.4824162717963829e-05,
"logits/chosen": -5.65625,
"logits/rejected": -2.828125,
"logps/chosen": -1272.0,
"logps/rejected": -2160.0,
"loss": 0.0133,
"rewards/accuracies": 1.0,
"rewards/chosen": -8.8125,
"rewards/margins": 7.4375,
"rewards/rejected": -16.25,
"step": 1920
},
{
"epoch": 2.020942408376963,
"grad_norm": 0.11874503801227174,
"learning_rate": 1.4546650115707192e-05,
"logits/chosen": -5.90625,
"logits/rejected": -3.65625,
"logps/chosen": -1280.0,
"logps/rejected": -2176.0,
"loss": 0.0136,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -9.125,
"rewards/margins": 7.90625,
"rewards/rejected": -17.0,
"step": 1930
},
{
"epoch": 2.031413612565445,
"grad_norm": 0.10142519243708019,
"learning_rate": 1.4270689843383111e-05,
"logits/chosen": -4.375,
"logits/rejected": -1.8359375,
"logps/chosen": -1400.0,
"logps/rejected": -2464.0,
"loss": 0.0156,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -9.9375,
"rewards/margins": 9.1875,
"rewards/rejected": -19.125,
"step": 1940
},
{
"epoch": 2.0418848167539267,
"grad_norm": 0.6941813800163106,
"learning_rate": 1.3996322881289347e-05,
"logits/chosen": -5.5,
"logits/rejected": -2.796875,
"logps/chosen": -1360.0,
"logps/rejected": -2448.0,
"loss": 0.0184,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -9.875,
"rewards/margins": 9.125,
"rewards/rejected": -19.0,
"step": 1950
},
{
"epoch": 2.0523560209424083,
"grad_norm": 0.16238337413970777,
"learning_rate": 1.3723589973115963e-05,
"logits/chosen": -5.65625,
"logits/rejected": -2.765625,
"logps/chosen": -1344.0,
"logps/rejected": -2560.0,
"loss": 0.015,
"rewards/accuracies": 1.0,
"rewards/chosen": -9.75,
"rewards/margins": 9.75,
"rewards/rejected": -19.5,
"step": 1960
},
{
"epoch": 2.06282722513089,
"grad_norm": 0.3624302685755531,
"learning_rate": 1.3452531619894837e-05,
"logits/chosen": -4.65625,
"logits/rejected": -2.484375,
"logps/chosen": -1464.0,
"logps/rejected": -2608.0,
"loss": 0.0092,
"rewards/accuracies": 1.0,
"rewards/chosen": -10.75,
"rewards/margins": 9.375,
"rewards/rejected": -20.125,
"step": 1970
},
{
"epoch": 2.073298429319372,
"grad_norm": 0.1704688503966962,
"learning_rate": 1.3183188073985245e-05,
"logits/chosen": -5.03125,
"logits/rejected": -2.546875,
"logps/chosen": -1504.0,
"logps/rejected": -2672.0,
"loss": 0.0177,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -11.25,
"rewards/margins": 9.8125,
"rewards/rejected": -21.125,
"step": 1980
},
{
"epoch": 2.0837696335078535,
"grad_norm": 0.13555602900626618,
"learning_rate": 1.2915599333096351e-05,
"logits/chosen": -4.875,
"logits/rejected": -2.8125,
"logps/chosen": -1632.0,
"logps/rejected": -2848.0,
"loss": 0.0088,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -12.1875,
"rewards/margins": 10.1875,
"rewards/rejected": -22.375,
"step": 1990
},
{
"epoch": 2.094240837696335,
"grad_norm": 0.20811722838434601,
"learning_rate": 1.264980513434752e-05,
"logits/chosen": -4.78125,
"logits/rejected": -1.7578125,
"logps/chosen": -1624.0,
"logps/rejected": -2832.0,
"loss": 0.0132,
"rewards/accuracies": 1.0,
"rewards/chosen": -11.9375,
"rewards/margins": 10.125,
"rewards/rejected": -22.0,
"step": 2000
},
{
"epoch": 2.094240837696335,
"eval_logits/chosen": -1.5390625,
"eval_logits/rejected": 0.3359375,
"eval_logps/chosen": -1856.0,
"eval_logps/rejected": -2040.0,
"eval_loss": 1.6895312070846558,
"eval_rewards/accuracies": 0.619140625,
"eval_rewards/chosen": -15.4375,
"eval_rewards/margins": 2.0625,
"eval_rewards/rejected": -17.5,
"eval_runtime": 58.1357,
"eval_samples_per_second": 34.402,
"eval_steps_per_second": 0.55,
"step": 2000
},
{
"epoch": 2.1047120418848166,
"grad_norm": 0.18572065082511588,
"learning_rate": 1.2385844948367324e-05,
"logits/chosen": -4.4375,
"logits/rejected": -2.109375,
"logps/chosen": -1552.0,
"logps/rejected": -2704.0,
"loss": 0.0148,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -11.8125,
"rewards/margins": 10.125,
"rewards/rejected": -22.0,
"step": 2010
},
{
"epoch": 2.115183246073298,
"grad_norm": 0.13824259124862698,
"learning_rate": 1.2123757973432114e-05,
"logits/chosen": -4.625,
"logits/rejected": -2.203125,
"logps/chosen": -1616.0,
"logps/rejected": -2672.0,
"loss": 0.0142,
"rewards/accuracies": 0.981249988079071,
"rewards/chosen": -12.0,
"rewards/margins": 9.75,
"rewards/rejected": -21.75,
"step": 2020
},
{
"epoch": 2.1256544502617802,
"grad_norm": 0.16457344724928352,
"learning_rate": 1.1863583129645106e-05,
"logits/chosen": -4.5,
"logits/rejected": -2.8125,
"logps/chosen": -1576.0,
"logps/rejected": -2896.0,
"loss": 0.0166,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -11.875,
"rewards/margins": 10.625,
"rewards/rejected": -22.5,
"step": 2030
},
{
"epoch": 2.136125654450262,
"grad_norm": 0.50619842855216,
"learning_rate": 1.1605359053156607e-05,
"logits/chosen": -5.15625,
