dctanner's picture
Model save
e291de3 verified
raw
history blame
52.3 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.999508116084604,
"eval_steps": 200,
"global_step": 1016,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 4.901960784313726e-08,
"logits/chosen": -2.0737838745117188,
"logits/rejected": -2.1456010341644287,
"logps/chosen": -95.6572265625,
"logps/rejected": -106.55765533447266,
"loss": 0.6931,
"rewards/accuracies": 0.0,
"rewards/chosen": 0.0,
"rewards/margins": 0.0,
"rewards/rejected": 0.0,
"step": 1
},
{
"epoch": 0.01,
"learning_rate": 4.901960784313725e-07,
"logits/chosen": -2.165830373764038,
"logits/rejected": -2.060776948928833,
"logps/chosen": -121.03773498535156,
"logps/rejected": -87.5294189453125,
"loss": 0.6918,
"rewards/accuracies": 0.5,
"rewards/chosen": 0.00047249632189050317,
"rewards/margins": 0.002704059472307563,
"rewards/rejected": -0.002231562975794077,
"step": 10
},
{
"epoch": 0.02,
"learning_rate": 9.80392156862745e-07,
"logits/chosen": -2.1349050998687744,
"logits/rejected": -2.016066312789917,
"logps/chosen": -130.94175720214844,
"logps/rejected": -105.7674789428711,
"loss": 0.6935,
"rewards/accuracies": 0.48750001192092896,
"rewards/chosen": -0.0016262540593743324,
"rewards/margins": -0.0006325626163743436,
"rewards/rejected": -0.0009936915012076497,
"step": 20
},
{
"epoch": 0.03,
"learning_rate": 1.4705882352941177e-06,
"logits/chosen": -2.2818872928619385,
"logits/rejected": -2.1805636882781982,
"logps/chosen": -121.03263854980469,
"logps/rejected": -104.84712982177734,
"loss": 0.6942,
"rewards/accuracies": 0.3499999940395355,
"rewards/chosen": -0.003329185303300619,
"rewards/margins": -0.002142944373190403,
"rewards/rejected": -0.0011862408136948943,
"step": 30
},
{
"epoch": 0.04,
"learning_rate": 1.96078431372549e-06,
"logits/chosen": -2.2757694721221924,
"logits/rejected": -2.156691074371338,
"logps/chosen": -126.5389633178711,
"logps/rejected": -105.20024108886719,
"loss": 0.6919,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": 0.0010745985200628638,
"rewards/margins": 0.002502765040844679,
"rewards/rejected": -0.0014281660551205277,
"step": 40
},
{
"epoch": 0.05,
"learning_rate": 2.450980392156863e-06,
"logits/chosen": -2.3002381324768066,
"logits/rejected": -2.206784725189209,
"logps/chosen": -124.18415832519531,
"logps/rejected": -98.63652801513672,
"loss": 0.6916,
"rewards/accuracies": 0.612500011920929,
"rewards/chosen": 0.0012203993974253535,
"rewards/margins": 0.003048995044082403,
"rewards/rejected": -0.0018285956466570497,
"step": 50
},
{
"epoch": 0.06,
"learning_rate": 2.9411764705882355e-06,
"logits/chosen": -2.3112730979919434,
"logits/rejected": -2.232532501220703,
"logps/chosen": -126.93055725097656,
"logps/rejected": -109.7610092163086,
"loss": 0.6914,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": 0.01018393412232399,
"rewards/margins": 0.0034624538384377956,
"rewards/rejected": 0.006721480283886194,
"step": 60
},
{
"epoch": 0.07,
"learning_rate": 3.431372549019608e-06,
"logits/chosen": -2.2930376529693604,
"logits/rejected": -2.187328815460205,
"logps/chosen": -119.68111419677734,
"logps/rejected": -93.93470001220703,
"loss": 0.6912,
"rewards/accuracies": 0.5874999761581421,
"rewards/chosen": 0.007645626552402973,
"rewards/margins": 0.003846182022243738,
"rewards/rejected": 0.0037994447629898787,
"step": 70
},
{
"epoch": 0.08,
"learning_rate": 3.92156862745098e-06,
"logits/chosen": -2.144854784011841,
"logits/rejected": -2.016728401184082,
"logps/chosen": -132.2415313720703,
"logps/rejected": -106.7207260131836,
"loss": 0.6902,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": 0.015773242339491844,
"rewards/margins": 0.006012483034282923,
"rewards/rejected": 0.009760759770870209,
"step": 80
},
{
"epoch": 0.09,
"learning_rate": 4.411764705882353e-06,
"logits/chosen": -2.095534324645996,
"logits/rejected": -1.9636192321777344,
"logps/chosen": -106.81976318359375,
"logps/rejected": -83.68408966064453,
"loss": 0.6887,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": 0.02147563174366951,
"rewards/margins": 0.009103062562644482,
"rewards/rejected": 0.012372570112347603,
"step": 90
},
{
"epoch": 0.1,
"learning_rate": 4.901960784313726e-06,
"logits/chosen": -2.0606656074523926,
"logits/rejected": -1.87876296043396,
"logps/chosen": -141.498779296875,
"logps/rejected": -105.37713623046875,
"loss": 0.6811,
"rewards/accuracies": 0.75,
"rewards/chosen": 0.04738330841064453,
"rewards/margins": 0.024673232808709145,
"rewards/rejected": 0.022710075601935387,
"step": 100
},
{
"epoch": 0.11,
"learning_rate": 4.9990549169459415e-06,
"logits/chosen": -2.2754921913146973,
"logits/rejected": -2.135282516479492,
"logps/chosen": -124.19816589355469,
"logps/rejected": -98.95077514648438,
"loss": 0.6777,
"rewards/accuracies": 0.7250000238418579,
"rewards/chosen": 0.056061066687107086,
"rewards/margins": 0.03164363652467728,
"rewards/rejected": 0.02441743016242981,
"step": 110
},
{
"epoch": 0.12,
"learning_rate": 4.995216741642263e-06,
"logits/chosen": -2.280646324157715,
"logits/rejected": -2.206347942352295,
"logps/chosen": -115.8212890625,
"logps/rejected": -96.82814025878906,
"loss": 0.6805,
"rewards/accuracies": 0.6625000238418579,
"rewards/chosen": 0.07867949455976486,
"rewards/margins": 0.02664627507328987,
"rewards/rejected": 0.05203322693705559,
"step": 120
},
{
"epoch": 0.13,
"learning_rate": 4.988430936991089e-06,
"logits/chosen": -2.2835030555725098,
