Delirium-v1 / trainer_state.json
sam-paech's picture
Upload folder using huggingface_hub
5dd4f27 verified
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.0,
"eval_steps": 400,
"global_step": 577,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.008665511265164644,
"grad_norm": 144.30636115944256,
"learning_rate": 8.620689655172414e-09,
"logps/chosen": -2.4541897773742676,
"logps/rejected": -0.7712920904159546,
"loss": 19.0474,
"rewards/accuracies": 0.0,
"rewards/chosen": -24.541898727416992,
"rewards/margins": -16.828975677490234,
"rewards/rejected": -7.712920188903809,
"step": 5
},
{
"epoch": 0.01733102253032929,
"grad_norm": 161.5257295296766,
"learning_rate": 1.724137931034483e-08,
"logps/chosen": -1.8905527591705322,
"logps/rejected": -0.8088749647140503,
"loss": 17.1265,
"rewards/accuracies": 0.4000000059604645,
"rewards/chosen": -18.90552520751953,
"rewards/margins": -10.81677532196045,
"rewards/rejected": -8.088749885559082,
"step": 10
},
{
"epoch": 0.025996533795493933,
"grad_norm": 137.887178762291,
"learning_rate": 2.586206896551724e-08,
"logps/chosen": -2.4456095695495605,
"logps/rejected": -0.7901965379714966,
"loss": 18.428,
"rewards/accuracies": 0.20000000298023224,
"rewards/chosen": -24.456096649169922,
"rewards/margins": -16.55413246154785,
"rewards/rejected": -7.901966094970703,
"step": 15
},
{
"epoch": 0.03466204506065858,
"grad_norm": 126.89326071068598,
"learning_rate": 3.448275862068966e-08,
"logps/chosen": -2.5534451007843018,
"logps/rejected": -1.1100285053253174,
"loss": 14.051,
"rewards/accuracies": 0.0,
"rewards/chosen": -25.53445053100586,
"rewards/margins": -14.434164047241211,
"rewards/rejected": -11.100285530090332,
"step": 20
},
{
"epoch": 0.043327556325823226,
"grad_norm": 192.70418536899953,
"learning_rate": 4.3103448275862064e-08,
"logps/chosen": -1.6850401163101196,
"logps/rejected": -0.7781764268875122,
"loss": 14.7994,
"rewards/accuracies": 0.20000000298023224,
"rewards/chosen": -16.850399017333984,
"rewards/margins": -9.068635940551758,
"rewards/rejected": -7.781764030456543,
"step": 25
},
{
"epoch": 0.05199306759098787,
"grad_norm": 150.3771160776878,
"learning_rate": 5.172413793103448e-08,
"logps/chosen": -2.7311158180236816,
"logps/rejected": -1.1124465465545654,
"loss": 17.0918,
"rewards/accuracies": 0.0,
"rewards/chosen": -27.3111572265625,
"rewards/margins": -16.186691284179688,
"rewards/rejected": -11.12446403503418,
"step": 30
},
{
"epoch": 0.060658578856152515,
"grad_norm": 119.59215071363475,
"learning_rate": 6.03448275862069e-08,
"logps/chosen": -2.034471273422241,
"logps/rejected": -0.7978214025497437,
"loss": 17.5941,
"rewards/accuracies": 0.4000000059604645,
"rewards/chosen": -20.344711303710938,
"rewards/margins": -12.366498947143555,
"rewards/rejected": -7.978213310241699,
"step": 35
},
{
"epoch": 0.06932409012131716,
"grad_norm": 156.50159769490176,
"learning_rate": 6.896551724137931e-08,
"logps/chosen": -2.3780651092529297,
"logps/rejected": -1.1195534467697144,
"loss": 16.6879,
"rewards/accuracies": 0.20000000298023224,
"rewards/chosen": -23.780651092529297,
"rewards/margins": -12.585118293762207,
"rewards/rejected": -11.195535659790039,
"step": 40
},
{
"epoch": 0.0779896013864818,
"grad_norm": 120.36498362349127,
"learning_rate": 7.758620689655172e-08,
"logps/chosen": -2.3653724193573,
"logps/rejected": -0.8212674260139465,
"loss": 16.0315,
"rewards/accuracies": 0.0,
"rewards/chosen": -23.65372085571289,
"rewards/margins": -15.441045761108398,
"rewards/rejected": -8.212674140930176,
"step": 45
},
{
"epoch": 0.08665511265164645,
"grad_norm": 140.48384963924636,
"learning_rate": 8.620689655172413e-08,
"logps/chosen": -2.0239596366882324,
"logps/rejected": -0.8580087423324585,
"loss": 17.0992,
"rewards/accuracies": 0.20000000298023224,
"rewards/chosen": -20.23959732055664,
"rewards/margins": -11.65950870513916,
"rewards/rejected": -8.580087661743164,
"step": 50
},
{
"epoch": 0.09532062391681109,
"grad_norm": 128.56326610568328,
"learning_rate": 9.482758620689655e-08,
"logps/chosen": -2.002194881439209,
"logps/rejected": -0.8230875134468079,
"loss": 15.9385,
"rewards/accuracies": 0.4000000059604645,
"rewards/chosen": -20.021947860717773,
"rewards/margins": -11.791071891784668,
"rewards/rejected": -8.230875015258789,
"step": 55
},
{
"epoch": 0.10398613518197573,
"grad_norm": 120.25035045719453,
"learning_rate": 9.999633596405632e-08,
"logps/chosen": -2.8458476066589355,
"logps/rejected": -0.928970992565155,
"loss": 17.6529,
"rewards/accuracies": 0.0,
"rewards/chosen": -28.458477020263672,
"rewards/margins": -19.168766021728516,
"rewards/rejected": -9.289709091186523,
"step": 60
},
{
"epoch": 0.11265164644714037,
"grad_norm": 110.05644440005858,
"learning_rate": 9.995512172662394e-08,
"logps/chosen": -2.2458930015563965,
"logps/rejected": -0.8278621435165405,
"loss": 15.5726,
"rewards/accuracies": 0.20000000298023224,
"rewards/chosen": -22.458932876586914,
"rewards/margins": -14.180310249328613,
"rewards/rejected": -8.278621673583984,
"step": 65
},
{
"epoch": 0.12131715771230503,
"grad_norm": 121.60204255752639,
"learning_rate": 9.986815108288271e-08,
"logps/chosen": -1.1772969961166382,
"logps/rejected": -0.8211766481399536,
"loss": 14.691,
"rewards/accuracies": 0.4000000059604645,
"rewards/chosen": -11.772969245910645,
"rewards/margins": -3.561204195022583,
"rewards/rejected": -8.211766242980957,
"step": 70
},
{
