Control-8B-DPO / trainer_state.json
Delta-Vector's picture
Upload 11 files
9a4a12e verified
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.9934426229508198,
"eval_steps": 500,
"global_step": 152,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.013114754098360656,
"grad_norm": NaN,
"learning_rate": 0.0,
"logits/chosen": -0.7201688289642334,
"logits/rejected": -0.25958356261253357,
"logps/chosen": -235.22982788085938,
"logps/rejected": -110.77284240722656,
"loss": 0.6931,
"rewards/accuracies": 0.0,
"rewards/chosen": 0.0,
"rewards/margins": 0.0,
"rewards/rejected": 0.0,
"step": 1
},
{
"epoch": 0.02622950819672131,
"grad_norm": NaN,
"learning_rate": 0.0,
"logits/chosen": -0.6440946459770203,
"logits/rejected": -0.8803070783615112,
"logps/chosen": -233.35064697265625,
"logps/rejected": -270.211669921875,
"loss": 0.6931,
"rewards/accuracies": 0.0,
"rewards/chosen": 0.0,
"rewards/margins": 0.0,
"rewards/rejected": 0.0,
"step": 2
},
{
"epoch": 0.03934426229508197,
"grad_norm": Infinity,
"learning_rate": 0.0,
"logits/chosen": -0.3529421091079712,
"logits/rejected": -0.46579116582870483,
"logps/chosen": -283.93499755859375,
"logps/rejected": -236.17298889160156,
"loss": 0.6931,
"rewards/accuracies": 0.0,
"rewards/chosen": 0.0,
"rewards/margins": 0.0,
"rewards/rejected": 0.0,
"step": 3
},
{
"epoch": 0.05245901639344262,
"grad_norm": 116.6382827758789,
"learning_rate": 3.125e-07,
"logits/chosen": -0.6149371862411499,
"logits/rejected": -0.6765552163124084,
"logps/chosen": -330.7442321777344,
"logps/rejected": -333.74853515625,
"loss": 0.6931,
"rewards/accuracies": 0.0,
"rewards/chosen": 0.0,
"rewards/margins": 0.0,
"rewards/rejected": 0.0,
"step": 4
},
{
"epoch": 0.06557377049180328,
"grad_norm": 96.14250946044922,
"learning_rate": 6.25e-07,
"logits/chosen": -0.48342880606651306,
"logits/rejected": -0.552554190158844,
"logps/chosen": -224.43968200683594,
"logps/rejected": -243.84121704101562,
"loss": 0.6931,
"rewards/accuracies": 0.0,
"rewards/chosen": 0.0,
"rewards/margins": 0.0,
"rewards/rejected": 0.0,
"step": 5
},
{
"epoch": 0.07868852459016394,
"grad_norm": 119.71466064453125,
"learning_rate": 9.375000000000001e-07,
"logits/chosen": -0.4438214600086212,
"logits/rejected": -0.41411423683166504,
"logps/chosen": -509.1705322265625,
"logps/rejected": -331.69927978515625,
"loss": 0.6942,
"rewards/accuracies": 0.5,
"rewards/chosen": -0.005420684814453125,
"rewards/margins": -0.0020191192161291838,
"rewards/rejected": -0.0034015655983239412,
"step": 6
},
{
"epoch": 0.09180327868852459,
"grad_norm": 89.92263793945312,
"learning_rate": 1.25e-06,
"logits/chosen": -0.5990754961967468,
"logits/rejected": -0.5052899718284607,
"logps/chosen": -353.8023376464844,
"logps/rejected": -293.9719543457031,
"loss": 0.6924,
"rewards/accuracies": 0.75,
"rewards/chosen": 0.0021602632477879524,
"rewards/margins": 0.001445007394067943,
"rewards/rejected": 0.0007152557955123484,
"step": 7
},
{
"epoch": 0.10491803278688525,
"grad_norm": 91.0330810546875,
"learning_rate": 1.5625e-06,
"logits/chosen": -0.7285553216934204,
"logits/rejected": -0.8203944563865662,
"logps/chosen": -201.29092407226562,
"logps/rejected": -310.0835876464844,
"loss": 0.6923,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.0027733801398426294,
"rewards/margins": 0.00168361677788198,
"rewards/rejected": -0.004456996917724609,
"step": 8
},
{
"epoch": 0.1180327868852459,
"grad_norm": 115.72259521484375,
"learning_rate": 1.8750000000000003e-06,
"logits/chosen": -0.6387780904769897,
"logits/rejected": -0.4966055750846863,
"logps/chosen": -286.18316650390625,
"logps/rejected": -174.62039184570312,
"loss": 0.6924,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.008039474487304688,
"rewards/margins": 0.0014192580711096525,
"rewards/rejected": -0.009458731859922409,
"step": 9
},
{
"epoch": 0.13114754098360656,
"grad_norm": 81.71347045898438,
"learning_rate": 2.1875000000000002e-06,
"logits/chosen": -0.43023377656936646,
"logits/rejected": -0.4917900860309601,
"logps/chosen": -123.80950164794922,
"logps/rejected": -231.78781127929688,
"loss": 0.6922,
"rewards/accuracies": 0.25,
"rewards/chosen": -0.013120556250214577,
"rewards/margins": 0.001862620934844017,
"rewards/rejected": -0.014983177185058594,
"step": 10
},
{
"epoch": 0.14426229508196722,
"grad_norm": 126.95443725585938,
"learning_rate": 2.5e-06,
"logits/chosen": -0.5431928634643555,
"logits/rejected": -0.6928932666778564,
"logps/chosen": -466.6590881347656,
"logps/rejected": -450.839111328125,
"loss": 0.6835,
"rewards/accuracies": 0.75,
"rewards/chosen": 0.003947257064282894,
"rewards/margins": 0.01957416534423828,
"rewards/rejected": -0.015626907348632812,
"step": 11
},
{
"epoch": 0.15737704918032788,
"grad_norm": 106.9858169555664,
"learning_rate": 2.8125e-06,
"logits/chosen": -0.8515354990959167,
"logits/rejected": -0.8400233387947083,
"logps/chosen": -496.6812438964844,
"logps/rejected": -367.9149169921875,
"loss": 0.6932,
"rewards/accuracies": 0.5,
"rewards/chosen": -0.013410568237304688,
"rewards/margins": 0.0008895890787243843,
"rewards/rejected": -0.014300156384706497,
"step": 12
},
{
"epoch": 0.17049180327868851,
"grad_norm": 94.6556167602539,
"learning_rate": 3.125e-06,
"logits/chosen": -0.8845179080963135,
"logits/rejected": -0.8273138403892517,
"logps/chosen": -332.0273742675781,
"logps/rejected": -263.0736389160156,
"loss": 0.6739,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.0007216455414891243,
"rewards/margins": 0.03937549516558647,
"rewards/rejected": -0.04009714350104332,
"step": 13
},
{
"epoch": 0.18360655737704917,
"grad_norm": 110.63664245605469,
"learning_rate": 3.4375e-06,
"logits/chosen": -0.49672091007232666,
"logits/rejected": -0.7252901792526245,
"logps/chosen": -182.361572265625,
"logps/rejected": -378.4778747558594,
"loss": 0.6764,
"rewards/accuracies": 0.5,
"rewards/chosen": -0.03787527605891228,
"rewards/margins": 0.0358370766043663,
"rewards/rejected": -0.07371234893798828,
"step": 14
},
{
"epoch": 0.19672131147540983,
"grad_norm": 76.27680969238281,
"learning_rate": 3.7500000000000005e-06,
"logits/chosen": -0.5162211656570435,
"logits/rejected": -0.4975786507129669,
"logps/chosen": -179.30902099609375,
"logps/rejected": -132.12823486328125,
"loss": 0.6905,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.037923429161310196,
"rewards/margins": 0.005405997391790152,
"rewards/rejected": -0.04332943260669708,
"step": 15
},
{
"epoch": 0.2098360655737705,
"grad_norm": 91.94597625732422,
"learning_rate": 4.0625000000000005e-06,
"logits/chosen": -0.3636321425437927,
"logits/rejected": -0.6324741840362549,
"logps/chosen": -153.548828125,
"logps/rejected": -371.84765625,
"loss": 0.713,
"rewards/accuracies": 0.25,
"rewards/chosen": -0.06001920625567436,
"rewards/margins": -0.038457777351140976,
"rewards/rejected": -0.021561432629823685,
"step": 16
},
{
"epoch": 0.22295081967213115,
"grad_norm": 75.85398864746094,
"learning_rate": 4.3750000000000005e-06,
"logits/chosen": -0.5712193250656128,
"logits/rejected": -0.3860137164592743,
"logps/chosen": -267.605224609375,
"logps/rejected": -104.8357162475586,
"loss": 0.6921,
"rewards/accuracies": 0.5,
"rewards/chosen": -0.022034931927919388,
"rewards/margins": 0.0022784238681197166,
"rewards/rejected": -0.02431335672736168,
"step": 17
},
{
"epoch": 0.2360655737704918,
"grad_norm": 117.28124237060547,
"learning_rate": 4.6875000000000004e-06,
"logits/chosen": -0.5285807251930237,
"logits/rejected": -0.3493229150772095,
"logps/chosen": -380.4576416015625,
"logps/rejected": -311.0014343261719,
"loss": 0.7359,
"rewards/accuracies": 0.5,
"rewards/chosen": -0.16734714806079865,
"rewards/margins": -0.07893352955579758,
"rewards/rejected": -0.08841361850500107,
"step": 18
},
{
"epoch": 0.24918032786885247,
"grad_norm": 90.01153564453125,
"learning_rate": 5e-06,
"logits/chosen": -0.6711679697036743,
"logits/rejected": -0.2690754532814026,
"logps/chosen": -462.41973876953125,
"logps/rejected": -297.78363037109375,
"loss": 0.6435,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.0683719664812088,
"rewards/margins": 0.10821684449911118,
"rewards/rejected": -0.17658883333206177,
"step": 19
},
{
"epoch": 0.26229508196721313,
"grad_norm": 73.38824462890625,
"learning_rate": 4.9632352941176475e-06,