"logits/rejected": -2.140094041824341,
"logps/chosen": -127.94944763183594,
"logps/rejected": -101.47581481933594,
"loss": 0.6635,
"rewards/accuracies": 0.75,
"rewards/chosen": 0.10174749791622162,
"rewards/margins": 0.06262228637933731,
"rewards/rejected": 0.039125215262174606,
"step": 130
},
{
"epoch": 0.14,
"learning_rate": 4.978705519144525e-06,
"logits/chosen": -2.18971586227417,
"logits/rejected": -2.0279135704040527,
"logps/chosen": -140.58509826660156,
"logps/rejected": -104.29924011230469,
"loss": 0.664,
"rewards/accuracies": 0.75,
"rewards/chosen": 0.07688155025243759,
"rewards/margins": 0.061921559274196625,
"rewards/rejected": 0.014959996566176414,
"step": 140
},
{
"epoch": 0.15,
"learning_rate": 4.966051976854862e-06,
"logits/chosen": -2.333808183670044,
"logits/rejected": -2.2144458293914795,
"logps/chosen": -111.2592544555664,
"logps/rejected": -89.93299865722656,
"loss": 0.6526,
"rewards/accuracies": 0.762499988079071,
"rewards/chosen": 0.11892716586589813,
"rewards/margins": 0.08827908337116241,
"rewards/rejected": 0.030648082494735718,
"step": 150
},
{
"epoch": 0.16,
"learning_rate": 4.950485257902782e-06,
"logits/chosen": -2.209681749343872,
"logits/rejected": -2.1354687213897705,
"logps/chosen": -122.34986877441406,
"logps/rejected": -97.61170959472656,
"loss": 0.6622,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": 0.09915992617607117,
"rewards/margins": 0.06800667941570282,
"rewards/rejected": 0.031153246760368347,
"step": 160
},
{
"epoch": 0.17,
"learning_rate": 4.932023751439358e-06,
"logits/chosen": -2.276695489883423,
"logits/rejected": -2.118220329284668,
"logps/chosen": -131.77114868164062,
"logps/rejected": -103.4157943725586,
"loss": 0.6666,
"rewards/accuracies": 0.6499999761581421,
"rewards/chosen": 0.10167907178401947,
"rewards/margins": 0.06214705854654312,
"rewards/rejected": 0.039532024413347244,
"step": 170
},
{
"epoch": 0.18,
"learning_rate": 4.9106892662627395e-06,
"logits/chosen": -2.347627878189087,
"logits/rejected": -2.223806858062744,
"logps/chosen": -125.82316589355469,
"logps/rejected": -102.32342529296875,
"loss": 0.6605,
"rewards/accuracies": 0.7124999761581421,
"rewards/chosen": 0.13379618525505066,
"rewards/margins": 0.07378261536359787,
"rewards/rejected": 0.06001356244087219,
"step": 180
},
{
"epoch": 0.19,
"learning_rate": 4.886507005055149e-06,
"logits/chosen": -2.299999713897705,
"logits/rejected": -2.124567985534668,
"logps/chosen": -135.9125518798828,
"logps/rejected": -102.34548950195312,
"loss": 0.6471,
"rewards/accuracies": 0.762499988079071,
"rewards/chosen": 0.12664374709129333,
"rewards/margins": 0.10224989801645279,
"rewards/rejected": 0.024393849074840546,
"step": 190
},
{
"epoch": 0.2,
"learning_rate": 4.859505534610658e-06,
"logits/chosen": -2.2595176696777344,
"logits/rejected": -2.167226552963257,
"logps/chosen": -115.63578033447266,
"logps/rejected": -97.40113830566406,
"loss": 0.652,
"rewards/accuracies": 0.6625000238418579,
"rewards/chosen": 0.07887722551822662,
"rewards/margins": 0.09930779039859772,
"rewards/rejected": -0.020430563017725945,
"step": 200
},
{
"epoch": 0.2,
"eval_logits/chosen": -2.1231298446655273,
"eval_logits/rejected": -2.0235936641693115,
"eval_logps/chosen": -120.2629165649414,
"eval_logps/rejected": -98.29754638671875,
"eval_loss": 0.6594940423965454,
"eval_rewards/accuracies": 0.6415094137191772,
"eval_rewards/chosen": 0.04982735216617584,
"eval_rewards/margins": 0.08182442933320999,
"eval_rewards/rejected": -0.031997084617614746,
"eval_runtime": 417.8564,
"eval_samples_per_second": 1.0,
"eval_steps_per_second": 0.127,
"step": 200
},
{
"epoch": 0.21,
"learning_rate": 4.829716752088893e-06,
"logits/chosen": -2.202777147293091,
"logits/rejected": -2.1421408653259277,
"logps/chosen": -107.44587707519531,
"logps/rejected": -101.29158020019531,
"loss": 0.6594,
"rewards/accuracies": 0.6875,
"rewards/chosen": 0.03508143126964569,
"rewards/margins": 0.07955195009708405,
"rewards/rejected": -0.04447052255272865,
"step": 210
},
{
"epoch": 0.22,
"learning_rate": 4.797175847334535e-06,
"logits/chosen": -2.217074155807495,
"logits/rejected": -2.0960960388183594,
"logps/chosen": -130.84152221679688,
"logps/rejected": -107.61418151855469,
"loss": 0.6708,
"rewards/accuracies": 0.6499999761581421,
"rewards/chosen": 0.0030643828213214874,
"rewards/margins": 0.05633828788995743,
"rewards/rejected": -0.05327390506863594,
"step": 220
},
{
"epoch": 0.23,
"learning_rate": 4.761921261307143e-06,
"logits/chosen": -2.2311947345733643,
"logits/rejected": -2.0887582302093506,
"logps/chosen": -123.64229583740234,
"logps/rejected": -103.31034851074219,
"loss": 0.655,
"rewards/accuracies": 0.6875,
"rewards/chosen": -0.009160916320979595,
"rewards/margins": 0.09632667899131775,
"rewards/rejected": -0.10548758506774902,
"step": 230
},
{
"epoch": 0.24,
"learning_rate": 4.723994640670377e-06,
"logits/chosen": -2.2253684997558594,
"logits/rejected": -2.0454909801483154,
"logps/chosen": -137.86196899414062,
"logps/rejected": -104.11041259765625,
"loss": 0.6146,
"rewards/accuracies": 0.8125,
"rewards/chosen": -0.011105505749583244,
"rewards/margins": 0.1803070306777954,
"rewards/rejected": -0.1914125233888626,
"step": 240
},
{
"epoch": 0.25,
"learning_rate": 4.68344078859431e-06,
"logits/chosen": -2.1830849647521973,
"logits/rejected": -2.1306838989257812,
"logps/chosen": -119.04063415527344,
"logps/rejected": -113.45658111572266,
"loss": 0.6407,