"epoch": 0.12998266897746968,
"grad_norm": 101.42661844850082,
"learning_rate": 9.973550369361562e-08,
"logps/chosen": -1.9986432790756226,
"logps/rejected": -1.0442060232162476,
"loss": 15.0518,
"rewards/accuracies": 0.4000000059604645,
"rewards/chosen": -19.986433029174805,
"rewards/margins": -9.54437255859375,
"rewards/rejected": -10.442060470581055,
"step": 75
},
{
"epoch": 0.1386481802426343,
"grad_norm": 152.87624500994318,
"learning_rate": 9.955730105723222e-08,
"logps/chosen": -1.9634087085723877,
"logps/rejected": -0.7768585085868835,
"loss": 17.3279,
"rewards/accuracies": 0.4000000059604645,
"rewards/chosen": -19.63408851623535,
"rewards/margins": -11.865501403808594,
"rewards/rejected": -7.768586158752441,
"step": 80
},
{
"epoch": 0.14731369150779897,
"grad_norm": 103.93615787493052,
"learning_rate": 9.93337063984821e-08,
"logps/chosen": -2.262453079223633,
"logps/rejected": -0.8849401473999023,
"loss": 15.6727,
"rewards/accuracies": 0.20000000298023224,
"rewards/chosen": -22.624530792236328,
"rewards/margins": -13.775128364562988,
"rewards/rejected": -8.849401473999023,
"step": 85
},
{
"epoch": 0.1559792027729636,
"grad_norm": 106.77837368888825,
"learning_rate": 9.906492451894921e-08,
"logps/chosen": -1.9127286672592163,
"logps/rejected": -1.1486315727233887,
"loss": 14.4561,
"rewards/accuracies": 0.4000000059604645,
"rewards/chosen": -19.127286911010742,
"rewards/margins": -7.640970706939697,
"rewards/rejected": -11.48631477355957,
"step": 90
},
{
"epoch": 0.16464471403812825,
"grad_norm": 115.47717214236427,
"learning_rate": 9.875120160946371e-08,
"logps/chosen": -1.957664132118225,
"logps/rejected": -0.7855945825576782,
"loss": 17.5329,
"rewards/accuracies": 0.4000000059604645,
"rewards/chosen": -19.576641082763672,
"rewards/margins": -11.720695495605469,
"rewards/rejected": -7.8559465408325195,
"step": 95
},
{
"epoch": 0.1733102253032929,
"grad_norm": 142.79264155311262,
"learning_rate": 9.83928250246034e-08,
"logps/chosen": -1.542560338973999,
"logps/rejected": -0.8671437501907349,
"loss": 11.9286,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -15.425603866577148,
"rewards/margins": -6.754166603088379,
"rewards/rejected": -8.67143726348877,
"step": 100
},
{
"epoch": 0.18197573656845753,
"grad_norm": 129.71279838398823,
"learning_rate": 9.799012301949136e-08,
"logps/chosen": -2.3877921104431152,
"logps/rejected": -0.9329522252082825,
"loss": 17.6579,
"rewards/accuracies": 0.0,
"rewards/chosen": -23.8779239654541,
"rewards/margins": -14.54840087890625,
"rewards/rejected": -9.329523086547852,
"step": 105
},
{
"epoch": 0.19064124783362218,
"grad_norm": 148.06599803902049,
"learning_rate": 9.754346444913042e-08,
"logps/chosen": -2.4009385108947754,
"logps/rejected": -1.1456081867218018,
"loss": 16.4529,
"rewards/accuracies": 0.20000000298023224,
"rewards/chosen": -24.009387969970703,
"rewards/margins": -12.553304672241211,
"rewards/rejected": -11.456080436706543,
"step": 110
},
{
"epoch": 0.19930675909878684,
"grad_norm": 160.51208384133946,
"learning_rate": 9.705325843055043e-08,
"logps/chosen": -1.5254523754119873,
"logps/rejected": -0.8646356463432312,
"loss": 12.0618,
"rewards/accuracies": 0.4000000059604645,
"rewards/chosen": -15.254524230957031,
"rewards/margins": -6.608166694641113,
"rewards/rejected": -8.646357536315918,
"step": 115
},
{
"epoch": 0.20797227036395147,
"grad_norm": 122.45823808036546,
"learning_rate": 9.651995396807742e-08,
"logps/chosen": -2.2533581256866455,
"logps/rejected": -0.908549964427948,
"loss": 15.2713,
"rewards/accuracies": 0.4000000059604645,
"rewards/chosen": -22.53358268737793,
"rewards/margins": -13.448083877563477,
"rewards/rejected": -9.08549976348877,
"step": 120
},
{
"epoch": 0.21663778162911612,
"grad_norm": 92.45529246500953,
"learning_rate": 9.594403954206789e-08,
"logps/chosen": -1.814674973487854,
"logps/rejected": -0.8079738616943359,
"loss": 14.6802,
"rewards/accuracies": 0.4000000059604645,
"rewards/chosen": -18.14674949645996,
"rewards/margins": -10.067010879516602,
"rewards/rejected": -8.07973861694336,
"step": 125
},
{
"epoch": 0.22530329289428075,
"grad_norm": 94.30521246591931,
"learning_rate": 9.53260426614852e-08,
"logps/chosen": -2.3024892807006836,
"logps/rejected": -1.008353590965271,
"loss": 17.7655,
"rewards/accuracies": 0.20000000298023224,
"rewards/chosen": -23.024892807006836,
"rewards/margins": -12.941357612609863,
"rewards/rejected": -10.083536148071289,
"step": 130
},
{
"epoch": 0.2339688041594454,
"grad_norm": 112.29975161277991,
"learning_rate": 9.466652938072753e-08,
"logps/chosen": -2.0203070640563965,
"logps/rejected": -1.113219141960144,
"loss": 13.6402,
"rewards/accuracies": 0.4000000059604645,
"rewards/chosen": -20.203073501586914,
"rewards/margins": -9.070881843566895,
"rewards/rejected": -11.13219165802002,
"step": 135
},
{
"epoch": 0.24263431542461006,
"grad_norm": 120.79443565778939,
"learning_rate": 9.396610378115025e-08,
"logps/chosen": -2.2065517902374268,
"logps/rejected": -0.9879514575004578,
"loss": 16.0808,
"rewards/accuracies": 0.20000000298023224,
"rewards/chosen": -22.06551742553711,
"rewards/margins": -12.186001777648926,
"rewards/rejected": -9.879514694213867,
"step": 140
},
{
"epoch": 0.2512998266897747,
"grad_norm": 114.12979842546811,
"learning_rate": 9.322540741775743e-08,
"logps/chosen": -1.9439926147460938,
"logps/rejected": -0.9498629570007324,
"loss": 14.4467,