"logits/chosen": -0.6787087917327881,
"logits/rejected": -0.6652618050575256,
"logps/chosen": -361.91131591796875,
"logps/rejected": -243.35589599609375,
"loss": 0.6319,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.04816999286413193,
"rewards/margins": 0.12730780243873596,
"rewards/rejected": -0.07913780957460403,
"step": 20
},
{
"epoch": 0.2754098360655738,
"grad_norm": 115.89125061035156,
"learning_rate": 4.9264705882352945e-06,
"logits/chosen": -0.3588840365409851,
"logits/rejected": -0.2116285115480423,
"logps/chosen": -400.4305419921875,
"logps/rejected": -362.8456726074219,
"loss": 0.6087,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.10708275437355042,
"rewards/margins": 0.1842678040266037,
"rewards/rejected": -0.07718505710363388,
"step": 21
},
{
"epoch": 0.28852459016393445,
"grad_norm": 78.77503204345703,
"learning_rate": 4.889705882352942e-06,
"logits/chosen": -0.8203290700912476,
"logits/rejected": -0.8240777254104614,
"logps/chosen": -166.8377685546875,
"logps/rejected": -170.83688354492188,
"loss": 0.7257,
"rewards/accuracies": 0.5,
"rewards/chosen": -0.06841621547937393,
"rewards/margins": -0.061698053032159805,
"rewards/rejected": -0.0067181577906012535,
"step": 22
},
{
"epoch": 0.3016393442622951,
"grad_norm": 121.3331527709961,
"learning_rate": 4.852941176470589e-06,
"logits/chosen": -0.2023768126964569,
"logits/rejected": 0.14346128702163696,
"logps/chosen": -404.2669677734375,
"logps/rejected": -286.7450256347656,
"loss": 0.6238,
"rewards/accuracies": 0.75,
"rewards/chosen": 0.12858295440673828,
"rewards/margins": 0.14947663247585297,
"rewards/rejected": -0.020893670618534088,
"step": 23
},
{
"epoch": 0.31475409836065577,
"grad_norm": 103.55362701416016,
"learning_rate": 4.816176470588236e-06,
"logits/chosen": -0.5769512057304382,
"logits/rejected": -0.7388426661491394,
"logps/chosen": -483.7005310058594,
"logps/rejected": -367.4910583496094,
"loss": 0.5843,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.05827799811959267,
"rewards/margins": 0.27695170044898987,
"rewards/rejected": -0.2186737060546875,
"step": 24
},
{
"epoch": 0.32786885245901637,
"grad_norm": 72.6626205444336,
"learning_rate": 4.779411764705883e-06,
"logits/chosen": -0.7255613803863525,
"logits/rejected": -0.884934663772583,
"logps/chosen": -183.0727081298828,
"logps/rejected": -202.55877685546875,
"loss": 0.6588,
"rewards/accuracies": 0.75,
"rewards/chosen": 0.0200653113424778,
"rewards/margins": 0.09606742858886719,
"rewards/rejected": -0.07600211352109909,
"step": 25
},
{
"epoch": 0.34098360655737703,
"grad_norm": 88.45654296875,
"learning_rate": 4.74264705882353e-06,
"logits/chosen": -0.5392031073570251,
"logits/rejected": -0.7600653171539307,
"logps/chosen": -405.0447998046875,
"logps/rejected": -448.3751220703125,
"loss": 0.5453,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.170909121632576,
"rewards/margins": 0.33048784732818604,
"rewards/rejected": -0.15957871079444885,
"step": 26
},
{
"epoch": 0.3540983606557377,
"grad_norm": 72.38986206054688,
"learning_rate": 4.705882352941177e-06,
"logits/chosen": -0.8080704212188721,
"logits/rejected": -0.4376976788043976,
"logps/chosen": -242.55665588378906,
"logps/rejected": -131.02154541015625,
"loss": 0.7378,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.07056399434804916,
"rewards/margins": -0.06801767647266388,
"rewards/rejected": -0.0025463104248046875,
"step": 27
},
{
"epoch": 0.36721311475409835,
"grad_norm": 89.0915298461914,
"learning_rate": 4.669117647058824e-06,
"logits/chosen": -0.7039719820022583,
"logits/rejected": -0.45313090085983276,
"logps/chosen": -357.1167297363281,
"logps/rejected": -254.34747314453125,
"loss": 0.6511,
"rewards/accuracies": 0.5,
"rewards/chosen": 0.1653575897216797,
"rewards/margins": 0.11446285247802734,
"rewards/rejected": 0.050894737243652344,
"step": 28
},
{
"epoch": 0.380327868852459,
"grad_norm": 72.88048553466797,
"learning_rate": 4.632352941176471e-06,
"logits/chosen": -0.4176654815673828,
"logits/rejected": -0.8021068572998047,
"logps/chosen": -118.46458435058594,
"logps/rejected": -266.1793518066406,
"loss": 0.6207,
"rewards/accuracies": 0.5,
"rewards/chosen": -0.03454189747571945,
"rewards/margins": 0.17219945788383484,
"rewards/rejected": -0.2067413479089737,
"step": 29
},
{
"epoch": 0.39344262295081966,
"grad_norm": 112.41997528076172,
"learning_rate": 4.595588235294118e-06,
"logits/chosen": -0.5515627264976501,
"logits/rejected": -0.3022332489490509,
"logps/chosen": -383.1882629394531,
"logps/rejected": -355.9986877441406,
"loss": 0.7112,
"rewards/accuracies": 0.25,
"rewards/chosen": 0.12149354070425034,
"rewards/margins": 0.018218979239463806,
"rewards/rejected": 0.10327454656362534,
"step": 30
},
{
"epoch": 0.4065573770491803,
"grad_norm": 99.6341781616211,
"learning_rate": 4.558823529411765e-06,
"logits/chosen": -0.32010310888290405,
"logits/rejected": -0.34287354350090027,
"logps/chosen": -433.3697509765625,
"logps/rejected": -427.46868896484375,
"loss": 0.5341,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.1381477415561676,
"rewards/margins": 0.3722812533378601,
"rewards/rejected": -0.2341335415840149,
"step": 31
},
{
"epoch": 0.419672131147541,
"grad_norm": 141.9149932861328,
"learning_rate": 4.522058823529412e-06,
"logits/chosen": -0.3783895969390869,
"logits/rejected": -0.22028671205043793,
"logps/chosen": -438.865478515625,
"logps/rejected": -378.45306396484375,
"loss": 0.7389,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.20297011733055115,
"rewards/margins": -0.03645037114620209,
"rewards/rejected": -0.16651973128318787,
"step": 32
},
{
"epoch": 0.43278688524590164,
"grad_norm": 83.45268249511719,
"learning_rate": 4.485294117647059e-06,
"logits/chosen": -0.4643877446651459,
"logits/rejected": -0.23612239956855774,
"logps/chosen": -323.2715148925781,
"logps/rejected": -257.071533203125,
"loss": 0.6477,
"rewards/accuracies": 0.5,
"rewards/chosen": 0.06585502624511719,
"rewards/margins": 0.09959708154201508,
"rewards/rejected": -0.03374204784631729,
"step": 33
},
{
"epoch": 0.4459016393442623,
"grad_norm": 123.2249526977539,
"learning_rate": 4.448529411764706e-06,
"logits/chosen": -0.7680925130844116,
"logits/rejected": -0.6171376705169678,
"logps/chosen": -480.37481689453125,
"logps/rejected": -430.70379638671875,
"loss": 0.7551,
"rewards/accuracies": 0.5,
"rewards/chosen": -0.217463880777359,
"rewards/margins": -0.08437519520521164,
"rewards/rejected": -0.13308867812156677,
"step": 34
},
{
"epoch": 0.45901639344262296,
"grad_norm": NaN,
"learning_rate": 4.448529411764706e-06,
"logits/chosen": -0.9053277969360352,
"logits/rejected": -0.691852331161499,
"logps/chosen": -409.5699462890625,
"logps/rejected": -475.8656005859375,
"loss": 0.8483,
"rewards/accuracies": 0.25,
"rewards/chosen": -0.18217086791992188,
"rewards/margins": -0.2824680209159851,
"rewards/rejected": 0.10029716044664383,
"step": 35
},
{
"epoch": 0.4721311475409836,
"grad_norm": 105.83935546875,
"learning_rate": 4.411764705882353e-06,
"logits/chosen": -0.28655514121055603,
"logits/rejected": -0.05234713852405548,
"logps/chosen": -563.8226318359375,
"logps/rejected": -375.774658203125,
"loss": 0.6024,
"rewards/accuracies": 0.75,
"rewards/chosen": 0.048460379242897034,
"rewards/margins": 0.28112107515335083,
"rewards/rejected": -0.2326606810092926,
"step": 36
},
{
"epoch": 0.4852459016393443,
"grad_norm": 68.51673126220703,
"learning_rate": 4.3750000000000005e-06,
"logits/chosen": -0.7692010402679443,
"logits/rejected": -0.9052639007568359,
"logps/chosen": -262.10467529296875,
"logps/rejected": -209.62413024902344,
"loss": 0.5991,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.09292706847190857,
"rewards/margins": 0.20516462624073029,
"rewards/rejected": -0.11223754286766052,
"step": 37
},
{
"epoch": 0.49836065573770494,
"grad_norm": 152.35870361328125,
"learning_rate": 4.3382352941176475e-06,
"logits/chosen": -0.4434085190296173,
"logits/rejected": -0.5462446212768555,
"logps/chosen": -325.6352844238281,
"logps/rejected": -386.6681823730469,
"loss": 1.048,
"rewards/accuracies": 0.5,
"rewards/chosen": -0.2829402983188629,
"rewards/margins": -0.5018165707588196,
"rewards/rejected": 0.21887625753879547,
"step": 38
},
{
"epoch": 0.5114754098360655,
"grad_norm": 128.87644958496094,