"rewards/accuracies": 0.6499999761581421,
"rewards/chosen": -0.10984460264444351,
"rewards/margins": 0.1269720494747162,
"rewards/rejected": -0.2368166148662567,
"step": 250
},
{
"epoch": 0.26,
"learning_rate": 4.6403076118289006e-06,
"logits/chosen": -2.151690721511841,
"logits/rejected": -2.0012199878692627,
"logps/chosen": -135.11477661132812,
"logps/rejected": -110.16157531738281,
"loss": 0.6217,
"rewards/accuracies": 0.699999988079071,
"rewards/chosen": -0.162130668759346,
"rewards/margins": 0.17287474870681763,
"rewards/rejected": -0.33500543236732483,
"step": 260
},
{
"epoch": 0.27,
"learning_rate": 4.5946460641111776e-06,
"logits/chosen": -2.3390426635742188,
"logits/rejected": -2.1155288219451904,
"logps/chosen": -132.8888397216797,
"logps/rejected": -110.8713607788086,
"loss": 0.5927,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.19292452931404114,
"rewards/margins": 0.24748054146766663,
"rewards/rejected": -0.44040507078170776,
"step": 270
},
{
"epoch": 0.28,
"learning_rate": 4.546510085972983e-06,
"logits/chosen": -2.3495311737060547,
"logits/rejected": -2.2263104915618896,
"logps/chosen": -143.91481018066406,
"logps/rejected": -119.75496673583984,
"loss": 0.6171,
"rewards/accuracies": 0.6625000238418579,
"rewards/chosen": -0.20113949477672577,
"rewards/margins": 0.19329218566417694,
"rewards/rejected": -0.3944316804409027,
"step": 280
},
{
"epoch": 0.29,
"learning_rate": 4.495956541020376e-06,
"logits/chosen": -2.3487744331359863,
"logits/rejected": -2.1971921920776367,
"logps/chosen": -150.26535034179688,
"logps/rejected": -133.6148681640625,
"loss": 0.6038,
"rewards/accuracies": 0.6625000238418579,
"rewards/chosen": -0.36308565735816956,
"rewards/margins": 0.2375185787677765,
"rewards/rejected": -0.6006041765213013,
"step": 290
},
{
"epoch": 0.3,
"learning_rate": 4.443045148759978e-06,
"logits/chosen": -2.3486955165863037,
"logits/rejected": -2.1962506771087646,
"logps/chosen": -159.32064819335938,
"logps/rejected": -126.43644714355469,
"loss": 0.5853,
"rewards/accuracies": 0.7250000238418579,
"rewards/chosen": -0.44245219230651855,
"rewards/margins": 0.2853606343269348,
"rewards/rejected": -0.7278127670288086,
"step": 300
},
{
"epoch": 0.3,
"learning_rate": 4.3878384140516025e-06,
"logits/chosen": -2.3736767768859863,
"logits/rejected": -2.249598741531372,
"logps/chosen": -146.68006896972656,
"logps/rejected": -135.25997924804688,
"loss": 0.5898,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": -0.41068369150161743,
"rewards/margins": 0.26723065972328186,
"rewards/rejected": -0.6779143214225769,
"step": 310
},
{
"epoch": 0.31,
"learning_rate": 4.330401553270522e-06,
"logits/chosen": -2.298905372619629,
"logits/rejected": -2.1949081420898438,
"logps/chosen": -146.1105194091797,
"logps/rejected": -136.11727905273438,
"loss": 0.6003,
"rewards/accuracies": 0.7124999761581421,
"rewards/chosen": -0.551005482673645,
"rewards/margins": 0.2657201588153839,
"rewards/rejected": -0.8167255520820618,
"step": 320
},
{
"epoch": 0.32,
"learning_rate": 4.2708024172665795e-06,
"logits/chosen": -2.402360439300537,
"logits/rejected": -2.2175159454345703,
"logps/chosen": -150.5807342529297,
"logps/rejected": -125.05631256103516,
"loss": 0.5465,
"rewards/accuracies": 0.824999988079071,
"rewards/chosen": -0.7037911415100098,
"rewards/margins": 0.3860488533973694,
"rewards/rejected": -1.089840054512024,
"step": 330
},
{
"epoch": 0.33,
"learning_rate": 4.209111411211174e-06,
"logits/chosen": -2.413839340209961,
"logits/rejected": -2.253542423248291,
"logps/chosen": -142.4720001220703,
"logps/rejected": -127.50254821777344,
"loss": 0.5395,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.7327712178230286,
"rewards/margins": 0.4077509045600891,
"rewards/rejected": -1.1405221223831177,
"step": 340
},
{
"epoch": 0.34,
"learning_rate": 4.145401411426788e-06,
"logits/chosen": -2.4574408531188965,
"logits/rejected": -2.3272013664245605,
"logps/chosen": -143.1080322265625,
"logps/rejected": -130.01637268066406,
"loss": 0.5795,
"rewards/accuracies": 0.7250000238418579,
"rewards/chosen": -0.7288345098495483,
"rewards/margins": 0.36709967255592346,
"rewards/rejected": -1.0959341526031494,
"step": 350
},
{
"epoch": 0.35,
"learning_rate": 4.079747679297314e-06,
"logits/chosen": -2.295055627822876,
"logits/rejected": -2.179664134979248,
"logps/chosen": -157.7494354248047,
"logps/rejected": -137.73941040039062,
"loss": 0.5973,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": -0.8279803395271301,
"rewards/margins": 0.3018009066581726,
"rewards/rejected": -1.1297812461853027,
"step": 360
},
{
"epoch": 0.36,
"learning_rate": 4.012227772360889e-06,
"logits/chosen": -2.2948107719421387,
"logits/rejected": -2.1263232231140137,
"logps/chosen": -167.21339416503906,
"logps/rejected": -150.82669067382812,
"loss": 0.5349,
"rewards/accuracies": 0.7875000238418579,
"rewards/chosen": -0.9286147356033325,
"rewards/margins": 0.44435009360313416,
"rewards/rejected": -1.3729647397994995,
"step": 370
},
{
"epoch": 0.37,
"learning_rate": 3.942921452690245e-06,
"logits/chosen": -2.3513195514678955,
"logits/rejected": -2.25927734375,
"logps/chosen": -173.61428833007812,
"logps/rejected": -163.9345703125,
"loss": 0.6164,
"rewards/accuracies": 0.7124999761581421,
"rewards/chosen": -1.3186113834381104,
"rewards/margins": 0.24015231430530548,
"rewards/rejected": -1.558763861656189,
"step": 380
},
{
"epoch": 0.38,
"learning_rate": 3.871910592668817e-06,
"logits/chosen": -2.433640480041504,