"rewards/accuracies": 0.4000000059604645,
"rewards/chosen": -19.439926147460938,
"rewards/margins": -9.94129753112793,
"rewards/rejected": -9.498628616333008,
"step": 145
},
{
"epoch": 0.25996533795493937,
"grad_norm": 125.45737006408432,
"learning_rate": 9.244511873156944e-08,
"logps/chosen": -1.816112756729126,
"logps/rejected": -1.040378451347351,
"loss": 13.5576,
"rewards/accuracies": 0.4000000059604645,
"rewards/chosen": -18.1611270904541,
"rewards/margins": -7.7573418617248535,
"rewards/rejected": -10.403783798217773,
"step": 150
},
{
"epoch": 0.268630849220104,
"grad_norm": 112.60264240438396,
"learning_rate": 9.16259524282046e-08,
"logps/chosen": -1.8842222690582275,
"logps/rejected": -1.0307762622833252,
"loss": 13.5764,
"rewards/accuracies": 0.4000000059604645,
"rewards/chosen": -18.842222213745117,
"rewards/margins": -8.534459114074707,
"rewards/rejected": -10.307764053344727,
"step": 155
},
{
"epoch": 0.2772963604852686,
"grad_norm": 122.00400288293022,
"learning_rate": 9.076865882324452e-08,
"logps/chosen": -1.9621737003326416,
"logps/rejected": -1.0562703609466553,
"loss": 17.3966,
"rewards/accuracies": 0.4000000059604645,
"rewards/chosen": -19.621734619140625,
"rewards/margins": -9.05903434753418,
"rewards/rejected": -10.562703132629395,
"step": 160
},
{
"epoch": 0.28596187175043325,
"grad_norm": 124.16081774837384,
"learning_rate": 8.987402315498223e-08,
"logps/chosen": -1.1465253829956055,
"logps/rejected": -1.0642750263214111,
"loss": 12.7749,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -11.465253829956055,
"rewards/margins": -0.8225023150444031,
"rewards/rejected": -10.64275074005127,
"step": 165
},
{
"epoch": 0.29462738301559793,
"grad_norm": 124.62572211999075,
"learning_rate": 8.89428648651831e-08,
"logps/chosen": -1.7204177379608154,
"logps/rejected": -1.1967977285385132,
"loss": 14.6418,
"rewards/accuracies": 0.4000000059604645,
"rewards/chosen": -17.204177856445312,
"rewards/margins": -5.23620080947876,
"rewards/rejected": -11.967977523803711,
"step": 170
},
{
"epoch": 0.30329289428076256,
"grad_norm": 112.56300891139004,
"learning_rate": 8.797603684851683e-08,
"logps/chosen": -2.6767685413360596,
"logps/rejected": -1.069800853729248,
"loss": 16.2711,
"rewards/accuracies": 0.0,
"rewards/chosen": -26.767688751220703,
"rewards/margins": -16.069677352905273,
"rewards/rejected": -10.69800853729248,
"step": 175
},
{
"epoch": 0.3119584055459272,
"grad_norm": 120.78807431486482,
"learning_rate": 8.697442467134845e-08,
"logps/chosen": -1.9326350688934326,
"logps/rejected": -1.0922300815582275,
"loss": 15.0838,
"rewards/accuracies": 0.4000000059604645,
"rewards/chosen": -19.326351165771484,
"rewards/margins": -8.40404987335205,
"rewards/rejected": -10.922300338745117,
"step": 180
},
{
"epoch": 0.32062391681109187,
"grad_norm": 110.86338987148085,
"learning_rate": 8.593894576060354e-08,
"logps/chosen": -1.953160285949707,
"logps/rejected": -1.1958823204040527,
"loss": 10.7819,
"rewards/accuracies": 0.4000000059604645,
"rewards/chosen": -19.53160285949707,
"rewards/margins": -7.572779178619385,
"rewards/rejected": -11.958823204040527,
"step": 185
},
{
"epoch": 0.3292894280762565,
"grad_norm": 123.90742849743228,
"learning_rate": 8.487054856345081e-08,
"logps/chosen": -1.568197250366211,
"logps/rejected": -1.1650375127792358,
"loss": 12.3005,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -15.681971549987793,
"rewards/margins": -4.03159761428833,
"rewards/rejected": -11.650375366210938,
"step": 190
},
{
"epoch": 0.3379549393414211,
"grad_norm": 126.6347080831252,
"learning_rate": 8.377021167857166e-08,
"logps/chosen": -1.8314793109893799,
"logps/rejected": -1.2466024160385132,
"loss": 14.669,
"rewards/accuracies": 0.4000000059604645,
"rewards/chosen": -18.31479263305664,
"rewards/margins": -5.848769187927246,
"rewards/rejected": -12.466025352478027,
"step": 195
},
{
"epoch": 0.3466204506065858,
"grad_norm": 121.64468314179587,
"learning_rate": 8.263894295981257e-08,
"logps/chosen": -2.2646422386169434,
"logps/rejected": -1.226609706878662,
"loss": 12.8032,
"rewards/accuracies": 0.20000000298023224,
"rewards/chosen": -22.64642333984375,
"rewards/margins": -10.380326271057129,
"rewards/rejected": -12.266097068786621,
"step": 200
},
{
"epoch": 0.35528596187175043,
"grad_norm": 126.85588315541666,
"learning_rate": 8.147777859304095e-08,
"logps/chosen": -1.198981523513794,
"logps/rejected": -1.345165491104126,
"loss": 12.7799,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -11.989816665649414,
"rewards/margins": 1.461838960647583,
"rewards/rejected": -13.451654434204102,
"step": 205
},
{
"epoch": 0.36395147313691506,
"grad_norm": 120.43560857917016,
"learning_rate": 8.028778214705058e-08,
"logps/chosen": -2.1712536811828613,
"logps/rejected": -1.199004888534546,
"loss": 11.8499,
"rewards/accuracies": 0.20000000298023224,
"rewards/chosen": -21.712539672851562,
"rewards/margins": -9.722488403320312,
"rewards/rejected": -11.990049362182617,
"step": 210
},
{
"epoch": 0.37261698440207974,
"grad_norm": 626.8379552865316,
"learning_rate": 7.907004359938548e-08,
"logps/chosen": -2.164957046508789,
"logps/rejected": -1.0314610004425049,
"loss": 14.2149,
"rewards/accuracies": 0.20000000298023224,
"rewards/chosen": -21.649572372436523,
"rewards/margins": -11.3349609375,
"rewards/rejected": -10.314610481262207,
"step": 215
},
{
"epoch": 0.38128249566724437,
