yihang7's picture
Model save
eee971a verified
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.0,
"eval_steps": 100,
"global_step": 2942,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 1.694915254237288e-09,
"logits/chosen": -3.241600751876831,
"logits/rejected": -2.8775925636291504,
"logps/chosen": -233.8565673828125,
"logps/rejected": -768.6746215820312,
"loss": 0.6931,
"rewards/accuracies": 0.0,
"rewards/chosen": 0.0,
"rewards/margins": 0.0,
"rewards/rejected": 0.0,
"step": 1
},
{
"epoch": 0.0,
"learning_rate": 1.6949152542372882e-08,
"logits/chosen": -3.183110475540161,
"logits/rejected": -3.2940425872802734,
"logps/chosen": -479.7237548828125,
"logps/rejected": -508.1022644042969,
"loss": 0.6914,
"rewards/accuracies": 0.4305555522441864,
"rewards/chosen": -0.0011810138821601868,
"rewards/margins": -0.001814355025999248,
"rewards/rejected": 0.0006333404453471303,
"step": 10
},
{
"epoch": 0.01,
"learning_rate": 3.3898305084745764e-08,
"logits/chosen": -3.196739912033081,
"logits/rejected": -3.3269875049591064,
"logps/chosen": -235.5388946533203,
"logps/rejected": -412.62353515625,
"loss": 0.6541,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": 0.04803984612226486,
"rewards/margins": 0.09699604660272598,
"rewards/rejected": -0.048956211656332016,
"step": 20
},
{
"epoch": 0.01,
"learning_rate": 5.0847457627118645e-08,
"logits/chosen": -3.134932041168213,
"logits/rejected": -3.2440593242645264,
"logps/chosen": -389.2142639160156,
"logps/rejected": -413.16461181640625,
"loss": 0.5485,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": 0.0791018083691597,
"rewards/margins": 0.25357091426849365,
"rewards/rejected": -0.17446911334991455,
"step": 30
},
{
"epoch": 0.01,
"learning_rate": 6.779661016949153e-08,
"logits/chosen": -3.1792006492614746,
"logits/rejected": -3.3407816886901855,
"logps/chosen": -251.45498657226562,
"logps/rejected": -452.0169982910156,
"loss": 0.4281,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": 0.3247135281562805,
"rewards/margins": 0.7740644812583923,
"rewards/rejected": -0.4493509829044342,
"step": 40
},
{
"epoch": 0.02,
"learning_rate": 8.47457627118644e-08,
"logits/chosen": -3.2240090370178223,
"logits/rejected": -3.144906520843506,
"logps/chosen": -191.65354919433594,
"logps/rejected": -543.9041748046875,
"loss": 0.3076,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": 0.7691439986228943,
"rewards/margins": 1.466748595237732,
"rewards/rejected": -0.6976046562194824,
"step": 50
},
{
"epoch": 0.02,
"learning_rate": 1.0169491525423729e-07,
"logits/chosen": -3.2373671531677246,
"logits/rejected": -3.348659038543701,
"logps/chosen": -271.8299560546875,
"logps/rejected": -320.3260803222656,
"loss": 0.2537,
"rewards/accuracies": 0.9624999761581421,
"rewards/chosen": 0.8796433210372925,
"rewards/margins": 1.8711017370224,
"rewards/rejected": -0.9914585947990417,
"step": 60
},
{
"epoch": 0.02,
"learning_rate": 1.1864406779661017e-07,
"logits/chosen": -3.1753342151641846,
"logits/rejected": -3.235198974609375,
"logps/chosen": -256.61114501953125,
"logps/rejected": -491.931640625,
"loss": 0.2074,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 0.8632518649101257,
"rewards/margins": 2.171854019165039,
"rewards/rejected": -1.3086023330688477,
"step": 70
},
{
"epoch": 0.03,
"learning_rate": 1.3559322033898305e-07,
"logits/chosen": -3.1588549613952637,
"logits/rejected": -3.2258479595184326,
"logps/chosen": -386.23126220703125,
"logps/rejected": -396.27880859375,
"loss": 0.1891,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": 0.9385740160942078,
"rewards/margins": 2.8416147232055664,
"rewards/rejected": -1.9030405282974243,
"step": 80
},
{
"epoch": 0.03,
"learning_rate": 1.5254237288135593e-07,
"logits/chosen": -3.170820951461792,
"logits/rejected": -3.3037006855010986,
"logps/chosen": -204.9710235595703,
"logps/rejected": -283.68133544921875,
"loss": 0.1613,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": 1.398224115371704,
"rewards/margins": 3.5001823902130127,
"rewards/rejected": -2.1019580364227295,
"step": 90
},
{
"epoch": 0.03,
"learning_rate": 1.694915254237288e-07,
"logits/chosen": -3.1509833335876465,
"logits/rejected": -3.2727856636047363,
"logps/chosen": -182.78797912597656,
"logps/rejected": -422.30426025390625,
"loss": 0.1608,
"rewards/accuracies": 0.9624999761581421,
"rewards/chosen": 1.3794167041778564,
"rewards/margins": 4.1482086181640625,
"rewards/rejected": -2.768791913986206,
"step": 100
},
{
"epoch": 0.03,
"eval_logits/chosen": -3.2044522762298584,
"eval_logits/rejected": -3.2204079627990723,
"eval_logps/chosen": -243.43142700195312,
"eval_logps/rejected": -450.3221740722656,
"eval_loss": 0.16541637480258942,
"eval_rewards/accuracies": 0.9570707082748413,
"eval_rewards/chosen": 1.2374151945114136,
"eval_rewards/margins": 3.846338987350464,
"eval_rewards/rejected": -2.6089231967926025,
"eval_runtime": 452.6529,
"eval_samples_per_second": 20.987,
"eval_steps_per_second": 0.656,
"step": 100
},
{
"epoch": 0.04,
"learning_rate": 1.8644067796610168e-07,
"logits/chosen": -3.1778359413146973,
"logits/rejected": -3.0348751544952393,
"logps/chosen": -191.72665405273438,
"logps/rejected": -667.9417724609375,
"loss": 0.1325,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 1.5171259641647339,
"rewards/margins": 3.9871017932891846,
"rewards/rejected": -2.4699759483337402,
"step": 110
},
{
"epoch": 0.04,
"learning_rate": 2.0338983050847458e-07,
"logits/chosen": -3.21209979057312,
"logits/rejected": -3.245016574859619,
"logps/chosen": -206.03567504882812,
"logps/rejected": -441.95758056640625,
"loss": 0.1188,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": 1.7967685461044312,
"rewards/margins": 4.877519607543945,
"rewards/rejected": -3.0807507038116455,
"step": 120
},
{
"epoch": 0.04,
"learning_rate": 2.2033898305084743e-07,
"logits/chosen": -3.121410369873047,
"logits/rejected": -3.2283928394317627,
"logps/chosen": -255.36746215820312,
"logps/rejected": -485.18017578125,
"loss": 0.2426,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": 1.4470430612564087,
"rewards/margins": 5.143087387084961,
"rewards/rejected": -3.6960442066192627,
"step": 130
},
{
"epoch": 0.05,
"learning_rate": 2.3728813559322033e-07,
"logits/chosen": -3.1555418968200684,
"logits/rejected": -3.1503801345825195,
"logps/chosen": -167.48965454101562,
"logps/rejected": -516.3311767578125,
"loss": 0.1002,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": 1.1309149265289307,
"rewards/margins": 4.941633701324463,
"rewards/rejected": -3.810718059539795,
"step": 140
},
{
"epoch": 0.05,
"learning_rate": 2.542372881355932e-07,
"logits/chosen": -3.143369674682617,
"logits/rejected": -3.172342538833618,
"logps/chosen": -197.7274169921875,
"logps/rejected": -469.4510803222656,
"loss": 0.1209,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": 1.3236812353134155,
"rewards/margins": 5.159438133239746,
"rewards/rejected": -3.8357574939727783,
"step": 150
},
{
"epoch": 0.05,
"learning_rate": 2.711864406779661e-07,
"logits/chosen": -3.113320827484131,
"logits/rejected": -3.179581642150879,
"logps/chosen": -190.3457794189453,
"logps/rejected": -422.5477600097656,
"loss": 0.1031,
"rewards/accuracies": 0.9624999761581421,
"rewards/chosen": 1.2049860954284668,
"rewards/margins": 5.337760925292969,
"rewards/rejected": -4.132774353027344,
"step": 160
},
{
"epoch": 0.06,
"learning_rate": 2.88135593220339e-07,
"logits/chosen": -3.1512656211853027,
"logits/rejected": -3.2015151977539062,
"logps/chosen": -291.2518615722656,
"logps/rejected": -427.808837890625,
"loss": 0.0989,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": 1.2022926807403564,
"rewards/margins": 6.1538920402526855,
"rewards/rejected": -4.951599597930908,
"step": 170
},
{
"epoch": 0.06,
"learning_rate": 3.0508474576271186e-07,
"logits/chosen": -3.0925116539001465,
"logits/rejected": -3.1885197162628174,
"logps/chosen": -316.8929138183594,
"logps/rejected": -568.46435546875,
"loss": 0.1677,
"rewards/accuracies": 0.9624999761581421,
"rewards/chosen": 1.3121305704116821,
"rewards/margins": 6.406019687652588,
"rewards/rejected": -5.093889236450195,
"step": 180
},
{
"epoch": 0.06,
"learning_rate": 3.220338983050847e-07,
"logits/chosen": -3.191131830215454,
"logits/rejected": -3.2551910877227783,
"logps/chosen": -200.42120361328125,
"logps/rejected": -473.00506591796875,
"loss": 0.0712,
"rewards/accuracies": 0.9624999761581421,
"rewards/chosen": 1.504188060760498,
"rewards/margins": 6.774598598480225,
"rewards/rejected": -5.270411491394043,
"step": 190
},
{
"epoch": 0.07,
"learning_rate": 3.389830508474576e-07,
"logits/chosen": -3.1266510486602783,
"logits/rejected": -3.2319884300231934,
"logps/chosen": -254.8800811767578,
"logps/rejected": -386.7709045410156,
"loss": 0.1349,
"rewards/accuracies": 0.9375,
"rewards/chosen": 0.716153621673584,
"rewards/margins": 6.651679992675781,
"rewards/rejected": -5.935526371002197,
"step": 200
},
{
"epoch": 0.07,
"eval_logits/chosen": -3.2216391563415527,
"eval_logits/rejected": -3.1898040771484375,
"eval_logps/chosen": -246.39938354492188,
"eval_logps/rejected": -487.68365478515625,
"eval_loss": 0.09607043862342834,
"eval_rewards/accuracies": 0.9755892157554626,
"eval_rewards/chosen": 0.9406206011772156,
"eval_rewards/margins": 7.285686492919922,
"eval_rewards/rejected": -6.345065593719482,
"eval_runtime": 451.6853,
"eval_samples_per_second": 21.032,
"eval_steps_per_second": 0.658,
"step": 200
},
{
"epoch": 0.07,
"learning_rate": 3.559322033898305e-07,
"logits/chosen": -3.162513256072998,
"logits/rejected": -3.0620617866516113,
"logps/chosen": -185.5023651123047,
"logps/rejected": -624.7681274414062,
"loss": 0.1602,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": 1.2032901048660278,
"rewards/margins": 7.2529778480529785,
"rewards/rejected": -6.04968786239624,
"step": 210
},
{
"epoch": 0.07,
"learning_rate": 3.7288135593220336e-07,
"logits/chosen": -3.16162109375,
"logits/rejected": -3.2060294151306152,
"logps/chosen": -306.1747131347656,
"logps/rejected": -416.8641662597656,
"loss": 0.0542,
"rewards/accuracies": 1.0,
"rewards/chosen": 1.0064268112182617,
"rewards/margins": 7.897356986999512,
"rewards/rejected": -6.890931129455566,
"step": 220
},
{
"epoch": 0.08,
"learning_rate": 3.898305084745763e-07,
"logits/chosen": -3.1517322063446045,
"logits/rejected": -3.153977394104004,
"logps/chosen": -265.21539306640625,
"logps/rejected": -504.4422912597656,
"loss": 0.0677,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 0.7571786642074585,
"rewards/margins": 7.3733367919921875,
"rewards/rejected": -6.616157531738281,
"step": 230
},
{
"epoch": 0.08,
"learning_rate": 4.0677966101694916e-07,
"logits/chosen": -3.0925240516662598,
"logits/rejected": -3.102187395095825,
"logps/chosen": -250.0133514404297,
"logps/rejected": -633.9852294921875,
"loss": 0.0745,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 0.48636823892593384,
"rewards/margins": 9.285165786743164,
"rewards/rejected": -8.798797607421875,
"step": 240
},
{
"epoch": 0.08,
"learning_rate": 4.23728813559322e-07,
"logits/chosen": -3.176499843597412,
"logits/rejected": -3.1582164764404297,
"logps/chosen": -195.40415954589844,
"logps/rejected": -477.47833251953125,
"loss": 0.3177,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": 1.0647075176239014,
"rewards/margins": 8.443204879760742,
"rewards/rejected": -7.378498077392578,
"step": 250
},
{
"epoch": 0.09,
"learning_rate": 4.4067796610169486e-07,
"logits/chosen": -3.1202080249786377,
"logits/rejected": -3.175307273864746,
"logps/chosen": -255.7227783203125,
"logps/rejected": -521.5419311523438,
"loss": 0.0741,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 1.017443299293518,
"rewards/margins": 9.245269775390625,
"rewards/rejected": -8.227825164794922,
"step": 260
},
{
"epoch": 0.09,
"learning_rate": 4.576271186440678e-07,
"logits/chosen": -3.1279916763305664,
"logits/rejected": -3.19228196144104,
"logps/chosen": -238.5277862548828,
"logps/rejected": -373.4498596191406,
"loss": 0.0622,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 0.6001161336898804,
"rewards/margins": 8.970643043518066,
"rewards/rejected": -8.370526313781738,
"step": 270
},
{
"epoch": 0.1,
"learning_rate": 4.7457627118644066e-07,
"logits/chosen": -3.114872694015503,
"logits/rejected": -3.2416419982910156,
"logps/chosen": -214.206298828125,
"logps/rejected": -393.0050354003906,
"loss": 0.0953,
"rewards/accuracies": 0.9624999761581421,
"rewards/chosen": 0.1830778867006302,
"rewards/margins": 10.744420051574707,
"rewards/rejected": -10.561342239379883,
"step": 280
},
{
"epoch": 0.1,
"learning_rate": 4.915254237288136e-07,
"logits/chosen": -3.1324026584625244,
"logits/rejected": -3.228682041168213,
"logps/chosen": -186.04457092285156,
"logps/rejected": -486.37286376953125,
"loss": 0.0347,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 0.23065343499183655,
"rewards/margins": 7.208334445953369,
"rewards/rejected": -6.9776811599731445,
"step": 290
},
{
"epoch": 0.1,
"learning_rate": 4.990555345674349e-07,
"logits/chosen": -3.194873094558716,
"logits/rejected": -3.2273032665252686,
"logps/chosen": -205.6195831298828,
"logps/rejected": -589.7122192382812,
"loss": 0.1065,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.19616279006004333,
"rewards/margins": 8.628320693969727,
"rewards/rejected": -8.432156562805176,
"step": 300
},
{
"epoch": 0.1,
"eval_logits/chosen": -3.2283244132995605,
"eval_logits/rejected": -3.1999356746673584,
"eval_logps/chosen": -258.00885009765625,
"eval_logps/rejected": -516.9434204101562,
"eval_loss": 0.10145324468612671,
"eval_rewards/accuracies": 0.9840067625045776,
"eval_rewards/chosen": -0.22032807767391205,
"eval_rewards/margins": 9.050712585449219,
"eval_rewards/rejected": -9.271040916442871,
"eval_runtime": 451.8175,
"eval_samples_per_second": 21.026,
"eval_steps_per_second": 0.657,
"step": 300
},
{
"epoch": 0.11,
"learning_rate": 4.971666037023044e-07,
"logits/chosen": -3.2539610862731934,
"logits/rejected": -3.189645290374756,
"logps/chosen": -218.11282348632812,
"logps/rejected": -482.13238525390625,
"loss": 0.1651,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.5336198210716248,
"rewards/margins": 8.224891662597656,
"rewards/rejected": -8.758512496948242,
"step": 310
},
{
"epoch": 0.11,
"learning_rate": 4.952776728371742e-07,
"logits/chosen": -3.1139378547668457,
"logits/rejected": -3.1811156272888184,
"logps/chosen": -395.83807373046875,
"logps/rejected": -422.86126708984375,
"loss": 0.0746,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.5153623819351196,
"rewards/margins": 8.82536506652832,
"rewards/rejected": -9.340726852416992,
"step": 320
},
{
"epoch": 0.11,
"learning_rate": 4.933887419720438e-07,
"logits/chosen": -3.185594320297241,
"logits/rejected": -3.141575574874878,
"logps/chosen": -274.31475830078125,
"logps/rejected": -605.5130004882812,
"loss": 0.0585,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.5278395414352417,
"rewards/margins": 9.545147895812988,
"rewards/rejected": -10.07298755645752,
"step": 330
},
{
"epoch": 0.12,
"learning_rate": 4.914998111069135e-07,
"logits/chosen": -3.2110514640808105,
"logits/rejected": -3.252934217453003,
"logps/chosen": -197.74810791015625,
"logps/rejected": -527.8907470703125,
"loss": 0.0403,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15357597172260284,
"rewards/margins": 10.004459381103516,
"rewards/rejected": -10.158035278320312,
"step": 340
},
{
"epoch": 0.12,
"learning_rate": 4.896108802417831e-07,
"logits/chosen": -3.2196593284606934,
"logits/rejected": -3.116806745529175,
"logps/chosen": -252.00607299804688,
"logps/rejected": -586.9210205078125,
"loss": 0.0643,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.3348214626312256,
"rewards/margins": 10.310604095458984,
"rewards/rejected": -10.645425796508789,
"step": 350
},
{
"epoch": 0.12,
"learning_rate": 4.877219493766528e-07,
"logits/chosen": -3.2220287322998047,
"logits/rejected": -3.3167202472686768,
"logps/chosen": -195.5267333984375,
"logps/rejected": -458.151123046875,
"loss": 0.0394,
"rewards/accuracies": 0.9624999761581421,
"rewards/chosen": -0.5834547877311707,
"rewards/margins": 13.117718696594238,
"rewards/rejected": -13.701173782348633,
"step": 360
},
{
"epoch": 0.13,
"learning_rate": 4.858330185115224e-07,
"logits/chosen": -3.207420825958252,
"logits/rejected": -3.1515700817108154,
