phi-2-ipo-chatml-lora-20k-40k-i1 / trainer_state.json
BraylonDash's picture
Model save
0caa5c2 verified
raw
history blame
31 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.0,
"eval_steps": 500,
"global_step": 625,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 7.936507936507937e-08,
"logits/chosen": 0.9578127264976501,
"logits/rejected": 0.7914565801620483,
"logps/chosen": -360.49114990234375,
"logps/rejected": -309.01898193359375,
"loss": 250000.0,
"rewards/accuracies": 0.0,
"rewards/chosen": 0.0,
"rewards/margins": 0.0,
"rewards/rejected": 0.0,
"step": 1
},
{
"epoch": 0.02,
"learning_rate": 7.936507936507937e-07,
"logits/chosen": 0.7728305459022522,
"logits/rejected": 0.768233597278595,
"logps/chosen": -306.1025390625,
"logps/rejected": -277.5509338378906,
"loss": 249993.8889,
"rewards/accuracies": 0.2986111044883728,
"rewards/chosen": -5.4183616157388315e-05,
"rewards/margins": -3.124285285593942e-05,
"rewards/rejected": -2.29407651204383e-05,
"step": 10
},
{
"epoch": 0.03,
"learning_rate": 1.5873015873015873e-06,
"logits/chosen": 0.7892485857009888,
"logits/rejected": 0.8077142834663391,
"logps/chosen": -343.6685791015625,
"logps/rejected": -315.8797912597656,
"loss": 249850.3,
"rewards/accuracies": 0.550000011920929,
"rewards/chosen": -0.0003508512454573065,
"rewards/margins": 0.0002387135464232415,
"rewards/rejected": -0.0005895647336728871,
"step": 20
},
{
"epoch": 0.05,
"learning_rate": 2.380952380952381e-06,
"logits/chosen": 0.7931613922119141,
"logits/rejected": 0.856336236000061,
"logps/chosen": -329.352294921875,
"logps/rejected": -310.99786376953125,
"loss": 249686.05,
"rewards/accuracies": 0.5562499761581421,
"rewards/chosen": -0.0014547129394486547,
"rewards/margins": 0.00029922794783487916,
"rewards/rejected": -0.001753940829075873,
"step": 30
},
{
"epoch": 0.06,
"learning_rate": 3.1746031746031746e-06,
"logits/chosen": 0.8100332021713257,
"logits/rejected": 0.7524169087409973,
"logps/chosen": -318.7286071777344,
"logps/rejected": -312.85418701171875,
"loss": 249475.5,
"rewards/accuracies": 0.5,
"rewards/chosen": -0.003864737693220377,
"rewards/margins": 0.0004840078763663769,
"rewards/rejected": -0.004348746035248041,
"step": 40
},
{
"epoch": 0.08,
"learning_rate": 3.968253968253968e-06,
"logits/chosen": 0.7530747652053833,
"logits/rejected": 0.8437241315841675,
"logps/chosen": -315.67999267578125,
"logps/rejected": -298.5876770019531,
"loss": 248612.35,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.008299140259623528,
"rewards/margins": 0.0013344786129891872,
"rewards/rejected": -0.009633619338274002,
"step": 50
},
{
"epoch": 0.1,
"learning_rate": 4.761904761904762e-06,
"logits/chosen": 0.7030428647994995,
"logits/rejected": 0.6946632266044617,
"logps/chosen": -321.80157470703125,
"logps/rejected": -302.8958740234375,
"loss": 248274.05,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": -0.016654204577207565,
"rewards/margins": 0.002316893544048071,
"rewards/rejected": -0.018971096724271774,
"step": 60
},
{
"epoch": 0.11,
"learning_rate": 4.998086282661188e-06,
"logits/chosen": 0.6814395189285278,
"logits/rejected": 0.7533793449401855,
"logps/chosen": -353.67669677734375,
"logps/rejected": -338.9217224121094,
"loss": 246213.4,
"rewards/accuracies": 0.5687500238418579,
"rewards/chosen": -0.026208389550447464,
"rewards/margins": 0.0031645516864955425,
"rewards/rejected": -0.029372945427894592,
"step": 70
},
{
"epoch": 0.13,
"learning_rate": 4.988720025682995e-06,
"logits/chosen": 0.66422039270401,
"logits/rejected": 0.6669551134109497,
"logps/chosen": -333.9057312011719,
"logps/rejected": -346.008544921875,
"loss": 244032.0,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -0.03818231448531151,
"rewards/margins": 0.00513039855286479,
"rewards/rejected": -0.043312713503837585,
"step": 80
},
{
"epoch": 0.14,
"learning_rate": 4.9715789537359126e-06,
"logits/chosen": 0.6255489587783813,
"logits/rejected": 0.6321488618850708,