"grad_norm": 161.48734947105442,
"learning_rate": 7.782567833797457e-08,
"logps/chosen": -2.232151508331299,
"logps/rejected": -1.2837927341461182,
"loss": 13.421,
"rewards/accuracies": 0.20000000298023224,
"rewards/chosen": -22.321514129638672,
"rewards/margins": -9.483587265014648,
"rewards/rejected": -12.837926864624023,
"step": 220
},
{
"epoch": 0.389948006932409,
"grad_norm": 151.19422846807146,
"learning_rate": 7.655582613949202e-08,
"logps/chosen": -2.078399658203125,
"logps/rejected": -1.2598081827163696,
"loss": 14.0057,
"rewards/accuracies": 0.4000000059604645,
"rewards/chosen": -20.783998489379883,
"rewards/margins": -8.185917854309082,
"rewards/rejected": -12.5980806350708,
"step": 225
},
{
"epoch": 0.3986135181975737,
"grad_norm": 114.95939011147192,
"learning_rate": 7.526165012537843e-08,
"logps/chosen": -2.0014336109161377,
"logps/rejected": -1.2770476341247559,
"loss": 12.8558,
"rewards/accuracies": 0.4000000059604645,
"rewards/chosen": -20.014333724975586,
"rewards/margins": -7.24385929107666,
"rewards/rejected": -12.770475387573242,
"step": 230
},
{
"epoch": 0.4072790294627383,
"grad_norm": 611.9655768877072,
"learning_rate": 7.394433569647934e-08,
"logps/chosen": -1.4717886447906494,
"logps/rejected": -1.286112666130066,
"loss": 11.6772,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -14.717885971069336,
"rewards/margins": -1.8567602634429932,
"rewards/rejected": -12.861126899719238,
"step": 235
},
{
"epoch": 0.41594454072790293,
"grad_norm": 122.39432963173844,
"learning_rate": 7.260508944727723e-08,
"logps/chosen": -2.4743523597717285,
"logps/rejected": -1.451537013053894,
"loss": 13.4604,
"rewards/accuracies": 0.20000000298023224,
"rewards/chosen": -24.74352264404297,
"rewards/margins": -10.22815227508545,
"rewards/rejected": -14.515371322631836,
"step": 240
},
{
"epoch": 0.4246100519930676,
"grad_norm": 178.54805378295796,
"learning_rate": 7.124513806071086e-08,
"logps/chosen": -1.7837231159210205,
"logps/rejected": -1.259641170501709,
"loss": 13.0229,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -17.837230682373047,
"rewards/margins": -5.240820407867432,
"rewards/rejected": -12.596410751342773,
"step": 245
},
{
"epoch": 0.43327556325823224,
"grad_norm": 301.11675694113546,
"learning_rate": 6.986572718459478e-08,
"logps/chosen": -2.424816131591797,
"logps/rejected": -1.6280561685562134,
"loss": 12.263,
"rewards/accuracies": 0.20000000298023224,
"rewards/chosen": -24.2481632232666,
"rewards/margins": -7.9676008224487305,
"rewards/rejected": -16.280563354492188,
"step": 250
},
{
"epoch": 0.44194107452339687,
"grad_norm": 151.1924896359916,
"learning_rate": 6.846812029066787e-08,
"logps/chosen": -2.284111499786377,
"logps/rejected": -1.4265038967132568,
"loss": 10.2145,
"rewards/accuracies": 0.20000000298023224,
"rewards/chosen": -22.841115951538086,
"rewards/margins": -8.576075553894043,
"rewards/rejected": -14.265039443969727,
"step": 255
},
{
"epoch": 0.4506065857885615,
"grad_norm": 117.0245050950957,
"learning_rate": 6.705359751731611e-08,
"logps/chosen": -2.581634998321533,
"logps/rejected": -1.299729585647583,
"loss": 11.8198,
"rewards/accuracies": 0.20000000298023224,
"rewards/chosen": -25.81635093688965,
"rewards/margins": -12.819055557250977,
"rewards/rejected": -12.997296333312988,
"step": 260
},
{
"epoch": 0.4592720970537262,
"grad_norm": 164.40913549983566,
"learning_rate": 6.562345449702951e-08,
"logps/chosen": -1.6220916509628296,
"logps/rejected": -1.5810987949371338,
"loss": 10.2127,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -16.220916748046875,
"rewards/margins": -0.40992775559425354,
"rewards/rejected": -15.81098747253418,
"step": 265
},
{
"epoch": 0.4679376083188908,
"grad_norm": 197.25533329293654,
"learning_rate": 6.417900116966714e-08,
"logps/chosen": -2.8826935291290283,
"logps/rejected": -1.366994857788086,
"loss": 11.5423,
"rewards/accuracies": 0.0,
"rewards/chosen": -28.826934814453125,
"rewards/margins": -15.156987190246582,
"rewards/rejected": -13.669949531555176,
"step": 270
},
{
"epoch": 0.47660311958405543,
"grad_norm": 152.00371643796973,
"learning_rate": 6.272156058261753e-08,
"logps/chosen": -1.8090347051620483,
"logps/rejected": -1.4397003650665283,
"loss": 9.7595,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -18.090347290039062,
"rewards/margins": -3.6933434009552,
"rewards/rejected": -14.397003173828125,
"step": 275
},
{
"epoch": 0.4852686308492201,
"grad_norm": 394.38126102668485,
"learning_rate": 6.125246767895286e-08,
"logps/chosen": -1.9543708562850952,
"logps/rejected": -1.5106321573257446,
"loss": 11.2695,
"rewards/accuracies": 0.4000000059604645,
"rewards/chosen": -19.54370880126953,
"rewards/margins": -4.437387943267822,
"rewards/rejected": -15.10632038116455,
"step": 280
},
{
"epoch": 0.49393414211438474,
"grad_norm": 142.92864697276755,
"learning_rate": 5.977306807468774e-08,
"logps/chosen": -2.7028214931488037,
"logps/rejected": -1.3628876209259033,
"loss": 11.5499,
"rewards/accuracies": 0.0,
"rewards/chosen": -27.028217315673828,
"rewards/margins": -13.399337768554688,
"rewards/rejected": -13.628877639770508,
"step": 285
},
{
"epoch": 0.5025996533795494,
"grad_norm": 178.01151675646489,
"learning_rate": 5.828471682626175e-08,
"logps/chosen": -2.410367012023926,
"logps/rejected": -1.658764123916626,
"loss": 10.0203,
"rewards/accuracies": 0.20000000298023224,
"rewards/chosen": -24.103670120239258,