"logits/rejected": -2.300931215286255,
"logps/chosen": -170.8921356201172,
"logps/rejected": -165.83462524414062,
"loss": 0.503,
"rewards/accuracies": 0.8125,
"rewards/chosen": -1.420196294784546,
"rewards/margins": 0.5389910340309143,
"rewards/rejected": -1.9591872692108154,
"step": 390
},
{
"epoch": 0.39,
"learning_rate": 3.799279078273921e-06,
"logits/chosen": -2.3466134071350098,
"logits/rejected": -2.157196044921875,
"logps/chosen": -177.853271484375,
"logps/rejected": -158.73776245117188,
"loss": 0.4905,
"rewards/accuracies": 0.762499988079071,
"rewards/chosen": -1.5086350440979004,
"rewards/margins": 0.6321147680282593,
"rewards/rejected": -2.140749454498291,
"step": 400
},
{
"epoch": 0.39,
"eval_logits/chosen": -2.195096015930176,
"eval_logits/rejected": -2.0950398445129395,
"eval_logps/chosen": -177.19459533691406,
"eval_logps/rejected": -168.9884490966797,
"eval_loss": 0.5550708174705505,
"eval_rewards/accuracies": 0.6957547068595886,
"eval_rewards/chosen": -1.6581227779388428,
"eval_rewards/margins": 0.4946018159389496,
"eval_rewards/rejected": -2.152724504470825,
"eval_runtime": 417.866,
"eval_samples_per_second": 1.0,
"eval_steps_per_second": 0.127,
"step": 400
},
{
"epoch": 0.4,
"learning_rate": 3.725112709981249e-06,
"logits/chosen": -2.20538592338562,
"logits/rejected": -2.059528112411499,
"logps/chosen": -192.91726684570312,
"logps/rejected": -185.90341186523438,
"loss": 0.5449,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": -1.8029924631118774,
"rewards/margins": 0.5775827765464783,
"rewards/rejected": -2.38057541847229,
"step": 410
},
{
"epoch": 0.41,
"learning_rate": 3.649499101407737e-06,
"logits/chosen": -2.363370895385742,
"logits/rejected": -2.2048511505126953,
"logps/chosen": -224.8810577392578,
"logps/rejected": -214.8511505126953,
"loss": 0.5825,
"rewards/accuracies": 0.6875,
"rewards/chosen": -2.7866973876953125,
"rewards/margins": 0.5168424844741821,
"rewards/rejected": -3.303539991378784,
"step": 420
},
{
"epoch": 0.42,
"learning_rate": 3.5725275758125564e-06,
"logits/chosen": -2.27677059173584,
"logits/rejected": -2.1246485710144043,
"logps/chosen": -200.77066040039062,
"logps/rejected": -190.68533325195312,
"loss": 0.4805,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -2.6391139030456543,
"rewards/margins": 0.7239618897438049,
"rewards/rejected": -3.3630757331848145,
"step": 430
},
{
"epoch": 0.43,
"learning_rate": 3.494289060578478e-06,
"logits/chosen": -2.3822944164276123,
"logits/rejected": -2.2915594577789307,
"logps/chosen": -169.66128540039062,
"logps/rejected": -174.19607543945312,
"loss": 0.4812,
"rewards/accuracies": 0.699999988079071,
"rewards/chosen": -1.6458194255828857,
"rewards/margins": 0.7625138163566589,
"rewards/rejected": -2.4083335399627686,
"step": 440
},
{
"epoch": 0.44,
"learning_rate": 3.414875979798272e-06,
"logits/chosen": -2.327730178833008,
"logits/rejected": -2.2062036991119385,
"logps/chosen": -194.1064453125,
"logps/rejected": -193.63107299804688,
"loss": 0.4394,
"rewards/accuracies": 0.8374999761581421,
"rewards/chosen": -1.4300081729888916,
"rewards/margins": 0.892192006111145,
"rewards/rejected": -2.322200059890747,
"step": 450
},
{
"epoch": 0.45,
"learning_rate": 3.3343821450930196e-06,
"logits/chosen": -2.454336166381836,
"logits/rejected": -2.2937960624694824,
"logps/chosen": -201.33657836914062,
"logps/rejected": -188.39266967773438,
"loss": 0.4758,
"rewards/accuracies": 0.75,
"rewards/chosen": -2.0642800331115723,
"rewards/margins": 0.8783187866210938,
"rewards/rejected": -2.942598819732666,
"step": 460
},
{
"epoch": 0.46,
"learning_rate": 3.252902644791325e-06,
"logits/chosen": -2.188424587249756,
"logits/rejected": -2.041105031967163,
"logps/chosen": -214.7936248779297,
"logps/rejected": -222.3824462890625,
"loss": 0.4536,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -2.7383971214294434,
"rewards/margins": 0.9385073781013489,
"rewards/rejected": -3.6769042015075684,
"step": 470
},
{
"epoch": 0.47,
"learning_rate": 3.170533731600339e-06,
"logits/chosen": -2.3067922592163086,
"logits/rejected": -2.1540229320526123,
"logps/chosen": -204.9595184326172,
"logps/rejected": -204.6565704345703,
"loss": 0.4887,
"rewards/accuracies": 0.75,
"rewards/chosen": -2.3199477195739746,
"rewards/margins": 0.8953973650932312,
"rewards/rejected": -3.2153449058532715,
"step": 480
},
{
"epoch": 0.48,
"learning_rate": 3.0873727089012816e-06,
"logits/chosen": -2.4507811069488525,
"logits/rejected": -2.357919931411743,
"logps/chosen": -212.8636932373047,
"logps/rejected": -208.5207977294922,
"loss": 0.5393,
"rewards/accuracies": 0.75,
"rewards/chosen": -2.4198460578918457,
"rewards/margins": 0.7457529306411743,
"rewards/rejected": -3.1655986309051514,
"step": 490
},
{
"epoch": 0.49,
"learning_rate": 3.0035178158038026e-06,
"logits/chosen": -2.2298638820648193,
"logits/rejected": -2.035947322845459,
"logps/chosen": -217.26846313476562,
"logps/rejected": -209.74526977539062,
"loss": 0.4886,
"rewards/accuracies": 0.7250000238418579,
"rewards/chosen": -2.4076380729675293,
"rewards/margins": 0.9488040208816528,
"rewards/rejected": -3.3564422130584717,
"step": 500
},
{
"epoch": 0.5,
"learning_rate": 2.919068111094937e-06,
"logits/chosen": -2.3187146186828613,
"logits/rejected": -2.193861484527588,
"logps/chosen": -185.0901641845703,
"logps/rejected": -195.90634155273438,
"loss": 0.552,
"rewards/accuracies": 0.6875,
"rewards/chosen": -2.132998466491699,