"learning_rate": 4.301470588235295e-06,
"logits/chosen": -0.32321175932884216,
"logits/rejected": -0.6219479441642761,
"logps/chosen": -220.80865478515625,
"logps/rejected": -679.1721801757812,
"loss": 0.6471,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.08989439904689789,
"rewards/margins": 0.1783585399389267,
"rewards/rejected": -0.268252968788147,
"step": 39
},
{
"epoch": 0.5245901639344263,
"grad_norm": 164.9374542236328,
"learning_rate": 4.264705882352942e-06,
"logits/chosen": -0.41436851024627686,
"logits/rejected": -0.3371754288673401,
"logps/chosen": -453.7709045410156,
"logps/rejected": -186.964599609375,
"loss": 0.7023,
"rewards/accuracies": 0.5,
"rewards/chosen": -0.012208174914121628,
"rewards/margins": 0.0052278414368629456,
"rewards/rejected": -0.01743602193892002,
"step": 40
},
{
"epoch": 0.5377049180327869,
"grad_norm": 85.42108917236328,
"learning_rate": 4.227941176470589e-06,
"logits/chosen": -0.3417915403842926,
"logits/rejected": -0.6106492280960083,
"logps/chosen": -410.3317565917969,
"logps/rejected": -356.129638671875,
"loss": 0.5101,
"rewards/accuracies": 0.75,
"rewards/chosen": 0.12232876569032669,
"rewards/margins": 0.4886293411254883,
"rewards/rejected": -0.3663005828857422,
"step": 41
},
{
"epoch": 0.5508196721311476,
"grad_norm": 92.59017181396484,
"learning_rate": 4.191176470588236e-06,
"logits/chosen": -0.3282807171344757,
"logits/rejected": -0.6609268188476562,
"logps/chosen": -273.1469421386719,
"logps/rejected": -394.88787841796875,
"loss": 0.5399,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.09656943380832672,
"rewards/margins": 0.3409425616264343,
"rewards/rejected": -0.2443731427192688,
"step": 42
},
{
"epoch": 0.5639344262295082,
"grad_norm": 113.27176666259766,
"learning_rate": 4.154411764705883e-06,
"logits/chosen": -0.6696023941040039,
"logits/rejected": -0.4758712649345398,
"logps/chosen": -259.9328918457031,
"logps/rejected": -257.8404846191406,
"loss": 0.7035,
"rewards/accuracies": 0.5,
"rewards/chosen": -0.07427978515625,
"rewards/margins": 0.009069837629795074,
"rewards/rejected": -0.08334961533546448,
"step": 43
},
{
"epoch": 0.5770491803278689,
"grad_norm": 115.66248321533203,
"learning_rate": 4.11764705882353e-06,
"logits/chosen": -0.5908689498901367,
"logits/rejected": -0.6039953231811523,
"logps/chosen": -264.3017578125,
"logps/rejected": -345.41534423828125,
"loss": 0.758,
"rewards/accuracies": 0.5,
"rewards/chosen": -0.16783800721168518,
"rewards/margins": -0.07998895645141602,
"rewards/rejected": -0.08784904330968857,
"step": 44
},
{
"epoch": 0.5901639344262295,
"grad_norm": 114.06060028076172,
"learning_rate": 4.080882352941177e-06,
"logits/chosen": -0.4014369249343872,
"logits/rejected": -0.31871065497398376,
"logps/chosen": -365.35687255859375,
"logps/rejected": -380.46661376953125,
"loss": 0.7162,
"rewards/accuracies": 0.75,
"rewards/chosen": 0.09307613968849182,
"rewards/margins": -0.03436946123838425,
"rewards/rejected": 0.12744559347629547,
"step": 45
},
{
"epoch": 0.6032786885245902,
"grad_norm": 138.62567138671875,
"learning_rate": 4.044117647058824e-06,
"logits/chosen": -0.638996958732605,
"logits/rejected": -0.39604443311691284,
"logps/chosen": -286.7137451171875,
"logps/rejected": -220.05154418945312,
"loss": 0.8332,
"rewards/accuracies": 0.5,
"rewards/chosen": -0.1774437129497528,
"rewards/margins": -0.23038703203201294,
"rewards/rejected": 0.05294332653284073,
"step": 46
},
{
"epoch": 0.6163934426229508,
"grad_norm": 57.410499572753906,
"learning_rate": 4.007352941176471e-06,
"logits/chosen": -0.6090636253356934,
"logits/rejected": -0.5290719866752625,
"logps/chosen": -137.49166870117188,
"logps/rejected": -139.84095764160156,
"loss": 0.5675,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.006125451996922493,
"rewards/margins": 0.29940930008888245,
"rewards/rejected": -0.2932838499546051,
"step": 47
},
{
"epoch": 0.6295081967213115,
"grad_norm": 74.56676483154297,
"learning_rate": 3.970588235294118e-06,
"logits/chosen": -0.7181419134140015,
"logits/rejected": -0.4697718024253845,
"logps/chosen": -189.2108154296875,
"logps/rejected": -129.29373168945312,
"loss": 0.6282,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.16447561979293823,
"rewards/margins": 0.13910141587257385,
"rewards/rejected": -0.3035770356655121,
"step": 48
},
{
"epoch": 0.6426229508196721,
"grad_norm": 71.27189636230469,
"learning_rate": 3.933823529411765e-06,
"logits/chosen": -0.5085806846618652,
"logits/rejected": -0.44473588466644287,
"logps/chosen": -442.37554931640625,
"logps/rejected": -340.0184326171875,
"loss": 0.4663,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.27928200364112854,
"rewards/margins": 0.581198513507843,
"rewards/rejected": -0.3019165098667145,
"step": 49
},
{
"epoch": 0.6557377049180327,
"grad_norm": 94.27779388427734,
"learning_rate": 3.897058823529412e-06,
"logits/chosen": -0.43181851506233215,
"logits/rejected": -0.2016599327325821,
"logps/chosen": -412.7520446777344,
"logps/rejected": -255.32366943359375,
"loss": 0.5602,
"rewards/accuracies": 0.75,
"rewards/chosen": 0.15803948044776917,
"rewards/margins": 0.30720824003219604,
"rewards/rejected": -0.14916877448558807,
"step": 50
},
{
"epoch": 0.6688524590163935,
"grad_norm": 105.73281860351562,
"learning_rate": 3.860294117647059e-06,
"logits/chosen": -0.8714503645896912,
"logits/rejected": -0.8866394758224487,
"logps/chosen": -234.35894775390625,
"logps/rejected": -381.7604064941406,
"loss": 0.679,
"rewards/accuracies": 0.5,
"rewards/chosen": 0.20015183091163635,
"rewards/margins": 0.0683051347732544,
"rewards/rejected": 0.13184669613838196,
"step": 51
},
{
"epoch": 0.6819672131147541,
"grad_norm": 114.59404754638672,
"learning_rate": 3.8235294117647055e-06,
"logits/chosen": -0.3797226846218109,
"logits/rejected": -0.28768518567085266,
"logps/chosen": -527.8758544921875,
"logps/rejected": -398.22833251953125,
"loss": 0.5557,
"rewards/accuracies": 0.75,
"rewards/chosen": 0.1556282937526703,
"rewards/margins": 0.3929787576198578,
"rewards/rejected": -0.2373504638671875,
"step": 52
},
{
"epoch": 0.6950819672131148,
"grad_norm": 79.82359313964844,
"learning_rate": 3.786764705882353e-06,
"logits/chosen": -0.36424490809440613,
"logits/rejected": -0.4016554355621338,
"logps/chosen": -264.1070556640625,
"logps/rejected": -228.17445373535156,
"loss": 0.5864,
"rewards/accuracies": 0.75,
"rewards/chosen": 0.08102244883775711,
"rewards/margins": 0.2558790147304535,
"rewards/rejected": -0.17485657334327698,
"step": 53
},
{
"epoch": 0.7081967213114754,
"grad_norm": 96.55302429199219,
"learning_rate": 3.7500000000000005e-06,
"logits/chosen": -0.2944083511829376,
"logits/rejected": -0.1153193861246109,
"logps/chosen": -548.8867797851562,
"logps/rejected": -476.7703857421875,
"loss": 0.4772,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.578123152256012,
"rewards/margins": 0.5144607424736023,
"rewards/rejected": 0.0636623352766037,
"step": 54
},
{
"epoch": 0.7213114754098361,
"grad_norm": 99.41586303710938,
"learning_rate": 3.7132352941176476e-06,
"logits/chosen": -0.28071513772010803,
"logits/rejected": -0.3517005443572998,
"logps/chosen": -269.35107421875,
"logps/rejected": -245.49374389648438,
"loss": 0.6535,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.32255783677101135,
"rewards/margins": 0.18369215726852417,
"rewards/rejected": -0.5062500238418579,
"step": 55
},
{
"epoch": 0.7344262295081967,
"grad_norm": 102.873046875,
"learning_rate": 3.6764705882352946e-06,
"logits/chosen": -0.5045206546783447,
"logits/rejected": -0.33617326617240906,
"logps/chosen": -446.717529296875,
"logps/rejected": -354.23199462890625,
"loss": 0.5308,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.11581029742956161,
"rewards/margins": 0.36529916524887085,
"rewards/rejected": -0.24948886036872864,
"step": 56
},
{
"epoch": 0.7475409836065574,
"grad_norm": 107.32563018798828,
"learning_rate": 3.6397058823529413e-06,
"logits/chosen": -0.7487730979919434,
"logits/rejected": -0.5132504105567932,
"logps/chosen": -297.9276428222656,
"logps/rejected": -272.4541015625,
"loss": 0.6717,
"rewards/accuracies": 0.75,
"rewards/chosen": 0.00452117994427681,
"rewards/margins": 0.049825482070446014,
"rewards/rejected": -0.045304298400878906,