"rewards/margins": -7.516029357910156,
"rewards/rejected": -16.5876407623291,
"step": 290
},
{
"epoch": 0.511265164644714,
"grad_norm": 441.7833165957171,
"learning_rate": 5.678877718937519e-08,
"logps/chosen": -1.7007253170013428,
"logps/rejected": -1.7013359069824219,
"loss": 10.809,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -17.00725555419922,
"rewards/margins": 0.006104087922722101,
"rewards/rejected": -17.01336097717285,
"step": 295
},
{
"epoch": 0.5199306759098787,
"grad_norm": 117.48808947391147,
"learning_rate": 5.52866193703147e-08,
"logps/chosen": -2.6699347496032715,
"logps/rejected": -1.7835134267807007,
"loss": 9.6309,
"rewards/accuracies": 0.0,
"rewards/chosen": -26.699350357055664,
"rewards/margins": -8.864214897155762,
"rewards/rejected": -17.835134506225586,
"step": 300
},
{
"epoch": 0.5285961871750433,
"grad_norm": 217.9261859040597,
"learning_rate": 5.3779619270912414e-08,
"logps/chosen": -1.99038827419281,
"logps/rejected": -1.9274762868881226,
"loss": 10.2615,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -19.90388298034668,
"rewards/margins": -0.6291202306747437,
"rewards/rejected": -19.274763107299805,
"step": 305
},
{
"epoch": 0.537261698440208,
"grad_norm": 205.3616408894159,
"learning_rate": 5.22691572282884e-08,
"logps/chosen": -2.596215009689331,
"logps/rejected": -2.011929750442505,
"loss": 9.5973,
"rewards/accuracies": 0.4000000059604645,
"rewards/chosen": -25.962148666381836,
"rewards/margins": -5.842852592468262,
"rewards/rejected": -20.11929702758789,
"step": 310
},
{
"epoch": 0.5459272097053726,
"grad_norm": 190.6871604383273,
"learning_rate": 5.0756616750530436e-08,
"logps/chosen": -1.3547618389129639,
"logps/rejected": -1.6575043201446533,
"loss": 8.561,
"rewards/accuracies": 1.0,
"rewards/chosen": -13.54761791229248,
"rewards/margins": 3.027425527572632,
"rewards/rejected": -16.575044631958008,
"step": 315
},
{
"epoch": 0.5545927209705372,
"grad_norm": 233.80398397633627,
"learning_rate": 4.924338324946955e-08,
"logps/chosen": -2.2690348625183105,
"logps/rejected": -1.7823927402496338,
"loss": 10.8419,
"rewards/accuracies": 0.20000000298023224,
"rewards/chosen": -22.690349578857422,
"rewards/margins": -4.8664231300354,
"rewards/rejected": -17.823925018310547,
"step": 320
},
{
"epoch": 0.5632582322357019,
"grad_norm": 271.18165460140943,
"learning_rate": 4.77308427717116e-08,
"logps/chosen": -2.08347749710083,
"logps/rejected": -1.5966533422470093,
"loss": 9.6404,
"rewards/accuracies": 0.4000000059604645,
"rewards/chosen": -20.83477783203125,
"rewards/margins": -4.868241786956787,
"rewards/rejected": -15.966534614562988,
"step": 325
},
{
"epoch": 0.5719237435008665,
"grad_norm": 242.574459310927,
"learning_rate": 4.622038072908758e-08,
"logps/chosen": -1.9583733081817627,
"logps/rejected": -1.7426979541778564,
"loss": 8.1746,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -19.58373260498047,
"rewards/margins": -2.156752824783325,
"rewards/rejected": -17.42698097229004,
"step": 330
},
{
"epoch": 0.5805892547660312,
"grad_norm": 253.3241055482191,
"learning_rate": 4.4713380629685306e-08,
"logps/chosen": -2.1484360694885254,
"logps/rejected": -2.0616002082824707,
"loss": 7.292,
"rewards/accuracies": 0.4000000059604645,
"rewards/chosen": -21.48436164855957,
"rewards/margins": -0.8683557510375977,
"rewards/rejected": -20.616004943847656,
"step": 335
},
{
"epoch": 0.5892547660311959,
"grad_norm": 252.56727877933076,
"learning_rate": 4.321122281062481e-08,
"logps/chosen": -2.473069906234741,
"logps/rejected": -1.9099452495574951,
"loss": 8.4532,
"rewards/accuracies": 0.20000000298023224,
"rewards/chosen": -24.73069953918457,
"rewards/margins": -5.631247043609619,
"rewards/rejected": -19.09945297241211,
"step": 340
},
{
"epoch": 0.5979202772963604,
"grad_norm": 193.98886710844522,
"learning_rate": 4.1715283173738244e-08,
"logps/chosen": -1.969444990158081,
"logps/rejected": -1.8921003341674805,
"loss": 7.5218,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -19.6944522857666,
"rewards/margins": -0.7734478116035461,
"rewards/rejected": -18.921003341674805,
"step": 345
},
{
"epoch": 0.6065857885615251,
"grad_norm": 237.02702622060875,
"learning_rate": 4.022693192531226e-08,
"logps/chosen": -2.34706974029541,
"logps/rejected": -1.886749029159546,
"loss": 6.8617,
"rewards/accuracies": 0.4000000059604645,
"rewards/chosen": -23.4706974029541,
"rewards/margins": -4.603204727172852,
"rewards/rejected": -18.86749267578125,
"step": 350
},
{
"epoch": 0.6152512998266898,
"grad_norm": 291.8573957521699,
"learning_rate": 3.874753232104714e-08,
"logps/chosen": -2.6580417156219482,
"logps/rejected": -1.7777878046035767,
"loss": 9.7241,
"rewards/accuracies": 0.20000000298023224,
"rewards/chosen": -26.580419540405273,
"rewards/margins": -8.802539825439453,
"rewards/rejected": -17.777877807617188,
"step": 355
},
{
"epoch": 0.6239168110918544,
"grad_norm": 189.51798627174236,
"learning_rate": 3.727843941738248e-08,
"logps/chosen": -2.312593936920166,
"logps/rejected": -2.0596203804016113,
"loss": 6.3336,
"rewards/accuracies": 0.4000000059604645,
"rewards/chosen": -23.125940322875977,
"rewards/margins": -2.529735565185547,
"rewards/rejected": -20.596202850341797,
"step": 360
},
{
"epoch": 0.6325823223570191,
"grad_norm": 286.87663861473686,
"learning_rate": 3.582099883033285e-08,
"logps/chosen": -2.689950704574585,
"logps/rejected": -2.024871826171875,