"rewards/margins": 0.827830970287323,
"rewards/rejected": -2.960829257965088,
"step": 510
},
{
"epoch": 0.51,
"learning_rate": 2.8341233562197895e-06,
"logits/chosen": -2.3116376399993896,
"logits/rejected": -2.246950626373291,
"logps/chosen": -172.0282440185547,
"logps/rejected": -177.47348022460938,
"loss": 0.5152,
"rewards/accuracies": 0.762499988079071,
"rewards/chosen": -1.7447055578231812,
"rewards/margins": 0.754051685333252,
"rewards/rejected": -2.4987568855285645,
"step": 520
},
{
"epoch": 0.52,
"learning_rate": 2.7487838974321352e-06,
"logits/chosen": -2.2577805519104004,
"logits/rejected": -2.143658399581909,
"logps/chosen": -178.53317260742188,
"logps/rejected": -182.5507049560547,
"loss": 0.4719,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -1.789046287536621,
"rewards/margins": 0.8771921396255493,
"rewards/rejected": -2.666238307952881,
"step": 530
},
{
"epoch": 0.53,
"learning_rate": 2.6631505472541997e-06,
"logits/chosen": -2.2621216773986816,
"logits/rejected": -2.1167335510253906,
"logps/chosen": -193.0259246826172,
"logps/rejected": -198.51705932617188,
"loss": 0.4442,
"rewards/accuracies": 0.762499988079071,
"rewards/chosen": -1.8945804834365845,
"rewards/margins": 1.0066120624542236,
"rewards/rejected": -2.9011926651000977,
"step": 540
},
{
"epoch": 0.54,
"learning_rate": 2.5773244653856173e-06,
"logits/chosen": -2.2354609966278076,
"logits/rejected": -2.1045310497283936,
"logps/chosen": -205.27163696289062,
"logps/rejected": -207.8455810546875,
"loss": 0.4791,
"rewards/accuracies": 0.7250000238418579,
"rewards/chosen": -2.161865711212158,
"rewards/margins": 0.9317368268966675,
"rewards/rejected": -3.0936026573181152,
"step": 550
},
{
"epoch": 0.55,
"learning_rate": 2.4914070392022717e-06,
"logits/chosen": -2.274534225463867,
"logits/rejected": -2.158811569213867,
"logps/chosen": -208.08349609375,
"logps/rejected": -211.2762451171875,
"loss": 0.4978,
"rewards/accuracies": 0.75,
"rewards/chosen": -2.0876078605651855,
"rewards/margins": 0.8888559341430664,
"rewards/rejected": -2.976463794708252,
"step": 560
},
{
"epoch": 0.56,
"learning_rate": 2.4054997639861778e-06,
"logits/chosen": -2.1874241828918457,
"logits/rejected": -2.0126781463623047,
"logps/chosen": -207.1369171142578,
"logps/rejected": -207.59768676757812,
"loss": 0.4732,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -2.0929489135742188,
"rewards/margins": 1.14609694480896,
"rewards/rejected": -3.2390456199645996,
"step": 570
},
{
"epoch": 0.57,
"learning_rate": 2.3197041230278905e-06,
"logits/chosen": -2.3066487312316895,
"logits/rejected": -2.18741512298584,
"logps/chosen": -204.58642578125,
"logps/rejected": -222.44973754882812,
"loss": 0.427,
"rewards/accuracies": 0.8125,
"rewards/chosen": -2.2493700981140137,
"rewards/margins": 1.2541824579238892,
"rewards/rejected": -3.5035526752471924,
"step": 580
},
{
"epoch": 0.58,
"learning_rate": 2.234121467743082e-06,
"logits/chosen": -2.3349661827087402,
"logits/rejected": -2.252894163131714,
"logps/chosen": -208.3444061279297,
"logps/rejected": -211.37295532226562,
"loss": 0.5968,
"rewards/accuracies": 0.7250000238418579,
"rewards/chosen": -2.635042905807495,
"rewards/margins": 0.8392454981803894,
"rewards/rejected": -3.4742884635925293,
"step": 590
},
{
"epoch": 0.59,
"learning_rate": 2.148852897944905e-06,
"logits/chosen": -2.3977198600769043,
"logits/rejected": -2.2428812980651855,
"logps/chosen": -208.42538452148438,
"logps/rejected": -219.8905487060547,
"loss": 0.4249,
"rewards/accuracies": 0.8374999761581421,
"rewards/chosen": -2.645089626312256,
"rewards/margins": 1.1129385232925415,
"rewards/rejected": -3.758028030395508,
"step": 600
},
{
"epoch": 0.59,
"eval_logits/chosen": -2.17734956741333,
"eval_logits/rejected": -2.078249454498291,
"eval_logps/chosen": -237.10446166992188,
"eval_logps/rejected": -241.38665771484375,
"eval_loss": 0.5327094793319702,
"eval_rewards/accuracies": 0.724056601524353,
"eval_rewards/chosen": -3.455418348312378,
"eval_rewards/margins": 0.8692519068717957,
"eval_rewards/rejected": -4.324670314788818,
"eval_runtime": 423.7708,
"eval_samples_per_second": 0.986,
"eval_steps_per_second": 0.125,
"step": 600
},
{
"epoch": 0.6,
"learning_rate": 2.063999142413574e-06,
"logits/chosen": -2.2915334701538086,
"logits/rejected": -2.1494946479797363,
"logps/chosen": -247.0482177734375,
"logps/rejected": -239.52462768554688,
"loss": 0.489,
"rewards/accuracies": 0.762499988079071,
"rewards/chosen": -3.5384364128112793,
"rewards/margins": 0.9590060114860535,
"rewards/rejected": -4.497443199157715,
"step": 610
},
{
"epoch": 0.61,
"learning_rate": 1.9796604399042547e-06,
"logits/chosen": -2.3757712841033936,
"logits/rejected": -2.2203211784362793,
"logps/chosen": -268.9779052734375,
"logps/rejected": -274.21771240234375,
"loss": 0.4595,
"rewards/accuracies": 0.7875000238418579,
"rewards/chosen": -4.319927215576172,
"rewards/margins": 1.1538012027740479,
"rewards/rejected": -5.473728179931641,
"step": 620
},
{
"epoch": 0.62,
"learning_rate": 1.8959364207338216e-06,
"logits/chosen": -2.345416784286499,
"logits/rejected": -2.180387020111084,
"logps/chosen": -245.0520477294922,
"logps/rejected": -258.79498291015625,
"loss": 0.4835,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": -4.156452655792236,
"rewards/margins": 1.1458499431610107,
"rewards/rejected": -5.302302360534668,
"step": 630
},
{
"epoch": 0.63,
"learning_rate": 1.8129259890863825e-06,
"logits/chosen": -2.3113701343536377,