"step": 57
},
{
"epoch": 0.760655737704918,
"grad_norm": 197.63619995117188,
"learning_rate": 3.6029411764705883e-06,
"logits/chosen": -0.3516054153442383,
"logits/rejected": -0.4525684714317322,
"logps/chosen": -527.5667724609375,
"logps/rejected": -721.8489990234375,
"loss": 0.9658,
"rewards/accuracies": 0.5,
"rewards/chosen": -0.026310719549655914,
"rewards/margins": -0.28759005665779114,
"rewards/rejected": 0.26127931475639343,
"step": 58
},
{
"epoch": 0.7737704918032787,
"grad_norm": 58.864341735839844,
"learning_rate": 3.566176470588236e-06,
"logits/chosen": -0.8907312750816345,
"logits/rejected": -0.5560203194618225,
"logps/chosen": -356.2855224609375,
"logps/rejected": -259.6497802734375,
"loss": 0.5038,
"rewards/accuracies": 0.75,
"rewards/chosen": 0.31428688764572144,
"rewards/margins": 0.4714818000793457,
"rewards/rejected": -0.15719491243362427,
"step": 59
},
{
"epoch": 0.7868852459016393,
"grad_norm": 77.59774017333984,
"learning_rate": 3.529411764705883e-06,
"logits/chosen": -0.5632966160774231,
"logits/rejected": -0.6254044771194458,
"logps/chosen": -523.7625732421875,
"logps/rejected": -410.9280700683594,
"loss": 0.3917,
"rewards/accuracies": 0.75,
"rewards/chosen": 0.16903266310691833,
"rewards/margins": 0.9428247213363647,
"rewards/rejected": -0.7737920880317688,
"step": 60
},
{
"epoch": 0.8,
"grad_norm": 125.0260009765625,
"learning_rate": 3.4926470588235295e-06,
"logits/chosen": -0.7440542578697205,
"logits/rejected": -0.703271210193634,
"logps/chosen": -259.8118591308594,
"logps/rejected": -273.839599609375,
"loss": 0.8524,
"rewards/accuracies": 0.25,
"rewards/chosen": -0.13262319564819336,
"rewards/margins": -0.27512502670288086,
"rewards/rejected": 0.1425018012523651,
"step": 61
},
{
"epoch": 0.8131147540983606,
"grad_norm": 76.26204681396484,
"learning_rate": 3.4558823529411766e-06,
"logits/chosen": -0.7221469283103943,
"logits/rejected": -0.7659784555435181,
"logps/chosen": -184.25143432617188,
"logps/rejected": -179.50430297851562,
"loss": 0.6865,
"rewards/accuracies": 0.5,
"rewards/chosen": 0.0025919005274772644,
"rewards/margins": 0.019242577254772186,
"rewards/rejected": -0.016650676727294922,
"step": 62
},
{
"epoch": 0.8262295081967214,
"grad_norm": 98.49539184570312,
"learning_rate": 3.419117647058824e-06,
"logits/chosen": -0.39552026987075806,
"logits/rejected": -0.5626529455184937,
"logps/chosen": -299.3195495605469,
"logps/rejected": -409.58978271484375,
"loss": 0.6728,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.10388422012329102,
"rewards/margins": 0.0691218301653862,
"rewards/rejected": -0.1730060577392578,
"step": 63
},
{
"epoch": 0.839344262295082,
"grad_norm": 121.93528747558594,
"learning_rate": 3.382352941176471e-06,
"logits/chosen": -0.3543356955051422,
"logits/rejected": -0.23809625208377838,
"logps/chosen": -189.68069458007812,
"logps/rejected": -349.9007263183594,
"loss": 0.8636,
"rewards/accuracies": 0.25,
"rewards/chosen": -0.3187013864517212,
"rewards/margins": -0.2557255029678345,
"rewards/rejected": -0.06297589093446732,
"step": 64
},
{
"epoch": 0.8524590163934426,
"grad_norm": 35.53516387939453,
"learning_rate": 3.3455882352941178e-06,
"logits/chosen": -0.9258323311805725,
"logits/rejected": -0.8565582036972046,
"logps/chosen": -272.64453125,
"logps/rejected": -173.0827178955078,
"loss": 0.5008,
"rewards/accuracies": 0.5,
"rewards/chosen": 0.3930124044418335,
"rewards/margins": 0.5905007123947144,
"rewards/rejected": -0.19748830795288086,
"step": 65
},
{
"epoch": 0.8655737704918033,
"grad_norm": 162.02955627441406,
"learning_rate": 3.308823529411765e-06,
"logits/chosen": -0.7951459288597107,
"logits/rejected": -0.9009961485862732,
"logps/chosen": -322.9057312011719,
"logps/rejected": -461.6522216796875,
"loss": 0.9941,
"rewards/accuracies": 0.25,
"rewards/chosen": -0.22350434958934784,
"rewards/margins": -0.43997135758399963,
"rewards/rejected": 0.2164669930934906,
"step": 66
},
{
"epoch": 0.8786885245901639,
"grad_norm": 49.46587371826172,
"learning_rate": 3.272058823529412e-06,
"logits/chosen": -0.8077062368392944,
"logits/rejected": -0.7711391448974609,
"logps/chosen": -334.55621337890625,
"logps/rejected": -304.2231750488281,
"loss": 0.4265,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.5165804028511047,
"rewards/margins": 0.7691186666488647,
"rewards/rejected": -0.2525383234024048,
"step": 67
},
{
"epoch": 0.8918032786885246,
"grad_norm": 113.88951110839844,
"learning_rate": 3.2352941176470594e-06,
"logits/chosen": -0.48406150937080383,
"logits/rejected": -0.550665020942688,
"logps/chosen": -252.42161560058594,
"logps/rejected": -426.91375732421875,
"loss": 0.7502,
"rewards/accuracies": 0.5,
"rewards/chosen": -0.13350410759449005,
"rewards/margins": -0.05722009390592575,
"rewards/rejected": -0.07628403604030609,
"step": 68
},
{
"epoch": 0.9049180327868852,
"grad_norm": 82.540771484375,
"learning_rate": 3.198529411764706e-06,
"logits/chosen": -0.6168251633644104,
"logits/rejected": -0.5766561031341553,
"logps/chosen": -355.02587890625,
"logps/rejected": -185.5700225830078,
"loss": 0.5665,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.07997532188892365,
"rewards/margins": 0.31432589888572693,
"rewards/rejected": -0.3943012058734894,
"step": 69
},
{
"epoch": 0.9180327868852459,
"grad_norm": 80.47476196289062,
"learning_rate": 3.161764705882353e-06,
"logits/chosen": -0.44328969717025757,
"logits/rejected": -0.23468244075775146,
"logps/chosen": -298.8565673828125,
"logps/rejected": -271.02874755859375,
"loss": 0.5233,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.3650921881198883,
"rewards/margins": 0.4301413893699646,
"rewards/rejected": -0.0650491714477539,
"step": 70
},
{
"epoch": 0.9311475409836065,
"grad_norm": 183.71018981933594,
"learning_rate": 3.125e-06,
"logits/chosen": -0.7024189233779907,
"logits/rejected": -0.3720092177391052,
"logps/chosen": -663.25341796875,
"logps/rejected": -509.05072021484375,
"loss": 0.6182,
"rewards/accuracies": 0.5,
"rewards/chosen": 0.02078094333410263,
"rewards/margins": 0.3649251461029053,
"rewards/rejected": -0.3441442549228668,
"step": 71
},
{
"epoch": 0.9442622950819672,
"grad_norm": 73.77120971679688,
"learning_rate": 3.0882352941176476e-06,
"logits/chosen": -0.7303851842880249,
"logits/rejected": -0.5275728702545166,
"logps/chosen": -273.287353515625,
"logps/rejected": -248.97366333007812,
"loss": 0.5902,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.15096759796142578,
"rewards/margins": 0.22571565210819244,
"rewards/rejected": -0.07474804669618607,
"step": 72
},
{
"epoch": 0.9573770491803278,
"grad_norm": 119.30152893066406,
"learning_rate": 3.0514705882352947e-06,
"logits/chosen": -0.7145899534225464,
"logits/rejected": -0.6755335330963135,
"logps/chosen": -358.4460144042969,
"logps/rejected": -413.93402099609375,
"loss": 0.715,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.5442755818367004,
"rewards/margins": -0.015435710549354553,
"rewards/rejected": -0.5288398861885071,
"step": 73
},
{
"epoch": 0.9704918032786886,
"grad_norm": Infinity,
"learning_rate": 3.0514705882352947e-06,
"logits/chosen": -0.37683844566345215,
"logits/rejected": -0.19479577243328094,
"logps/chosen": -191.80364990234375,
"logps/rejected": -239.4271240234375,
"loss": 0.6901,
"rewards/accuracies": 0.5,
"rewards/chosen": -0.12151336669921875,
"rewards/margins": 0.011674299836158752,
"rewards/rejected": -0.1331876814365387,
"step": 74
},
{
"epoch": 0.9836065573770492,
"grad_norm": 37.73300552368164,
"learning_rate": 3.0147058823529413e-06,
"logits/chosen": -0.7108013033866882,
"logits/rejected": -0.8427099585533142,
"logps/chosen": -237.83132934570312,
"logps/rejected": -235.86509704589844,
"loss": 0.5368,
"rewards/accuracies": 0.5,
"rewards/chosen": 0.3780634105205536,
"rewards/margins": 0.5749028921127319,
"rewards/rejected": -0.19683943688869476,
"step": 75
},
{
"epoch": 0.9967213114754099,
"grad_norm": 50.56627655029297,
"learning_rate": 2.9779411764705884e-06,
"logits/chosen": -0.3920336663722992,
"logits/rejected": -0.5925789475440979,
"logps/chosen": -352.024658203125,
"logps/rejected": -319.0198669433594,
"loss": 0.3911,
"rewards/accuracies": 0.75,
"rewards/chosen": 0.4176803529262543,
"rewards/margins": 0.8605508804321289,
"rewards/rejected": -0.44287049770355225,