"logps/chosen": -204.42086791992188,
"logps/rejected": -775.6949462890625,
"loss": 0.1166,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -0.6016790270805359,
"rewards/margins": 10.775850296020508,
"rewards/rejected": -11.377527236938477,
"step": 370
},
{
"epoch": 0.13,
"learning_rate": 4.839440876463921e-07,
"logits/chosen": -3.261585235595703,
"logits/rejected": -3.1481096744537354,
"logps/chosen": -224.3874053955078,
"logps/rejected": -682.3482055664062,
"loss": 0.0561,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.30562129616737366,
"rewards/margins": 11.442914962768555,
"rewards/rejected": -11.748537063598633,
"step": 380
},
{
"epoch": 0.13,
"learning_rate": 4.820551567812618e-07,
"logits/chosen": -3.1784658432006836,
"logits/rejected": -3.128221273422241,
"logps/chosen": -276.6926574707031,
"logps/rejected": -459.2041931152344,
"loss": 0.0602,
"rewards/accuracies": 0.9624999761581421,
"rewards/chosen": -0.6345301866531372,
"rewards/margins": 11.814035415649414,
"rewards/rejected": -12.448564529418945,
"step": 390
},
{
"epoch": 0.14,
"learning_rate": 4.801662259161314e-07,
"logits/chosen": -3.2648723125457764,
"logits/rejected": -3.161604404449463,
"logps/chosen": -217.42861938476562,
"logps/rejected": -682.0848388671875,
"loss": 0.0876,
"rewards/accuracies": 0.9624999761581421,
"rewards/chosen": -0.6267936825752258,
"rewards/margins": 13.1732177734375,
"rewards/rejected": -13.800012588500977,
"step": 400
},
{
"epoch": 0.14,
"eval_logits/chosen": -3.275331974029541,
"eval_logits/rejected": -3.206566095352173,
"eval_logps/chosen": -270.2173767089844,
"eval_logps/rejected": -561.2250366210938,
"eval_loss": 0.059712644666433334,
"eval_rewards/accuracies": 0.9865319728851318,
"eval_rewards/chosen": -1.4411805868148804,
"eval_rewards/margins": 12.258034706115723,
"eval_rewards/rejected": -13.699216842651367,
"eval_runtime": 451.4664,
"eval_samples_per_second": 21.043,
"eval_steps_per_second": 0.658,
"step": 400
},
{
"epoch": 0.14,
"learning_rate": 4.782772950510011e-07,
"logits/chosen": -3.2593846321105957,
"logits/rejected": -3.095431327819824,
"logps/chosen": -216.7010498046875,
"logps/rejected": -542.0743408203125,
"loss": 0.0473,
"rewards/accuracies": 0.9624999761581421,
"rewards/chosen": -0.9579986333847046,
"rewards/margins": 10.643548965454102,
"rewards/rejected": -11.601548194885254,
"step": 410
},
{
"epoch": 0.14,
"learning_rate": 4.7638836418587073e-07,
"logits/chosen": -3.1919872760772705,
"logits/rejected": -3.1519837379455566,
"logps/chosen": -222.6879425048828,
"logps/rejected": -716.9598388671875,
"loss": 0.0867,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -1.237994909286499,
"rewards/margins": 13.0872220993042,
"rewards/rejected": -14.325218200683594,
"step": 420
},
{
"epoch": 0.15,
"learning_rate": 4.7449943332074044e-07,
"logits/chosen": -3.1929614543914795,
"logits/rejected": -3.092634677886963,
"logps/chosen": -203.4615020751953,
"logps/rejected": -651.5323486328125,
"loss": 0.0356,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -0.7688810229301453,
"rewards/margins": 12.568059921264648,
"rewards/rejected": -13.336939811706543,
"step": 430
},
{
"epoch": 0.15,
"learning_rate": 4.7261050245561014e-07,
"logits/chosen": -3.2217884063720703,
"logits/rejected": -3.214984178543091,
"logps/chosen": -209.8181915283203,
"logps/rejected": -500.721923828125,
"loss": 0.039,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -1.114246129989624,
"rewards/margins": 12.425156593322754,
"rewards/rejected": -13.539402961730957,
"step": 440
},
{
"epoch": 0.15,
"learning_rate": 4.7072157159047975e-07,
"logits/chosen": -3.191771984100342,
"logits/rejected": -3.340414524078369,
"logps/chosen": -251.2989501953125,
"logps/rejected": -453.38812255859375,
"loss": 0.0678,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -1.3671858310699463,
"rewards/margins": 13.762568473815918,
"rewards/rejected": -15.129753112792969,
"step": 450
},
{
"epoch": 0.16,
"learning_rate": 4.6883264072534946e-07,
"logits/chosen": -3.229691982269287,
"logits/rejected": -3.2845585346221924,
"logps/chosen": -275.047119140625,
"logps/rejected": -483.4437561035156,
"loss": 0.0405,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.8710290789604187,
"rewards/margins": 13.83751106262207,
"rewards/rejected": -14.708539962768555,
"step": 460
},
{
"epoch": 0.16,
"learning_rate": 4.6694370986021906e-07,
"logits/chosen": -3.261676073074341,
"logits/rejected": -3.308655261993408,
"logps/chosen": -364.6744689941406,
"logps/rejected": -348.94195556640625,
"loss": 0.2232,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -1.5191086530685425,
"rewards/margins": 12.976631164550781,
"rewards/rejected": -14.49573802947998,
"step": 470
},
{
"epoch": 0.16,
"learning_rate": 4.6505477899508877e-07,
"logits/chosen": -3.34987211227417,
"logits/rejected": -3.1941416263580322,
"logps/chosen": -215.4343719482422,
"logps/rejected": -669.6368408203125,
"loss": 0.1063,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -1.3562166690826416,
"rewards/margins": 15.658292770385742,
"rewards/rejected": -17.014511108398438,
"step": 480
},
{
"epoch": 0.17,
"learning_rate": 4.631658481299584e-07,
"logits/chosen": -3.346367359161377,
"logits/rejected": -3.2884230613708496,
"logps/chosen": -229.24880981445312,
"logps/rejected": -600.4979858398438,
"loss": 0.0584,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.6670745015144348,
"rewards/margins": 18.63577651977539,
"rewards/rejected": -19.3028507232666,
"step": 490
},
{
"epoch": 0.17,
"learning_rate": 4.612769172648281e-07,
"logits/chosen": -3.2286019325256348,
"logits/rejected": -3.272526264190674,
"logps/chosen": -331.59771728515625,
"logps/rejected": -384.60089111328125,
"loss": 0.304,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.12641265988349915,
"rewards/margins": 16.35049819946289,
"rewards/rejected": -16.224084854125977,
"step": 500
},
{
"epoch": 0.17,
"eval_logits/chosen": -3.2601242065429688,
"eval_logits/rejected": -3.20926833152771,
"eval_logps/chosen": -258.48223876953125,
"eval_logps/rejected": -596.7301635742188,
"eval_loss": 0.08744455128908157,
"eval_rewards/accuracies": 0.9890572428703308,
"eval_rewards/chosen": -0.26766717433929443,
"eval_rewards/margins": 16.982051849365234,
"eval_rewards/rejected": -17.249719619750977,
"eval_runtime": 452.2491,
"eval_samples_per_second": 21.006,
"eval_steps_per_second": 0.657,
"step": 500
},
{
"epoch": 0.17,
"learning_rate": 4.5938798639969773e-07,
"logits/chosen": -3.1900811195373535,
"logits/rejected": -3.165273904800415,
"logps/chosen": -310.04840087890625,
"logps/rejected": -706.6512451171875,
"loss": 0.0779,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.08271221071481705,
"rewards/margins": 16.163110733032227,
"rewards/rejected": -16.24582290649414,
"step": 510
},
{
"epoch": 0.18,
"learning_rate": 4.574990555345674e-07,
"logits/chosen": -3.233689546585083,
"logits/rejected": -3.1376094818115234,
"logps/chosen": -227.1941680908203,
"logps/rejected": -826.2154541015625,
"loss": 0.3105,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -1.272916555404663,
"rewards/margins": 36.48918151855469,
"rewards/rejected": -37.76210021972656,
"step": 520
},
{
"epoch": 0.18,
"learning_rate": 4.556101246694371e-07,
"logits/chosen": -3.3540146350860596,
"logits/rejected": -3.1351585388183594,
"logps/chosen": -206.82461547851562,
"logps/rejected": -887.9490966796875,
"loss": 0.1344,
"rewards/accuracies": 0.9624999761581421,
"rewards/chosen": -0.5983909368515015,
"rewards/margins": 37.151283264160156,
"rewards/rejected": -37.749671936035156,
"step": 530
},
{
"epoch": 0.18,
"learning_rate": 4.5372119380430675e-07,
"logits/chosen": -3.2952301502227783,
"logits/rejected": -3.2790439128875732,
"logps/chosen": -196.64913940429688,
"logps/rejected": -452.2648010253906,
"loss": 0.0358,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.035662006586790085,
"rewards/margins": 12.354232788085938,
"rewards/rejected": -12.389894485473633,
"step": 540
},
{
"epoch": 0.19,
"learning_rate": 4.518322629391764e-07,
"logits/chosen": -3.339197874069214,
"logits/rejected": -3.2826008796691895,
"logps/chosen": -215.4700469970703,
"logps/rejected": -526.360595703125,
"loss": 0.0365,
"rewards/accuracies": 0.9624999761581421,
"rewards/chosen": -0.8227304220199585,
"rewards/margins": 13.043925285339355,
"rewards/rejected": -13.866655349731445,
"step": 550
},
{
"epoch": 0.19,
"learning_rate": 4.4994333207404607e-07,
"logits/chosen": -3.238922119140625,
"logits/rejected": -3.2724907398223877,
"logps/chosen": -336.5367126464844,
"logps/rejected": -491.58416748046875,
"loss": 0.0928,
"rewards/accuracies": 0.9624999761581421,
"rewards/chosen": -1.3954570293426514,
"rewards/margins": 12.810078620910645,
"rewards/rejected": -14.205537796020508,
"step": 560
},
{
"epoch": 0.19,
"learning_rate": 4.480544012089157e-07,
"logits/chosen": -3.1199750900268555,
"logits/rejected": -3.2144436836242676,
"logps/chosen": -372.39630126953125,
"logps/rejected": -409.29168701171875,
"loss": 0.0147,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.8765771985054016,
"rewards/margins": 15.976984977722168,
"rewards/rejected": -16.85356330871582,
"step": 570
},
{
"epoch": 0.2,
"learning_rate": 4.461654703437854e-07,
"logits/chosen": -3.1464171409606934,
"logits/rejected": -3.1977782249450684,
"logps/chosen": -329.33721923828125,
"logps/rejected": -579.863525390625,
"loss": 0.0306,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.042082421481609344,
"rewards/margins": 18.116785049438477,
"rewards/rejected": -18.158864974975586,
"step": 580
},
{
"epoch": 0.2,
"learning_rate": 4.442765394786551e-07,
"logits/chosen": -3.2647204399108887,
"logits/rejected": -3.173513889312744,
"logps/chosen": -210.3753662109375,
"logps/rejected": -446.36175537109375,
"loss": 0.1479,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.5811902284622192,
"rewards/margins": 14.033243179321289,
"rewards/rejected": -14.614431381225586,
"step": 590
},
{
"epoch": 0.2,
"learning_rate": 4.423876086135247e-07,
"logits/chosen": -3.1431689262390137,
"logits/rejected": -3.110055685043335,
"logps/chosen": -322.7793884277344,
"logps/rejected": -717.8176879882812,
"loss": 0.1206,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.4665941298007965,
"rewards/margins": 16.126140594482422,
"rewards/rejected": -16.592737197875977,
"step": 600
},
{
"epoch": 0.2,
"eval_logits/chosen": -3.2024269104003906,
"eval_logits/rejected": -3.168919801712036,
"eval_logps/chosen": -260.0578308105469,
"eval_logps/rejected": -580.7472534179688,
"eval_loss": 0.06855383515357971,
"eval_rewards/accuracies": 0.9890572428703308,
"eval_rewards/chosen": -0.42522314190864563,
"eval_rewards/margins": 15.226205825805664,
"eval_rewards/rejected": -15.65142822265625,
"eval_runtime": 462.2117,
"eval_samples_per_second": 20.553,
"eval_steps_per_second": 0.643,
"step": 600
},
{
"epoch": 0.21,
"learning_rate": 4.404986777483944e-07,
"logits/chosen": -3.0911104679107666,
"logits/rejected": -3.2177834510803223,
"logps/chosen": -253.5253448486328,
"logps/rejected": -505.5723571777344,
"loss": 0.0857,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.4803188443183899,
"rewards/margins": 14.278889656066895,
"rewards/rejected": -14.759210586547852,
"step": 610
},
{
"epoch": 0.21,
"learning_rate": 4.3860974688326405e-07,
"logits/chosen": -3.17213773727417,
"logits/rejected": -3.153881072998047,
"logps/chosen": -341.3065185546875,
"logps/rejected": -633.9520874023438,
"loss": 0.0714,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.8739015460014343,
"rewards/margins": 14.758657455444336,
"rewards/rejected": -15.632558822631836,
"step": 620
},
{
"epoch": 0.21,
"learning_rate": 4.367208160181337e-07,
"logits/chosen": -3.1356098651885986,
"logits/rejected": -3.269990921020508,
"logps/chosen": -300.65411376953125,
"logps/rejected": -361.2610778808594,
"loss": 0.114,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.36743250489234924,
"rewards/margins": 14.380537033081055,
"rewards/rejected": -14.747968673706055,
"step": 630
},
{
"epoch": 0.22,
"learning_rate": 4.348318851530034e-07,
"logits/chosen": -3.2255866527557373,
"logits/rejected": -3.2463996410369873,
"logps/chosen": -273.51776123046875,
"logps/rejected": -633.8323364257812,
"loss": 0.4839,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.413738876581192,
"rewards/margins": 17.93083381652832,
"rewards/rejected": -18.344573974609375,
"step": 640
},
{
"epoch": 0.22,
"learning_rate": 4.32942954287873e-07,
"logits/chosen": -3.2475905418395996,
"logits/rejected": -3.3233604431152344,
"logps/chosen": -264.77996826171875,
"logps/rejected": -494.20855712890625,
"loss": 0.0405,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -1.034877061843872,
"rewards/margins": 19.03714942932129,
"rewards/rejected": -20.072025299072266,
"step": 650
},
{
"epoch": 0.22,
"learning_rate": 4.3105402342274273e-07,
"logits/chosen": -3.2771410942077637,
"logits/rejected": -3.288753032684326,
"logps/chosen": -217.06277465820312,
"logps/rejected": -617.0843505859375,
"loss": 0.0595,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.8464560508728027,
"rewards/margins": 17.9014949798584,
"rewards/rejected": -18.74795150756836,
"step": 660
},
{
"epoch": 0.23,
"learning_rate": 4.2916509255761233e-07,
"logits/chosen": -3.221287488937378,
"logits/rejected": -3.186659812927246,
"logps/chosen": -258.8490295410156,
"logps/rejected": -667.2969970703125,
"loss": 0.0473,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -1.5467805862426758,
"rewards/margins": 15.2776460647583,
"rewards/rejected": -16.82442855834961,
"step": 670
},
{
"epoch": 0.23,
"learning_rate": 4.2727616169248204e-07,
"logits/chosen": -3.259873151779175,
"logits/rejected": -3.214630126953125,
"logps/chosen": -230.55038452148438,
"logps/rejected": -636.394287109375,
"loss": 0.0119,
"rewards/accuracies": 1.0,
"rewards/chosen": -1.225334644317627,
"rewards/margins": 16.479169845581055,
"rewards/rejected": -17.70450210571289,
"step": 680
},
{
"epoch": 0.23,
"learning_rate": 4.253872308273517e-07,
"logits/chosen": -3.2281653881073,
"logits/rejected": -3.0544328689575195,
"logps/chosen": -301.4876403808594,
"logps/rejected": -634.31689453125,
"loss": 0.0151,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.3511894941329956,
"rewards/margins": 17.707965850830078,
"rewards/rejected": -18.05915641784668,
"step": 690
},
{
"epoch": 0.24,
"learning_rate": 4.2349829996222135e-07,
"logits/chosen": -3.2615528106689453,
"logits/rejected": -3.146155834197998,
"logps/chosen": -257.6978759765625,
"logps/rejected": -651.7713012695312,
"loss": 0.0176,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.4299692213535309,
"rewards/margins": 16.86044692993164,
"rewards/rejected": -17.290416717529297,
"step": 700
},
{
"epoch": 0.24,
"eval_logits/chosen": -3.2957923412323,
"eval_logits/rejected": -3.2304515838623047,
"eval_logps/chosen": -262.88763427734375,
"eval_logps/rejected": -599.524169921875,
"eval_loss": 0.06300165504217148,
"eval_rewards/accuracies": 0.9932659864425659,
"eval_rewards/chosen": -0.7082026600837708,
"eval_rewards/margins": 16.8209228515625,
"eval_rewards/rejected": -17.529123306274414,
"eval_runtime": 459.6358,
"eval_samples_per_second": 20.669,
"eval_steps_per_second": 0.646,
"step": 700
},
{
"epoch": 0.24,
"learning_rate": 4.2160936909709106e-07,
"logits/chosen": -3.19771409034729,
"logits/rejected": -3.223461866378784,
"logps/chosen": -260.47796630859375,
"logps/rejected": -475.68341064453125,
"loss": 0.0625,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.8059867024421692,
"rewards/margins": 14.349054336547852,
"rewards/rejected": -15.155041694641113,
"step": 710
},
{
"epoch": 0.24,
"learning_rate": 4.1972043823196066e-07,
"logits/chosen": -3.252122402191162,
"logits/rejected": -3.2310726642608643,
"logps/chosen": -230.65432739257812,
"logps/rejected": -538.1376953125,
"loss": 0.0743,
"rewards/accuracies": 1.0,
"rewards/chosen": -1.229381799697876,
"rewards/margins": 19.127460479736328,
"rewards/rejected": -20.356840133666992,
"step": 720
},
{
"epoch": 0.25,
"learning_rate": 4.1783150736683037e-07,
"logits/chosen": -3.29115629196167,
"logits/rejected": -3.1870217323303223,
"logps/chosen": -289.22088623046875,
"logps/rejected": -615.3138427734375,
"loss": 0.0194,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -1.0578917264938354,