"logits/rejected": -2.200329303741455,
"logps/chosen": -258.2785949707031,
"logps/rejected": -273.8367614746094,
"loss": 0.5494,
"rewards/accuracies": 0.675000011920929,
"rewards/chosen": -3.8280506134033203,
"rewards/margins": 1.0454813241958618,
"rewards/rejected": -4.873531818389893,
"step": 640
},
{
"epoch": 0.64,
"learning_rate": 1.7307272061765738e-06,
"logits/chosen": -2.332291841506958,
"logits/rejected": -2.238374948501587,
"logps/chosen": -245.2693328857422,
"logps/rejected": -258.13922119140625,
"loss": 0.4966,
"rewards/accuracies": 0.762499988079071,
"rewards/chosen": -3.4873454570770264,
"rewards/margins": 1.1437709331512451,
"rewards/rejected": -4.6311163902282715,
"step": 650
},
{
"epoch": 0.65,
"learning_rate": 1.649437174408685e-06,
"logits/chosen": -2.2653393745422363,
"logits/rejected": -2.1328892707824707,
"logps/chosen": -225.3964385986328,
"logps/rejected": -243.22384643554688,
"loss": 0.5442,
"rewards/accuracies": 0.75,
"rewards/chosen": -2.7976624965667725,
"rewards/margins": 1.2127068042755127,
"rewards/rejected": -4.010369300842285,
"step": 660
},
{
"epoch": 0.66,
"learning_rate": 1.569151922668422e-06,
"logits/chosen": -2.367928981781006,
"logits/rejected": -2.253605365753174,
"logps/chosen": -201.37025451660156,
"logps/rejected": -218.9671630859375,
"loss": 0.5362,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": -2.52730131149292,
"rewards/margins": 1.1274640560150146,
"rewards/rejected": -3.6547648906707764,
"step": 670
},
{
"epoch": 0.67,
"learning_rate": 1.4899662928828428e-06,
"logits/chosen": -2.309407949447632,
"logits/rejected": -2.1246845722198486,
"logps/chosen": -186.25376892089844,
"logps/rejected": -205.3202667236328,
"loss": 0.4534,
"rewards/accuracies": 0.8125,
"rewards/chosen": -2.198615550994873,
"rewards/margins": 1.3432915210723877,
"rewards/rejected": -3.5419068336486816,
"step": 680
},
{
"epoch": 0.68,
"learning_rate": 1.4119738279824507e-06,
"logits/chosen": -2.233764886856079,
"logits/rejected": -2.0911850929260254,
"logps/chosen": -194.0051727294922,
"logps/rejected": -202.19174194335938,
"loss": 0.5816,
"rewards/accuracies": 0.7250000238418579,
"rewards/chosen": -2.272157907485962,
"rewards/margins": 0.9171239137649536,
"rewards/rejected": -3.189281463623047,
"step": 690
},
{
"epoch": 0.69,
"learning_rate": 1.3352666613978152e-06,
"logits/chosen": -2.1566410064697266,
"logits/rejected": -2.075209379196167,
"logps/chosen": -198.47071838378906,
"logps/rejected": -202.29379272460938,
"loss": 0.5355,
"rewards/accuracies": 0.7250000238418579,
"rewards/chosen": -2.325533151626587,
"rewards/margins": 0.8646724820137024,
"rewards/rejected": -3.1902058124542236,
"step": 700
},
{
"epoch": 0.7,
"learning_rate": 1.2599354082212523e-06,
"logits/chosen": -2.222766876220703,
"logits/rejected": -2.0928432941436768,
"logps/chosen": -190.0797119140625,
"logps/rejected": -198.69985961914062,
"loss": 0.5152,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": -2.1240830421447754,
"rewards/margins": 0.9723888635635376,
"rewards/rejected": -3.0964715480804443,
"step": 710
},
{
"epoch": 0.71,
"learning_rate": 1.186069058162127e-06,
"logits/chosen": -2.3559908866882324,
"logits/rejected": -2.249436140060425,
"logps/chosen": -197.04818725585938,
"logps/rejected": -201.38693237304688,
"loss": 0.5537,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": -2.438502788543701,
"rewards/margins": 0.8104500770568848,
"rewards/rejected": -3.248952865600586,
"step": 720
},
{
"epoch": 0.72,
"learning_rate": 1.113754870422254e-06,
"logits/chosen": -2.40106201171875,
"logits/rejected": -2.2911598682403564,
"logps/chosen": -198.04409790039062,
"logps/rejected": -214.9007568359375,
"loss": 0.5454,
"rewards/accuracies": 0.75,
"rewards/chosen": -2.3586318492889404,
"rewards/margins": 0.8204809427261353,
"rewards/rejected": -3.179112672805786,
"step": 730
},
{
"epoch": 0.73,
"learning_rate": 1.0430782706155545e-06,
"logits/chosen": -2.4164676666259766,
"logits/rejected": -2.2673213481903076,
"logps/chosen": -204.253173828125,
"logps/rejected": -209.43753051757812,
"loss": 0.45,
"rewards/accuracies": 0.7250000238418579,
"rewards/chosen": -2.2196271419525146,
"rewards/margins": 1.149065613746643,
"rewards/rejected": -3.3686928749084473,
"step": 740
},
{
"epoch": 0.74,
"learning_rate": 9.741227498537615e-07,
"logits/chosen": -2.457432270050049,
"logits/rejected": -2.321898937225342,
"logps/chosen": -194.0103302001953,
"logps/rejected": -204.46145629882812,
"loss": 0.4596,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": -2.4977972507476807,
"rewards/margins": 1.0296275615692139,
"rewards/rejected": -3.5274243354797363,
"step": 750
},
{
"epoch": 0.75,
"learning_rate": 9.069697661173668e-07,
"logits/chosen": -2.3087573051452637,
"logits/rejected": -2.210942268371582,
"logps/chosen": -205.71200561523438,
"logps/rejected": -217.53921508789062,
"loss": 0.6425,
"rewards/accuracies": 0.6875,
"rewards/chosen": -2.667379379272461,
"rewards/margins": 0.8116022348403931,
"rewards/rejected": -3.4789810180664062,
"step": 760
},
{
"epoch": 0.76,
"learning_rate": 8.416986480283434e-07,
"logits/chosen": -2.2393643856048584,
"logits/rejected": -2.096648693084717,
"logps/chosen": -205.7020721435547,
"logps/rejected": -220.5606689453125,
"loss": 0.5619,
"rewards/accuracies": 0.699999988079071,
"rewards/chosen": -2.711803913116455,
"rewards/margins": 1.0723652839660645,
"rewards/rejected": -3.7841694355010986,
"step": 770
},