"logits/rejected": -2.09375,
"logps/chosen": -1584.0,
"logps/rejected": -2768.0,
"loss": 0.0143,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -11.625,
"rewards/margins": 10.0625,
"rewards/rejected": -21.625,
"step": 2040
},
{
"epoch": 2.1465968586387434,
"grad_norm": 0.09341262208205552,
"learning_rate": 1.1349124090426644e-05,
"logits/chosen": -4.375,
"logits/rejected": -1.7421875,
"logps/chosen": -1536.0,
"logps/rejected": -2896.0,
"loss": 0.0111,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -11.3125,
"rewards/margins": 11.25,
"rewards/rejected": -22.5,
"step": 2050
},
{
"epoch": 2.157068062827225,
"grad_norm": 0.07358157022422468,
"learning_rate": 1.1094916292530406e-05,
"logits/chosen": -4.0,
"logits/rejected": -2.140625,
"logps/chosen": -1552.0,
"logps/rejected": -2688.0,
"loss": 0.0205,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -11.4375,
"rewards/margins": 9.6875,
"rewards/rejected": -21.125,
"step": 2060
},
{
"epoch": 2.167539267015707,
"grad_norm": 0.2188424266492573,
"learning_rate": 1.0842773409507623e-05,
"logits/chosen": -4.0,
"logits/rejected": -2.109375,
"logps/chosen": -1488.0,
"logps/rejected": -2768.0,
"loss": 0.0132,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -11.1875,
"rewards/margins": 10.375,
"rewards/rejected": -21.625,
"step": 2070
},
{
"epoch": 2.1780104712041886,
"grad_norm": 0.12046304272345514,
"learning_rate": 1.0592732884756754e-05,
"logits/chosen": -4.65625,
"logits/rejected": -2.671875,
"logps/chosen": -1576.0,
"logps/rejected": -2912.0,
"loss": 0.0117,
"rewards/accuracies": 0.981249988079071,
"rewards/chosen": -11.625,
"rewards/margins": 10.4375,
"rewards/rejected": -22.125,
"step": 2080
},
{
"epoch": 2.18848167539267,
"grad_norm": 0.2096910787929615,
"learning_rate": 1.0344831849474507e-05,
"logits/chosen": -4.34375,
"logits/rejected": -1.84375,
"logps/chosen": -1624.0,
"logps/rejected": -2864.0,
"loss": 0.0096,
"rewards/accuracies": 1.0,
"rewards/chosen": -12.125,
"rewards/margins": 10.3125,
"rewards/rejected": -22.5,
"step": 2090
},
{
"epoch": 2.1989528795811517,
"grad_norm": 0.09853978908547517,
"learning_rate": 1.0099107117141879e-05,
"logits/chosen": -3.828125,
"logits/rejected": -1.1796875,
"logps/chosen": -1576.0,
"logps/rejected": -2800.0,
"loss": 0.0613,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -12.0625,
"rewards/margins": 10.375,
"rewards/rejected": -22.375,
"step": 2100
},
{
"epoch": 2.1989528795811517,
"eval_logits/chosen": -1.0625,
"eval_logits/rejected": 0.75390625,
"eval_logps/chosen": -1896.0,
"eval_logps/rejected": -2096.0,
"eval_loss": 1.7890390157699585,
"eval_rewards/accuracies": 0.619140625,
"eval_rewards/chosen": -15.8125,
"eval_rewards/margins": 2.203125,
"eval_rewards/rejected": -18.0,
"eval_runtime": 58.1512,
"eval_samples_per_second": 34.393,
"eval_steps_per_second": 0.55,
"step": 2100
},
{
"epoch": 2.2094240837696333,
"grad_norm": 0.9269743969402358,
"learning_rate": 9.855595178057334e-06,
"logits/chosen": -3.546875,
"logits/rejected": -1.5625,
"logps/chosen": -1688.0,
"logps/rejected": -2992.0,
"loss": 0.0113,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -12.75,
"rewards/margins": 11.1875,
"rewards/rejected": -24.0,
"step": 2110
},
{
"epoch": 2.2198952879581153,
"grad_norm": 0.17831469262169247,
"learning_rate": 9.614332193917949e-06,
"logits/chosen": -3.484375,
"logits/rejected": -1.0859375,
"logps/chosen": -1664.0,
"logps/rejected": -2976.0,
"loss": 0.0223,
"rewards/accuracies": 1.0,
"rewards/chosen": -12.375,
"rewards/margins": 11.0,
"rewards/rejected": -23.375,
"step": 2120
},
{
"epoch": 2.230366492146597,
"grad_norm": 0.48727204777736,
"learning_rate": 9.375353992449385e-06,
"logits/chosen": -4.09375,
"logits/rejected": -2.484375,
"logps/chosen": -1672.0,
"logps/rejected": -2928.0,
"loss": 0.0082,
"rewards/accuracies": 1.0,
"rewards/chosen": -12.75,
"rewards/margins": 10.6875,
"rewards/rejected": -23.5,
"step": 2130
},
{
"epoch": 2.2408376963350785,
"grad_norm": 0.13914719685219426,
"learning_rate": 9.138696062085442e-06,
"logits/chosen": -3.8125,
"logits/rejected": -1.890625,
"logps/chosen": -1624.0,
"logps/rejected": -3040.0,
"loss": 0.0106,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -12.3125,
"rewards/margins": 11.875,
"rewards/rejected": -24.125,
"step": 2140
},
{
"epoch": 2.25130890052356,
"grad_norm": 0.08289961396099146,
"learning_rate": 8.904393546698007e-06,
"logits/chosen": -3.8125,
"logits/rejected": -1.53125,
"logps/chosen": -1672.0,
"logps/rejected": -2992.0,
"loss": 0.0147,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -12.625,
"rewards/margins": 11.125,
"rewards/rejected": -23.75,
"step": 2150
},
{
"epoch": 2.261780104712042,
"grad_norm": 0.3089019892632263,
"learning_rate": 8.672481240378142e-06,
"logits/chosen": -2.96875,
"logits/rejected": -1.234375,