"rewards/margins": 16.45914649963379,
"rewards/rejected": -17.51703643798828,
"step": 730
},
{
"epoch": 0.25,
"learning_rate": 4.1594257650170003e-07,
"logits/chosen": -3.2585701942443848,
"logits/rejected": -3.274698257446289,
"logps/chosen": -291.2822570800781,
"logps/rejected": -534.2890625,
"loss": 0.0079,
"rewards/accuracies": 1.0,
"rewards/chosen": -1.3135229349136353,
"rewards/margins": 16.959497451782227,
"rewards/rejected": -18.273021697998047,
"step": 740
},
{
"epoch": 0.25,
"learning_rate": 4.140536456365697e-07,
"logits/chosen": -3.2825417518615723,
"logits/rejected": -3.2054946422576904,
"logps/chosen": -204.35067749023438,
"logps/rejected": -632.234375,
"loss": 0.0207,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.6541246771812439,
"rewards/margins": 18.069482803344727,
"rewards/rejected": -18.723609924316406,
"step": 750
},
{
"epoch": 0.26,
"learning_rate": 4.1216471477143934e-07,
"logits/chosen": -3.3405938148498535,
"logits/rejected": -3.2676734924316406,
"logps/chosen": -214.34793090820312,
"logps/rejected": -491.17803955078125,
"loss": 0.0314,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.9016032218933105,
"rewards/margins": 17.351411819458008,
"rewards/rejected": -18.253013610839844,
"step": 760
},
{
"epoch": 0.26,
"learning_rate": 4.10275783906309e-07,
"logits/chosen": -3.272359848022461,
"logits/rejected": -3.3219470977783203,
"logps/chosen": -323.8666687011719,
"logps/rejected": -581.1951904296875,
"loss": 0.009,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.8133894205093384,
"rewards/margins": 15.973344802856445,
"rewards/rejected": -16.786733627319336,
"step": 770
},
{
"epoch": 0.27,
"learning_rate": 4.0838685304117865e-07,
"logits/chosen": -3.353651762008667,
"logits/rejected": -3.259639263153076,
"logps/chosen": -336.8185729980469,
"logps/rejected": -695.9039306640625,
"loss": 0.2987,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -1.0787861347198486,
"rewards/margins": 17.26241683959961,
"rewards/rejected": -18.34119987487793,
"step": 780
},
{
"epoch": 0.27,
"learning_rate": 4.0649792217604836e-07,
"logits/chosen": -3.4104390144348145,
"logits/rejected": -3.359205961227417,
"logps/chosen": -263.760498046875,
"logps/rejected": -435.3912658691406,
"loss": 0.0309,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.8803040385246277,
"rewards/margins": 15.400744438171387,
"rewards/rejected": -16.281049728393555,
"step": 790
},
{
"epoch": 0.27,
"learning_rate": 4.04608991310918e-07,
"logits/chosen": -3.4024837017059326,
"logits/rejected": -3.3290324211120605,
"logps/chosen": -274.2511291503906,
"logps/rejected": -561.3245239257812,
"loss": 0.0461,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -1.1979191303253174,
"rewards/margins": 17.333105087280273,
"rewards/rejected": -18.531024932861328,
"step": 800
},
{
"epoch": 0.27,
"eval_logits/chosen": -3.5158255100250244,
"eval_logits/rejected": -3.3936259746551514,
"eval_logps/chosen": -268.3476867675781,
"eval_logps/rejected": -636.7913818359375,
"eval_loss": 0.034081265330314636,
"eval_rewards/accuracies": 0.9932659864425659,
"eval_rewards/chosen": -1.2542104721069336,
"eval_rewards/margins": 20.001632690429688,
"eval_rewards/rejected": -21.255842208862305,
"eval_runtime": 459.1675,
"eval_samples_per_second": 20.69,
"eval_steps_per_second": 0.647,
"step": 800
},
{
"epoch": 0.28,
"learning_rate": 4.0272006044578767e-07,
"logits/chosen": -3.4222965240478516,
"logits/rejected": -3.3901965618133545,
"logps/chosen": -271.127685546875,
"logps/rejected": -702.6803588867188,
"loss": 0.0138,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.3194767236709595,
"rewards/margins": 20.698434829711914,
"rewards/rejected": -21.01791000366211,
"step": 810
},
{
"epoch": 0.28,
"learning_rate": 4.0083112958065733e-07,
"logits/chosen": -3.3571228981018066,
"logits/rejected": -3.378098964691162,
"logps/chosen": -354.65338134765625,
"logps/rejected": -460.4593811035156,
"loss": 0.0081,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.2477954924106598,
"rewards/margins": 16.9534854888916,
"rewards/rejected": -16.705690383911133,
"step": 820
},
{
"epoch": 0.28,
"learning_rate": 3.98942198715527e-07,
"logits/chosen": -3.4696907997131348,
"logits/rejected": -3.3343029022216797,
"logps/chosen": -200.86773681640625,
"logps/rejected": -668.7745361328125,
"loss": 0.0249,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.9370819926261902,
"rewards/margins": 18.672664642333984,
"rewards/rejected": -19.609745025634766,
"step": 830
},
{
"epoch": 0.29,
"learning_rate": 3.970532678503967e-07,
"logits/chosen": -3.428400754928589,
"logits/rejected": -3.392636775970459,
"logps/chosen": -278.60614013671875,
"logps/rejected": -547.3959350585938,
"loss": 0.0232,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.5795613527297974,
"rewards/margins": 18.447797775268555,
"rewards/rejected": -19.027359008789062,
"step": 840
},
{
"epoch": 0.29,
"learning_rate": 3.951643369852663e-07,
"logits/chosen": -3.3655319213867188,
"logits/rejected": -3.2096476554870605,
"logps/chosen": -277.06854248046875,
"logps/rejected": -800.4194946289062,
"loss": 0.0132,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.36317068338394165,
"rewards/margins": 18.841623306274414,
"rewards/rejected": -19.204795837402344,
"step": 850
},
{
"epoch": 0.29,
"learning_rate": 3.93275406120136e-07,
"logits/chosen": -3.3958353996276855,
"logits/rejected": -3.364926815032959,
"logps/chosen": -260.45867919921875,
"logps/rejected": -638.11328125,
"loss": 0.0255,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.29275327920913696,
"rewards/margins": 20.899669647216797,
"rewards/rejected": -21.192420959472656,
"step": 860
},
{
"epoch": 0.3,
"learning_rate": 3.913864752550056e-07,
"logits/chosen": -3.555651903152466,
"logits/rejected": -3.362936019897461,
"logps/chosen": -336.64990234375,
"logps/rejected": -678.8165283203125,
"loss": 0.0281,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.29712626338005066,
"rewards/margins": 20.055971145629883,
"rewards/rejected": -19.758846282958984,
"step": 870
},
{
"epoch": 0.3,
"learning_rate": 3.894975443898753e-07,
"logits/chosen": -3.5221924781799316,
"logits/rejected": -3.4192261695861816,
"logps/chosen": -184.4869384765625,
"logps/rejected": -517.7220458984375,
"loss": 0.0357,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.2702915668487549,
"rewards/margins": 17.328258514404297,
"rewards/rejected": -17.05797004699707,
"step": 880
},
{
"epoch": 0.3,
"learning_rate": 3.87608613524745e-07,
"logits/chosen": -3.3060905933380127,
"logits/rejected": -3.3240628242492676,
"logps/chosen": -249.1829833984375,
"logps/rejected": -423.19183349609375,
"loss": 0.0244,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.4387660622596741,
"rewards/margins": 15.609105110168457,
"rewards/rejected": -15.17033863067627,
"step": 890
},
{
"epoch": 0.31,
"learning_rate": 3.857196826596146e-07,
"logits/chosen": -3.432100296020508,
"logits/rejected": -3.3346171379089355,
"logps/chosen": -213.72622680664062,
"logps/rejected": -522.6224365234375,
"loss": 0.0185,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.36038094758987427,
"rewards/margins": 17.75735855102539,
"rewards/rejected": -17.39698028564453,
"step": 900
},
{
"epoch": 0.31,
"eval_logits/chosen": -3.494093418121338,
"eval_logits/rejected": -3.3745272159576416,
"eval_logps/chosen": -252.02415466308594,
"eval_logps/rejected": -596.7079467773438,
"eval_loss": 0.029099902138113976,
"eval_rewards/accuracies": 0.996632993221283,
"eval_rewards/chosen": 0.3781438171863556,
"eval_rewards/margins": 17.625646591186523,
"eval_rewards/rejected": -17.24750328063965,
"eval_runtime": 458.9428,
"eval_samples_per_second": 20.7,
"eval_steps_per_second": 0.647,
"step": 900
},
{
"epoch": 0.31,
"learning_rate": 3.8383075179448433e-07,
"logits/chosen": -3.4732635021209717,
"logits/rejected": -3.32080340385437,
"logps/chosen": -299.28045654296875,
"logps/rejected": -412.11737060546875,
"loss": 0.0154,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.5949827432632446,
"rewards/margins": 14.9349365234375,
"rewards/rejected": -14.339953422546387,
"step": 910
},
{
"epoch": 0.31,
"learning_rate": 3.8194182092935394e-07,
"logits/chosen": -3.5183627605438232,
"logits/rejected": -3.418959856033325,
"logps/chosen": -206.648193359375,
"logps/rejected": -540.2952880859375,
"loss": 0.0068,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.14846962690353394,
"rewards/margins": 18.112279891967773,
"rewards/rejected": -17.963809967041016,
"step": 920
},
{
"epoch": 0.32,
"learning_rate": 3.8005289006422365e-07,
"logits/chosen": -3.4474635124206543,
"logits/rejected": -3.4216742515563965,
"logps/chosen": -245.8963165283203,
"logps/rejected": -637.1114501953125,
"loss": 0.0291,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 0.09156160801649094,
"rewards/margins": 20.223173141479492,
"rewards/rejected": -20.13161277770996,
"step": 930
},
{
"epoch": 0.32,
"learning_rate": 3.7816395919909325e-07,
"logits/chosen": -3.43304443359375,
"logits/rejected": -3.3371405601501465,
"logps/chosen": -205.36312866210938,
"logps/rejected": -577.2255249023438,
"loss": 0.0663,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.1940477043390274,
"rewards/margins": 17.12152099609375,
"rewards/rejected": -17.31557273864746,
"step": 940
},
{
"epoch": 0.32,
"learning_rate": 3.7627502833396296e-07,
"logits/chosen": -3.4164633750915527,
"logits/rejected": -3.334111452102661,
"logps/chosen": -252.3224334716797,
"logps/rejected": -522.1593017578125,
"loss": 0.0247,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.06414094567298889,
"rewards/margins": 16.263051986694336,
"rewards/rejected": -16.327190399169922,
"step": 950
},
{
"epoch": 0.33,
"learning_rate": 3.7438609746883267e-07,
"logits/chosen": -3.3646836280822754,
"logits/rejected": -3.2790799140930176,
"logps/chosen": -303.45623779296875,
"logps/rejected": -564.6026611328125,
"loss": 0.138,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 0.1330849528312683,
"rewards/margins": 15.452998161315918,
"rewards/rejected": -15.319913864135742,
"step": 960
},
{
"epoch": 0.33,
"learning_rate": 3.7249716660370227e-07,
"logits/chosen": -3.5046660900115967,
"logits/rejected": -3.309417724609375,
"logps/chosen": -193.89120483398438,
"logps/rejected": -745.4686279296875,
"loss": 0.0171,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.28295058012008667,
"rewards/margins": 19.531200408935547,
"rewards/rejected": -19.814151763916016,
"step": 970
},
{
"epoch": 0.33,
"learning_rate": 3.70608235738572e-07,
"logits/chosen": -3.4429755210876465,
"logits/rejected": -3.297156572341919,
"logps/chosen": -311.4560852050781,
"logps/rejected": -706.6754150390625,
"loss": 0.0115,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.6654003858566284,
"rewards/margins": 20.311656951904297,
"rewards/rejected": -19.646259307861328,
"step": 980
},
{
"epoch": 0.34,
"learning_rate": 3.687193048734416e-07,
"logits/chosen": -3.448244094848633,
"logits/rejected": -3.379380464553833,
"logps/chosen": -186.96884155273438,
"logps/rejected": -682.4005126953125,
"loss": 0.038,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.3271837532520294,
"rewards/margins": 18.818578720092773,
"rewards/rejected": -18.491397857666016,
"step": 990
},
{
"epoch": 0.34,
"learning_rate": 3.668303740083113e-07,
"logits/chosen": -3.448310136795044,
"logits/rejected": -3.360579013824463,
"logps/chosen": -269.36297607421875,
"logps/rejected": -513.7430419921875,
"loss": 0.0219,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.5543234944343567,
"rewards/margins": 17.900325775146484,
"rewards/rejected": -17.34600067138672,
"step": 1000
},
{
"epoch": 0.34,
"eval_logits/chosen": -3.4703125953674316,
"eval_logits/rejected": -3.3235578536987305,
"eval_logps/chosen": -256.8191223144531,
"eval_logps/rejected": -620.40966796875,
"eval_loss": 0.024830101057887077,
"eval_rewards/accuracies": 0.9957912564277649,
"eval_rewards/chosen": -0.10135477781295776,
"eval_rewards/margins": 19.51631736755371,
"eval_rewards/rejected": -19.6176700592041,
"eval_runtime": 459.9759,
"eval_samples_per_second": 20.653,
"eval_steps_per_second": 0.646,
"step": 1000
},
{
"epoch": 0.34,
"learning_rate": 3.6494144314318094e-07,
"logits/chosen": -3.414768695831299,
"logits/rejected": -3.3657734394073486,
"logps/chosen": -279.57891845703125,
"logps/rejected": -491.0535583496094,
"loss": 0.0365,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.23943224549293518,
"rewards/margins": 14.393826484680176,
"rewards/rejected": -14.154393196105957,
"step": 1010
},
{
"epoch": 0.35,
"learning_rate": 3.630525122780506e-07,
"logits/chosen": -3.4718196392059326,
"logits/rejected": -3.3277339935302734,
"logps/chosen": -190.89041137695312,
"logps/rejected": -588.8604736328125,
"loss": 0.0648,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.5905485153198242,
"rewards/margins": 20.05153465270996,
"rewards/rejected": -20.642086029052734,
"step": 1020
},
{
"epoch": 0.35,
"learning_rate": 3.6116358141292026e-07,
"logits/chosen": -3.4760546684265137,
"logits/rejected": -3.3435845375061035,
"logps/chosen": -203.86050415039062,
"logps/rejected": -592.036865234375,
"loss": 0.0302,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.7946874499320984,
"rewards/margins": 24.075565338134766,
"rewards/rejected": -24.87025260925293,
"step": 1030
},
{
"epoch": 0.35,
"learning_rate": 3.592746505477899e-07,
"logits/chosen": -3.3917622566223145,
"logits/rejected": -3.2945053577423096,
"logps/chosen": -364.39324951171875,
"logps/rejected": -807.2418212890625,
"loss": 0.0895,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.8276251554489136,
"rewards/margins": 28.354557037353516,
"rewards/rejected": -29.182178497314453,
"step": 1040
},
{
"epoch": 0.36,
"learning_rate": 3.573857196826596e-07,
"logits/chosen": -3.579265594482422,
"logits/rejected": -3.4756500720977783,
"logps/chosen": -352.31451416015625,
"logps/rejected": -720.6471557617188,
"loss": 0.1426,
"rewards/accuracies": 1.0,
"rewards/chosen": -1.192226529121399,
"rewards/margins": 35.108890533447266,
"rewards/rejected": -36.301116943359375,
"step": 1050
},
{
"epoch": 0.36,
"learning_rate": 3.554967888175293e-07,
"logits/chosen": -3.6324493885040283,
"logits/rejected": -3.4453964233398438,
"logps/chosen": -275.65631103515625,
"logps/rejected": -759.9483642578125,
"loss": 0.0578,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -1.29346764087677,
"rewards/margins": 34.53205490112305,
"rewards/rejected": -35.825523376464844,
"step": 1060
},
{
"epoch": 0.36,
"learning_rate": 3.5360785795239893e-07,
"logits/chosen": -3.4921135902404785,
"logits/rejected": -3.409318208694458,
"logps/chosen": -326.6520080566406,
"logps/rejected": -557.5526733398438,
"loss": 0.0153,
"rewards/accuracies": 1.0,
"rewards/chosen": -1.002267599105835,
"rewards/margins": 22.072717666625977,
"rewards/rejected": -23.07498550415039,
"step": 1070
},
{
"epoch": 0.37,
"learning_rate": 3.517189270872686e-07,
"logits/chosen": -3.4556102752685547,
"logits/rejected": -3.42388916015625,
"logps/chosen": -266.01983642578125,
"logps/rejected": -539.4820556640625,
"loss": 0.1283,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.8836309313774109,
"rewards/margins": 21.98915672302246,
"rewards/rejected": -22.872791290283203,
"step": 1080
},
{
"epoch": 0.37,
"learning_rate": 3.4982999622213824e-07,
"logits/chosen": -3.318305253982544,
"logits/rejected": -3.3859639167785645,
"logps/chosen": -262.29730224609375,
"logps/rejected": -423.01531982421875,
"loss": 0.0238,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.06474297493696213,
"rewards/margins": 20.654415130615234,
"rewards/rejected": -20.719158172607422,
"step": 1090
},
{
"epoch": 0.37,
"learning_rate": 3.479410653570079e-07,
"logits/chosen": -3.4146296977996826,
"logits/rejected": -3.3967928886413574,
"logps/chosen": -263.7212219238281,
"logps/rejected": -663.2433471679688,
"loss": 0.0193,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.18020686507225037,
"rewards/margins": 23.35451889038086,
"rewards/rejected": -23.174312591552734,
"step": 1100
},
{
"epoch": 0.37,
"eval_logits/chosen": -3.512702226638794,
"eval_logits/rejected": -3.369966745376587,
"eval_logps/chosen": -253.3647918701172,
"eval_logps/rejected": -652.9178466796875,
"eval_loss": 0.04759080708026886,
"eval_rewards/accuracies": 0.9949495196342468,