{
"epoch": 0.77,
"learning_rate": 7.783865011382876e-07,
"logits/chosen": -2.2971372604370117,
"logits/rejected": -2.1558918952941895,
"logps/chosen": -213.41958618164062,
"logps/rejected": -214.00326538085938,
"loss": 0.5043,
"rewards/accuracies": 0.699999988079071,
"rewards/chosen": -2.363002300262451,
"rewards/margins": 1.088847279548645,
"rewards/rejected": -3.4518496990203857,
"step": 780
},
{
"epoch": 0.78,
"learning_rate": 7.171081168427205e-07,
"logits/chosen": -2.331343173980713,
"logits/rejected": -2.165480613708496,
"logps/chosen": -221.4240264892578,
"logps/rejected": -226.91650390625,
"loss": 0.473,
"rewards/accuracies": 0.762499988079071,
"rewards/chosen": -2.2688238620758057,
"rewards/margins": 1.2109057903289795,
"rewards/rejected": -3.479729413986206,
"step": 790
},
{
"epoch": 0.79,
"learning_rate": 6.579358840291064e-07,
"logits/chosen": -2.2774598598480225,
"logits/rejected": -2.1744918823242188,
"logps/chosen": -209.7314453125,
"logps/rejected": -224.2617645263672,
"loss": 0.5858,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": -2.5639281272888184,
"rewards/margins": 0.7768001556396484,
"rewards/rejected": -3.340728282928467,
"step": 800
},
{
"epoch": 0.79,
"eval_logits/chosen": -2.159109354019165,
"eval_logits/rejected": -2.058640241622925,
"eval_logps/chosen": -205.4982452392578,
"eval_logps/rejected": -212.2717742919922,
"eval_loss": 0.5206710696220398,
"eval_rewards/accuracies": 0.7334905862808228,
"eval_rewards/chosen": -2.507232666015625,
"eval_rewards/margins": 0.9439911842346191,
"eval_rewards/rejected": -3.4512243270874023,
"eval_runtime": 423.3516,
"eval_samples_per_second": 0.987,
"eval_steps_per_second": 0.125,
"step": 800
},
{
"epoch": 0.8,
"learning_rate": 6.00939703563006e-07,
"logits/chosen": -2.3462395668029785,
"logits/rejected": -2.2609333992004395,
"logps/chosen": -191.94070434570312,
"logps/rejected": -208.1775665283203,
"loss": 0.5793,
"rewards/accuracies": 0.699999988079071,
"rewards/chosen": -2.5009541511535645,
"rewards/margins": 0.8846192359924316,
"rewards/rejected": -3.385573625564575,
"step": 810
},
{
"epoch": 0.81,
"learning_rate": 5.461869057133412e-07,
"logits/chosen": -2.386518955230713,
"logits/rejected": -2.2898330688476562,
"logps/chosen": -198.1256866455078,
"logps/rejected": -216.5961456298828,
"loss": 0.4895,
"rewards/accuracies": 0.699999988079071,
"rewards/chosen": -2.2799556255340576,
"rewards/margins": 1.0353368520736694,
"rewards/rejected": -3.3152928352355957,
"step": 820
},
{
"epoch": 0.82,
"learning_rate": 4.937421706143497e-07,
"logits/chosen": -2.2601094245910645,
"logits/rejected": -2.1199710369110107,
"logps/chosen": -208.54898071289062,
"logps/rejected": -210.0823974609375,
"loss": 0.5261,
"rewards/accuracies": 0.75,
"rewards/chosen": -2.4491305351257324,
"rewards/margins": 0.9635556936264038,
"rewards/rejected": -3.412686586380005,
"step": 830
},
{
"epoch": 0.83,
"learning_rate": 4.43667451858166e-07,
"logits/chosen": -2.262453079223633,
"logits/rejected": -2.0967936515808105,
"logps/chosen": -188.5206298828125,
"logps/rejected": -202.87240600585938,
"loss": 0.4361,
"rewards/accuracies": 0.7875000238418579,
"rewards/chosen": -2.2963976860046387,
"rewards/margins": 1.180870771408081,
"rewards/rejected": -3.4772682189941406,
"step": 840
},
{
"epoch": 0.84,
"learning_rate": 3.9602190330830484e-07,
"logits/chosen": -2.2342655658721924,
"logits/rejected": -2.130017042160034,
"logps/chosen": -200.30343627929688,
"logps/rejected": -220.6678466796875,
"loss": 0.4885,
"rewards/accuracies": 0.7124999761581421,
"rewards/chosen": -2.533165454864502,
"rewards/margins": 1.0824253559112549,
"rewards/rejected": -3.6155905723571777,
"step": 850
},
{
"epoch": 0.85,
"learning_rate": 3.5086180922049295e-07,
"logits/chosen": -2.3764655590057373,
"logits/rejected": -2.235109329223633,
"logps/chosen": -224.97146606445312,
"logps/rejected": -221.2453155517578,
"loss": 0.4675,
"rewards/accuracies": 0.762499988079071,
"rewards/chosen": -2.4995951652526855,
"rewards/margins": 1.0894619226455688,
"rewards/rejected": -3.589057445526123,
"step": 860
},
{
"epoch": 0.86,
"learning_rate": 3.0824051775340895e-07,
"logits/chosen": -2.380509853363037,
"logits/rejected": -2.275191068649292,
"logps/chosen": -180.95895385742188,
"logps/rejected": -195.82479858398438,
"loss": 0.547,
"rewards/accuracies": 0.6875,
"rewards/chosen": -2.3850598335266113,
"rewards/margins": 0.8363308906555176,
"rewards/rejected": -3.22139048576355,
"step": 870
},
{
"epoch": 0.87,
"learning_rate": 2.6820837794786336e-07,
"logits/chosen": -2.2321319580078125,
"logits/rejected": -2.1590046882629395,
"logps/chosen": -204.28173828125,
"logps/rejected": -214.01327514648438,
"loss": 0.6701,
"rewards/accuracies": 0.5249999761581421,
"rewards/chosen": -2.5836868286132812,
"rewards/margins": 0.4722086787223816,
"rewards/rejected": -3.0558953285217285,
"step": 880
},
{
"epoch": 0.88,
"learning_rate": 2.3081268024887694e-07,
"logits/chosen": -2.222381114959717,
"logits/rejected": -2.051706075668335,
"logps/chosen": -199.63970947265625,
"logps/rejected": -208.906005859375,
"loss": 0.4139,
"rewards/accuracies": 0.8125,
"rewards/chosen": -2.3457841873168945,
"rewards/margins": 1.2535767555236816,
"rewards/rejected": -3.599360942840576,
"step": 890
},
{
"epoch": 0.89,
"learning_rate": 1.9609760064091044e-07,
"logits/chosen": -2.3242409229278564,
"logits/rejected": -2.2560477256774902,
"logps/chosen": -205.27554321289062,