"logps/chosen": -1552.0,
"logps/rejected": -2800.0,
"loss": 0.0162,
"rewards/accuracies": 1.0,
"rewards/chosen": -12.0625,
"rewards/margins": 10.375,
"rewards/rejected": -22.5,
"step": 2160
},
{
"epoch": 2.2722513089005236,
"grad_norm": 0.21930433159710241,
"learning_rate": 8.44299358226919e-06,
"logits/chosen": -3.390625,
"logits/rejected": -1.28125,
"logps/chosen": -1608.0,
"logps/rejected": -2944.0,
"loss": 0.0144,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -12.125,
"rewards/margins": 10.875,
"rewards/rejected": -23.0,
"step": 2170
},
{
"epoch": 2.282722513089005,
"grad_norm": 0.2865059163431868,
"learning_rate": 8.215964651452457e-06,
"logits/chosen": -2.921875,
"logits/rejected": -0.63671875,
"logps/chosen": -1584.0,
"logps/rejected": -2928.0,
"loss": 0.0118,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -12.0,
"rewards/margins": 10.875,
"rewards/rejected": -22.875,
"step": 2180
},
{
"epoch": 2.2931937172774868,
"grad_norm": 0.3657961778127388,
"learning_rate": 7.991428161886502e-06,
"logits/chosen": -3.703125,
"logits/rejected": -1.4921875,
"logps/chosen": -1584.0,
"logps/rejected": -2768.0,
"loss": 0.008,
"rewards/accuracies": 1.0,
"rewards/chosen": -11.8125,
"rewards/margins": 10.5,
"rewards/rejected": -22.375,
"step": 2190
},
{
"epoch": 2.303664921465969,
"grad_norm": 0.14259571783431596,
"learning_rate": 7.76941745740061e-06,
"logits/chosen": -3.375,
"logits/rejected": -1.546875,
"logps/chosen": -1712.0,
"logps/rejected": -2976.0,
"loss": 0.0101,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -12.875,
"rewards/margins": 11.0,
"rewards/rejected": -23.875,
"step": 2200
},
{
"epoch": 2.303664921465969,
"eval_logits/chosen": -0.44140625,
"eval_logits/rejected": 1.25,
"eval_logps/chosen": -1928.0,
"eval_logps/rejected": -2128.0,
"eval_loss": 1.7494844198226929,
"eval_rewards/accuracies": 0.62109375,
"eval_rewards/chosen": -16.125,
"eval_rewards/margins": 2.203125,
"eval_rewards/rejected": -18.375,
"eval_runtime": 58.1492,
"eval_samples_per_second": 34.394,
"eval_steps_per_second": 0.55,
"step": 2200
},
{
"epoch": 2.3141361256544504,
"grad_norm": 0.3018564091193485,
"learning_rate": 7.549965506743123e-06,
"logits/chosen": -2.90625,
"logits/rejected": -1.53125,
"logps/chosen": -1664.0,
"logps/rejected": -2896.0,
"loss": 0.0065,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -12.8125,
"rewards/margins": 10.6875,
"rewards/rejected": -23.5,
"step": 2210
},
{
"epoch": 2.324607329842932,
"grad_norm": 0.054910636671413646,
"learning_rate": 7.333104898685631e-06,
"logits/chosen": -2.8125,
"logits/rejected": -1.234375,
"logps/chosen": -1664.0,
"logps/rejected": -2912.0,
"loss": 0.0114,
"rewards/accuracies": 1.0,
"rewards/chosen": -12.75,
"rewards/margins": 10.9375,
"rewards/rejected": -23.625,
"step": 2220
},
{
"epoch": 2.3350785340314135,
"grad_norm": 0.1526158300005802,
"learning_rate": 7.118867837183518e-06,
"logits/chosen": -3.625,
"logits/rejected": -1.1484375,
"logps/chosen": -1808.0,
"logps/rejected": -3136.0,
"loss": 0.0097,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -13.4375,
"rewards/margins": 11.8125,
"rewards/rejected": -25.25,
"step": 2230
},
{
"epoch": 2.345549738219895,
"grad_norm": 0.1512150013581962,
"learning_rate": 6.9072861365936045e-06,
"logits/chosen": -3.625,
"logits/rejected": -1.765625,
"logps/chosen": -1856.0,
"logps/rejected": -3296.0,
"loss": 0.0079,
"rewards/accuracies": 1.0,
"rewards/chosen": -14.0,
"rewards/margins": 12.3125,
"rewards/rejected": -26.25,
"step": 2240
},
{
"epoch": 2.356020942408377,
"grad_norm": 0.1346446956940355,
"learning_rate": 6.6983912169497005e-06,
"logits/chosen": -2.9375,
"logits/rejected": -0.984375,
"logps/chosen": -1792.0,
"logps/rejected": -3136.0,
"loss": 0.0116,
"rewards/accuracies": 1.0,
"rewards/chosen": -13.5,
"rewards/margins": 11.3125,
"rewards/rejected": -24.875,
"step": 2250
},
{
"epoch": 2.3664921465968587,
"grad_norm": 0.6587835209525754,
"learning_rate": 6.49221409929677e-06,
"logits/chosen": -2.296875,
"logits/rejected": -0.169921875,
"logps/chosen": -1720.0,
"logps/rejected": -2928.0,
"loss": 0.021,
"rewards/accuracies": 0.981249988079071,
"rewards/chosen": -13.375,
"rewards/margins": 10.6875,
"rewards/rejected": -24.0,
"step": 2260
},
{
"epoch": 2.3769633507853403,
"grad_norm": 1.3040415144386621,
"learning_rate": 6.2887854010842075e-06,
"logits/chosen": -2.5,
"logits/rejected": -1.1328125,
"logps/chosen": -1680.0,
"logps/rejected": -3040.0,
"loss": 0.0148,
"rewards/accuracies": 0.96875,
"rewards/chosen": -13.0625,
"rewards/margins": 11.6875,
"rewards/rejected": -24.75,
"step": 2270
},
{
"epoch": 2.387434554973822,
"grad_norm": 0.0647087622813948,
"learning_rate": 6.0881353316191385e-06,
"logits/chosen": -3.75,
"logits/rejected": -1.8515625,
"logps/chosen": -1640.0,
"logps/rejected": -2992.0,