"loss": 9.0664,
"rewards/accuracies": 0.0,
"rewards/chosen": -26.89950942993164,
"rewards/margins": -6.650788307189941,
"rewards/rejected": -20.248720169067383,
"step": 365
},
{
"epoch": 0.6412478336221837,
"grad_norm": 206.15397080557344,
"learning_rate": 3.437654550297049e-08,
"logps/chosen": -1.957179069519043,
"logps/rejected": -2.0333800315856934,
"loss": 5.0065,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -19.57179069519043,
"rewards/margins": 0.7620084881782532,
"rewards/rejected": -20.333799362182617,
"step": 370
},
{
"epoch": 0.6499133448873483,
"grad_norm": 265.4555794317019,
"learning_rate": 3.294640248268389e-08,
"logps/chosen": -2.076646566390991,
"logps/rejected": -2.0849030017852783,
"loss": 6.0048,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -20.766464233398438,
"rewards/margins": 0.08256340026855469,
"rewards/rejected": -20.849027633666992,
"step": 375
},
{
"epoch": 0.658578856152513,
"grad_norm": 244.98209798309657,
"learning_rate": 3.153187970933213e-08,
"logps/chosen": -2.385413646697998,
"logps/rejected": -2.327801465988159,
"loss": 5.5495,
"rewards/accuracies": 0.4000000059604645,
"rewards/chosen": -23.854137420654297,
"rewards/margins": -0.576123058795929,
"rewards/rejected": -23.27801513671875,
"step": 380
},
{
"epoch": 0.6672443674176777,
"grad_norm": 243.82242299590067,
"learning_rate": 3.0134272815405224e-08,
"logps/chosen": -2.1494534015655518,
"logps/rejected": -2.3645691871643066,
"loss": 6.1118,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -21.49453353881836,
"rewards/margins": 2.1511597633361816,
"rewards/rejected": -23.645694732666016,
"step": 385
},
{
"epoch": 0.6759098786828422,
"grad_norm": 306.8365924711833,
"learning_rate": 2.8754861939289133e-08,
"logps/chosen": -2.2285687923431396,
"logps/rejected": -2.242079257965088,
"loss": 5.9324,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -22.285686492919922,
"rewards/margins": 0.1351058930158615,
"rewards/rejected": -22.420795440673828,
"step": 390
},
{
"epoch": 0.6845753899480069,
"grad_norm": 292.0875551239737,
"learning_rate": 2.7394910552722773e-08,
"logps/chosen": -2.4236674308776855,
"logps/rejected": -2.342806339263916,
"loss": 5.984,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -24.236675262451172,
"rewards/margins": -0.8086126446723938,
"rewards/rejected": -23.428064346313477,
"step": 395
},
{
"epoch": 0.6932409012131716,
"grad_norm": 330.0702338474899,
"learning_rate": 2.6055664303520652e-08,
"logps/chosen": -2.262308359146118,
"logps/rejected": -2.4734067916870117,
"loss": 5.5632,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -22.623083114624023,
"rewards/margins": 2.1109836101531982,
"rewards/rejected": -24.734067916870117,
"step": 400
},
{
"epoch": 0.6932409012131716,
"eval_logps/chosen": -2.345215082168579,
"eval_logps/rejected": -2.3579459190368652,
"eval_loss": 6.394889831542969,
"eval_rewards/accuracies": 0.5,
"eval_rewards/chosen": -23.452152252197266,
"eval_rewards/margins": 0.12730884552001953,
"eval_rewards/rejected": -23.57946014404297,
"eval_runtime": 4.1708,
"eval_samples_per_second": 2.398,
"eval_steps_per_second": 0.48,
"step": 400
},
{
"epoch": 0.7019064124783362,
"grad_norm": 432.18725823455117,
"learning_rate": 2.4738349874621583e-08,
"logps/chosen": -2.2886712551116943,
"logps/rejected": -2.3248345851898193,
"loss": 5.6469,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -22.8867130279541,
"rewards/margins": 0.36163291335105896,
"rewards/rejected": -23.248348236083984,
"step": 405
},
{
"epoch": 0.7105719237435009,
"grad_norm": 289.0106970167949,
"learning_rate": 2.3444173860507965e-08,
"logps/chosen": -2.2404701709747314,
"logps/rejected": -2.36967134475708,
"loss": 4.7682,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -22.40470314025879,
"rewards/margins": 1.2920114994049072,
"rewards/rejected": -23.696712493896484,
"step": 410
},
{
"epoch": 0.7192374350086655,
"grad_norm": 336.7530764224563,
"learning_rate": 2.2174321662025424e-08,
"logps/chosen": -2.565544843673706,
"logps/rejected": -2.5797317028045654,
"loss": 5.0399,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -25.65545082092285,
"rewards/margins": 0.1418682038784027,
"rewards/rejected": -25.797317504882812,
"step": 415
},
{
"epoch": 0.7279029462738301,
"grad_norm": 236.60136142580333,
"learning_rate": 2.092995640061454e-08,
"logps/chosen": -2.1976778507232666,
"logps/rejected": -2.399247169494629,
"loss": 3.4771,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -21.976778030395508,
"rewards/margins": 2.015695095062256,
"rewards/rejected": -23.99247169494629,
"step": 420
},
{
"epoch": 0.7365684575389948,
"grad_norm": 285.99685463404285,
"learning_rate": 1.9712217852949407e-08,
"logps/chosen": -1.974400281906128,
"logps/rejected": -2.2366092205047607,
"loss": 3.8414,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -19.744003295898438,
"rewards/margins": 2.6220881938934326,
"rewards/rejected": -22.366092681884766,
"step": 425
},
{
"epoch": 0.7452339688041595,
"grad_norm": 386.7211621912297,
"learning_rate": 1.852222140695906e-08,
"logps/chosen": -2.8100500106811523,
"logps/rejected": -2.9486374855041504,
"loss": 3.7648,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -28.100500106811523,
"rewards/margins": 1.3858760595321655,
"rewards/rejected": -29.486377716064453,
"step": 430
},
{
"epoch": 0.7538994800693241,
"grad_norm": 435.7919496767513,