"step": 76
},
{
"epoch": 1.0098360655737706,
"grad_norm": 31.756732940673828,
"learning_rate": 2.9411764705882355e-06,
"logits/chosen": -0.9694571495056152,
"logits/rejected": -0.7683681845664978,
"logps/chosen": -397.28277587890625,
"logps/rejected": -506.4493103027344,
"loss": 0.2859,
"rewards/accuracies": 1.0,
"rewards/chosen": 3.0198912620544434,
"rewards/margins": 6.584094524383545,
"rewards/rejected": -3.5642035007476807,
"step": 77
},
{
"epoch": 1.022950819672131,
"grad_norm": 15.996756553649902,
"learning_rate": 2.904411764705883e-06,
"logits/chosen": -0.4139990210533142,
"logits/rejected": -0.5289357900619507,
"logps/chosen": -157.25994873046875,
"logps/rejected": -292.02252197265625,
"loss": 0.0868,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.7793710827827454,
"rewards/margins": 3.51633358001709,
"rewards/rejected": -2.73696231842041,
"step": 78
},
{
"epoch": 1.0360655737704918,
"grad_norm": 4.163623809814453,
"learning_rate": 2.8676470588235296e-06,
"logits/chosen": -0.8043258190155029,
"logits/rejected": -0.7802602052688599,
"logps/chosen": -419.2426452636719,
"logps/rejected": -306.1612548828125,
"loss": 0.0156,
"rewards/accuracies": 1.0,
"rewards/chosen": 2.816192626953125,
"rewards/margins": 5.203756332397461,
"rewards/rejected": -2.3875632286071777,
"step": 79
},
{
"epoch": 1.0491803278688525,
"grad_norm": 13.230965614318848,
"learning_rate": 2.8308823529411766e-06,
"logits/chosen": -0.6542176008224487,
"logits/rejected": -0.3940190374851227,
"logps/chosen": -421.56170654296875,
"logps/rejected": -406.03619384765625,
"loss": 0.1049,
"rewards/accuracies": 1.0,
"rewards/chosen": 3.0519039630889893,
"rewards/margins": 5.3741607666015625,
"rewards/rejected": -2.322256565093994,
"step": 80
},
{
"epoch": 1.0622950819672132,
"grad_norm": 31.09598731994629,
"learning_rate": 2.7941176470588237e-06,
"logits/chosen": -0.7882400751113892,
"logits/rejected": -0.755602240562439,
"logps/chosen": -143.84951782226562,
"logps/rejected": -207.64418029785156,
"loss": 0.3054,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.5518492460250854,
"rewards/margins": 1.3898382186889648,
"rewards/rejected": -0.8379888534545898,
"step": 81
},
{
"epoch": 1.0754098360655737,
"grad_norm": 7.787254810333252,
"learning_rate": 2.757352941176471e-06,
"logits/chosen": -0.347468763589859,
"logits/rejected": -0.34622734785079956,
"logps/chosen": -263.9562072753906,
"logps/rejected": -390.29901123046875,
"loss": 0.0387,
"rewards/accuracies": 1.0,
"rewards/chosen": 2.743457555770874,
"rewards/margins": 6.276088714599609,
"rewards/rejected": -3.5326313972473145,
"step": 82
},
{
"epoch": 1.0885245901639344,
"grad_norm": 12.739912986755371,
"learning_rate": 2.720588235294118e-06,
"logits/chosen": -0.5520642399787903,
"logits/rejected": -0.632639467716217,
"logps/chosen": -542.7330322265625,
"logps/rejected": -477.4020080566406,
"loss": 0.0857,
"rewards/accuracies": 1.0,
"rewards/chosen": 3.9007229804992676,
"rewards/margins": 8.109764099121094,
"rewards/rejected": -4.209041595458984,
"step": 83
},
{
"epoch": 1.1016393442622952,
"grad_norm": 39.59391784667969,
"learning_rate": 2.683823529411765e-06,
"logits/chosen": -0.5291910767555237,
"logits/rejected": -0.47945737838745117,
"logps/chosen": -385.80242919921875,
"logps/rejected": -362.5047912597656,
"loss": 0.1411,
"rewards/accuracies": 1.0,
"rewards/chosen": 1.6711763143539429,
"rewards/margins": 3.675436496734619,
"rewards/rejected": -2.004260301589966,
"step": 84
},
{
"epoch": 1.1147540983606556,
"grad_norm": 34.75082015991211,
"learning_rate": 2.647058823529412e-06,
"logits/chosen": -0.8582125902175903,
"logits/rejected": -0.961124837398529,
"logps/chosen": -127.2431640625,
"logps/rejected": -181.58657836914062,
"loss": 0.3015,
"rewards/accuracies": 1.0,
"rewards/chosen": 1.1156275272369385,
"rewards/margins": 2.283085346221924,
"rewards/rejected": -1.1674578189849854,
"step": 85
},
{
"epoch": 1.1278688524590164,
"grad_norm": 18.05373191833496,
"learning_rate": 2.610294117647059e-06,
"logits/chosen": -0.8542808890342712,
"logits/rejected": -0.6588492393493652,
"logps/chosen": -249.22372436523438,
"logps/rejected": -176.06527709960938,
"loss": 0.1498,
"rewards/accuracies": 1.0,
"rewards/chosen": 1.803849697113037,
"rewards/margins": 2.9191579818725586,
"rewards/rejected": -1.115308165550232,
"step": 86
},
{
"epoch": 1.140983606557377,
"grad_norm": 5.507705211639404,
"learning_rate": 2.5735294117647057e-06,
"logits/chosen": -0.4963078498840332,
"logits/rejected": -0.8181336522102356,
"logps/chosen": -108.07630920410156,
"logps/rejected": -320.96392822265625,
"loss": 0.0256,
"rewards/accuracies": 1.0,
"rewards/chosen": 1.0501946210861206,
"rewards/margins": 4.599462032318115,
"rewards/rejected": -3.549267292022705,
"step": 87
},
{
"epoch": 1.1540983606557378,
"grad_norm": 12.978127479553223,
"learning_rate": 2.536764705882353e-06,
"logits/chosen": -0.8022123575210571,
"logits/rejected": -0.8916893005371094,
"logps/chosen": -98.68746948242188,
"logps/rejected": -129.98577880859375,
"loss": 0.2735,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.5756351947784424,
"rewards/margins": 1.4945859909057617,
"rewards/rejected": -0.9189507961273193,
"step": 88
},
{
"epoch": 1.1672131147540983,
"grad_norm": 8.445465087890625,
"learning_rate": 2.5e-06,
"logits/chosen": -0.238316610455513,
"logits/rejected": -0.21277102828025818,
"logps/chosen": -355.0726318359375,
"logps/rejected": -468.03424072265625,
"loss": 0.0439,
"rewards/accuracies": 1.0,
"rewards/chosen": 3.785343885421753,
"rewards/margins": 8.177167892456055,
"rewards/rejected": -4.391823768615723,
"step": 89
},
{
"epoch": 1.180327868852459,
"grad_norm": 119.01290130615234,
"learning_rate": 2.4632352941176473e-06,
"logits/chosen": -0.38539862632751465,
"logits/rejected": -0.2583141326904297,
"logps/chosen": -479.6025695800781,
"logps/rejected": -433.03045654296875,
"loss": 0.2972,
"rewards/accuracies": 0.75,
"rewards/chosen": 2.800222873687744,
"rewards/margins": 6.087710380554199,
"rewards/rejected": -3.287487506866455,
"step": 90
},
{
"epoch": 1.1934426229508197,
"grad_norm": 31.620267868041992,
"learning_rate": 2.4264705882352943e-06,
"logits/chosen": -0.5990991592407227,
"logits/rejected": -0.4338776469230652,
"logps/chosen": -287.31878662109375,
"logps/rejected": -425.55963134765625,
"loss": 0.2764,
"rewards/accuracies": 1.0,
"rewards/chosen": 1.9531813859939575,
"rewards/margins": 4.420433044433594,
"rewards/rejected": -2.4672515392303467,
"step": 91
},
{
"epoch": 1.2065573770491804,
"grad_norm": 5.715668201446533,
"learning_rate": 2.3897058823529414e-06,
"logits/chosen": -0.5918264389038086,
"logits/rejected": -0.34481585025787354,
"logps/chosen": -462.36181640625,
"logps/rejected": -291.8490295410156,
"loss": 0.0246,
"rewards/accuracies": 1.0,
"rewards/chosen": 4.236080169677734,
"rewards/margins": 6.656676292419434,
"rewards/rejected": -2.420595645904541,
"step": 92
},
{
"epoch": 1.219672131147541,
"grad_norm": 10.8475923538208,
"learning_rate": 2.3529411764705885e-06,
"logits/chosen": -0.48867395520210266,
"logits/rejected": -0.2037501335144043,
"logps/chosen": -248.76011657714844,
"logps/rejected": -278.86505126953125,
"loss": 0.0703,
"rewards/accuracies": 1.0,
"rewards/chosen": 2.4005353450775146,
"rewards/margins": 3.532731056213379,
"rewards/rejected": -1.1321957111358643,
"step": 93
},
{
"epoch": 1.2327868852459016,
"grad_norm": 17.5373477935791,
"learning_rate": 2.3161764705882355e-06,
"logits/chosen": -0.800743043422699,
"logits/rejected": -0.3216054141521454,
"logps/chosen": -332.1054382324219,
"logps/rejected": -262.9385070800781,
"loss": 0.0781,
"rewards/accuracies": 1.0,
"rewards/chosen": 3.3670170307159424,
"rewards/margins": 5.165108680725098,
"rewards/rejected": -1.7980915307998657,
"step": 94
},
{
"epoch": 1.2459016393442623,
"grad_norm": 17.166118621826172,
"learning_rate": 2.2794117647058826e-06,
"logits/chosen": -0.5876657366752625,
"logits/rejected": -0.30134695768356323,
"logps/chosen": -438.53070068359375,
"logps/rejected": -283.11053466796875,
"loss": 0.0878,
"rewards/accuracies": 1.0,
"rewards/chosen": 5.845110893249512,
"rewards/margins": 7.538168430328369,
"rewards/rejected": -1.6930572986602783,
"step": 95
},
{
"epoch": 1.2590163934426228,