"loss": 0.0109,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -12.25,
"rewards/margins": 11.625,
"rewards/rejected": -23.875,
"step": 2280
},
{
"epoch": 2.3979057591623034,
"grad_norm": 0.20521838341116405,
"learning_rate": 5.890293687580381e-06,
"logits/chosen": -2.71875,
"logits/rejected": -0.87890625,
"logps/chosen": -1696.0,
"logps/rejected": -2928.0,
"loss": 0.0098,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -12.8125,
"rewards/margins": 11.0,
"rewards/rejected": -23.75,
"step": 2290
},
{
"epoch": 2.4083769633507854,
"grad_norm": 0.4074950411969442,
"learning_rate": 5.695289848593532e-06,
"logits/chosen": -3.890625,
"logits/rejected": -2.421875,
"logps/chosen": -1576.0,
"logps/rejected": -2832.0,
"loss": 0.0138,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -12.0,
"rewards/margins": 10.625,
"rewards/rejected": -22.625,
"step": 2300
},
{
"epoch": 2.4083769633507854,
"eval_logits/chosen": -0.7890625,
"eval_logits/rejected": 1.0234375,
"eval_logps/chosen": -1880.0,
"eval_logps/rejected": -2064.0,
"eval_loss": 1.7595937252044678,
"eval_rewards/accuracies": 0.61328125,
"eval_rewards/chosen": -15.625,
"eval_rewards/margins": 2.203125,
"eval_rewards/rejected": -17.75,
"eval_runtime": 58.1919,
"eval_samples_per_second": 34.369,
"eval_steps_per_second": 0.55,
"step": 2300
},
{
"epoch": 2.418848167539267,
"grad_norm": 0.1836092276273333,
"learning_rate": 5.50315277286812e-06,
"logits/chosen": -3.890625,
"logits/rejected": -1.890625,
"logps/chosen": -1584.0,
"logps/rejected": -2864.0,
"loss": 0.013,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -12.0625,
"rewards/margins": 11.5,
"rewards/rejected": -23.5,
"step": 2310
},
{
"epoch": 2.4293193717277486,
"grad_norm": 0.23398770391718288,
"learning_rate": 5.313910992897281e-06,
"logits/chosen": -2.90625,
"logits/rejected": -0.80078125,
"logps/chosen": -1592.0,
"logps/rejected": -2944.0,
"loss": 0.008,
"rewards/accuracies": 1.0,
"rewards/chosen": -12.0625,
"rewards/margins": 11.25,
"rewards/rejected": -23.25,
"step": 2320
},
{
"epoch": 2.4397905759162306,
"grad_norm": 0.17563380231523723,
"learning_rate": 5.127592611220658e-06,
"logits/chosen": -3.0,
"logits/rejected": -1.34375,
"logps/chosen": -1704.0,
"logps/rejected": -3072.0,
"loss": 0.0184,
"rewards/accuracies": 0.981249988079071,
"rewards/chosen": -12.875,
"rewards/margins": 11.375,
"rewards/rejected": -24.25,
"step": 2330
},
{
"epoch": 2.450261780104712,
"grad_norm": 0.8529276803071487,
"learning_rate": 4.944225296251159e-06,
"logits/chosen": -3.53125,
"logits/rejected": -1.9375,
"logps/chosen": -1584.0,
"logps/rejected": -2944.0,
"loss": 0.0104,
"rewards/accuracies": 1.0,
"rewards/chosen": -12.25,
"rewards/margins": 11.5625,
"rewards/rejected": -23.875,
"step": 2340
},
{
"epoch": 2.4607329842931938,
"grad_norm": 0.30651326714210586,
"learning_rate": 4.763836278166172e-06,
"logits/chosen": -4.28125,
"logits/rejected": -1.6875,
"logps/chosen": -1560.0,
"logps/rejected": -2784.0,
"loss": 0.0129,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -11.6875,
"rewards/margins": 10.6875,
"rewards/rejected": -22.375,
"step": 2350
},
{
"epoch": 2.4712041884816753,
"grad_norm": 0.04117812370730546,
"learning_rate": 4.586452344863876e-06,
"logits/chosen": -4.0625,
"logits/rejected": -1.8828125,
"logps/chosen": -1648.0,
"logps/rejected": -2928.0,
"loss": 0.0113,
"rewards/accuracies": 0.981249988079071,
"rewards/chosen": -12.3125,
"rewards/margins": 11.0625,
"rewards/rejected": -23.375,
"step": 2360
},
{
"epoch": 2.481675392670157,
"grad_norm": 0.21358131298816088,
"learning_rate": 4.412099837985192e-06,
"logits/chosen": -3.453125,
"logits/rejected": -1.1328125,
"logps/chosen": -1648.0,
"logps/rejected": -3040.0,
"loss": 0.0111,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -12.375,
"rewards/margins": 11.75,
"rewards/rejected": -24.125,
"step": 2370
},
{
"epoch": 2.492146596858639,
"grad_norm": 2.0477892172682424,
"learning_rate": 4.240804649002089e-06,
"logits/chosen": -3.953125,
"logits/rejected": -1.4140625,
"logps/chosen": -1560.0,
"logps/rejected": -2784.0,
"loss": 0.0155,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -12.0,
"rewards/margins": 11.0,
"rewards/rejected": -23.0,
"step": 2380
},
{
"epoch": 2.5026178010471205,
"grad_norm": 1.779051556458814,
"learning_rate": 4.072592215372598e-06,
"logits/chosen": -4.53125,
"logits/rejected": -2.1875,
"logps/chosen": -1712.0,
"logps/rejected": -3056.0,
"loss": 0.0178,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -12.8125,
"rewards/margins": 12.25,
"rewards/rejected": -25.0,
"step": 2390
},
{
"epoch": 2.513089005235602,
"grad_norm": 0.6478507152699798,
"learning_rate": 3.907487516763389e-06,
"logits/chosen": -3.59375,
"logits/rejected": -1.8359375,
"logps/chosen": -1632.0,
"logps/rejected": -2944.0,