"eval_rewards/chosen": 0.24408026039600372,
"eval_rewards/margins": 23.112565994262695,
"eval_rewards/rejected": -22.868486404418945,
"eval_runtime": 459.4583,
"eval_samples_per_second": 20.677,
"eval_steps_per_second": 0.646,
"step": 1100
},
{
"epoch": 0.38,
"learning_rate": 3.460521344918776e-07,
"logits/chosen": -3.4741005897521973,
"logits/rejected": -3.3692924976348877,
"logps/chosen": -177.2493896484375,
"logps/rejected": -684.059326171875,
"loss": 0.0304,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.10438014566898346,
"rewards/margins": 19.24637794494629,
"rewards/rejected": -19.141998291015625,
"step": 1110
},
{
"epoch": 0.38,
"learning_rate": 3.441632036267472e-07,
"logits/chosen": -3.519453525543213,
"logits/rejected": -3.450646162033081,
"logps/chosen": -206.3206787109375,
"logps/rejected": -591.7532348632812,
"loss": 0.1613,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.19403085112571716,
"rewards/margins": 18.250011444091797,
"rewards/rejected": -18.055980682373047,
"step": 1120
},
{
"epoch": 0.38,
"learning_rate": 3.422742727616169e-07,
"logits/chosen": -3.4171173572540283,
"logits/rejected": -3.402754545211792,
"logps/chosen": -285.92596435546875,
"logps/rejected": -420.0677795410156,
"loss": 0.0851,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.29394811391830444,
"rewards/margins": 17.166831970214844,
"rewards/rejected": -16.872886657714844,
"step": 1130
},
{
"epoch": 0.39,
"learning_rate": 3.403853418964866e-07,
"logits/chosen": -3.395599842071533,
"logits/rejected": -3.352531909942627,
"logps/chosen": -267.9078674316406,
"logps/rejected": -602.1204833984375,
"loss": 0.0109,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.25522318482398987,
"rewards/margins": 24.411819458007812,
"rewards/rejected": -24.667041778564453,
"step": 1140
},
{
"epoch": 0.39,
"learning_rate": 3.3849641103135623e-07,
"logits/chosen": -3.4090888500213623,
"logits/rejected": -3.3088366985321045,
"logps/chosen": -302.02484130859375,
"logps/rejected": -429.8524475097656,
"loss": 0.078,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.5110098123550415,
"rewards/margins": 17.73603057861328,
"rewards/rejected": -18.247039794921875,
"step": 1150
},
{
"epoch": 0.39,
"learning_rate": 3.3660748016622594e-07,
"logits/chosen": -3.284252882003784,
"logits/rejected": -3.364304304122925,
"logps/chosen": -329.56622314453125,
"logps/rejected": -570.0607299804688,
"loss": 0.0307,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.14421747624874115,
"rewards/margins": 23.145597457885742,
"rewards/rejected": -23.00138282775879,
"step": 1160
},
{
"epoch": 0.4,
"learning_rate": 3.3471854930109554e-07,
"logits/chosen": -3.2979743480682373,
"logits/rejected": -3.361307144165039,
"logps/chosen": -186.98870849609375,
"logps/rejected": -614.3134765625,
"loss": 0.074,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 0.3126475214958191,
"rewards/margins": 21.26107406616211,
"rewards/rejected": -20.948427200317383,
"step": 1170
},
{
"epoch": 0.4,
"learning_rate": 3.3282961843596525e-07,
"logits/chosen": -3.3583598136901855,
"logits/rejected": -3.2492566108703613,
"logps/chosen": -241.7067108154297,
"logps/rejected": -792.7431640625,
"loss": 0.0112,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 0.38423070311546326,
"rewards/margins": 24.471080780029297,
"rewards/rejected": -24.086851119995117,
"step": 1180
},
{
"epoch": 0.4,
"learning_rate": 3.3094068757083485e-07,
"logits/chosen": -3.342761516571045,
"logits/rejected": -3.278207778930664,
"logps/chosen": -275.2469482421875,
"logps/rejected": -528.4226684570312,
"loss": 0.0103,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.5847635269165039,
"rewards/margins": 23.485641479492188,
"rewards/rejected": -22.900880813598633,
"step": 1190
},
{
"epoch": 0.41,
"learning_rate": 3.2905175670570456e-07,
"logits/chosen": -3.3539364337921143,
"logits/rejected": -3.304426670074463,
"logps/chosen": -338.57464599609375,
"logps/rejected": -586.469970703125,
"loss": 0.0153,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.03903156518936157,
"rewards/margins": 19.734745025634766,
"rewards/rejected": -19.77377700805664,
"step": 1200
},
{
"epoch": 0.41,
"eval_logits/chosen": -3.443300485610962,
"eval_logits/rejected": -3.3281137943267822,
"eval_logps/chosen": -253.46900939941406,
"eval_logps/rejected": -634.9552612304688,
"eval_loss": 0.03437602147459984,
"eval_rewards/accuracies": 0.9957912564277649,
"eval_rewards/chosen": 0.23365840315818787,
"eval_rewards/margins": 21.305892944335938,
"eval_rewards/rejected": -21.072233200073242,
"eval_runtime": 460.0736,
"eval_samples_per_second": 20.649,
"eval_steps_per_second": 0.646,
"step": 1200
},
{
"epoch": 0.41,
"learning_rate": 3.271628258405742e-07,
"logits/chosen": -3.3404109477996826,
"logits/rejected": -3.3278732299804688,
"logps/chosen": -266.605712890625,
"logps/rejected": -432.56243896484375,
"loss": 0.0312,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.16924318671226501,
"rewards/margins": 20.001480102539062,
"rewards/rejected": -19.832239151000977,
"step": 1210
},
{
"epoch": 0.41,
"learning_rate": 3.252738949754439e-07,
"logits/chosen": -3.325308322906494,
"logits/rejected": -3.2718875408172607,
"logps/chosen": -239.7605743408203,
"logps/rejected": -689.46240234375,
"loss": 0.0095,
"rewards/accuracies": 1.0,
"rewards/chosen": 1.257117748260498,
"rewards/margins": 22.868335723876953,
"rewards/rejected": -21.611217498779297,
"step": 1220
},
{
"epoch": 0.42,
"learning_rate": 3.233849641103136e-07,
"logits/chosen": -3.409510850906372,
"logits/rejected": -3.2908267974853516,
"logps/chosen": -186.55523681640625,
"logps/rejected": -838.28271484375,
"loss": 0.0083,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.19115009903907776,
"rewards/margins": 21.021907806396484,
"rewards/rejected": -20.830759048461914,
"step": 1230
},
{
"epoch": 0.42,
"learning_rate": 3.214960332451832e-07,
"logits/chosen": -3.3691024780273438,
"logits/rejected": -3.2960734367370605,
"logps/chosen": -276.49310302734375,
"logps/rejected": -673.3426513671875,
"loss": 0.027,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -0.20441074669361115,
"rewards/margins": 21.827564239501953,
"rewards/rejected": -22.03197479248047,
"step": 1240
},
{
"epoch": 0.42,
"learning_rate": 3.196071023800529e-07,
"logits/chosen": -3.3508994579315186,
"logits/rejected": -3.2464585304260254,
"logps/chosen": -262.32366943359375,
"logps/rejected": -801.7515869140625,
"loss": 0.0091,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.30456364154815674,
"rewards/margins": 18.644428253173828,
"rewards/rejected": -18.948993682861328,
"step": 1250
},
{
"epoch": 0.43,
"learning_rate": 3.1771817151492255e-07,
"logits/chosen": -3.3253204822540283,
"logits/rejected": -3.182976245880127,
"logps/chosen": -271.783203125,
"logps/rejected": -792.1323852539062,
"loss": 0.0606,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.4825393557548523,
"rewards/margins": 23.283037185668945,
"rewards/rejected": -22.800498962402344,
"step": 1260
},
{
"epoch": 0.43,
"learning_rate": 3.158292406497922e-07,
"logits/chosen": -3.281102418899536,
"logits/rejected": -3.305457353591919,
"logps/chosen": -303.48016357421875,
"logps/rejected": -653.7716674804688,
"loss": 0.0389,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.34954649209976196,
"rewards/margins": 18.922035217285156,
"rewards/rejected": -18.57248306274414,
"step": 1270
},
{
"epoch": 0.44,
"learning_rate": 3.1394030978466186e-07,
"logits/chosen": -3.4600696563720703,
"logits/rejected": -3.3820204734802246,
"logps/chosen": -226.12631225585938,
"logps/rejected": -523.9305419921875,
"loss": 0.034,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.39240407943725586,
"rewards/margins": 21.62295913696289,
"rewards/rejected": -22.015361785888672,
"step": 1280
},
{
"epoch": 0.44,
"learning_rate": 3.120513789195315e-07,
"logits/chosen": -3.4715068340301514,
"logits/rejected": -3.343583583831787,
"logps/chosen": -208.66445922851562,
"logps/rejected": -671.6092529296875,
"loss": 0.0211,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.2728857100009918,
"rewards/margins": 23.407129287719727,
"rewards/rejected": -23.68001365661621,
"step": 1290
},
{
"epoch": 0.44,
"learning_rate": 3.1016244805440117e-07,
"logits/chosen": -3.3316524028778076,
"logits/rejected": -3.259505033493042,
"logps/chosen": -252.8136749267578,
"logps/rejected": -625.1239013671875,
"loss": 0.1011,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.11937759071588516,
"rewards/margins": 17.959081649780273,
"rewards/rejected": -18.078458786010742,
"step": 1300
},
{
"epoch": 0.44,
"eval_logits/chosen": -3.2942728996276855,
"eval_logits/rejected": -3.2086024284362793,
"eval_logps/chosen": -251.94058227539062,
"eval_logps/rejected": -619.3322143554688,
"eval_loss": 0.03203802928328514,
"eval_rewards/accuracies": 0.994107723236084,
"eval_rewards/chosen": 0.3865027129650116,
"eval_rewards/margins": 19.89643096923828,
"eval_rewards/rejected": -19.50992774963379,
"eval_runtime": 459.9792,
"eval_samples_per_second": 20.653,
"eval_steps_per_second": 0.646,
"step": 1300
},
{
"epoch": 0.45,
"learning_rate": 3.082735171892709e-07,
"logits/chosen": -3.283964157104492,
"logits/rejected": -3.3076889514923096,
"logps/chosen": -200.6337432861328,
"logps/rejected": -526.4237060546875,
"loss": 0.1087,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.0861460343003273,
"rewards/margins": 19.039958953857422,
"rewards/rejected": -19.126102447509766,
"step": 1310
},
{
"epoch": 0.45,
"learning_rate": 3.0638458632414054e-07,
"logits/chosen": -3.3101184368133545,
"logits/rejected": -3.2691454887390137,
"logps/chosen": -248.6476593017578,
"logps/rejected": -730.8521728515625,
"loss": 0.0054,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.3345467150211334,
"rewards/margins": 21.247516632080078,
"rewards/rejected": -20.912973403930664,
"step": 1320
},
{
"epoch": 0.45,
"learning_rate": 3.044956554590102e-07,
"logits/chosen": -3.3269455432891846,
"logits/rejected": -3.3831734657287598,
"logps/chosen": -250.7469024658203,
"logps/rejected": -490.6947326660156,
"loss": 0.0732,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.0001120984525186941,
"rewards/margins": 21.17990493774414,
"rewards/rejected": -21.180017471313477,
"step": 1330
},
{
"epoch": 0.46,
"learning_rate": 3.0260672459387985e-07,
"logits/chosen": -3.445746660232544,
"logits/rejected": -3.3551487922668457,
"logps/chosen": -194.33663940429688,
"logps/rejected": -587.1473388671875,
"loss": 0.0121,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.2983397841453552,
"rewards/margins": 21.926280975341797,
"rewards/rejected": -22.224620819091797,
"step": 1340
},
{
"epoch": 0.46,
"learning_rate": 3.007177937287495e-07,
"logits/chosen": -3.3813576698303223,
"logits/rejected": -3.2786917686462402,
"logps/chosen": -200.80307006835938,
"logps/rejected": -760.9759521484375,
"loss": 0.0063,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.03139156848192215,
"rewards/margins": 25.731273651123047,
"rewards/rejected": -25.762670516967773,
"step": 1350
},
{
"epoch": 0.46,
"learning_rate": 2.988288628636192e-07,
"logits/chosen": -3.431568145751953,
"logits/rejected": -3.312471389770508,
"logps/chosen": -261.0316162109375,
"logps/rejected": -559.0184326171875,
"loss": 0.057,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.4650034010410309,
"rewards/margins": 19.260265350341797,
"rewards/rejected": -18.795259475708008,
"step": 1360
},
{
"epoch": 0.47,
"learning_rate": 2.969399319984888e-07,
"logits/chosen": -3.423460006713867,
"logits/rejected": -3.2820401191711426,
"logps/chosen": -222.3740692138672,
"logps/rejected": -875.9083862304688,
"loss": 0.0138,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.7400540709495544,
"rewards/margins": 24.780017852783203,
"rewards/rejected": -25.520071029663086,
"step": 1370
},
{
"epoch": 0.47,
"learning_rate": 2.950510011333585e-07,
"logits/chosen": -3.4326465129852295,
"logits/rejected": -3.351288318634033,
"logps/chosen": -222.06887817382812,
"logps/rejected": -691.3297729492188,
"loss": 0.0427,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.06572739779949188,
"rewards/margins": 22.379207611083984,
"rewards/rejected": -22.313480377197266,
"step": 1380
},
{
"epoch": 0.47,
"learning_rate": 2.9316207026822813e-07,
"logits/chosen": -3.4397430419921875,
"logits/rejected": -3.294466733932495,
"logps/chosen": -224.94287109375,
"logps/rejected": -672.6677856445312,
"loss": 0.0298,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.15702833235263824,
"rewards/margins": 21.14691734313965,
"rewards/rejected": -21.303943634033203,
"step": 1390
},
{
"epoch": 0.48,
"learning_rate": 2.9127313940309784e-07,
"logits/chosen": -3.423677921295166,
"logits/rejected": -3.4165546894073486,
"logps/chosen": -191.03762817382812,
"logps/rejected": -612.9573364257812,
"loss": 0.0085,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.3937866985797882,
"rewards/margins": 22.95120620727539,
"rewards/rejected": -23.34499740600586,
"step": 1400
},
{
"epoch": 0.48,
"eval_logits/chosen": -3.505523920059204,
"eval_logits/rejected": -3.3688061237335205,
"eval_logps/chosen": -259.40972900390625,
"eval_logps/rejected": -670.28564453125,
"eval_loss": 0.01642242632806301,
"eval_rewards/accuracies": 0.9957912564277649,
"eval_rewards/chosen": -0.36041346192359924,
"eval_rewards/margins": 24.244855880737305,
"eval_rewards/rejected": -24.605268478393555,
"eval_runtime": 459.6598,
"eval_samples_per_second": 20.667,
"eval_steps_per_second": 0.646,
"step": 1400
},
{
"epoch": 0.48,
"learning_rate": 2.8938420853796754e-07,
"logits/chosen": -3.4179089069366455,
"logits/rejected": -3.3702964782714844,
"logps/chosen": -280.6731262207031,
"logps/rejected": -737.249267578125,
"loss": 0.008,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.025803815573453903,
"rewards/margins": 28.388423919677734,
"rewards/rejected": -28.414230346679688,
"step": 1410
},
{
"epoch": 0.48,
"learning_rate": 2.8749527767283715e-07,
"logits/chosen": -3.369783878326416,
"logits/rejected": -3.376282215118408,
"logps/chosen": -268.02752685546875,
"logps/rejected": -681.7142944335938,
"loss": 0.0091,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.40402960777282715,
"rewards/margins": 21.377483367919922,
"rewards/rejected": -21.78151512145996,
"step": 1420
},
{
"epoch": 0.49,
"learning_rate": 2.8560634680770686e-07,
"logits/chosen": -3.4302501678466797,
"logits/rejected": -3.362534761428833,
"logps/chosen": -268.420166015625,
"logps/rejected": -660.0174560546875,
"loss": 0.0133,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.9907222986221313,
"rewards/margins": 23.22337532043457,
"rewards/rejected": -24.214096069335938,
"step": 1430
},
{
"epoch": 0.49,
"learning_rate": 2.8371741594257646e-07,
"logits/chosen": -3.4184436798095703,
"logits/rejected": -3.3809783458709717,
"logps/chosen": -341.3996887207031,
"logps/rejected": -584.3710327148438,
"loss": 0.1015,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.279394268989563,
"rewards/margins": 25.102916717529297,
"rewards/rejected": -25.382308959960938,
"step": 1440
},
{
"epoch": 0.49,
"learning_rate": 2.8182848507744617e-07,
"logits/chosen": -3.5210273265838623,
"logits/rejected": -3.411952495574951,
"logps/chosen": -221.7975311279297,
"logps/rejected": -520.7232666015625,
"loss": 0.011,
"rewards/accuracies": 1.0,
"rewards/chosen": -1.764832854270935,
"rewards/margins": 22.8917236328125,
"rewards/rejected": -24.65655517578125,
"step": 1450
},
{
"epoch": 0.5,
"learning_rate": 2.799395542123158e-07,
"logits/chosen": -3.3874526023864746,
"logits/rejected": -3.3241126537323,
"logps/chosen": -336.1128845214844,
"logps/rejected": -680.0740356445312,
"loss": 0.0197,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.9126793742179871,
"rewards/margins": 21.444177627563477,
"rewards/rejected": -22.35685920715332,
"step": 1460
},
{
"epoch": 0.5,
"learning_rate": 2.780506233471855e-07,
"logits/chosen": -3.438469409942627,
"logits/rejected": -3.3389315605163574,
"logps/chosen": -200.33642578125,
"logps/rejected": -602.5469360351562,
"loss": 0.0139,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -1.0372685194015503,
"rewards/margins": 25.210084915161133,
"rewards/rejected": -26.247350692749023,
"step": 1470
},
{
"epoch": 0.5,
"learning_rate": 2.7616169248205513e-07,