"logps/rejected": -201.7273712158203,
"loss": 0.5173,
"rewards/accuracies": 0.7124999761581421,
"rewards/chosen": -2.428300142288208,
"rewards/margins": 0.8561038970947266,
"rewards/rejected": -3.2844040393829346,
"step": 900
},
{
"epoch": 0.9,
"learning_rate": 1.6410414846224992e-07,
"logits/chosen": -2.220360517501831,
"logits/rejected": -2.109575033187866,
"logps/chosen": -201.75045776367188,
"logps/rejected": -214.3638458251953,
"loss": 0.4565,
"rewards/accuracies": 0.7875000238418579,
"rewards/chosen": -2.569467067718506,
"rewards/margins": 1.1193673610687256,
"rewards/rejected": -3.6888339519500732,
"step": 910
},
{
"epoch": 0.91,
"learning_rate": 1.348701179601819e-07,
"logits/chosen": -2.401984691619873,
"logits/rejected": -2.2650654315948486,
"logps/chosen": -215.353515625,
"logps/rejected": -225.1014404296875,
"loss": 0.4529,
"rewards/accuracies": 0.7875000238418579,
"rewards/chosen": -2.38716983795166,
"rewards/margins": 1.1682064533233643,
"rewards/rejected": -3.5553765296936035,
"step": 920
},
{
"epoch": 0.91,
"learning_rate": 1.0843004364420151e-07,
"logits/chosen": -2.2123489379882812,
"logits/rejected": -2.100048065185547,
"logps/chosen": -215.25479125976562,
"logps/rejected": -229.445556640625,
"loss": 0.5697,
"rewards/accuracies": 0.6875,
"rewards/chosen": -2.393723964691162,
"rewards/margins": 0.8360812067985535,
"rewards/rejected": -3.2298049926757812,
"step": 930
},
{
"epoch": 0.92,
"learning_rate": 8.481515948997931e-08,
"logits/chosen": -2.3680367469787598,
"logits/rejected": -2.26953387260437,
"logps/chosen": -216.99514770507812,
"logps/rejected": -212.17263793945312,
"loss": 0.6119,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": -2.478694200515747,
"rewards/margins": 0.7769169807434082,
"rewards/rejected": -3.255610942840576,
"step": 940
},
{
"epoch": 0.93,
"learning_rate": 6.4053362042297e-08,
"logits/chosen": -2.2625975608825684,
"logits/rejected": -2.104025363922119,
"logps/chosen": -205.9175262451172,
"logps/rejected": -217.52304077148438,
"loss": 0.4583,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": -2.4613893032073975,
"rewards/margins": 1.17227303981781,
"rewards/rejected": -3.633662462234497,
"step": 950
},
{
"epoch": 0.94,
"learning_rate": 4.616917746052163e-08,
"logits/chosen": -2.3516554832458496,
"logits/rejected": -2.214130401611328,
"logps/chosen": -203.63438415527344,
"logps/rejected": -211.27334594726562,
"loss": 0.546,
"rewards/accuracies": 0.75,
"rewards/chosen": -2.5478570461273193,
"rewards/margins": 0.990521252155304,
"rewards/rejected": -3.5383784770965576,
"step": 960
},
{
"epoch": 0.95,
"learning_rate": 3.118373254556412e-08,
"logits/chosen": -2.3878164291381836,
"logits/rejected": -2.2502613067626953,
"logps/chosen": -198.12130737304688,
"logps/rejected": -199.36111450195312,
"loss": 0.4535,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": -2.448812484741211,
"rewards/margins": 1.0348026752471924,
"rewards/rejected": -3.483614683151245,
"step": 970
},
{
"epoch": 0.96,
"learning_rate": 1.9114729782535037e-08,
"logits/chosen": -2.4108872413635254,
"logits/rejected": -2.297142505645752,
"logps/chosen": -194.57632446289062,
"logps/rejected": -202.7489776611328,
"loss": 0.4981,
"rewards/accuracies": 0.699999988079071,
"rewards/chosen": -2.350036859512329,
"rewards/margins": 0.9002410173416138,
"rewards/rejected": -3.2502777576446533,
"step": 980
},
{
"epoch": 0.97,
"learning_rate": 9.97642642858815e-09,
"logits/chosen": -2.28024959564209,
"logits/rejected": -2.139801502227783,
"logps/chosen": -211.3317413330078,
"logps/rejected": -211.1747283935547,
"loss": 0.5021,
"rewards/accuracies": 0.6875,
"rewards/chosen": -2.55255126953125,
"rewards/margins": 0.9248201251029968,
"rewards/rejected": -3.4773712158203125,
"step": 990
},
{
"epoch": 0.98,
"learning_rate": 3.779617670651436e-09,
"logits/chosen": -2.248671054840088,
"logits/rejected": -2.1480166912078857,
"logps/chosen": -217.71707153320312,
"logps/rejected": -225.20040893554688,
"loss": 0.6128,
"rewards/accuracies": 0.699999988079071,
"rewards/chosen": -2.73987078666687,
"rewards/margins": 0.9156008958816528,
"rewards/rejected": -3.6554713249206543,
"step": 1000
},
{
"epoch": 0.98,
"eval_logits/chosen": -2.1620335578918457,
"eval_logits/rejected": -2.062356948852539,
"eval_logps/chosen": -206.584716796875,
"eval_logps/rejected": -214.93492126464844,
"eval_loss": 0.521207869052887,
"eval_rewards/accuracies": 0.7405660152435303,
"eval_rewards/chosen": -2.5398268699645996,
"eval_rewards/margins": 0.9912916421890259,
"eval_rewards/rejected": -3.531118631362915,
"eval_runtime": 423.4333,
"eval_samples_per_second": 0.987,
"eval_steps_per_second": 0.125,
"step": 1000
},
{
"epoch": 0.99,
"learning_rate": 5.316238729444201e-10,
"logits/chosen": -2.234221935272217,
"logits/rejected": -2.0627448558807373,
"logps/chosen": -209.44503784179688,
"logps/rejected": -219.24606323242188,
"loss": 0.4684,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": -2.601386547088623,
"rewards/margins": 1.190915584564209,
"rewards/rejected": -3.792301893234253,
"step": 1010
},
{
"epoch": 1.0,
"step": 1016,
"total_flos": 0.0,
"train_loss": 0.5583291621658746,
"train_runtime": 16785.8838,
"train_samples_per_second": 0.484,
"train_steps_per_second": 0.061
}
],
"logging_steps": 10,
"max_steps": 1016,
"num_input_tokens_seen": 0,
"num_train_epochs": 1,
"save_steps": 100,
"total_flos": 0.0,
"train_batch_size": 4,
"trial_name": null,
"trial_params": null
}