"loss": 0.0121,
"rewards/accuracies": 0.981249988079071,
"rewards/chosen": -12.4375,
"rewards/margins": 11.0625,
"rewards/rejected": -23.5,
"step": 2400
},
{
"epoch": 2.513089005235602,
"eval_logits/chosen": -1.1796875,
"eval_logits/rejected": 0.6640625,
"eval_logps/chosen": -1880.0,
"eval_logps/rejected": -2080.0,
"eval_loss": 1.79115629196167,
"eval_rewards/accuracies": 0.615234375,
"eval_rewards/chosen": -15.625,
"eval_rewards/margins": 2.21875,
"eval_rewards/rejected": -17.875,
"eval_runtime": 58.1229,
"eval_samples_per_second": 34.41,
"eval_steps_per_second": 0.551,
"step": 2400
},
{
"epoch": 2.5235602094240837,
"grad_norm": 0.49694642857928184,
"learning_rate": 3.745515071340286e-06,
"logits/chosen": -3.734375,
"logits/rejected": -1.53125,
"logps/chosen": -1688.0,
"logps/rejected": -3024.0,
"loss": 0.013,
"rewards/accuracies": 1.0,
"rewards/chosen": -12.6875,
"rewards/margins": 11.5,
"rewards/rejected": -24.25,
"step": 2410
},
{
"epoch": 2.5340314136125652,
"grad_norm": 0.05803984103664332,
"learning_rate": 3.586698932127236e-06,
"logits/chosen": -4.3125,
"logits/rejected": -1.765625,
"logps/chosen": -1608.0,
"logps/rejected": -2880.0,
"loss": 0.0243,
"rewards/accuracies": 1.0,
"rewards/chosen": -12.125,
"rewards/margins": 11.25,
"rewards/rejected": -23.375,
"step": 2420
},
{
"epoch": 2.5445026178010473,
"grad_norm": 0.20879994672298605,
"learning_rate": 3.4310626834344744e-06,
"logits/chosen": -4.25,
"logits/rejected": -2.5,
"logps/chosen": -1696.0,
"logps/rejected": -3040.0,
"loss": 0.0128,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -12.75,
"rewards/margins": 11.625,
"rewards/rejected": -24.375,
"step": 2430
},
{
"epoch": 2.554973821989529,
"grad_norm": 0.030528572632281526,
"learning_rate": 3.2786294373562343e-06,
"logits/chosen": -3.6875,
"logits/rejected": -1.2578125,
"logps/chosen": -1680.0,
"logps/rejected": -3120.0,
"loss": 0.0072,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -12.625,
"rewards/margins": 12.125,
"rewards/rejected": -24.75,
"step": 2440
},
{
"epoch": 2.5654450261780104,
"grad_norm": 1.0337403674474577,
"learning_rate": 3.1294218303385526e-06,
"logits/chosen": -3.8125,
"logits/rejected": -2.0,
"logps/chosen": -1752.0,
"logps/rejected": -3232.0,
"loss": 0.0103,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -13.125,
"rewards/margins": 12.0625,
"rewards/rejected": -25.125,
"step": 2450
},
{
"epoch": 2.5759162303664924,
"grad_norm": 0.11292627728791665,
"learning_rate": 2.9834620198177483e-06,
"logits/chosen": -3.59375,
"logits/rejected": -1.4609375,
"logps/chosen": -1696.0,
"logps/rejected": -3008.0,
"loss": 0.0153,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -12.9375,
"rewards/margins": 11.5,
"rewards/rejected": -24.375,
"step": 2460
},
{
"epoch": 2.5863874345549736,
"grad_norm": 0.11282488048534844,
"learning_rate": 2.840771680930068e-06,
"logits/chosen": -4.28125,
"logits/rejected": -2.03125,
"logps/chosen": -1768.0,
"logps/rejected": -3360.0,
"loss": 0.0093,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -13.375,
"rewards/margins": 13.1875,
"rewards/rejected": -26.625,
"step": 2470
},
{
"epoch": 2.5968586387434556,
"grad_norm": 0.23597744448152033,
"learning_rate": 2.701372003292868e-06,
"logits/chosen": -3.09375,
"logits/rejected": -1.0703125,
"logps/chosen": -1608.0,
"logps/rejected": -2944.0,
"loss": 0.0212,
"rewards/accuracies": 0.981249988079071,
"rewards/chosen": -12.25,
"rewards/margins": 11.25,
"rewards/rejected": -23.5,
"step": 2480
},
{
"epoch": 2.607329842931937,
"grad_norm": 0.28556432311127383,
"learning_rate": 2.56528368785795e-06,
"logits/chosen": -3.3125,
"logits/rejected": -1.21875,
"logps/chosen": -1616.0,
"logps/rejected": -2832.0,
"loss": 0.0094,
"rewards/accuracies": 1.0,
"rewards/chosen": -12.0625,
"rewards/margins": 10.8125,
"rewards/rejected": -22.875,
"step": 2490
},
{
"epoch": 2.6178010471204187,
"grad_norm": 0.43089010851035947,
"learning_rate": 2.432526943837471e-06,
"logits/chosen": -3.828125,
"logits/rejected": -0.9296875,
"logps/chosen": -1584.0,
"logps/rejected": -2960.0,
"loss": 0.0107,
"rewards/accuracies": 1.0,
"rewards/chosen": -12.125,
"rewards/margins": 11.375,
"rewards/rejected": -23.5,
"step": 2500
},
{
"epoch": 2.6178010471204187,
"eval_logits/chosen": -0.98828125,
"eval_logits/rejected": 0.828125,
"eval_logps/chosen": -1896.0,
"eval_logps/rejected": -2080.0,
"eval_loss": 1.7926875352859497,
"eval_rewards/accuracies": 0.61328125,
"eval_rewards/chosen": -15.75,
"eval_rewards/margins": 2.1875,
"eval_rewards/rejected": -18.0,
"eval_runtime": 58.1114,
"eval_samples_per_second": 34.417,
"eval_steps_per_second": 0.551,
"step": 2500
},
{
"epoch": 2.6282722513089007,
"grad_norm": 0.17250399643318115,
"learning_rate": 2.3031214857028446e-06,
"logits/chosen": -3.375,
"logits/rejected": -1.625,