"logits/chosen": -3.3702118396759033,
"logits/rejected": -3.3400275707244873,
"logps/chosen": -205.5651397705078,
"logps/rejected": -772.1658935546875,
"loss": 0.0084,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -1.8077577352523804,
"rewards/margins": 37.4057731628418,
"rewards/rejected": -39.213531494140625,
"step": 1480
},
{
"epoch": 0.51,
"learning_rate": 2.742727616169248e-07,
"logits/chosen": -3.3958728313446045,
"logits/rejected": -3.326359510421753,
"logps/chosen": -254.0430450439453,
"logps/rejected": -525.82568359375,
"loss": 0.0084,
"rewards/accuracies": 1.0,
"rewards/chosen": -1.1838462352752686,
"rewards/margins": 29.607269287109375,
"rewards/rejected": -30.791118621826172,
"step": 1490
},
{
"epoch": 0.51,
"learning_rate": 2.723838307517945e-07,
"logits/chosen": -3.3379979133605957,
"logits/rejected": -3.2290759086608887,
"logps/chosen": -330.87176513671875,
"logps/rejected": -727.2958984375,
"loss": 0.0057,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1591452956199646,
"rewards/margins": 32.132423400878906,
"rewards/rejected": -32.29157257080078,
"step": 1500
},
{
"epoch": 0.51,
"eval_logits/chosen": -3.4454987049102783,
"eval_logits/rejected": -3.298642158508301,
"eval_logps/chosen": -264.3898010253906,
"eval_logps/rejected": -762.0860595703125,
"eval_loss": 0.011499395594000816,
"eval_rewards/accuracies": 0.996632993221283,
"eval_rewards/chosen": -0.8584219813346863,
"eval_rewards/margins": 32.92688751220703,
"eval_rewards/rejected": -33.785308837890625,
"eval_runtime": 460.1896,
"eval_samples_per_second": 20.644,
"eval_steps_per_second": 0.645,
"step": 1500
},
{
"epoch": 0.51,
"learning_rate": 2.7049489988666416e-07,
"logits/chosen": -3.3881945610046387,
"logits/rejected": -3.26824951171875,
"logps/chosen": -310.400634765625,
"logps/rejected": -629.2418212890625,
"loss": 0.0191,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -1.2945066690444946,
"rewards/margins": 28.674571990966797,
"rewards/rejected": -29.96907615661621,
"step": 1510
},
{
"epoch": 0.52,
"learning_rate": 2.686059690215338e-07,
"logits/chosen": -3.300244092941284,
"logits/rejected": -3.251279830932617,
"logps/chosen": -273.0030212402344,
"logps/rejected": -812.864990234375,
"loss": 0.0678,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.8532045483589172,
"rewards/margins": 32.147369384765625,
"rewards/rejected": -33.00057601928711,
"step": 1520
},
{
"epoch": 0.52,
"learning_rate": 2.6671703815640347e-07,
"logits/chosen": -3.333925247192383,
"logits/rejected": -3.2999801635742188,
"logps/chosen": -247.5169219970703,
"logps/rejected": -541.9185791015625,
"loss": 0.0593,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.754341721534729,
"rewards/margins": 25.133520126342773,
"rewards/rejected": -25.887863159179688,
"step": 1530
},
{
"epoch": 0.52,
"learning_rate": 2.648281072912731e-07,
"logits/chosen": -3.4177489280700684,
"logits/rejected": -3.2177319526672363,
"logps/chosen": -210.8187713623047,
"logps/rejected": -762.6394653320312,
"loss": 0.0087,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.5407046675682068,
"rewards/margins": 20.313051223754883,
"rewards/rejected": -20.853755950927734,
"step": 1540
},
{
"epoch": 0.53,
"learning_rate": 2.629391764261428e-07,
"logits/chosen": -3.3744702339172363,
"logits/rejected": -3.3114571571350098,
"logps/chosen": -278.4391784667969,
"logps/rejected": -541.5650634765625,
"loss": 0.0069,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.8419182896614075,
"rewards/margins": 19.7562313079834,
"rewards/rejected": -20.598148345947266,
"step": 1550
},
{
"epoch": 0.53,
"learning_rate": 2.610502455610125e-07,
"logits/chosen": -3.3633506298065186,
"logits/rejected": -3.3126094341278076,
"logps/chosen": -352.4827575683594,
"logps/rejected": -738.1781005859375,
"loss": 0.1372,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.72525554895401,
"rewards/margins": 22.449813842773438,
"rewards/rejected": -23.175067901611328,
"step": 1560
},
{
"epoch": 0.53,
"learning_rate": 2.591613146958821e-07,
"logits/chosen": -3.4918789863586426,
"logits/rejected": -3.401181697845459,
"logps/chosen": -248.28060913085938,
"logps/rejected": -687.9749145507812,
"loss": 0.0074,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.5774227380752563,
"rewards/margins": 24.075176239013672,
"rewards/rejected": -24.652597427368164,
"step": 1570
},
{
"epoch": 0.54,
"learning_rate": 2.572723838307518e-07,
"logits/chosen": -3.4203097820281982,
"logits/rejected": -3.419062852859497,
"logps/chosen": -262.29730224609375,
"logps/rejected": -591.0722045898438,
"loss": 0.0102,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19718003273010254,
"rewards/margins": 21.856969833374023,
"rewards/rejected": -22.05415153503418,
"step": 1580
},
{
"epoch": 0.54,
"learning_rate": 2.5538345296562145e-07,
"logits/chosen": -3.4380905628204346,
"logits/rejected": -3.3508377075195312,
"logps/chosen": -261.0439147949219,
"logps/rejected": -576.0928955078125,
"loss": 0.0285,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15919718146324158,
"rewards/margins": 24.959590911865234,
"rewards/rejected": -25.11878776550293,
"step": 1590
},
{
"epoch": 0.54,
"learning_rate": 2.534945221004911e-07,
"logits/chosen": -3.292537212371826,
"logits/rejected": -3.3550610542297363,
"logps/chosen": -380.9278869628906,
"logps/rejected": -537.5498657226562,
"loss": 0.0082,
"rewards/accuracies": 1.0,
"rewards/chosen": 1.219398856163025,
"rewards/margins": 21.093488693237305,
"rewards/rejected": -19.874088287353516,
"step": 1600
},
{
"epoch": 0.54,
"eval_logits/chosen": -3.481621742248535,
"eval_logits/rejected": -3.3371667861938477,
"eval_logps/chosen": -259.466796875,
"eval_logps/rejected": -648.6592407226562,
"eval_loss": 0.05247886851429939,
"eval_rewards/accuracies": 0.997474730014801,
"eval_rewards/chosen": -0.3661208748817444,
"eval_rewards/margins": 22.07651138305664,
"eval_rewards/rejected": -22.442630767822266,
"eval_runtime": 459.9365,
"eval_samples_per_second": 20.655,
"eval_steps_per_second": 0.646,
"step": 1600
},
{
"epoch": 0.55,
"learning_rate": 2.516055912353608e-07,
"logits/chosen": -3.3571033477783203,
"logits/rejected": -3.2502875328063965,
"logps/chosen": -277.83837890625,
"logps/rejected": -724.0299072265625,
"loss": 0.0104,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.8198422193527222,
"rewards/margins": 20.712133407592773,
"rewards/rejected": -21.53197479248047,
"step": 1610
},
{
"epoch": 0.55,
"learning_rate": 2.497166603702304e-07,
"logits/chosen": -3.3982932567596436,
"logits/rejected": -3.273448944091797,
"logps/chosen": -262.4102478027344,
"logps/rejected": -759.910888671875,
"loss": 0.0153,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.7558759450912476,
"rewards/margins": 27.525121688842773,
"rewards/rejected": -28.2810001373291,
"step": 1620
},
{
"epoch": 0.55,
"learning_rate": 2.4782772950510013e-07,
"logits/chosen": -3.3453495502471924,
"logits/rejected": -3.254922389984131,
"logps/chosen": -278.51287841796875,
"logps/rejected": -550.3945922851562,
"loss": 0.0642,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.517066478729248,
"rewards/margins": 19.643749237060547,
"rewards/rejected": -20.160816192626953,
"step": 1630
},
{
"epoch": 0.56,
"learning_rate": 2.459387986399698e-07,
"logits/chosen": -3.198303699493408,
"logits/rejected": -3.2498583793640137,
"logps/chosen": -262.79925537109375,
"logps/rejected": -609.2022705078125,
"loss": 0.0064,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.8170091509819031,
"rewards/margins": 23.31692123413086,
"rewards/rejected": -24.133930206298828,
"step": 1640
},
{
"epoch": 0.56,
"learning_rate": 2.4404986777483944e-07,
"logits/chosen": -3.2812728881835938,
"logits/rejected": -3.2436320781707764,
"logps/chosen": -318.45599365234375,
"logps/rejected": -511.8076171875,
"loss": 0.0135,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.28894877433776855,
"rewards/margins": 20.95669937133789,
"rewards/rejected": -20.667749404907227,
"step": 1650
},
{
"epoch": 0.56,
"learning_rate": 2.421609369097091e-07,
"logits/chosen": -3.3857626914978027,
"logits/rejected": -3.2701289653778076,
"logps/chosen": -277.44281005859375,
"logps/rejected": -628.5038452148438,
"loss": 0.0101,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.5810535550117493,
"rewards/margins": 24.878406524658203,
"rewards/rejected": -25.459457397460938,
"step": 1660
},
{
"epoch": 0.57,
"learning_rate": 2.4027200604457875e-07,
"logits/chosen": -3.3911805152893066,
"logits/rejected": -3.285238742828369,
"logps/chosen": -245.50717163085938,
"logps/rejected": -631.26611328125,
"loss": 0.0045,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.19214244186878204,
"rewards/margins": 23.272733688354492,
"rewards/rejected": -23.46487808227539,
"step": 1670
},
{
"epoch": 0.57,
"learning_rate": 2.383830751794484e-07,
"logits/chosen": -3.3616461753845215,
"logits/rejected": -3.2842185497283936,
"logps/chosen": -346.10980224609375,
"logps/rejected": -848.2496948242188,
"loss": 0.0029,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1335320770740509,
"rewards/margins": 26.36749839782715,
"rewards/rejected": -26.50103187561035,
"step": 1680
},
{
"epoch": 0.57,
"learning_rate": 2.364941443143181e-07,
"logits/chosen": -3.392210006713867,
"logits/rejected": -3.366656541824341,
"logps/chosen": -214.9364471435547,
"logps/rejected": -542.2637939453125,
"loss": 0.0033,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.6650384664535522,
"rewards/margins": 20.180553436279297,
"rewards/rejected": -20.845592498779297,
"step": 1690
},
{
"epoch": 0.58,
"learning_rate": 2.3460521344918775e-07,
"logits/chosen": -3.4405808448791504,
"logits/rejected": -3.241579055786133,
"logps/chosen": -228.2866973876953,
"logps/rejected": -684.4705810546875,
"loss": 0.0128,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.8028446435928345,
"rewards/margins": 30.044261932373047,
"rewards/rejected": -30.847110748291016,
"step": 1700
},
{
"epoch": 0.58,
"eval_logits/chosen": -3.4487576484680176,
"eval_logits/rejected": -3.310159206390381,
"eval_logps/chosen": -260.0583801269531,
"eval_logps/rejected": -667.2958374023438,
"eval_loss": 0.05137615278363228,
"eval_rewards/accuracies": 0.9957912564277649,
"eval_rewards/chosen": -0.42527905106544495,
"eval_rewards/margins": 23.88100814819336,
"eval_rewards/rejected": -24.30628776550293,
"eval_runtime": 459.8654,
"eval_samples_per_second": 20.658,
"eval_steps_per_second": 0.646,
"step": 1700
},
{
"epoch": 0.58,
"learning_rate": 2.327162825840574e-07,
"logits/chosen": -3.330414295196533,
"logits/rejected": -3.2635841369628906,
"logps/chosen": -326.1940612792969,
"logps/rejected": -795.8521728515625,
"loss": 0.0068,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.16869966685771942,
"rewards/margins": 20.82526206970215,
"rewards/rejected": -20.993961334228516,
"step": 1710
},
{
"epoch": 0.58,
"learning_rate": 2.3082735171892708e-07,
"logits/chosen": -3.359579086303711,
"logits/rejected": -3.1651229858398438,
"logps/chosen": -249.9299774169922,
"logps/rejected": -1015.4552001953125,
"loss": 0.0058,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.624853789806366,
"rewards/margins": 25.351341247558594,
"rewards/rejected": -25.9761962890625,
"step": 1720
},
{
"epoch": 0.59,
"learning_rate": 2.2893842085379674e-07,
"logits/chosen": -3.4433200359344482,
"logits/rejected": -3.279329776763916,
"logps/chosen": -218.9633026123047,
"logps/rejected": -560.0067138671875,
"loss": 0.0046,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.8118854761123657,
"rewards/margins": 23.033185958862305,
"rewards/rejected": -23.845069885253906,
"step": 1730
},
{
"epoch": 0.59,
"learning_rate": 2.2704948998866642e-07,
"logits/chosen": -3.2414298057556152,
"logits/rejected": -3.2678425312042236,
"logps/chosen": -395.7589111328125,
"logps/rejected": -570.04345703125,
"loss": 0.1014,
"rewards/accuracies": 0.9624999761581421,
"rewards/chosen": -1.5239030122756958,
"rewards/margins": 21.01217269897461,
"rewards/rejected": -22.536075592041016,
"step": 1740
},
{
"epoch": 0.59,
"learning_rate": 2.2516055912353608e-07,
"logits/chosen": -3.3212122917175293,
"logits/rejected": -3.2659897804260254,
"logps/chosen": -303.695556640625,
"logps/rejected": -606.518310546875,
"loss": 0.0063,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.16412094235420227,
"rewards/margins": 30.60677146911621,
"rewards/rejected": -30.442651748657227,
"step": 1750
},
{
"epoch": 0.6,
"learning_rate": 2.2327162825840573e-07,
"logits/chosen": -3.2955994606018066,
"logits/rejected": -3.330519199371338,
"logps/chosen": -313.94586181640625,
"logps/rejected": -576.6971435546875,
"loss": 0.0037,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.5809284448623657,
"rewards/margins": 21.45393180847168,
"rewards/rejected": -22.034860610961914,
"step": 1760
},
{
"epoch": 0.6,
"learning_rate": 2.213826973932754e-07,
"logits/chosen": -3.3576037883758545,
"logits/rejected": -3.1798653602600098,
"logps/chosen": -205.84744262695312,
"logps/rejected": -822.9035034179688,
"loss": 0.0085,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.8852649927139282,
"rewards/margins": 27.4652156829834,
"rewards/rejected": -28.350482940673828,
"step": 1770
},
{
"epoch": 0.61,
"learning_rate": 2.1949376652814505e-07,
"logits/chosen": -3.352653980255127,
"logits/rejected": -3.3083279132843018,
"logps/chosen": -281.2442932128906,
"logps/rejected": -563.5089111328125,
"loss": 0.0066,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -1.149827241897583,
"rewards/margins": 20.38897132873535,
"rewards/rejected": -21.53879737854004,
"step": 1780
},
{
"epoch": 0.61,
"learning_rate": 2.176048356630147e-07,
"logits/chosen": -3.310255765914917,
"logits/rejected": -3.2874767780303955,
"logps/chosen": -390.83807373046875,
"logps/rejected": -587.5394897460938,
"loss": 0.006,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.459928035736084,
"rewards/margins": 24.047969818115234,
"rewards/rejected": -24.507898330688477,
"step": 1790
},
{
"epoch": 0.61,
"learning_rate": 2.157159047978844e-07,
"logits/chosen": -3.4679012298583984,
"logits/rejected": -3.2024967670440674,
"logps/chosen": -222.7393035888672,
"logps/rejected": -737.6758422851562,
"loss": 0.0018,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.8875459432601929,
"rewards/margins": 23.48456573486328,
"rewards/rejected": -24.372106552124023,
"step": 1800
},
{
"epoch": 0.61,
"eval_logits/chosen": -3.415928602218628,
"eval_logits/rejected": -3.2893950939178467,
"eval_logps/chosen": -259.3686828613281,
"eval_logps/rejected": -665.7247314453125,
"eval_loss": 0.03561777248978615,
"eval_rewards/accuracies": 0.996632993221283,
"eval_rewards/chosen": -0.3563132882118225,
"eval_rewards/margins": 23.792869567871094,
"eval_rewards/rejected": -24.149181365966797,
"eval_runtime": 460.2637,
"eval_samples_per_second": 20.64,
"eval_steps_per_second": 0.645,
"step": 1800
},
{
"epoch": 0.62,
"learning_rate": 2.1382697393275407e-07,
"logits/chosen": -3.35868501663208,
"logits/rejected": -3.294783353805542,
"logps/chosen": -320.07965087890625,
"logps/rejected": -738.1607666015625,
"loss": 0.0021,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.3287244737148285,
"rewards/margins": 26.954553604125977,
"rewards/rejected": -27.283273696899414,
"step": 1810
},
{
"epoch": 0.62,
"learning_rate": 2.1193804306762372e-07,
"logits/chosen": -3.297001600265503,
"logits/rejected": -3.2148499488830566,
"logps/chosen": -269.9095153808594,
"logps/rejected": -863.6824340820312,
"loss": 0.0036,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.7257425785064697,
"rewards/margins": 20.9593448638916,
"rewards/rejected": -21.685089111328125,
"step": 1820
},
{
"epoch": 0.62,
"learning_rate": 2.1004911220249338e-07,
"logits/chosen": -3.3534445762634277,
"logits/rejected": -3.3217594623565674,
"logps/chosen": -259.9757080078125,
"logps/rejected": -613.6614990234375,
"loss": 0.0067,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.3644919991493225,
"rewards/margins": 22.092815399169922,
"rewards/rejected": -21.728322982788086,
"step": 1830
},
{
"epoch": 0.63,
"learning_rate": 2.0816018133736303e-07,
"logits/chosen": -3.335651397705078,
"logits/rejected": -3.2650818824768066,
"logps/chosen": -335.0670166015625,
"logps/rejected": -782.08349609375,
"loss": 0.0073,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.0012598276371136308,