"grad_norm": 23.068361282348633,
"learning_rate": 2.2426470588235296e-06,
"logits/chosen": -0.6558918356895447,
"logits/rejected": -0.3275706171989441,
"logps/chosen": -370.03094482421875,
"logps/rejected": -453.90203857421875,
"loss": 0.1287,
"rewards/accuracies": 1.0,
"rewards/chosen": 2.0850746631622314,
"rewards/margins": 4.879607677459717,
"rewards/rejected": -2.7945332527160645,
"step": 96
},
{
"epoch": 1.2721311475409836,
"grad_norm": 5.994075298309326,
"learning_rate": 2.2058823529411767e-06,
"logits/chosen": -0.6101892590522766,
"logits/rejected": -0.5373432040214539,
"logps/chosen": -347.5174560546875,
"logps/rejected": -322.55023193359375,
"loss": 0.0318,
"rewards/accuracies": 1.0,
"rewards/chosen": 2.3236427307128906,
"rewards/margins": 4.85604190826416,
"rewards/rejected": -2.5323991775512695,
"step": 97
},
{
"epoch": 1.2852459016393443,
"grad_norm": 14.375985145568848,
"learning_rate": 2.1691176470588238e-06,
"logits/chosen": -0.6097806692123413,
"logits/rejected": -0.5758112072944641,
"logps/chosen": -341.9951477050781,
"logps/rejected": -290.09136962890625,
"loss": 0.1422,
"rewards/accuracies": 1.0,
"rewards/chosen": 2.372694253921509,
"rewards/margins": 5.112575531005859,
"rewards/rejected": -2.7398810386657715,
"step": 98
},
{
"epoch": 1.298360655737705,
"grad_norm": 10.561140060424805,
"learning_rate": 2.132352941176471e-06,
"logits/chosen": -0.7798110842704773,
"logits/rejected": -0.5334391593933105,
"logps/chosen": -122.71707916259766,
"logps/rejected": -138.75685119628906,
"loss": 0.224,
"rewards/accuracies": 0.75,
"rewards/chosen": 0.45279568433761597,
"rewards/margins": 2.1160295009613037,
"rewards/rejected": -1.6632338762283325,
"step": 99
},
{
"epoch": 1.3114754098360657,
"grad_norm": 8.718037605285645,
"learning_rate": 2.095588235294118e-06,
"logits/chosen": -0.5125188827514648,
"logits/rejected": -0.35307127237319946,
"logps/chosen": -180.09408569335938,
"logps/rejected": -273.54046630859375,
"loss": 0.0465,
"rewards/accuracies": 1.0,
"rewards/chosen": 1.8067989349365234,
"rewards/margins": 5.022237777709961,
"rewards/rejected": -3.2154390811920166,
"step": 100
},
{
"epoch": 1.3245901639344262,
"grad_norm": 11.257354736328125,
"learning_rate": 2.058823529411765e-06,
"logits/chosen": -0.48480626940727234,
"logits/rejected": -0.4184003174304962,
"logps/chosen": -349.5723876953125,
"logps/rejected": -458.3537902832031,
"loss": 0.0341,
"rewards/accuracies": 1.0,
"rewards/chosen": 2.9435834884643555,
"rewards/margins": 6.52362060546875,
"rewards/rejected": -3.5800368785858154,
"step": 101
},
{
"epoch": 1.337704918032787,
"grad_norm": 8.118950843811035,
"learning_rate": 2.022058823529412e-06,
"logits/chosen": -0.6667636632919312,
"logits/rejected": -0.3804593086242676,
"logps/chosen": -331.16015625,
"logps/rejected": -266.6229248046875,
"loss": 0.1422,
"rewards/accuracies": 1.0,
"rewards/chosen": 4.118168830871582,
"rewards/margins": 5.915567398071289,
"rewards/rejected": -1.797398328781128,
"step": 102
},
{
"epoch": 1.3508196721311476,
"grad_norm": 14.03177547454834,
"learning_rate": 1.985294117647059e-06,
"logits/chosen": -0.34312984347343445,
"logits/rejected": -0.5378394722938538,
"logps/chosen": -205.9506378173828,
"logps/rejected": -219.46360778808594,
"loss": 0.113,
"rewards/accuracies": 1.0,
"rewards/chosen": 2.023257255554199,
"rewards/margins": 3.4092535972595215,
"rewards/rejected": -1.3859964609146118,
"step": 103
},
{
"epoch": 1.3639344262295081,
"grad_norm": 2.7447023391723633,
"learning_rate": 1.948529411764706e-06,
"logits/chosen": -0.3522460162639618,
"logits/rejected": -0.4929681122303009,
"logps/chosen": -295.0723571777344,
"logps/rejected": -344.48834228515625,
"loss": 0.0114,
"rewards/accuracies": 1.0,
"rewards/chosen": 2.613781690597534,
"rewards/margins": 6.21614408493042,
"rewards/rejected": -3.602362871170044,
"step": 104
},
{
"epoch": 1.3770491803278688,
"grad_norm": 3.9517393112182617,
"learning_rate": 1.9117647058823528e-06,
"logits/chosen": -0.5514786243438721,
"logits/rejected": -0.515537440776825,
"logps/chosen": -283.73504638671875,
"logps/rejected": -354.9101257324219,
"loss": 0.0147,
"rewards/accuracies": 1.0,
"rewards/chosen": 3.183412551879883,
"rewards/margins": 6.139578819274902,
"rewards/rejected": -2.9561657905578613,
"step": 105
},
{
"epoch": 1.3901639344262295,
"grad_norm": 32.080848693847656,
"learning_rate": 1.8750000000000003e-06,
"logits/chosen": -0.34524667263031006,
"logits/rejected": -0.2804412245750427,
"logps/chosen": -487.1918029785156,
"logps/rejected": -459.1595764160156,
"loss": 0.1497,
"rewards/accuracies": 1.0,
"rewards/chosen": 3.4442732334136963,
"rewards/margins": 6.211602210998535,
"rewards/rejected": -2.767328977584839,
"step": 106
},
{
"epoch": 1.40327868852459,
"grad_norm": 6.278583526611328,
"learning_rate": 1.8382352941176473e-06,
"logits/chosen": -1.0492626428604126,
"logits/rejected": -0.637208878993988,
"logps/chosen": -508.1822509765625,
"logps/rejected": -264.3661804199219,
"loss": 0.0256,
"rewards/accuracies": 1.0,
"rewards/chosen": 3.434685230255127,
"rewards/margins": 5.93399715423584,
"rewards/rejected": -2.4993114471435547,
"step": 107
},
{
"epoch": 1.4163934426229507,
"grad_norm": 3.8192379474639893,
"learning_rate": 1.8014705882352942e-06,
"logits/chosen": -0.49397462606430054,
"logits/rejected": -0.39174699783325195,
"logps/chosen": -250.099853515625,
"logps/rejected": -342.03033447265625,
"loss": 0.0194,
"rewards/accuracies": 1.0,
"rewards/chosen": 1.3859384059906006,
"rewards/margins": 4.495996475219727,
"rewards/rejected": -3.110058307647705,
"step": 108
},
{
"epoch": 1.4295081967213115,
"grad_norm": 5.807373523712158,
"learning_rate": 1.7647058823529414e-06,
"logits/chosen": -0.584580659866333,
"logits/rejected": -0.15329986810684204,
"logps/chosen": -442.9632873535156,
"logps/rejected": -318.378173828125,
"loss": 0.0165,
"rewards/accuracies": 1.0,
"rewards/chosen": 2.1344716548919678,
"rewards/margins": 5.192463397979736,
"rewards/rejected": -3.0579919815063477,
"step": 109
},
{
"epoch": 1.4426229508196722,
"grad_norm": 3.5901246070861816,
"learning_rate": 1.7279411764705883e-06,
"logits/chosen": -0.5349779725074768,
"logits/rejected": -0.2637510299682617,
"logps/chosen": -414.6310729980469,
"logps/rejected": -428.139404296875,
"loss": 0.0136,
"rewards/accuracies": 1.0,
"rewards/chosen": 2.6161444187164307,
"rewards/margins": 7.400443077087402,
"rewards/rejected": -4.784298896789551,
"step": 110
},
{
"epoch": 1.455737704918033,
"grad_norm": 17.46812629699707,
"learning_rate": 1.6911764705882356e-06,
"logits/chosen": -0.6205960512161255,
"logits/rejected": -0.6515633463859558,
"logps/chosen": -233.56886291503906,
"logps/rejected": -327.51611328125,
"loss": 0.1822,
"rewards/accuracies": 1.0,
"rewards/chosen": 1.1055248975753784,
"rewards/margins": 4.226891994476318,
"rewards/rejected": -3.1213672161102295,
"step": 111
},
{
"epoch": 1.4688524590163934,
"grad_norm": 11.27214241027832,
"learning_rate": 1.6544117647058824e-06,
"logits/chosen": -0.3313244581222534,
"logits/rejected": -0.22155167162418365,
"logps/chosen": -391.53466796875,
"logps/rejected": -230.6190643310547,
"loss": 0.1142,
"rewards/accuracies": 1.0,
"rewards/chosen": 2.988477945327759,
"rewards/margins": 5.188130855560303,
"rewards/rejected": -2.199653148651123,
"step": 112
},
{
"epoch": 1.481967213114754,
"grad_norm": 9.786858558654785,
"learning_rate": 1.6176470588235297e-06,
"logits/chosen": -0.5634331703186035,
"logits/rejected": -0.4348165690898895,
"logps/chosen": -252.67152404785156,
"logps/rejected": -352.4822998046875,
"loss": 0.0467,
"rewards/accuracies": 1.0,
"rewards/chosen": 1.8047726154327393,
"rewards/margins": 6.142679691314697,
"rewards/rejected": -4.337907314300537,
"step": 113
},
{
"epoch": 1.4950819672131148,
"grad_norm": 4.376170635223389,
"learning_rate": 1.5808823529411765e-06,
"logits/chosen": -0.15515179932117462,
"logits/rejected": -0.1870616376399994,
"logps/chosen": -289.219970703125,
"logps/rejected": -400.1667175292969,
"loss": 0.0207,
"rewards/accuracies": 1.0,
"rewards/chosen": 2.1680092811584473,
"rewards/margins": 4.854602336883545,
"rewards/rejected": -2.6865930557250977,
"step": 114
},
{
"epoch": 1.5081967213114753,
"grad_norm": 3.8763139247894287,
"learning_rate": 1.5441176470588238e-06,