"logps/chosen": -1640.0,
"logps/rejected": -2992.0,
"loss": 0.009,
"rewards/accuracies": 1.0,
"rewards/chosen": -12.25,
"rewards/margins": 11.6875,
"rewards/rejected": -23.875,
"step": 2510
},
{
"epoch": 2.6387434554973823,
"grad_norm": 0.20025747919767373,
"learning_rate": 2.177086530257139e-06,
"logits/chosen": -4.09375,
"logits/rejected": -1.6953125,
"logps/chosen": -1664.0,
"logps/rejected": -2896.0,
"loss": 0.0124,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -12.6875,
"rewards/margins": 10.75,
"rewards/rejected": -23.375,
"step": 2520
},
{
"epoch": 2.649214659685864,
"grad_norm": 0.17168462038841525,
"learning_rate": 2.054440793781354e-06,
"logits/chosen": -4.03125,
"logits/rejected": -1.390625,
"logps/chosen": -1632.0,
"logps/rejected": -2912.0,
"loss": 0.0137,
"rewards/accuracies": 0.981249988079071,
"rewards/chosen": -12.3125,
"rewards/margins": 11.1875,
"rewards/rejected": -23.5,
"step": 2530
},
{
"epoch": 2.6596858638743455,
"grad_norm": 0.13197617819970092,
"learning_rate": 1.9352024892550587e-06,
"logits/chosen": -3.4375,
"logits/rejected": -1.2109375,
"logps/chosen": -1624.0,
"logps/rejected": -2832.0,
"loss": 0.0113,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -12.25,
"rewards/margins": 10.8125,
"rewards/rejected": -23.125,
"step": 2540
},
{
"epoch": 2.670157068062827,
"grad_norm": 0.41515504064850234,
"learning_rate": 1.8193893236517379e-06,
"logits/chosen": -3.984375,
"logits/rejected": -1.7734375,
"logps/chosen": -1672.0,
"logps/rejected": -3008.0,
"loss": 0.0154,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -12.5,
"rewards/margins": 11.4375,
"rewards/rejected": -24.0,
"step": 2550
},
{
"epoch": 2.680628272251309,
"grad_norm": 0.33541000500884766,
"learning_rate": 1.7070184953092956e-06,
"logits/chosen": -4.25,
"logits/rejected": -1.1171875,
"logps/chosen": -1520.0,
"logps/rejected": -2784.0,
"loss": 0.0107,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -11.1875,
"rewards/margins": 11.25,
"rewards/rejected": -22.375,
"step": 2560
},
{
"epoch": 2.6910994764397906,
"grad_norm": 0.5505498224261427,
"learning_rate": 1.5981066913760918e-06,
"logits/chosen": -3.21875,
"logits/rejected": -0.69140625,
"logps/chosen": -1512.0,
"logps/rejected": -2736.0,
"loss": 0.008,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -11.4375,
"rewards/margins": 10.625,
"rewards/rejected": -22.0,
"step": 2570
},
{
"epoch": 2.701570680628272,
"grad_norm": 0.12939805986134525,
"learning_rate": 1.492670085332884e-06,
"logits/chosen": -4.625,
"logits/rejected": -2.046875,
"logps/chosen": -1616.0,
"logps/rejected": -3056.0,
"loss": 0.009,
"rewards/accuracies": 1.0,
"rewards/chosen": -12.125,
"rewards/margins": 11.8125,
"rewards/rejected": -23.875,
"step": 2580
},
{
"epoch": 2.712041884816754,
"grad_norm": 0.29263191177602443,
"learning_rate": 1.3907243345910787e-06,
"logits/chosen": -3.984375,
"logits/rejected": -1.3828125,
"logps/chosen": -1672.0,
"logps/rejected": -2944.0,
"loss": 0.01,
"rewards/accuracies": 1.0,
"rewards/chosen": -12.3125,
"rewards/margins": 11.1875,
"rewards/rejected": -23.5,
"step": 2590
},
{
"epoch": 2.7225130890052354,
"grad_norm": 0.27391534818018626,
"learning_rate": 1.292284578167538e-06,
"logits/chosen": -3.96875,
"logits/rejected": -1.578125,
"logps/chosen": -1608.0,
"logps/rejected": -2960.0,
"loss": 0.0145,
"rewards/accuracies": 1.0,
"rewards/chosen": -12.125,
"rewards/margins": 11.4375,
"rewards/rejected": -23.625,
"step": 2600
},
{
"epoch": 2.7225130890052354,
"eval_logits/chosen": -1.1328125,
"eval_logits/rejected": 0.703125,
"eval_logps/chosen": -1864.0,
"eval_logps/rejected": -2048.0,
"eval_loss": 1.7577812671661377,
"eval_rewards/accuracies": 0.619140625,
"eval_rewards/chosen": -15.5,
"eval_rewards/margins": 2.203125,
"eval_rewards/rejected": -17.625,
"eval_runtime": 58.1336,
"eval_samples_per_second": 34.404,
"eval_steps_per_second": 0.55,
"step": 2600
},
{
"epoch": 2.7329842931937174,
"grad_norm": 0.40970976352308147,
"learning_rate": 1.1973654344364925e-06,
"logits/chosen": -3.78125,
"logits/rejected": -2.078125,
"logps/chosen": -1584.0,
"logps/rejected": -2912.0,
"loss": 0.0091,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -11.625,
"rewards/margins": 11.0625,
"rewards/rejected": -22.75,
"step": 2610
},
{
"epoch": 2.743455497382199,
"grad_norm": 0.13853094862183052,
"learning_rate": 1.1059809989586583e-06,
"logits/chosen": -3.84375,
"logits/rejected": -1.1796875,
"logps/chosen": -1576.0,
"logps/rejected": -3040.0,
"loss": 0.0086,
"rewards/accuracies": 1.0,
"rewards/chosen": -11.75,
"rewards/margins": 12.0,
"rewards/rejected": -23.75,
"step": 2620
},
{
"epoch": 2.7539267015706805,
"grad_norm": 0.07818300406614886,
"learning_rate": 1.018144842388044e-06,
"logits/chosen": -3.546875,
"logits/rejected": -0.443359375,
"logps/chosen": -1560.0,