"rewards/margins": 27.04427719116211,
"rewards/rejected": -27.043018341064453,
"step": 1840
},
{
"epoch": 0.63,
"learning_rate": 2.0627125047223271e-07,
"logits/chosen": -3.357959032058716,
"logits/rejected": -3.278252363204956,
"logps/chosen": -266.217529296875,
"logps/rejected": -641.0094604492188,
"loss": 0.0517,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.38472968339920044,
"rewards/margins": 23.43660545349121,
"rewards/rejected": -23.821334838867188,
"step": 1850
},
{
"epoch": 0.63,
"learning_rate": 2.0438231960710237e-07,
"logits/chosen": -3.4566192626953125,
"logits/rejected": -3.297550678253174,
"logps/chosen": -265.8533630371094,
"logps/rejected": -815.4766845703125,
"loss": 0.0097,
"rewards/accuracies": 1.0,
"rewards/chosen": -1.3703159093856812,
"rewards/margins": 29.9644832611084,
"rewards/rejected": -31.334796905517578,
"step": 1860
},
{
"epoch": 0.64,
"learning_rate": 2.0249338874197203e-07,
"logits/chosen": -3.4326648712158203,
"logits/rejected": -3.4103198051452637,
"logps/chosen": -268.6964416503906,
"logps/rejected": -670.5267333984375,
"loss": 0.0043,
"rewards/accuracies": 1.0,
"rewards/chosen": -1.2914245128631592,
"rewards/margins": 30.470592498779297,
"rewards/rejected": -31.762014389038086,
"step": 1870
},
{
"epoch": 0.64,
"learning_rate": 2.0060445787684168e-07,
"logits/chosen": -3.5192794799804688,
"logits/rejected": -3.332420825958252,
"logps/chosen": -209.93576049804688,
"logps/rejected": -745.6237182617188,
"loss": 0.0053,
"rewards/accuracies": 1.0,
"rewards/chosen": -1.117185354232788,
"rewards/margins": 33.509422302246094,
"rewards/rejected": -34.62660598754883,
"step": 1880
},
{
"epoch": 0.64,
"learning_rate": 1.9871552701171136e-07,
"logits/chosen": -3.4282066822052,
"logits/rejected": -3.368283748626709,
"logps/chosen": -285.8004150390625,
"logps/rejected": -652.3516235351562,
"loss": 0.0065,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.227819561958313,
"rewards/margins": 26.842388153076172,
"rewards/rejected": -27.070209503173828,
"step": 1890
},
{
"epoch": 0.65,
"learning_rate": 1.9682659614658105e-07,
"logits/chosen": -3.4334263801574707,
"logits/rejected": -3.336756467819214,
"logps/chosen": -335.62725830078125,
"logps/rejected": -702.3507080078125,
"loss": 0.0105,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.38077887892723083,
"rewards/margins": 29.044042587280273,
"rewards/rejected": -29.424823760986328,
"step": 1900
},
{
"epoch": 0.65,
"eval_logits/chosen": -3.534825086593628,
"eval_logits/rejected": -3.3839564323425293,
"eval_logps/chosen": -265.37176513671875,
"eval_logps/rejected": -763.1902465820312,
"eval_loss": 0.03813723102211952,
"eval_rewards/accuracies": 0.9957912564277649,
"eval_rewards/chosen": -0.9566193222999573,
"eval_rewards/margins": 32.939109802246094,
"eval_rewards/rejected": -33.895729064941406,
"eval_runtime": 460.0454,
"eval_samples_per_second": 20.65,
"eval_steps_per_second": 0.646,
"step": 1900
},
{
"epoch": 0.65,
"learning_rate": 1.949376652814507e-07,
"logits/chosen": -3.396048069000244,
"logits/rejected": -3.275059461593628,
"logps/chosen": -252.13223266601562,
"logps/rejected": -747.9521484375,
"loss": 0.0043,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.7216171026229858,
"rewards/margins": 32.15826416015625,
"rewards/rejected": -32.87987518310547,
"step": 1910
},
{
"epoch": 0.65,
"learning_rate": 1.9304873441632036e-07,
"logits/chosen": -3.5610382556915283,
"logits/rejected": -3.372722625732422,
"logps/chosen": -199.77662658691406,
"logps/rejected": -584.327392578125,
"loss": 0.2531,
"rewards/accuracies": 1.0,
"rewards/chosen": -1.1961396932601929,
"rewards/margins": 29.65005874633789,
"rewards/rejected": -30.8461971282959,
"step": 1920
},
{
"epoch": 0.66,
"learning_rate": 1.9115980355119001e-07,
"logits/chosen": -3.369832992553711,
"logits/rejected": -3.314237117767334,
"logps/chosen": -215.8191680908203,
"logps/rejected": -815.2225952148438,
"loss": 0.0611,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.3732140064239502,
"rewards/margins": 27.058151245117188,
"rewards/rejected": -27.431365966796875,
"step": 1930
},
{
"epoch": 0.66,
"learning_rate": 1.8927087268605967e-07,
"logits/chosen": -3.4612975120544434,
"logits/rejected": -3.288508176803589,
"logps/chosen": -218.56201171875,
"logps/rejected": -906.2979736328125,
"loss": 0.0062,
"rewards/accuracies": 1.0,
"rewards/chosen": -1.040290117263794,
"rewards/margins": 31.08365249633789,
"rewards/rejected": -32.12394714355469,
"step": 1940
},
{
"epoch": 0.66,
"learning_rate": 1.8738194182092935e-07,
"logits/chosen": -3.348356246948242,
"logits/rejected": -3.2337405681610107,
"logps/chosen": -391.52935791015625,
"logps/rejected": -651.825927734375,
"loss": 0.0055,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.7969802021980286,
"rewards/margins": 28.728500366210938,
"rewards/rejected": -27.9315185546875,
"step": 1950
},
{
"epoch": 0.67,
"learning_rate": 1.85493010955799e-07,
"logits/chosen": -3.409937620162964,
"logits/rejected": -3.3914718627929688,
"logps/chosen": -265.65386962890625,
"logps/rejected": -596.263916015625,
"loss": 0.0173,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.3935322165489197,
"rewards/margins": 24.100194931030273,
"rewards/rejected": -24.49372673034668,
"step": 1960
},
{
"epoch": 0.67,
"learning_rate": 1.8360408009066866e-07,
"logits/chosen": -3.3308022022247314,
"logits/rejected": -3.2177085876464844,
"logps/chosen": -282.54046630859375,
"logps/rejected": -823.97412109375,
"loss": 0.007,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.41234469413757324,
"rewards/margins": 22.745586395263672,
"rewards/rejected": -23.157930374145508,
"step": 1970
},
{
"epoch": 0.67,
"learning_rate": 1.8171514922553835e-07,
"logits/chosen": -3.446866512298584,
"logits/rejected": -3.372515916824341,
"logps/chosen": -276.74468994140625,
"logps/rejected": -599.4273071289062,
"loss": 0.0043,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.35287076234817505,
"rewards/margins": 21.927663803100586,
"rewards/rejected": -22.280534744262695,
"step": 1980
},
{
"epoch": 0.68,
"learning_rate": 1.79826218360408e-07,
"logits/chosen": -3.317155361175537,
"logits/rejected": -3.3876793384552,
"logps/chosen": -318.00860595703125,
"logps/rejected": -510.11151123046875,
"loss": 0.0041,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.5086533427238464,
"rewards/margins": 24.184043884277344,
"rewards/rejected": -24.692697525024414,
"step": 1990
},
{
"epoch": 0.68,
"learning_rate": 1.7793728749527768e-07,
"logits/chosen": -3.3701987266540527,
"logits/rejected": -3.286350727081299,
"logps/chosen": -267.9636535644531,
"logps/rejected": -580.6959838867188,
"loss": 0.006,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -0.3567093014717102,
"rewards/margins": 24.048980712890625,
"rewards/rejected": -24.40568733215332,
"step": 2000
},
{
"epoch": 0.68,
"eval_logits/chosen": -3.4805188179016113,
"eval_logits/rejected": -3.337117910385132,
"eval_logps/chosen": -257.2082824707031,
"eval_logps/rejected": -686.7160034179688,
"eval_loss": 0.0072143604047596455,
"eval_rewards/accuracies": 0.997474730014801,
"eval_rewards/chosen": -0.14026859402656555,
"eval_rewards/margins": 26.10803985595703,
"eval_rewards/rejected": -26.248306274414062,
"eval_runtime": 460.6018,
"eval_samples_per_second": 20.625,
"eval_steps_per_second": 0.645,
"step": 2000
},
{
"epoch": 0.68,
"learning_rate": 1.7604835663014734e-07,
"logits/chosen": -3.408255100250244,
"logits/rejected": -3.403702974319458,
"logps/chosen": -203.19618225097656,
"logps/rejected": -665.8862915039062,
"loss": 0.0229,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.6091235280036926,
"rewards/margins": 24.911001205444336,
"rewards/rejected": -25.520126342773438,
"step": 2010
},
{
"epoch": 0.69,
"learning_rate": 1.74159425765017e-07,
"logits/chosen": -3.443638324737549,
"logits/rejected": -3.3524250984191895,
"logps/chosen": -209.58401489257812,
"logps/rejected": -749.86181640625,
"loss": 0.0053,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.21705660223960876,
"rewards/margins": 27.513622283935547,
"rewards/rejected": -27.730676651000977,
"step": 2020
},
{
"epoch": 0.69,
"learning_rate": 1.7227049489988665e-07,
"logits/chosen": -3.3942184448242188,
"logits/rejected": -3.311061143875122,
"logps/chosen": -317.77410888671875,
"logps/rejected": -552.6033325195312,
"loss": 0.014,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -0.5486845970153809,
"rewards/margins": 22.28003692626953,
"rewards/rejected": -22.82872200012207,
"step": 2030
},
{
"epoch": 0.69,
"learning_rate": 1.703815640347563e-07,
"logits/chosen": -3.5390067100524902,
"logits/rejected": -3.3566482067108154,
"logps/chosen": -201.36410522460938,
"logps/rejected": -666.7164306640625,
"loss": 0.0079,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.4069572389125824,
"rewards/margins": 28.751266479492188,
"rewards/rejected": -29.158227920532227,
"step": 2040
},
{
"epoch": 0.7,
"learning_rate": 1.6849263316962596e-07,
"logits/chosen": -3.3728034496307373,
"logits/rejected": -3.3503639698028564,
"logps/chosen": -267.21881103515625,
"logps/rejected": -626.498046875,
"loss": 0.0103,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.2131740152835846,
"rewards/margins": 27.280838012695312,
"rewards/rejected": -27.067663192749023,
"step": 2050
},
{
"epoch": 0.7,
"learning_rate": 1.6660370230449564e-07,
"logits/chosen": -3.4185855388641357,
"logits/rejected": -3.3411049842834473,
"logps/chosen": -272.4405212402344,
"logps/rejected": -621.2442626953125,
"loss": 0.0033,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.10709935426712036,
"rewards/margins": 26.78093910217285,
"rewards/rejected": -26.673839569091797,
"step": 2060
},
{
"epoch": 0.7,
"learning_rate": 1.6471477143936533e-07,
"logits/chosen": -3.4380111694335938,
"logits/rejected": -3.3767757415771484,
"logps/chosen": -248.4059295654297,
"logps/rejected": -633.5654296875,
"loss": 0.0096,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.7223705649375916,
"rewards/margins": 27.201065063476562,
"rewards/rejected": -26.478694915771484,
"step": 2070
},
{
"epoch": 0.71,
"learning_rate": 1.6282584057423498e-07,
"logits/chosen": -3.524405002593994,
"logits/rejected": -3.448660373687744,
"logps/chosen": -197.21920776367188,
"logps/rejected": -635.7800903320312,
"loss": 0.0211,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.5619519948959351,
"rewards/margins": 32.2112922668457,
"rewards/rejected": -32.77324676513672,
"step": 2080
},
{
"epoch": 0.71,
"learning_rate": 1.6093690970910464e-07,
"logits/chosen": -3.3876919746398926,
"logits/rejected": -3.3305823802948,
"logps/chosen": -287.99945068359375,
"logps/rejected": -709.9437255859375,
"loss": 0.0045,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.21997538208961487,
"rewards/margins": 31.6516170501709,
"rewards/rejected": -31.431640625,
"step": 2090
},
{
"epoch": 0.71,
"learning_rate": 1.590479788439743e-07,
"logits/chosen": -3.3942599296569824,
"logits/rejected": -3.2963924407958984,
"logps/chosen": -314.06060791015625,
"logps/rejected": -908.1524658203125,
"loss": 0.0026,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1970634162425995,
"rewards/margins": 28.25901222229004,
"rewards/rejected": -28.456073760986328,
"step": 2100
},
{
"epoch": 0.71,
"eval_logits/chosen": -3.497438669204712,
"eval_logits/rejected": -3.355656623840332,
"eval_logps/chosen": -257.67596435546875,
"eval_logps/rejected": -714.7033081054688,
"eval_loss": 0.010176397860050201,
"eval_rewards/accuracies": 0.996632993221283,
"eval_rewards/chosen": -0.18703816831111908,
"eval_rewards/margins": 28.85999298095703,
"eval_rewards/rejected": -29.047029495239258,
"eval_runtime": 460.179,
"eval_samples_per_second": 20.644,
"eval_steps_per_second": 0.645,
"step": 2100
},
{
"epoch": 0.72,
"learning_rate": 1.5715904797884398e-07,
"logits/chosen": -3.422182559967041,
"logits/rejected": -3.3635239601135254,
"logps/chosen": -237.2140350341797,
"logps/rejected": -596.4075317382812,
"loss": 0.0114,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.42542657256126404,
"rewards/margins": 25.331525802612305,
"rewards/rejected": -25.7569522857666,
"step": 2110
},
{
"epoch": 0.72,
"learning_rate": 1.5527011711371363e-07,
"logits/chosen": -3.46155047416687,
"logits/rejected": -3.287555694580078,
"logps/chosen": -221.025634765625,
"logps/rejected": -760.5899047851562,
"loss": 0.0066,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.587006688117981,
"rewards/margins": 23.792795181274414,
"rewards/rejected": -24.379802703857422,
"step": 2120
},
{
"epoch": 0.72,
"learning_rate": 1.533811862485833e-07,
"logits/chosen": -3.4075939655303955,
"logits/rejected": -3.2999377250671387,
"logps/chosen": -274.0217590332031,
"logps/rejected": -778.3746337890625,
"loss": 0.0048,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -1.0100147724151611,
"rewards/margins": 29.110565185546875,
"rewards/rejected": -30.120580673217773,
"step": 2130
},
{
"epoch": 0.73,
"learning_rate": 1.5149225538345294e-07,
"logits/chosen": -3.4074718952178955,
"logits/rejected": -3.328861951828003,
"logps/chosen": -234.0587921142578,
"logps/rejected": -573.584228515625,
"loss": 0.0481,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.8820997476577759,
"rewards/margins": 25.29749870300293,
"rewards/rejected": -26.17959976196289,
"step": 2140
},
{
"epoch": 0.73,
"learning_rate": 1.496033245183226e-07,
"logits/chosen": -3.433687925338745,
"logits/rejected": -3.2161545753479004,
"logps/chosen": -210.7549285888672,
"logps/rejected": -769.3878173828125,
"loss": 0.0106,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.381535142660141,
"rewards/margins": 30.15536117553711,
"rewards/rejected": -30.536895751953125,
"step": 2150
},
{
"epoch": 0.73,
"learning_rate": 1.477143936531923e-07,
"logits/chosen": -3.316527843475342,
"logits/rejected": -3.3601183891296387,
"logps/chosen": -334.7134094238281,
"logps/rejected": -479.4236755371094,
"loss": 0.0101,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.2200862169265747,
"rewards/margins": 26.942337036132812,
"rewards/rejected": -27.162424087524414,
"step": 2160
},
{
"epoch": 0.74,
"learning_rate": 1.4582546278806196e-07,
"logits/chosen": -3.372246503829956,
"logits/rejected": -3.286343812942505,
"logps/chosen": -269.2080993652344,
"logps/rejected": -628.6188354492188,
"loss": 0.0026,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.732342541217804,
"rewards/margins": 29.969751358032227,
"rewards/rejected": -30.702091217041016,
"step": 2170
},
{
"epoch": 0.74,
"learning_rate": 1.4393653192293162e-07,
"logits/chosen": -3.4120821952819824,
"logits/rejected": -3.2668113708496094,
"logps/chosen": -272.43658447265625,
"logps/rejected": -670.0343627929688,
"loss": 0.0172,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.10129819065332413,
"rewards/margins": 27.09328842163086,
"rewards/rejected": -27.194589614868164,
"step": 2180
},
{
"epoch": 0.74,
"learning_rate": 1.4204760105780127e-07,
"logits/chosen": -3.430737257003784,
"logits/rejected": -3.3516769409179688,
"logps/chosen": -270.27667236328125,
"logps/rejected": -783.06884765625,
"loss": 0.0057,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.26623135805130005,
"rewards/margins": 37.538352966308594,
"rewards/rejected": -37.80458450317383,
"step": 2190
},
{
"epoch": 0.75,
"learning_rate": 1.4015867019267093e-07,
"logits/chosen": -3.218029499053955,
"logits/rejected": -3.3251090049743652,
"logps/chosen": -504.22454833984375,
"logps/rejected": -690.3221435546875,
"loss": 0.0038,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.57887864112854,
"rewards/margins": 31.29214096069336,
"rewards/rejected": -31.871023178100586,
"step": 2200
},
{
"epoch": 0.75,
"eval_logits/chosen": -3.5046215057373047,
"eval_logits/rejected": -3.3551418781280518,
"eval_logps/chosen": -260.6087341308594,
"eval_logps/rejected": -723.0064086914062,
"eval_loss": 0.00784117542207241,
"eval_rewards/accuracies": 0.996632993221283,
"eval_rewards/chosen": -0.48031851649284363,
"eval_rewards/margins": 29.39702606201172,
"eval_rewards/rejected": -29.877344131469727,
"eval_runtime": 460.3396,
"eval_samples_per_second": 20.637,
"eval_steps_per_second": 0.645,
"step": 2200
},
{
"epoch": 0.75,
"learning_rate": 1.382697393275406e-07,
"logits/chosen": -3.2150256633758545,
"logits/rejected": -3.264436721801758,