"logits/chosen": -0.2717803716659546,
"logits/rejected": -0.17628008127212524,
"logps/chosen": -546.1766357421875,
"logps/rejected": -451.3502502441406,
"loss": 0.0166,
"rewards/accuracies": 1.0,
"rewards/chosen": 2.681325912475586,
"rewards/margins": 5.725113868713379,
"rewards/rejected": -3.0437874794006348,
"step": 115
},
{
"epoch": 1.521311475409836,
"grad_norm": 30.37358283996582,
"learning_rate": 1.5073529411764707e-06,
"logits/chosen": -0.9199023246765137,
"logits/rejected": -0.8192650079727173,
"logps/chosen": -247.5410919189453,
"logps/rejected": -390.4881896972656,
"loss": 0.1679,
"rewards/accuracies": 1.0,
"rewards/chosen": 1.817814588546753,
"rewards/margins": 4.526091575622559,
"rewards/rejected": -2.7082767486572266,
"step": 116
},
{
"epoch": 1.5344262295081967,
"grad_norm": 17.32929229736328,
"learning_rate": 1.4705882352941177e-06,
"logits/chosen": -0.4218984842300415,
"logits/rejected": -0.411405086517334,
"logps/chosen": -202.14004516601562,
"logps/rejected": -220.29254150390625,
"loss": 0.1733,
"rewards/accuracies": 1.0,
"rewards/chosen": 1.8182326555252075,
"rewards/margins": 3.9761695861816406,
"rewards/rejected": -2.1579370498657227,
"step": 117
},
{
"epoch": 1.5475409836065572,
"grad_norm": 15.47387981414795,
"learning_rate": 1.4338235294117648e-06,
"logits/chosen": -0.6655436754226685,
"logits/rejected": -0.9528688192367554,
"logps/chosen": -110.75199890136719,
"logps/rejected": -216.50701904296875,
"loss": 0.0995,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.448855996131897,
"rewards/margins": 2.9766037464141846,
"rewards/rejected": -2.5277481079101562,
"step": 118
},
{
"epoch": 1.5606557377049182,
"grad_norm": 56.997802734375,
"learning_rate": 1.3970588235294119e-06,
"logits/chosen": -0.3543809652328491,
"logits/rejected": -0.5634450912475586,
"logps/chosen": -307.3863525390625,
"logps/rejected": -422.4675598144531,
"loss": 0.19,
"rewards/accuracies": 0.75,
"rewards/chosen": 1.821000337600708,
"rewards/margins": 5.365086555480957,
"rewards/rejected": -3.5440866947174072,
"step": 119
},
{
"epoch": 1.5737704918032787,
"grad_norm": 5.717536926269531,
"learning_rate": 1.360294117647059e-06,
"logits/chosen": -0.394744873046875,
"logits/rejected": -0.46669620275497437,
"logps/chosen": -162.78341674804688,
"logps/rejected": -354.51654052734375,
"loss": 0.0251,
"rewards/accuracies": 1.0,
"rewards/chosen": 1.3121498823165894,
"rewards/margins": 5.8546037673950195,
"rewards/rejected": -4.542453765869141,
"step": 120
},
{
"epoch": 1.5868852459016394,
"grad_norm": 102.92301177978516,
"learning_rate": 1.323529411764706e-06,
"logits/chosen": -0.8797488212585449,
"logits/rejected": -0.5256719589233398,
"logps/chosen": -400.2969055175781,
"logps/rejected": -228.6015625,
"loss": 0.3735,
"rewards/accuracies": 0.75,
"rewards/chosen": 1.3958312273025513,
"rewards/margins": 2.500825881958008,
"rewards/rejected": -1.1049946546554565,
"step": 121
},
{
"epoch": 1.6,
"grad_norm": 12.96183967590332,
"learning_rate": 1.2867647058823528e-06,
"logits/chosen": -0.1941753625869751,
"logits/rejected": -0.2740371525287628,
"logps/chosen": -146.65438842773438,
"logps/rejected": -286.4040832519531,
"loss": 0.0833,
"rewards/accuracies": 1.0,
"rewards/chosen": 1.6577894687652588,
"rewards/margins": 5.093594074249268,
"rewards/rejected": -3.4358043670654297,
"step": 122
},
{
"epoch": 1.6131147540983606,
"grad_norm": 1.308056354522705,
"learning_rate": 1.25e-06,
"logits/chosen": -0.2991219460964203,
"logits/rejected": -0.608604371547699,
"logps/chosen": -332.8055419921875,
"logps/rejected": -501.82794189453125,
"loss": 0.0056,
"rewards/accuracies": 1.0,
"rewards/chosen": 1.9228206872940063,
"rewards/margins": 6.52579927444458,
"rewards/rejected": -4.602978706359863,
"step": 123
},
{
"epoch": 1.6262295081967213,
"grad_norm": 6.536846160888672,
"learning_rate": 1.2132352941176472e-06,
"logits/chosen": -0.5168918371200562,
"logits/rejected": -0.8077197670936584,
"logps/chosen": -200.44891357421875,
"logps/rejected": -283.05535888671875,
"loss": 0.0305,
"rewards/accuracies": 1.0,
"rewards/chosen": 1.2009633779525757,
"rewards/margins": 4.113708019256592,
"rewards/rejected": -2.9127449989318848,
"step": 124
},
{
"epoch": 1.639344262295082,
"grad_norm": 14.031917572021484,
"learning_rate": 1.1764705882352942e-06,
"logits/chosen": -0.2985928952693939,
"logits/rejected": -0.5871812701225281,
"logps/chosen": -203.19163513183594,
"logps/rejected": -404.0279846191406,
"loss": 0.1122,
"rewards/accuracies": 1.0,
"rewards/chosen": 1.369646668434143,
"rewards/margins": 5.168557643890381,
"rewards/rejected": -3.7989113330841064,
"step": 125
},
{
"epoch": 1.6524590163934425,
"grad_norm": 1.0998114347457886,
"learning_rate": 1.1397058823529413e-06,
"logits/chosen": -0.9684606790542603,
"logits/rejected": -0.8049026727676392,
"logps/chosen": -247.9980010986328,
"logps/rejected": -212.36807250976562,
"loss": 0.178,
"rewards/accuracies": 0.75,
"rewards/chosen": 2.4171135425567627,
"rewards/margins": 5.950634002685547,
"rewards/rejected": -3.533520460128784,
"step": 126
},
{
"epoch": 1.6655737704918034,
"grad_norm": 1.8453718423843384,
"learning_rate": 1.1029411764705884e-06,
"logits/chosen": -0.4717595875263214,
"logits/rejected": -0.13389728963375092,
"logps/chosen": -400.8570251464844,
"logps/rejected": -322.6037902832031,
"loss": 0.0069,
"rewards/accuracies": 1.0,
"rewards/chosen": 3.168325901031494,
"rewards/margins": 6.738244533538818,
"rewards/rejected": -3.569918155670166,
"step": 127
},
{
"epoch": 1.678688524590164,
"grad_norm": 9.327398300170898,
"learning_rate": 1.0661764705882354e-06,
"logits/chosen": -0.45057347416877747,
"logits/rejected": -0.22152967751026154,
"logps/chosen": -261.6713562011719,
"logps/rejected": -318.07391357421875,
"loss": 0.0558,
"rewards/accuracies": 1.0,
"rewards/chosen": 2.15228271484375,
"rewards/margins": 5.979676723480225,
"rewards/rejected": -3.8273940086364746,
"step": 128
},
{
"epoch": 1.6918032786885244,
"grad_norm": 21.2039737701416,
"learning_rate": 1.0294117647058825e-06,
"logits/chosen": -0.5791776180267334,
"logits/rejected": -0.8412386178970337,
"logps/chosen": -132.2311553955078,
"logps/rejected": -418.17657470703125,
"loss": 0.1523,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.6422145962715149,
"rewards/margins": 6.080798625946045,
"rewards/rejected": -5.438583850860596,
"step": 129
},
{
"epoch": 1.7049180327868854,
"grad_norm": 0.3313053250312805,
"learning_rate": 9.926470588235295e-07,
"logits/chosen": -0.7310227155685425,
"logits/rejected": -0.6085512638092041,
"logps/chosen": -517.57373046875,
"logps/rejected": -645.620849609375,
"loss": 0.001,
"rewards/accuracies": 1.0,
"rewards/chosen": 3.5617308616638184,
"rewards/margins": 8.085519790649414,
"rewards/rejected": -4.5237884521484375,
"step": 130
},
{
"epoch": 1.7180327868852459,
"grad_norm": 71.060546875,
"learning_rate": 9.558823529411764e-07,
"logits/chosen": -0.6338833570480347,
"logits/rejected": -0.897128701210022,
"logps/chosen": -193.55189514160156,
"logps/rejected": -340.1580810546875,
"loss": 0.3464,
"rewards/accuracies": 0.75,
"rewards/chosen": 0.23160991072654724,
"rewards/margins": 1.533915400505066,
"rewards/rejected": -1.3023054599761963,
"step": 131
},
{
"epoch": 1.7311475409836066,
"grad_norm": 11.899687767028809,
"learning_rate": 9.191176470588237e-07,
"logits/chosen": -0.8149244785308838,
"logits/rejected": -0.5785297155380249,
"logps/chosen": -213.12173461914062,
"logps/rejected": -237.8140411376953,
"loss": 0.0893,
"rewards/accuracies": 1.0,
"rewards/chosen": 1.6387001276016235,
"rewards/margins": 3.045469284057617,
"rewards/rejected": -1.4067691564559937,
"step": 132
},
{
"epoch": 1.7442622950819673,
"grad_norm": 1.8859055042266846,
"learning_rate": 8.823529411764707e-07,
"logits/chosen": -0.40556299686431885,
"logits/rejected": -0.577975332736969,
"logps/chosen": -514.1160888671875,
"logps/rejected": -692.885009765625,
"loss": 0.0032,
"rewards/accuracies": 1.0,
"rewards/chosen": 2.4865522384643555,
"rewards/margins": 9.726764678955078,
"rewards/rejected": -7.240212440490723,
"step": 133
},
{
"epoch": 1.7573770491803278,
"grad_norm": 15.04454517364502,
"learning_rate": 8.455882352941178e-07,
"logits/chosen": -0.7795289754867554,