"logps/rejected": -2880.0,
"loss": 0.0112,
"rewards/accuracies": 1.0,
"rewards/chosen": -11.3125,
"rewards/margins": 11.375,
"rewards/rejected": -22.625,
"step": 2630
},
{
"epoch": 2.7643979057591626,
"grad_norm": 0.08306384491731265,
"learning_rate": 9.338700084567109e-07,
"logits/chosen": -3.671875,
"logits/rejected": -1.625,
"logps/chosen": -1544.0,
"logps/rejected": -2896.0,
"loss": 0.0101,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -11.5625,
"rewards/margins": 11.375,
"rewards/rejected": -22.875,
"step": 2640
},
{
"epoch": 2.774869109947644,
"grad_norm": 0.4347739459977572,
"learning_rate": 8.531690120377606e-07,
"logits/chosen": -3.75,
"logits/rejected": -1.609375,
"logps/chosen": -1728.0,
"logps/rejected": -3024.0,
"loss": 0.0064,
"rewards/accuracies": 1.0,
"rewards/chosen": -12.875,
"rewards/margins": 11.3125,
"rewards/rejected": -24.125,
"step": 2650
},
{
"epoch": 2.7853403141361257,
"grad_norm": 0.3933843674722983,
"learning_rate": 7.760538372868636e-07,
"logits/chosen": -4.15625,
"logits/rejected": -1.71875,
"logps/chosen": -1592.0,
"logps/rejected": -2832.0,
"loss": 0.0137,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -12.0,
"rewards/margins": 10.8125,
"rewards/rejected": -22.75,
"step": 2660
},
{
"epoch": 2.7958115183246073,
"grad_norm": 0.7900234425544058,
"learning_rate": 7.025359358626166e-07,
"logits/chosen": -4.3125,
"logits/rejected": -1.25,
"logps/chosen": -1688.0,
"logps/rejected": -2960.0,
"loss": 0.0077,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -12.4375,
"rewards/margins": 11.5625,
"rewards/rejected": -24.0,
"step": 2670
},
{
"epoch": 2.806282722513089,
"grad_norm": 0.05818131899881143,
"learning_rate": 6.326262252259463e-07,
"logits/chosen": -3.953125,
"logits/rejected": -1.09375,
"logps/chosen": -1632.0,
"logps/rejected": -2992.0,
"loss": 0.0079,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -12.125,
"rewards/margins": 11.5625,
"rewards/rejected": -23.625,
"step": 2680
},
{
"epoch": 2.816753926701571,
"grad_norm": 0.21300390820618748,
"learning_rate": 5.66335087018871e-07,
"logits/chosen": -3.953125,
"logits/rejected": -1.9140625,
"logps/chosen": -1640.0,
"logps/rejected": -3024.0,
"loss": 0.0054,
"rewards/accuracies": 1.0,
"rewards/chosen": -12.125,
"rewards/margins": 11.125,
"rewards/rejected": -23.25,
"step": 2690
},
{
"epoch": 2.8272251308900525,
"grad_norm": 1.0643333773667172,
"learning_rate": 5.036723655228226e-07,
"logits/chosen": -3.546875,
"logits/rejected": -1.6484375,
"logps/chosen": -1576.0,
"logps/rejected": -2912.0,
"loss": 0.0133,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -12.0625,
"rewards/margins": 11.25,
"rewards/rejected": -23.375,
"step": 2700
},
{
"epoch": 2.8272251308900525,
"eval_logits/chosen": -0.99609375,
"eval_logits/rejected": 0.828125,
"eval_logps/chosen": -1880.0,
"eval_logps/rejected": -2080.0,
"eval_loss": 1.7674453258514404,
"eval_rewards/accuracies": 0.615234375,
"eval_rewards/chosen": -15.625,
"eval_rewards/margins": 2.203125,
"eval_rewards/rejected": -17.875,
"eval_runtime": 58.1554,
"eval_samples_per_second": 34.391,
"eval_steps_per_second": 0.55,
"step": 2700
},
{
"epoch": 2.837696335078534,
"grad_norm": 0.19759741658830307,
"learning_rate": 4.446473661967432e-07,
"logits/chosen": -4.125,
"logits/rejected": -1.8203125,
"logps/chosen": -1648.0,
"logps/rejected": -2864.0,
"loss": 0.011,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -12.5,
"rewards/margins": 10.8125,
"rewards/rejected": -23.25,
"step": 2710
},
{
"epoch": 2.8481675392670156,
"grad_norm": 0.39113186838089087,
"learning_rate": 3.892688542952411e-07,
"logits/chosen": -3.1875,
"logits/rejected": -0.9375,
"logps/chosen": -1528.0,
"logps/rejected": -2848.0,
"loss": 0.0111,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -11.5,
"rewards/margins": 11.25,
"rewards/rejected": -22.75,
"step": 2720
},
{
"epoch": 2.858638743455497,
"grad_norm": 0.08968545307460961,
"learning_rate": 3.3754505356694e-07,
"logits/chosen": -4.40625,
"logits/rejected": -1.859375,
"logps/chosen": -1696.0,
"logps/rejected": -2896.0,
"loss": 0.0114,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -12.6875,
"rewards/margins": 10.4375,
"rewards/rejected": -23.125,
"step": 2730
},
{
"epoch": 2.869109947643979,
"grad_norm": 0.5286398821182616,
"learning_rate": 2.8948364503322276e-07,
"logits/chosen": -3.53125,
"logits/rejected": -0.9921875,
"logps/chosen": -1568.0,
"logps/rejected": -2800.0,
"loss": 0.0197,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -11.9375,
"rewards/margins": 10.875,
"rewards/rejected": -22.875,
"step": 2740
},
{
"epoch": 2.8795811518324608,
"grad_norm": 0.09107393145696067,
"learning_rate": 2.4509176584762706e-07,
"logits/chosen": -3.703125,
"logits/rejected": -1.0390625,
"logps/chosen": -1568.0,
"logps/rejected": -2832.0,