"logps/chosen": -429.79412841796875,
"logps/rejected": -691.61767578125,
"loss": 0.0024,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.6222277283668518,
"rewards/margins": 24.18454360961914,
"rewards/rejected": -24.806772232055664,
"step": 2210
},
{
"epoch": 0.75,
"learning_rate": 1.3638080846241027e-07,
"logits/chosen": -3.491255283355713,
"logits/rejected": -3.2765109539031982,
"logps/chosen": -193.16854858398438,
"logps/rejected": -634.7061767578125,
"loss": 0.0178,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.7472248673439026,
"rewards/margins": 27.92132568359375,
"rewards/rejected": -28.66855239868164,
"step": 2220
},
{
"epoch": 0.76,
"learning_rate": 1.3449187759727992e-07,
"logits/chosen": -3.4371280670166016,
"logits/rejected": -3.394758701324463,
"logps/chosen": -200.95811462402344,
"logps/rejected": -568.9024047851562,
"loss": 0.0106,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.9408467411994934,
"rewards/margins": 25.23788833618164,
"rewards/rejected": -26.178735733032227,
"step": 2230
},
{
"epoch": 0.76,
"learning_rate": 1.3260294673214958e-07,
"logits/chosen": -3.448991060256958,
"logits/rejected": -3.209871292114258,
"logps/chosen": -215.184326171875,
"logps/rejected": -723.2108154296875,
"loss": 0.0051,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.9981611371040344,
"rewards/margins": 30.49850845336914,
"rewards/rejected": -31.496667861938477,
"step": 2240
},
{
"epoch": 0.76,
"learning_rate": 1.3071401586701926e-07,
"logits/chosen": -3.442951202392578,
"logits/rejected": -3.4533190727233887,
"logps/chosen": -226.30252075195312,
"logps/rejected": -531.5767822265625,
"loss": 0.0063,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.7389757633209229,
"rewards/margins": 25.352436065673828,
"rewards/rejected": -26.091405868530273,
"step": 2250
},
{
"epoch": 0.77,
"learning_rate": 1.2882508500188894e-07,
"logits/chosen": -3.467149257659912,
"logits/rejected": -3.3945045471191406,
"logps/chosen": -211.5034637451172,
"logps/rejected": -699.1697387695312,
"loss": 0.0066,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.5048588514328003,
"rewards/margins": 28.00215721130371,
"rewards/rejected": -28.50701904296875,
"step": 2260
},
{
"epoch": 0.77,
"learning_rate": 1.269361541367586e-07,
"logits/chosen": -3.3899827003479004,
"logits/rejected": -3.4136524200439453,
"logps/chosen": -312.1116943359375,
"logps/rejected": -670.1107177734375,
"loss": 0.0031,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.6437836289405823,
"rewards/margins": 30.72330665588379,
"rewards/rejected": -31.367090225219727,
"step": 2270
},
{
"epoch": 0.77,
"learning_rate": 1.2504722327162826e-07,
"logits/chosen": -3.4552528858184814,
"logits/rejected": -3.3502261638641357,
"logps/chosen": -222.96029663085938,
"logps/rejected": -634.1143798828125,
"loss": 0.0051,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.35589832067489624,
"rewards/margins": 26.0993595123291,
"rewards/rejected": -26.455257415771484,
"step": 2280
},
{
"epoch": 0.78,
"learning_rate": 1.231582924064979e-07,
"logits/chosen": -3.4500694274902344,
"logits/rejected": -3.356391191482544,
"logps/chosen": -226.2766876220703,
"logps/rejected": -721.9498901367188,
"loss": 0.0202,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.8104637265205383,
"rewards/margins": 28.76261329650879,
"rewards/rejected": -29.573078155517578,
"step": 2290
},
{
"epoch": 0.78,
"learning_rate": 1.2126936154136757e-07,
"logits/chosen": -3.4218978881835938,
"logits/rejected": -3.3485636711120605,
"logps/chosen": -249.322265625,
"logps/rejected": -696.6746826171875,
"loss": 0.0011,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.3162160813808441,
"rewards/margins": 24.947269439697266,
"rewards/rejected": -25.263486862182617,
"step": 2300
},
{
"epoch": 0.78,
"eval_logits/chosen": -3.4948203563690186,
"eval_logits/rejected": -3.345940589904785,
"eval_logps/chosen": -260.5770263671875,
"eval_logps/rejected": -708.5813598632812,
"eval_loss": 0.007498822640627623,
"eval_rewards/accuracies": 0.996632993221283,
"eval_rewards/chosen": -0.4771437644958496,
"eval_rewards/margins": 27.957693099975586,
"eval_rewards/rejected": -28.434837341308594,
"eval_runtime": 461.2562,
"eval_samples_per_second": 20.596,
"eval_steps_per_second": 0.644,
"step": 2300
},
{
"epoch": 0.79,
"learning_rate": 1.1938043067623725e-07,
"logits/chosen": -3.370896816253662,
"logits/rejected": -3.3326168060302734,
"logps/chosen": -347.14337158203125,
"logps/rejected": -689.7429809570312,
"loss": 0.0081,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.18211320042610168,
"rewards/margins": 27.03213119506836,
"rewards/rejected": -26.850021362304688,
"step": 2310
},
{
"epoch": 0.79,
"learning_rate": 1.1749149981110692e-07,
"logits/chosen": -3.3958702087402344,
"logits/rejected": -3.369236707687378,
"logps/chosen": -246.0237579345703,
"logps/rejected": -561.996826171875,
"loss": 0.0076,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.559197723865509,
"rewards/margins": 20.26263427734375,
"rewards/rejected": -20.82183074951172,
"step": 2320
},
{
"epoch": 0.79,
"learning_rate": 1.1560256894597657e-07,
"logits/chosen": -3.3493895530700684,
"logits/rejected": -3.370333433151245,
"logps/chosen": -424.63336181640625,
"logps/rejected": -620.3344116210938,
"loss": 0.0134,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.06108301132917404,
"rewards/margins": 25.891162872314453,
"rewards/rejected": -25.952245712280273,
"step": 2330
},
{
"epoch": 0.8,
"learning_rate": 1.1371363808084623e-07,
"logits/chosen": -3.4243323802948,
"logits/rejected": -3.247948169708252,
"logps/chosen": -254.7300262451172,
"logps/rejected": -1042.1585693359375,
"loss": 0.0018,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.3867115378379822,
"rewards/margins": 35.40499496459961,
"rewards/rejected": -35.01828384399414,
"step": 2340
},
{
"epoch": 0.8,
"learning_rate": 1.118247072157159e-07,
"logits/chosen": -3.4594218730926514,
"logits/rejected": -3.3651282787323,
"logps/chosen": -242.59677124023438,
"logps/rejected": -538.2279663085938,
"loss": 0.0067,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.49601420760154724,
"rewards/margins": 23.69955062866211,
"rewards/rejected": -24.19556427001953,
"step": 2350
},
{
"epoch": 0.8,
"learning_rate": 1.0993577635058557e-07,
"logits/chosen": -3.511322498321533,
"logits/rejected": -3.356180191040039,
"logps/chosen": -196.69949340820312,
"logps/rejected": -571.466796875,
"loss": 0.0076,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.02175600454211235,
"rewards/margins": 25.043487548828125,
"rewards/rejected": -25.021728515625,
"step": 2360
},
{
"epoch": 0.81,
"learning_rate": 1.0804684548545522e-07,
"logits/chosen": -3.4236786365509033,
"logits/rejected": -3.338402509689331,
"logps/chosen": -247.4124755859375,
"logps/rejected": -755.9312133789062,
"loss": 0.0036,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.05236620828509331,
"rewards/margins": 24.615774154663086,
"rewards/rejected": -24.668140411376953,
"step": 2370
},
{
"epoch": 0.81,
"learning_rate": 1.0615791462032489e-07,
"logits/chosen": -3.413336992263794,
"logits/rejected": -3.3395888805389404,
"logps/chosen": -274.93182373046875,
"logps/rejected": -602.7622680664062,
"loss": 0.0037,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.4528873562812805,
"rewards/margins": 26.461483001708984,
"rewards/rejected": -26.91436767578125,
"step": 2380
},
{
"epoch": 0.81,
"learning_rate": 1.0426898375519455e-07,
"logits/chosen": -3.4481735229492188,
"logits/rejected": -3.3065543174743652,
"logps/chosen": -273.5448913574219,
"logps/rejected": -812.5321044921875,
"loss": 0.1677,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.46621403098106384,
"rewards/margins": 29.433242797851562,
"rewards/rejected": -29.899456024169922,
"step": 2390
},
{
"epoch": 0.82,
"learning_rate": 1.0238005289006423e-07,
"logits/chosen": -3.498518466949463,
"logits/rejected": -3.352489948272705,
"logps/chosen": -229.6658172607422,
"logps/rejected": -674.809814453125,
"loss": 0.0033,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.506945788860321,
"rewards/margins": 24.02678680419922,
"rewards/rejected": -24.533733367919922,
"step": 2400
},
{
"epoch": 0.82,
"eval_logits/chosen": -3.4949593544006348,
"eval_logits/rejected": -3.3488945960998535,
"eval_logps/chosen": -257.8039245605469,
"eval_logps/rejected": -704.2630615234375,
"eval_loss": 0.00467069773003459,
"eval_rewards/accuracies": 0.9983165264129639,
"eval_rewards/chosen": -0.19983212649822235,
"eval_rewards/margins": 27.803180694580078,
"eval_rewards/rejected": -28.00301170349121,
"eval_runtime": 460.1772,
"eval_samples_per_second": 20.644,
"eval_steps_per_second": 0.645,
"step": 2400
},
{
"epoch": 0.82,
"learning_rate": 1.0049112202493389e-07,
"logits/chosen": -3.458670139312744,
"logits/rejected": -3.34519624710083,
"logps/chosen": -229.53012084960938,
"logps/rejected": -653.47021484375,
"loss": 0.9205,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.46636518836021423,
"rewards/margins": 31.985393524169922,
"rewards/rejected": -32.45175552368164,
"step": 2410
},
{
"epoch": 0.82,
"learning_rate": 9.860219115980354e-08,
"logits/chosen": -3.354882001876831,
"logits/rejected": -3.3362224102020264,
"logps/chosen": -298.93890380859375,
"logps/rejected": -575.4816284179688,
"loss": 0.0022,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.2504803538322449,
"rewards/margins": 25.459463119506836,
"rewards/rejected": -25.208984375,
"step": 2420
},
{
"epoch": 0.83,
"learning_rate": 9.671326029467321e-08,
"logits/chosen": -3.409818649291992,
"logits/rejected": -3.294722080230713,
"logps/chosen": -264.5761413574219,
"logps/rejected": -753.7411499023438,
"loss": 0.0166,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.6481723189353943,
"rewards/margins": 29.749752044677734,
"rewards/rejected": -30.39792251586914,
"step": 2430
},
{
"epoch": 0.83,
"learning_rate": 9.482432942954287e-08,
"logits/chosen": -3.398315906524658,
"logits/rejected": -3.2785720825195312,
"logps/chosen": -275.75506591796875,
"logps/rejected": -881.2058715820312,
"loss": 0.0051,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.4502645432949066,
"rewards/margins": 33.76361846923828,
"rewards/rejected": -34.21388626098633,
"step": 2440
},
{
"epoch": 0.83,
"learning_rate": 9.293539856441255e-08,
"logits/chosen": -3.4217476844787598,
"logits/rejected": -3.160505533218384,
"logps/chosen": -211.6370849609375,
"logps/rejected": -842.2235107421875,
"loss": 0.0173,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.4303048253059387,
"rewards/margins": 34.46601486206055,
"rewards/rejected": -34.89632034301758,
"step": 2450
},
{
"epoch": 0.84,
"learning_rate": 9.10464676992822e-08,
"logits/chosen": -3.360055446624756,
"logits/rejected": -3.2219321727752686,
"logps/chosen": -197.63504028320312,
"logps/rejected": -700.8560791015625,
"loss": 0.0038,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.5080915689468384,
"rewards/margins": 28.26936912536621,
"rewards/rejected": -28.7774600982666,
"step": 2460
},
{
"epoch": 0.84,
"learning_rate": 8.915753683415186e-08,
"logits/chosen": -3.324397325515747,
"logits/rejected": -3.2371230125427246,
"logps/chosen": -265.84197998046875,
"logps/rejected": -706.1388549804688,
"loss": 0.0212,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.3217003345489502,
"rewards/margins": 31.357486724853516,
"rewards/rejected": -31.035785675048828,
"step": 2470
},
{
"epoch": 0.84,
"learning_rate": 8.726860596902153e-08,
"logits/chosen": -3.324913501739502,
"logits/rejected": -3.31086802482605,
"logps/chosen": -289.2586364746094,
"logps/rejected": -618.4796142578125,
"loss": 0.0052,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.31292805075645447,
"rewards/margins": 25.791996002197266,
"rewards/rejected": -25.47906494140625,
"step": 2480
},
{
"epoch": 0.85,
"learning_rate": 8.53796751038912e-08,
"logits/chosen": -3.358564853668213,
"logits/rejected": -3.3472542762756348,
"logps/chosen": -252.2576141357422,
"logps/rejected": -531.1395263671875,
"loss": 0.0094,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.5012627840042114,
"rewards/margins": 24.36186408996582,
"rewards/rejected": -24.86312484741211,
"step": 2490
},
{
"epoch": 0.85,
"learning_rate": 8.349074423876085e-08,
"logits/chosen": -3.2991740703582764,
"logits/rejected": -3.3889808654785156,
"logps/chosen": -312.4256591796875,
"logps/rejected": -605.0225830078125,
"loss": 0.0051,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.34426942467689514,
"rewards/margins": 31.89084243774414,
"rewards/rejected": -31.546573638916016,
"step": 2500
},
{
"epoch": 0.85,
"eval_logits/chosen": -3.4427988529205322,
"eval_logits/rejected": -3.3025379180908203,
"eval_logps/chosen": -258.5765075683594,
"eval_logps/rejected": -716.5906372070312,
"eval_loss": 0.004813064821064472,
"eval_rewards/accuracies": 0.9991582632064819,
"eval_rewards/chosen": -0.2770873010158539,
"eval_rewards/margins": 28.958681106567383,
"eval_rewards/rejected": -29.23576545715332,
"eval_runtime": 460.0317,
"eval_samples_per_second": 20.651,
"eval_steps_per_second": 0.646,
"step": 2500
},
{
"epoch": 0.85,
"learning_rate": 8.160181337363052e-08,
"logits/chosen": -3.3120341300964355,
"logits/rejected": -3.2296226024627686,
"logps/chosen": -280.31353759765625,
"logps/rejected": -694.3499755859375,
"loss": 0.0019,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.37215790152549744,
"rewards/margins": 28.98830795288086,
"rewards/rejected": -29.3604679107666,
"step": 2510
},
{
"epoch": 0.86,
"learning_rate": 7.971288250850018e-08,
"logits/chosen": -3.4131808280944824,
"logits/rejected": -3.1928412914276123,
"logps/chosen": -205.73983764648438,
"logps/rejected": -883.18017578125,
"loss": 0.0071,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.5982518196105957,
"rewards/margins": 27.798303604125977,
"rewards/rejected": -28.39655113220215,
"step": 2520
},
{
"epoch": 0.86,
"learning_rate": 7.782395164336985e-08,
"logits/chosen": -3.383462429046631,
"logits/rejected": -3.292088747024536,
"logps/chosen": -337.71136474609375,
"logps/rejected": -649.876220703125,
"loss": 0.3124,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -2.490572690963745,
"rewards/margins": 20.789615631103516,
"rewards/rejected": -23.280189514160156,
"step": 2530
},
{
"epoch": 0.86,
"learning_rate": 7.593502077823952e-08,
"logits/chosen": -3.3705170154571533,
"logits/rejected": -3.262524127960205,
"logps/chosen": -210.7335205078125,
"logps/rejected": -573.064453125,
"loss": 0.2135,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.23895084857940674,
"rewards/margins": 23.423107147216797,
"rewards/rejected": -23.662057876586914,
"step": 2540
},
{
"epoch": 0.87,
"learning_rate": 7.404608991310917e-08,
"logits/chosen": -3.3921730518341064,
"logits/rejected": -3.3298392295837402,
"logps/chosen": -256.29339599609375,
"logps/rejected": -685.7628784179688,
"loss": 0.0059,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.020406579598784447,
"rewards/margins": 28.857458114624023,
"rewards/rejected": -28.837055206298828,
"step": 2550
},
{
"epoch": 0.87,
"learning_rate": 7.215715904797884e-08,
"logits/chosen": -3.4744045734405518,
"logits/rejected": -3.286008834838867,
"logps/chosen": -214.1748046875,
"logps/rejected": -831.2682495117188,
"loss": 0.0113,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.8809574842453003,
"rewards/margins": 33.92351150512695,
"rewards/rejected": -34.804466247558594,
"step": 2560
},
{
"epoch": 0.87,
"learning_rate": 7.02682281828485e-08,
"logits/chosen": -3.3835952281951904,
"logits/rejected": -3.3155086040496826,
"logps/chosen": -275.273193359375,
"logps/rejected": -767.3925170898438,
"loss": 0.0098,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.5842048525810242,
"rewards/margins": 30.7226505279541,
"rewards/rejected": -31.30685806274414,
"step": 2570
},
{
"epoch": 0.88,
"learning_rate": 6.837929731771818e-08,
"logits/chosen": -3.317415952682495,
"logits/rejected": -3.3041164875030518,
"logps/chosen": -258.95086669921875,
"logps/rejected": -488.6026916503906,
"loss": 0.0068,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17078134417533875,
"rewards/margins": 23.854570388793945,
"rewards/rejected": -24.025352478027344,
"step": 2580
},
{
"epoch": 0.88,
"learning_rate": 6.649036645258783e-08,
"logits/chosen": -3.344968795776367,