"learning_rate": 1.736105704018744e-08,
"logps/chosen": -2.645963668823242,
"logps/rejected": -2.6360223293304443,
"loss": 3.8238,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -26.45963478088379,
"rewards/margins": -0.09941063076257706,
"rewards/rejected": -26.3602237701416,
"step": 435
},
{
"epoch": 0.7625649913344887,
"grad_norm": 300.3158959034186,
"learning_rate": 1.622978832142833e-08,
"logps/chosen": -2.481766700744629,
"logps/rejected": -2.7485013008117676,
"loss": 3.9495,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -24.817668914794922,
"rewards/margins": 2.6673457622528076,
"rewards/rejected": -27.48501205444336,
"step": 440
},
{
"epoch": 0.7712305025996534,
"grad_norm": 267.00928393815593,
"learning_rate": 1.51294514365492e-08,
"logps/chosen": -2.0402495861053467,
"logps/rejected": -2.3271498680114746,
"loss": 3.2962,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -20.402498245239258,
"rewards/margins": 2.8689987659454346,
"rewards/rejected": -23.271495819091797,
"step": 445
},
{
"epoch": 0.779896013864818,
"grad_norm": 440.98075927307,
"learning_rate": 1.4061054239396452e-08,
"logps/chosen": -2.774311065673828,
"logps/rejected": -2.5872533321380615,
"loss": 4.3524,
"rewards/accuracies": 0.4000000059604645,
"rewards/chosen": -27.743112564086914,
"rewards/margins": -1.8705768585205078,
"rewards/rejected": -25.872533798217773,
"step": 450
},
{
"epoch": 0.7885615251299827,
"grad_norm": 249.3269527978748,
"learning_rate": 1.302557532865155e-08,
"logps/chosen": -2.2388012409210205,
"logps/rejected": -2.5778894424438477,
"loss": 3.1082,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -22.388011932373047,
"rewards/margins": 3.390883207321167,
"rewards/rejected": -25.778894424438477,
"step": 455
},
{
"epoch": 0.7972270363951474,
"grad_norm": 266.5651240717741,
"learning_rate": 1.2023963151483163e-08,
"logps/chosen": -2.2709708213806152,
"logps/rejected": -2.752668857574463,
"loss": 2.9738,
"rewards/accuracies": 1.0,
"rewards/chosen": -22.7097110748291,
"rewards/margins": 4.816981792449951,
"rewards/rejected": -27.52669334411621,
"step": 460
},
{
"epoch": 0.8058925476603119,
"grad_norm": 242.9598196737481,
"learning_rate": 1.1057135134816897e-08,
"logps/chosen": -2.198323965072632,
"logps/rejected": -2.5944952964782715,
"loss": 3.1251,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -21.98324203491211,
"rewards/margins": 3.96171236038208,
"rewards/rejected": -25.9449520111084,
"step": 465
},
{
"epoch": 0.8145580589254766,
"grad_norm": 352.7538054645251,
"learning_rate": 1.012597684501777e-08,
"logps/chosen": -2.405252456665039,
"logps/rejected": -2.7467832565307617,
"loss": 2.9631,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -24.052526473999023,
"rewards/margins": 3.415306806564331,
"rewards/rejected": -27.467830657958984,
"step": 470
},
{
"epoch": 0.8232235701906413,
"grad_norm": 419.88510678787657,
"learning_rate": 9.231341176755487e-09,
"logps/chosen": -2.7878851890563965,
"logps/rejected": -2.999274730682373,
"loss": 3.0581,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -27.87885093688965,
"rewards/margins": 2.113895893096924,
"rewards/rejected": -29.992748260498047,
"step": 475
},
{
"epoch": 0.8318890814558059,
"grad_norm": 374.57650865628955,
"learning_rate": 8.3740475717954e-09,
"logps/chosen": -2.368149518966675,
"logps/rejected": -2.6514530181884766,
"loss": 2.767,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -23.681495666503906,
"rewards/margins": 2.8330326080322266,
"rewards/rejected": -26.5145263671875,
"step": 480
},
{
"epoch": 0.8405545927209706,
"grad_norm": 298.2316161579765,
"learning_rate": 7.554881268430563e-09,
"logps/chosen": -1.9999809265136719,
"logps/rejected": -2.5488007068634033,
"loss": 2.6938,
"rewards/accuracies": 1.0,
"rewards/chosen": -19.99980926513672,
"rewards/margins": 5.488197326660156,
"rewards/rejected": -25.488004684448242,
"step": 485
},
{
"epoch": 0.8492201039861352,
"grad_norm": 289.378607710532,
"learning_rate": 6.774592582242567e-09,
"logps/chosen": -2.1875576972961426,
"logps/rejected": -2.7966582775115967,
"loss": 2.2383,
"rewards/accuracies": 1.0,
"rewards/chosen": -21.87557601928711,
"rewards/margins": 6.091004371643066,
"rewards/rejected": -27.966583251953125,
"step": 490
},
{
"epoch": 0.8578856152512998,
"grad_norm": 284.6211316368398,
"learning_rate": 6.033896218849766e-09,
"logps/chosen": -2.5260894298553467,
"logps/rejected": -2.4975428581237793,
"loss": 2.7359,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -25.260896682739258,
"rewards/margins": -0.2854675352573395,
"rewards/rejected": -24.97542953491211,
"step": 495
},
{
"epoch": 0.8665511265164645,
"grad_norm": 450.49404572669596,
"learning_rate": 5.3334706192724786e-09,
"logps/chosen": -2.397015333175659,
"logps/rejected": -2.851841449737549,
"loss": 3.193,
"rewards/accuracies": 1.0,
"rewards/chosen": -23.97015380859375,
"rewards/margins": 4.548262596130371,
"rewards/rejected": -28.518413543701172,
"step": 500
},
{
"epoch": 0.8752166377816292,
"grad_norm": 382.86348627674346,
"learning_rate": 4.673957338514811e-09,
"logps/chosen": -2.497405529022217,
"logps/rejected": -2.8319692611694336,
"loss": 2.438,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -24.974056243896484,
"rewards/margins": 3.3456368446350098,
"rewards/rejected": -28.319692611694336,
"step": 505
},
{
"epoch": 0.8838821490467937,