"logits/rejected": -0.8595843315124512,
"logps/chosen": -117.66368103027344,
"logps/rejected": -155.0380859375,
"loss": 0.1567,
"rewards/accuracies": 1.0,
"rewards/chosen": 1.0589184761047363,
"rewards/margins": 2.530170440673828,
"rewards/rejected": -1.4712518453598022,
"step": 134
},
{
"epoch": 1.7704918032786885,
"grad_norm": 79.06763458251953,
"learning_rate": 8.088235294117648e-07,
"logits/chosen": -0.7521681189537048,
"logits/rejected": -0.5849143862724304,
"logps/chosen": -370.17303466796875,
"logps/rejected": -297.7393493652344,
"loss": 0.3304,
"rewards/accuracies": 0.75,
"rewards/chosen": 1.5731346607208252,
"rewards/margins": 4.559065341949463,
"rewards/rejected": -2.9859304428100586,
"step": 135
},
{
"epoch": 1.7836065573770492,
"grad_norm": 75.73023223876953,
"learning_rate": 7.720588235294119e-07,
"logits/chosen": -0.24577771127223969,
"logits/rejected": -0.2834129333496094,
"logps/chosen": -169.0641632080078,
"logps/rejected": -227.2678985595703,
"loss": 0.1953,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.9874727725982666,
"rewards/margins": 2.466601848602295,
"rewards/rejected": -1.4791290760040283,
"step": 136
},
{
"epoch": 1.7967213114754097,
"grad_norm": 13.235738754272461,
"learning_rate": 7.352941176470589e-07,
"logits/chosen": -0.8364167213439941,
"logits/rejected": -0.7191005349159241,
"logps/chosen": -281.6129150390625,
"logps/rejected": -327.17169189453125,
"loss": 0.0731,
"rewards/accuracies": 1.0,
"rewards/chosen": 2.5926880836486816,
"rewards/margins": 5.672669410705566,
"rewards/rejected": -3.0799813270568848,
"step": 137
},
{
"epoch": 1.8098360655737706,
"grad_norm": 5.092410087585449,
"learning_rate": 6.985294117647059e-07,
"logits/chosen": -0.6490945219993591,
"logits/rejected": -0.9161443710327148,
"logps/chosen": -366.2134704589844,
"logps/rejected": -429.5161437988281,
"loss": 0.0178,
"rewards/accuracies": 1.0,
"rewards/chosen": 1.8624721765518188,
"rewards/margins": 6.365274906158447,
"rewards/rejected": -4.50280237197876,
"step": 138
},
{
"epoch": 1.8229508196721311,
"grad_norm": 3.215587615966797,
"learning_rate": 6.61764705882353e-07,
"logits/chosen": -0.7531751990318298,
"logits/rejected": -0.30548983812332153,
"logps/chosen": -386.60504150390625,
"logps/rejected": -275.23779296875,
"loss": 0.0127,
"rewards/accuracies": 1.0,
"rewards/chosen": 3.7106332778930664,
"rewards/margins": 6.189945220947266,
"rewards/rejected": -2.4793121814727783,
"step": 139
},
{
"epoch": 1.8360655737704918,
"grad_norm": 0.22684346139431,
"learning_rate": 6.25e-07,
"logits/chosen": -0.38512834906578064,
"logits/rejected": -0.39937111735343933,
"logps/chosen": -506.0502624511719,
"logps/rejected": -478.14990234375,
"loss": 0.0006,
"rewards/accuracies": 1.0,
"rewards/chosen": 4.004444599151611,
"rewards/margins": 9.403310775756836,
"rewards/rejected": -5.398866653442383,
"step": 140
},
{
"epoch": 1.8491803278688526,
"grad_norm": 8.121055603027344,
"learning_rate": 5.882352941176471e-07,
"logits/chosen": -0.5020557045936584,
"logits/rejected": -0.6416282653808594,
"logps/chosen": -369.1741027832031,
"logps/rejected": -440.558837890625,
"loss": 0.0314,
"rewards/accuracies": 1.0,
"rewards/chosen": 3.697706699371338,
"rewards/margins": 7.363080024719238,
"rewards/rejected": -3.6653730869293213,
"step": 141
},
{
"epoch": 1.862295081967213,
"grad_norm": 25.55944061279297,
"learning_rate": 5.514705882352942e-07,
"logits/chosen": -0.6144732236862183,
"logits/rejected": -0.5314571857452393,
"logps/chosen": -412.43310546875,
"logps/rejected": -425.11572265625,
"loss": 0.0594,
"rewards/accuracies": 1.0,
"rewards/chosen": 2.2833995819091797,
"rewards/margins": 5.651876926422119,
"rewards/rejected": -3.3684768676757812,
"step": 142
},
{
"epoch": 1.8754098360655738,
"grad_norm": 2.3017585277557373,
"learning_rate": 5.147058823529412e-07,
"logits/chosen": -0.7556811571121216,
"logits/rejected": -0.8270020484924316,
"logps/chosen": -401.3725280761719,
"logps/rejected": -348.38067626953125,
"loss": 0.181,
"rewards/accuracies": 0.75,
"rewards/chosen": 2.7437126636505127,
"rewards/margins": 7.136537551879883,
"rewards/rejected": -4.392825126647949,
"step": 143
},
{
"epoch": 1.8885245901639345,
"grad_norm": 1.4070271253585815,
"learning_rate": 4.779411764705882e-07,
"logits/chosen": -0.24664568901062012,
"logits/rejected": -0.640694797039032,
"logps/chosen": -327.0250549316406,
"logps/rejected": -863.6223754882812,
"loss": 0.0034,
"rewards/accuracies": 1.0,
"rewards/chosen": 1.61453115940094,
"rewards/margins": 6.990232467651367,
"rewards/rejected": -5.375700950622559,
"step": 144
},
{
"epoch": 1.901639344262295,
"grad_norm": 37.87571334838867,
"learning_rate": 4.4117647058823536e-07,
"logits/chosen": -0.7307272553443909,
"logits/rejected": -0.5720311403274536,
"logps/chosen": -393.85491943359375,
"logps/rejected": -344.6254577636719,
"loss": 0.352,
"rewards/accuracies": 0.75,
"rewards/chosen": 1.1627728939056396,
"rewards/margins": 3.641770124435425,
"rewards/rejected": -2.478997230529785,
"step": 145
},
{
"epoch": 1.9147540983606557,
"grad_norm": 15.04370403289795,
"learning_rate": 4.044117647058824e-07,
"logits/chosen": -0.842681884765625,
"logits/rejected": -0.4292832314968109,
"logps/chosen": -297.94207763671875,
"logps/rejected": -240.36907958984375,
"loss": 0.0534,
"rewards/accuracies": 1.0,
"rewards/chosen": 3.2723472118377686,
"rewards/margins": 5.443933963775635,
"rewards/rejected": -2.171586751937866,
"step": 146
},
{
"epoch": 1.9278688524590164,
"grad_norm": 15.202438354492188,
"learning_rate": 3.6764705882352943e-07,
"logits/chosen": -0.6476624011993408,
"logits/rejected": -0.49806222319602966,
"logps/chosen": -224.61077880859375,
"logps/rejected": -169.60565185546875,
"loss": 0.1546,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.8692402839660645,
"rewards/margins": 2.598930597305298,
"rewards/rejected": -1.7296903133392334,
"step": 147
},
{
"epoch": 1.940983606557377,
"grad_norm": 7.096276760101318,
"learning_rate": 3.308823529411765e-07,
"logits/chosen": -0.7730454802513123,
"logits/rejected": -0.4797229766845703,
"logps/chosen": -321.5426025390625,
"logps/rejected": -219.32769775390625,
"loss": 0.0219,
"rewards/accuracies": 1.0,
"rewards/chosen": 2.728835344314575,
"rewards/margins": 5.427863597869873,
"rewards/rejected": -2.699028253555298,
"step": 148
},
{
"epoch": 1.9540983606557378,
"grad_norm": 66.5175552368164,
"learning_rate": 2.9411764705882356e-07,
"logits/chosen": -0.31179729104042053,
"logits/rejected": -0.28874772787094116,
"logps/chosen": -279.7017822265625,
"logps/rejected": -289.318115234375,
"loss": 0.2663,
"rewards/accuracies": 0.75,
"rewards/chosen": 1.061594009399414,
"rewards/margins": 3.8362441062927246,
"rewards/rejected": -2.7746498584747314,
"step": 149
},
{
"epoch": 1.9672131147540983,
"grad_norm": 32.35987091064453,
"learning_rate": 2.573529411764706e-07,
"logits/chosen": -0.6153096556663513,
"logits/rejected": -0.725881814956665,
"logps/chosen": -301.1914367675781,
"logps/rejected": -272.52825927734375,
"loss": 0.233,
"rewards/accuracies": 1.0,
"rewards/chosen": 1.4337717294692993,
"rewards/margins": 3.663452625274658,
"rewards/rejected": -2.2296810150146484,
"step": 150
},
{
"epoch": 1.980327868852459,
"grad_norm": 0.4960515797138214,
"learning_rate": 2.2058823529411768e-07,
"logits/chosen": -0.30176612734794617,
"logits/rejected": 0.019010625779628754,
"logps/chosen": -270.0059814453125,
"logps/rejected": -376.6888732910156,
"loss": 0.0017,
"rewards/accuracies": 1.0,
"rewards/chosen": 1.2527967691421509,
"rewards/margins": 7.02215576171875,
"rewards/rejected": -5.7693586349487305,
"step": 151
},
{
"epoch": 1.9934426229508198,
"grad_norm": 29.579833984375,
"learning_rate": 1.8382352941176472e-07,
"logits/chosen": -0.753585696220398,
"logits/rejected": -0.5951805114746094,
"logps/chosen": -178.24880981445312,
"logps/rejected": -202.90585327148438,
"loss": 0.1476,
"rewards/accuracies": 1.0,
"rewards/chosen": 1.0991493463516235,
"rewards/margins": 2.3130767345428467,
"rewards/rejected": -1.2139275074005127,
"step": 152
}
],
"logging_steps": 1,
"max_steps": 152,
"num_input_tokens_seen": 0,
"num_train_epochs": 2,
"save_steps": 500,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 0.0,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}