"logits/rejected": -3.2897567749023438,
"logps/chosen": -314.9465026855469,
"logps/rejected": -535.2317504882812,
"loss": 0.0015,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.07202013581991196,
"rewards/margins": 28.91226577758789,
"rewards/rejected": -28.984283447265625,
"step": 2590
},
{
"epoch": 0.88,
"learning_rate": 6.460143558745749e-08,
"logits/chosen": -3.419276475906372,
"logits/rejected": -3.2689127922058105,
"logps/chosen": -205.4881134033203,
"logps/rejected": -798.7406005859375,
"loss": 0.0074,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.08600559085607529,
"rewards/margins": 38.14585494995117,
"rewards/rejected": -38.059852600097656,
"step": 2600
},
{
"epoch": 0.88,
"eval_logits/chosen": -3.4804773330688477,
"eval_logits/rejected": -3.332041025161743,
"eval_logps/chosen": -257.89501953125,
"eval_logps/rejected": -720.7188720703125,
"eval_loss": 0.004375319927930832,
"eval_rewards/accuracies": 0.997474730014801,
"eval_rewards/chosen": -0.20894265174865723,
"eval_rewards/margins": 29.43964385986328,
"eval_rewards/rejected": -29.64858627319336,
"eval_runtime": 460.0382,
"eval_samples_per_second": 20.65,
"eval_steps_per_second": 0.646,
"step": 2600
},
{
"epoch": 0.89,
"learning_rate": 6.271250472232716e-08,
"logits/chosen": -3.394615888595581,
"logits/rejected": -3.2852377891540527,
"logps/chosen": -266.36767578125,
"logps/rejected": -717.4393310546875,
"loss": 0.0151,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 0.6554498672485352,
"rewards/margins": 26.674755096435547,
"rewards/rejected": -26.019306182861328,
"step": 2610
},
{
"epoch": 0.89,
"learning_rate": 6.082357385719683e-08,
"logits/chosen": -3.381989002227783,
"logits/rejected": -3.3894596099853516,
"logps/chosen": -342.04913330078125,
"logps/rejected": -546.2716674804688,
"loss": 0.0047,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.6526740193367004,
"rewards/margins": 21.150623321533203,
"rewards/rejected": -21.803295135498047,
"step": 2620
},
{
"epoch": 0.89,
"learning_rate": 5.893464299206649e-08,
"logits/chosen": -3.4155497550964355,
"logits/rejected": -3.2923424243927,
"logps/chosen": -223.69140625,
"logps/rejected": -731.6795043945312,
"loss": 0.0058,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.3694123923778534,
"rewards/margins": 28.902706146240234,
"rewards/rejected": -29.27211570739746,
"step": 2630
},
{
"epoch": 0.9,
"learning_rate": 5.704571212693615e-08,
"logits/chosen": -3.330949306488037,
"logits/rejected": -3.3261523246765137,
"logps/chosen": -248.51528930664062,
"logps/rejected": -587.5772705078125,
"loss": 0.0021,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.7009295225143433,
"rewards/margins": 24.234630584716797,
"rewards/rejected": -24.935558319091797,
"step": 2640
},
{
"epoch": 0.9,
"learning_rate": 5.5156781261805816e-08,
"logits/chosen": -3.4656805992126465,
"logits/rejected": -3.3214333057403564,
"logps/chosen": -191.90036010742188,
"logps/rejected": -628.0974731445312,
"loss": 0.001,
"rewards/accuracies": 1.0,
"rewards/chosen": -1.088071584701538,
"rewards/margins": 31.438037872314453,
"rewards/rejected": -32.5261116027832,
"step": 2650
},
{
"epoch": 0.9,
"learning_rate": 5.326785039667548e-08,
"logits/chosen": -3.4162774085998535,
"logits/rejected": -3.4043285846710205,
"logps/chosen": -270.454345703125,
"logps/rejected": -554.826171875,
"loss": 0.001,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.00885864533483982,
"rewards/margins": 29.386489868164062,
"rewards/rejected": -29.395349502563477,
"step": 2660
},
{
"epoch": 0.91,
"learning_rate": 5.137891953154514e-08,
"logits/chosen": -3.5397841930389404,
"logits/rejected": -3.299877166748047,
"logps/chosen": -220.99899291992188,
"logps/rejected": -832.1795654296875,
"loss": 0.0054,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.49379825592041016,
"rewards/margins": 30.90732192993164,
"rewards/rejected": -31.4011173248291,
"step": 2670
},
{
"epoch": 0.91,
"learning_rate": 4.948998866641481e-08,
"logits/chosen": -3.422128200531006,
"logits/rejected": -3.272432327270508,
"logps/chosen": -266.85186767578125,
"logps/rejected": -875.5877075195312,
"loss": 0.0027,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.2771952152252197,
"rewards/margins": 30.630329132080078,
"rewards/rejected": -30.90752601623535,
"step": 2680
},
{
"epoch": 0.91,
"learning_rate": 4.760105780128447e-08,
"logits/chosen": -3.4342148303985596,
"logits/rejected": -3.2355589866638184,
"logps/chosen": -297.9632873535156,
"logps/rejected": -929.0408935546875,
"loss": 0.0018,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.417285293340683,
"rewards/margins": 34.31209182739258,
"rewards/rejected": -34.729373931884766,
"step": 2690
},
{
"epoch": 0.92,
"learning_rate": 4.5712126936154134e-08,
"logits/chosen": -3.488246202468872,
"logits/rejected": -3.3506011962890625,
"logps/chosen": -173.14797973632812,
"logps/rejected": -620.3333740234375,
"loss": 0.0032,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.3520640432834625,
"rewards/margins": 29.699413299560547,
"rewards/rejected": -30.05147933959961,
"step": 2700
},
{
"epoch": 0.92,
"eval_logits/chosen": -3.482171058654785,
"eval_logits/rejected": -3.3307948112487793,
"eval_logps/chosen": -257.48095703125,
"eval_logps/rejected": -726.024169921875,
"eval_loss": 0.004117514006793499,
"eval_rewards/accuracies": 0.997474730014801,
"eval_rewards/chosen": -0.16753698885440826,
"eval_rewards/margins": 30.011581420898438,
"eval_rewards/rejected": -30.179115295410156,
"eval_runtime": 459.6071,
"eval_samples_per_second": 20.67,
"eval_steps_per_second": 0.646,
"step": 2700
},
{
"epoch": 0.92,
"learning_rate": 4.3823196071023796e-08,
"logits/chosen": -3.4520740509033203,
"logits/rejected": -3.2903473377227783,
"logps/chosen": -204.6480712890625,
"logps/rejected": -706.2102661132812,
"loss": 0.0015,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.4886886179447174,
"rewards/margins": 32.48357009887695,
"rewards/rejected": -32.972259521484375,
"step": 2710
},
{
"epoch": 0.92,
"learning_rate": 4.1934265205893465e-08,
"logits/chosen": -3.3444876670837402,
"logits/rejected": -3.3331406116485596,
"logps/chosen": -345.8191223144531,
"logps/rejected": -787.9623413085938,
"loss": 0.0022,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.5544703006744385,
"rewards/margins": 30.112964630126953,
"rewards/rejected": -29.558496475219727,
"step": 2720
},
{
"epoch": 0.93,
"learning_rate": 4.004533434076313e-08,
"logits/chosen": -3.449005126953125,
"logits/rejected": -3.3541347980499268,
"logps/chosen": -205.72506713867188,
"logps/rejected": -782.6888427734375,
"loss": 0.0019,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.2865965962409973,
"rewards/margins": 26.386667251586914,
"rewards/rejected": -26.673263549804688,
"step": 2730
},
{
"epoch": 0.93,
"learning_rate": 3.815640347563279e-08,
"logits/chosen": -3.330857038497925,
"logits/rejected": -3.237947463989258,
"logps/chosen": -281.54742431640625,
"logps/rejected": -995.6536865234375,
"loss": 0.0023,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.840170681476593,
"rewards/margins": 30.559356689453125,
"rewards/rejected": -31.399526596069336,
"step": 2740
},
{
"epoch": 0.93,
"learning_rate": 3.626747261050245e-08,
"logits/chosen": -3.4661128520965576,
"logits/rejected": -3.30942964553833,
"logps/chosen": -212.7915496826172,
"logps/rejected": -827.4059448242188,
"loss": 0.0112,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.515861988067627,
"rewards/margins": 28.297542572021484,
"rewards/rejected": -28.813405990600586,
"step": 2750
},
{
"epoch": 0.94,
"learning_rate": 3.4378541745372115e-08,
"logits/chosen": -3.424384355545044,
"logits/rejected": -3.36513090133667,
"logps/chosen": -288.58001708984375,
"logps/rejected": -790.6275634765625,
"loss": 0.0089,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.2595888078212738,
"rewards/margins": 32.224891662597656,
"rewards/rejected": -31.965301513671875,
"step": 2760
},
{
"epoch": 0.94,
"learning_rate": 3.2489610880241784e-08,
"logits/chosen": -3.450979232788086,
"logits/rejected": -3.3050060272216797,
"logps/chosen": -198.88816833496094,
"logps/rejected": -727.8658447265625,
"loss": 0.0016,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.09249986708164215,
"rewards/margins": 30.053543090820312,
"rewards/rejected": -29.96103858947754,
"step": 2770
},
{
"epoch": 0.94,
"learning_rate": 3.0600680015111446e-08,
"logits/chosen": -3.4271132946014404,
"logits/rejected": -3.3997604846954346,
"logps/chosen": -195.2742919921875,
"logps/rejected": -512.84619140625,
"loss": 0.0051,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.01265418529510498,
"rewards/margins": 30.05193519592285,
"rewards/rejected": -30.064586639404297,
"step": 2780
},
{
"epoch": 0.95,
"learning_rate": 2.871174914998111e-08,
"logits/chosen": -3.450589656829834,
"logits/rejected": -3.260341167449951,
"logps/chosen": -221.8329315185547,
"logps/rejected": -699.6900634765625,
"loss": 0.0042,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.08895153552293777,
"rewards/margins": 34.512027740478516,
"rewards/rejected": -34.423072814941406,
"step": 2790
},
{
"epoch": 0.95,
"learning_rate": 2.682281828485077e-08,
"logits/chosen": -3.405123472213745,
"logits/rejected": -3.2812092304229736,
"logps/chosen": -212.4518585205078,
"logps/rejected": -927.1981201171875,
"loss": 0.0023,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.3205551505088806,
"rewards/margins": 29.993661880493164,
"rewards/rejected": -30.314218521118164,
"step": 2800
},
{
"epoch": 0.95,
"eval_logits/chosen": -3.4751133918762207,
"eval_logits/rejected": -3.3266823291778564,
"eval_logps/chosen": -255.20126342773438,
"eval_logps/rejected": -718.1399536132812,
"eval_loss": 0.003812924027442932,
"eval_rewards/accuracies": 0.9983165264129639,
"eval_rewards/chosen": 0.06043152138590813,
"eval_rewards/margins": 29.45113182067871,
"eval_rewards/rejected": -29.390703201293945,
"eval_runtime": 459.6851,
"eval_samples_per_second": 20.666,
"eval_steps_per_second": 0.646,
"step": 2800
},
{
"epoch": 0.96,
"learning_rate": 2.4933887419720436e-08,
"logits/chosen": -3.470411777496338,
"logits/rejected": -3.371030330657959,
"logps/chosen": -203.39833068847656,
"logps/rejected": -554.4922485351562,
"loss": 0.0043,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.4396284222602844,
"rewards/margins": 28.658405303955078,
"rewards/rejected": -29.0980281829834,
"step": 2810
},
{
"epoch": 0.96,
"learning_rate": 2.30449565545901e-08,
"logits/chosen": -3.403038740158081,
"logits/rejected": -3.3339126110076904,
"logps/chosen": -218.4235076904297,
"logps/rejected": -501.0118713378906,
"loss": 0.0107,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.3079354763031006,
"rewards/margins": 22.659008026123047,
"rewards/rejected": -22.96694564819336,
"step": 2820
},
{
"epoch": 0.96,
"learning_rate": 2.1156025689459764e-08,
"logits/chosen": -3.360138416290283,
"logits/rejected": -3.204312562942505,
"logps/chosen": -275.4970703125,
"logps/rejected": -807.7415771484375,
"loss": 0.0047,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.563244104385376,
"rewards/margins": 28.03116226196289,
"rewards/rejected": -27.467914581298828,
"step": 2830
},
{
"epoch": 0.97,
"learning_rate": 1.926709482432943e-08,
"logits/chosen": -3.4086861610412598,
"logits/rejected": -3.273714542388916,
"logps/chosen": -241.6406707763672,
"logps/rejected": -834.037109375,
"loss": 0.0142,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 0.5851477384567261,
"rewards/margins": 38.93583679199219,
"rewards/rejected": -38.350685119628906,
"step": 2840
},
{
"epoch": 0.97,
"learning_rate": 1.7378163959199092e-08,
"logits/chosen": -3.443998336791992,
"logits/rejected": -3.288005828857422,
"logps/chosen": -259.77630615234375,
"logps/rejected": -637.4151611328125,
"loss": 0.0042,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.25292807817459106,
"rewards/margins": 29.613880157470703,
"rewards/rejected": -29.360952377319336,
"step": 2850
},
{
"epoch": 0.97,
"learning_rate": 1.5489233094068758e-08,
"logits/chosen": -3.4653220176696777,
"logits/rejected": -3.3464725017547607,
"logps/chosen": -222.8519287109375,
"logps/rejected": -588.8685913085938,
"loss": 0.0037,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.2632875442504883,
"rewards/margins": 26.466629028320312,
"rewards/rejected": -26.729915618896484,
"step": 2860
},
{
"epoch": 0.98,
"learning_rate": 1.3600302228938419e-08,
"logits/chosen": -3.376246929168701,
"logits/rejected": -3.2897162437438965,
"logps/chosen": -312.7208251953125,
"logps/rejected": -748.8970336914062,
"loss": 0.0038,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 0.849084734916687,
"rewards/margins": 28.013254165649414,
"rewards/rejected": -27.164173126220703,
"step": 2870
},
{
"epoch": 0.98,
"learning_rate": 1.1711371363808084e-08,
"logits/chosen": -3.442333936691284,
"logits/rejected": -3.225363254547119,
"logps/chosen": -266.2436828613281,
"logps/rejected": -922.4913330078125,
"loss": 0.0051,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.09482870250940323,
"rewards/margins": 27.245941162109375,
"rewards/rejected": -27.340768814086914,
"step": 2880
},
{
"epoch": 0.98,
"learning_rate": 9.822440498677748e-09,
"logits/chosen": -3.2727837562561035,
"logits/rejected": -3.270648241043091,
"logps/chosen": -396.17156982421875,
"logps/rejected": -819.6370239257812,
"loss": 0.0042,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 0.2518737316131592,
"rewards/margins": 33.05315017700195,
"rewards/rejected": -32.80127716064453,
"step": 2890
},
{
"epoch": 0.99,
"learning_rate": 7.933509633547412e-09,
"logits/chosen": -3.4724514484405518,
"logits/rejected": -3.2815041542053223,
"logps/chosen": -211.4152374267578,
"logps/rejected": -643.9974365234375,
"loss": 0.003,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16061940789222717,
"rewards/margins": 27.16357421875,
"rewards/rejected": -27.324193954467773,
"step": 2900
},
{
"epoch": 0.99,
"eval_logits/chosen": -3.4722537994384766,
"eval_logits/rejected": -3.3257179260253906,
"eval_logps/chosen": -254.35963439941406,
"eval_logps/rejected": -714.0263671875,
"eval_loss": 0.003982194699347019,
"eval_rewards/accuracies": 0.9983165264129639,
"eval_rewards/chosen": 0.1445969194173813,
"eval_rewards/margins": 29.123931884765625,
"eval_rewards/rejected": -28.979337692260742,
"eval_runtime": 459.5744,
"eval_samples_per_second": 20.671,
"eval_steps_per_second": 0.646,
"step": 2900
},
{
"epoch": 0.99,
"learning_rate": 6.044578768417076e-09,
"logits/chosen": -3.3734288215637207,
"logits/rejected": -3.2156894207000732,
"logps/chosen": -206.44552612304688,
"logps/rejected": -797.2452392578125,
"loss": 0.0135,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.3702377378940582,
"rewards/margins": 27.78318214416504,
"rewards/rejected": -28.153417587280273,
"step": 2910
},
{
"epoch": 0.99,
"learning_rate": 4.15564790328674e-09,
"logits/chosen": -3.3592917919158936,
"logits/rejected": -3.2777762413024902,
"logps/chosen": -216.9835968017578,
"logps/rejected": -824.8385620117188,
"loss": 0.0014,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.20981796085834503,
"rewards/margins": 28.391815185546875,
"rewards/rejected": -28.601633071899414,
"step": 2920
},
{
"epoch": 1.0,
"learning_rate": 2.2667170381564033e-09,
"logits/chosen": -3.416182041168213,
"logits/rejected": -3.359605312347412,
"logps/chosen": -203.88467407226562,
"logps/rejected": -526.6511840820312,
"loss": 0.0036,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.26971641182899475,
"rewards/margins": 26.130962371826172,
"rewards/rejected": -26.400676727294922,
"step": 2930
},
{
"epoch": 1.0,
"learning_rate": 3.7778617302606723e-10,
"logits/chosen": -3.350004196166992,
"logits/rejected": -3.2843971252441406,
"logps/chosen": -274.75946044921875,
"logps/rejected": -673.1805419921875,
"loss": 0.0035,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.04258955270051956,
"rewards/margins": 28.780603408813477,
"rewards/rejected": -28.82318687438965,
"step": 2940
},
{
"epoch": 1.0,
"step": 2942,
"total_flos": 0.0,
"train_loss": 0.0552501158515984,
"train_runtime": 31216.715,
"train_samples_per_second": 6.032,
"train_steps_per_second": 0.094
}
],
"logging_steps": 10,
"max_steps": 2942,
"num_train_epochs": 1,
"save_steps": 500,
"total_flos": 0.0,
"trial_name": null,
"trial_params": null
}