"grad_norm": 169.16569455598628,
"learning_rate": 4.055960457932106e-09,
"logps/chosen": -2.318751811981201,
"logps/rejected": -2.7315309047698975,
"loss": 2.5218,
"rewards/accuracies": 1.0,
"rewards/chosen": -23.187519073486328,
"rewards/margins": 4.127790451049805,
"rewards/rejected": -27.315311431884766,
"step": 510
},
{
"epoch": 0.8925476603119584,
"grad_norm": 290.65862840125214,
"learning_rate": 3.4800460319225734e-09,
"logps/chosen": -2.376023054122925,
"logps/rejected": -2.822336196899414,
"loss": 3.0831,
"rewards/accuracies": 1.0,
"rewards/chosen": -23.760231018066406,
"rewards/margins": 4.463130950927734,
"rewards/rejected": -28.22336196899414,
"step": 515
},
{
"epoch": 0.901213171577123,
"grad_norm": 473.51636465627206,
"learning_rate": 2.9467415694495624e-09,
"logps/chosen": -2.086275100708008,
"logps/rejected": -2.509833574295044,
"loss": 2.7674,
"rewards/accuracies": 1.0,
"rewards/chosen": -20.86275291442871,
"rewards/margins": 4.235583782196045,
"rewards/rejected": -25.098339080810547,
"step": 520
},
{
"epoch": 0.9098786828422877,
"grad_norm": 195.68912447867717,
"learning_rate": 2.4565355508695816e-09,
"logps/chosen": -2.3413867950439453,
"logps/rejected": -3.0255393981933594,
"loss": 2.3523,
"rewards/accuracies": 1.0,
"rewards/chosen": -23.413867950439453,
"rewards/margins": 6.841525077819824,
"rewards/rejected": -30.255395889282227,
"step": 525
},
{
"epoch": 0.9185441941074524,
"grad_norm": 165.55622486042316,
"learning_rate": 2.0098769805086456e-09,
"logps/chosen": -2.413383722305298,
"logps/rejected": -2.902367115020752,
"loss": 2.0526,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -24.133834838867188,
"rewards/margins": 4.889834880828857,
"rewards/rejected": -29.023672103881836,
"step": 530
},
{
"epoch": 0.9272097053726169,
"grad_norm": 271.2253606637343,
"learning_rate": 1.6071749753965912e-09,
"logps/chosen": -2.4116291999816895,
"logps/rejected": -3.0955259799957275,
"loss": 2.1614,
"rewards/accuracies": 1.0,
"rewards/chosen": -24.116291046142578,
"rewards/margins": 6.838971138000488,
"rewards/rejected": -30.95526123046875,
"step": 535
},
{
"epoch": 0.9358752166377816,
"grad_norm": 222.81738320230136,
"learning_rate": 1.2487983905362932e-09,
"logps/chosen": -2.4959144592285156,
"logps/rejected": -2.7256417274475098,
"loss": 2.1184,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -24.959142684936523,
"rewards/margins": 2.297271251678467,
"rewards/rejected": -27.25641441345215,
"step": 540
},
{
"epoch": 0.9445407279029463,
"grad_norm": 214.85286607093462,
"learning_rate": 9.350754810507845e-10,
"logps/chosen": -2.820622682571411,
"logps/rejected": -3.021366596221924,
"loss": 2.2908,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -28.206228256225586,
"rewards/margins": 2.007439136505127,
"rewards/rejected": -30.213665008544922,
"step": 545
},
{
"epoch": 0.9532062391681109,
"grad_norm": 211.8074649095587,
"learning_rate": 6.662936015178977e-10,
"logps/chosen": -2.4099273681640625,
"logps/rejected": -2.8419957160949707,
"loss": 3.6358,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -24.099275588989258,
"rewards/margins": 4.3206787109375,
"rewards/rejected": -28.419952392578125,
"step": 550
},
{
"epoch": 0.9618717504332756,
"grad_norm": 378.71477984576666,
"learning_rate": 4.4269894276779406e-10,
"logps/chosen": -2.685300588607788,
"logps/rejected": -3.123471736907959,
"loss": 2.9005,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -26.85300636291504,
"rewards/margins": 4.381711006164551,
"rewards/rejected": -31.234716415405273,
"step": 555
},
{
"epoch": 0.9705372616984402,
"grad_norm": 288.4807564220741,
"learning_rate": 2.6449630638438193e-10,
"logps/chosen": -2.6586508750915527,
"logps/rejected": -2.9211390018463135,
"loss": 2.7155,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -26.58650779724121,
"rewards/margins": 2.624882221221924,
"rewards/rejected": -29.211389541625977,
"step": 560
},
{
"epoch": 0.9792027729636048,
"grad_norm": 271.5646477670488,
"learning_rate": 1.3184891711727765e-10,
"logps/chosen": -2.1007838249206543,
"logps/rejected": -2.534923553466797,
"loss": 2.8816,
"rewards/accuracies": 1.0,
"rewards/chosen": -21.007837295532227,
"rewards/margins": 4.341399192810059,
"rewards/rejected": -25.3492374420166,
"step": 565
},
{
"epoch": 0.9878682842287695,
"grad_norm": 354.88743197085506,
"learning_rate": 4.487827337604222e-11,
"logps/chosen": -2.3186147212982178,
"logps/rejected": -3.0023269653320312,
"loss": 2.7364,
"rewards/accuracies": 1.0,
"rewards/chosen": -23.186147689819336,
"rewards/margins": 6.837121486663818,
"rewards/rejected": -30.023269653320312,
"step": 570
},
{
"epoch": 0.9965337954939342,
"grad_norm": 184.69149045968697,
"learning_rate": 3.664035943679033e-12,
"logps/chosen": -2.4100213050842285,
"logps/rejected": -2.7757315635681152,
"loss": 3.0952,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -24.100210189819336,
"rewards/margins": 3.6571030616760254,
"rewards/rejected": -27.757314682006836,
"step": 575
},
{
"epoch": 1.0,
"step": 577,
"total_flos": 0.0,
"train_loss": 9.667842355096774,
"train_runtime": 2221.9837,
"train_samples_per_second": 2.075,
"train_steps_per_second": 0.26
}
],
"logging_steps": 5,
"max_steps": 577,
"num_input_tokens_seen": 0,
"num_train_epochs": 1,
"save_steps": 1000000,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 0.0,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}