"loss": 0.0079,
"rewards/accuracies": 1.0,
"rewards/chosen": -11.8125,
"rewards/margins": 10.6875,
"rewards/rejected": -22.5,
"step": 2750
},
{
"epoch": 2.8900523560209423,
"grad_norm": 0.27231956515941225,
"learning_rate": 2.0437600823595694e-07,
"logits/chosen": -3.90625,
"logits/rejected": -1.546875,
"logps/chosen": -1664.0,
"logps/rejected": -2944.0,
"loss": 0.0062,
"rewards/accuracies": 1.0,
"rewards/chosen": -12.1875,
"rewards/margins": 11.0,
"rewards/rejected": -23.25,
"step": 2760
},
{
"epoch": 2.900523560209424,
"grad_norm": 0.07986771421213888,
"learning_rate": 1.6734241851733e-07,
"logits/chosen": -3.953125,
"logits/rejected": -1.5859375,
"logps/chosen": -1712.0,
"logps/rejected": -3024.0,
"loss": 0.0092,
"rewards/accuracies": 1.0,
"rewards/chosen": -12.75,
"rewards/margins": 11.375,
"rewards/rejected": -24.125,
"step": 2770
},
{
"epoch": 2.9109947643979055,
"grad_norm": 0.36212437590778634,
"learning_rate": 1.339964962062984e-07,
"logits/chosen": -3.828125,
"logits/rejected": -2.03125,
"logps/chosen": -1616.0,
"logps/rejected": -2880.0,
"loss": 0.0087,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -12.4375,
"rewards/margins": 10.9375,
"rewards/rejected": -23.375,
"step": 2780
},
{
"epoch": 2.9214659685863875,
"grad_norm": 0.1511722662252906,
"learning_rate": 1.0434319319617136e-07,
"logits/chosen": -3.75,
"logits/rejected": -1.078125,
"logps/chosen": -1640.0,
"logps/rejected": -2816.0,
"loss": 0.01,
"rewards/accuracies": 1.0,
"rewards/chosen": -12.4375,
"rewards/margins": 10.4375,
"rewards/rejected": -22.875,
"step": 2790
},
{
"epoch": 2.931937172774869,
"grad_norm": 0.24166413552466873,
"learning_rate": 7.838691302365363e-08,
"logits/chosen": -3.546875,
"logits/rejected": -1.2578125,
"logps/chosen": -1608.0,
"logps/rejected": -2976.0,
"loss": 0.0114,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -12.0,
"rewards/margins": 11.4375,
"rewards/rejected": -23.5,
"step": 2800
},
{
"epoch": 2.931937172774869,
"eval_logits/chosen": -0.9921875,
"eval_logits/rejected": 0.83203125,
"eval_logps/chosen": -1888.0,
"eval_logps/rejected": -2080.0,
"eval_loss": 1.7740156650543213,
"eval_rewards/accuracies": 0.6171875,
"eval_rewards/chosen": -15.6875,
"eval_rewards/margins": 2.203125,
"eval_rewards/rejected": -17.875,
"eval_runtime": 58.1383,
"eval_samples_per_second": 34.401,
"eval_steps_per_second": 0.55,
"step": 2800
},
{
"epoch": 2.9424083769633507,
"grad_norm": 0.6900016457542163,
"learning_rate": 5.613151021490459e-08,
"logits/chosen": -3.34375,
"logits/rejected": -1.71875,
"logps/chosen": -1536.0,
"logps/rejected": -2896.0,
"loss": 0.0086,
"rewards/accuracies": 1.0,
"rewards/chosen": -11.9375,
"rewards/margins": 11.25,
"rewards/rejected": -23.25,
"step": 2810
},
{
"epoch": 2.9528795811518327,
"grad_norm": 0.03448366750749116,
"learning_rate": 3.7580289713159965e-08,
"logits/chosen": -3.859375,
"logits/rejected": -0.98828125,
"logps/chosen": -1648.0,
"logps/rejected": -3008.0,
"loss": 0.009,
"rewards/accuracies": 1.0,
"rewards/chosen": -12.25,
"rewards/margins": 11.8125,
"rewards/rejected": -24.125,
"step": 2820
},
{
"epoch": 2.9633507853403143,
"grad_norm": 0.1652917416859425,
"learning_rate": 2.2736006387927277e-08,
"logits/chosen": -3.765625,
"logits/rejected": -1.1015625,
"logps/chosen": -1648.0,
"logps/rejected": -2848.0,
"loss": 0.01,
"rewards/accuracies": 0.981249988079071,
"rewards/chosen": -12.3125,
"rewards/margins": 10.75,
"rewards/rejected": -23.0,
"step": 2830
},
{
"epoch": 2.973821989528796,
"grad_norm": 0.4262082897283008,
"learning_rate": 1.1600864625893604e-08,
"logits/chosen": -3.625,
"logits/rejected": -1.7109375,
"logps/chosen": -1616.0,
"logps/rejected": -3008.0,
"loss": 0.0106,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -12.125,
"rewards/margins": 11.5625,
"rewards/rejected": -23.625,
"step": 2840
},
{
"epoch": 2.9842931937172774,
"grad_norm": 0.06359823598729338,
"learning_rate": 4.176518003573549e-09,
"logits/chosen": -4.40625,
"logits/rejected": -1.8046875,
"logps/chosen": -1608.0,
"logps/rejected": -2864.0,
"loss": 0.0176,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -12.125,
"rewards/margins": 10.9375,
"rewards/rejected": -23.0,
"step": 2850
},
{
"epoch": 2.994764397905759,
"grad_norm": 0.32828869709920777,
"learning_rate": 4.640690417528903e-10,
"logits/chosen": -3.84375,
"logits/rejected": -1.6953125,
"logps/chosen": -1592.0,
"logps/rejected": -2816.0,
"loss": 0.0172,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -12.125,
"rewards/margins": 10.9375,
"rewards/rejected": -23.0,
"step": 2860
},
{
"epoch": 3.0,
"step": 2865,
"total_flos": 0.0,
"train_loss": 0.17748919836981758,
"train_runtime": 12763.6539,
"train_samples_per_second": 14.366,
"train_steps_per_second": 0.224
}
],
"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
}