"logps/chosen": -334.863525390625,
"logps/rejected": -334.16644287109375,
"loss": 240875.95,
"rewards/accuracies": 0.6187499761581421,
"rewards/chosen": -0.047401346266269684,
"rewards/margins": 0.010096731595695019,
"rewards/rejected": -0.05749807506799698,
"step": 90
},
{
"epoch": 0.16,
"learning_rate": 4.946716615897932e-06,
"logits/chosen": 0.5756382942199707,
"logits/rejected": 0.5615967512130737,
"logps/chosen": -333.1421813964844,
"logps/rejected": -330.11920166015625,
"loss": 240338.95,
"rewards/accuracies": 0.5562499761581421,
"rewards/chosen": -0.0577034130692482,
"rewards/margins": 0.00923432968556881,
"rewards/rejected": -0.06693773716688156,
"step": 100
},
{
"epoch": 0.18,
"learning_rate": 4.9142106826480114e-06,
"logits/chosen": 0.4540364742279053,
"logits/rejected": 0.5108239054679871,
"logps/chosen": -352.08721923828125,
"logps/rejected": -346.1022644042969,
"loss": 240647.4,
"rewards/accuracies": 0.5062500238418579,
"rewards/chosen": -0.06690040230751038,
"rewards/margins": 0.007805522531270981,
"rewards/rejected": -0.07470591366291046,
"step": 110
},
{
"epoch": 0.19,
"learning_rate": 4.874162703221823e-06,
"logits/chosen": 0.395730584859848,
"logits/rejected": 0.42432737350463867,
"logps/chosen": -387.7721252441406,
"logps/rejected": -377.1489562988281,
"loss": 235908.05,
"rewards/accuracies": 0.6187499761581421,
"rewards/chosen": -0.09094537049531937,
"rewards/margins": 0.01256307028234005,
"rewards/rejected": -0.10350845009088516,
"step": 120
},
{
"epoch": 0.21,
"learning_rate": 4.826697788369752e-06,
"logits/chosen": 0.3145865797996521,
"logits/rejected": 0.3371677100658417,
"logps/chosen": -404.41522216796875,
"logps/rejected": -429.7151794433594,
"loss": 229710.4,
"rewards/accuracies": 0.643750011920929,
"rewards/chosen": -0.115650475025177,
"rewards/margins": 0.024767927825450897,
"rewards/rejected": -0.1404184103012085,
"step": 130
},
{
"epoch": 0.22,
"learning_rate": 4.7719642195082224e-06,
"logits/chosen": 0.17865605652332306,
"logits/rejected": 0.24005897343158722,
"logps/chosen": -409.31573486328125,
"logps/rejected": -430.16961669921875,
"loss": 226331.25,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.1381843388080597,
"rewards/margins": 0.028940195217728615,
"rewards/rejected": -0.16712453961372375,
"step": 140
},
{
"epoch": 0.24,
"learning_rate": 4.710132985485355e-06,
"logits/chosen": 0.05765807628631592,
"logits/rejected": 0.11190862953662872,
"logps/chosen": -517.4808959960938,
"logps/rejected": -503.72503662109375,
"loss": 217724.3,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.1879872977733612,
"rewards/margins": 0.04196537286043167,
"rewards/rejected": -0.22995266318321228,
"step": 150
},
{
"epoch": 0.26,
"learning_rate": 4.641397248408122e-06,
"logits/chosen": 0.02083182893693447,
"logits/rejected": 0.06791719049215317,
"logps/chosen": -501.41314697265625,
"logps/rejected": -518.4356689453125,
"loss": 218921.525,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.22430023550987244,
"rewards/margins": 0.047038402408361435,
"rewards/rejected": -0.27133864164352417,
"step": 160
},
{
"epoch": 0.27,
"learning_rate": 4.5659717401997655e-06,
"logits/chosen": 0.05980447679758072,
"logits/rejected": 0.13415992259979248,
"logps/chosen": -510.23504638671875,
"logps/rejected": -534.2890014648438,
"loss": 224117.425,
"rewards/accuracies": 0.59375,
"rewards/chosen": -0.20021386444568634,
"rewards/margins": 0.04722369834780693,
"rewards/rejected": -0.24743756651878357,
"step": 170
},
{
"epoch": 0.29,
"learning_rate": 4.4840920917726425e-06,
"logits/chosen": 0.0785370022058487,
"logits/rejected": 0.09097695350646973,
"logps/chosen": -595.0328369140625,
"logps/rejected": -597.6621704101562,
"loss": 213793.725,
"rewards/accuracies": 0.6312500238418579,
"rewards/chosen": -0.28900542855262756,
"rewards/margins": 0.04301437363028526,
"rewards/rejected": -0.3320198357105255,
"step": 180
},
{
"epoch": 0.3,
"learning_rate": 4.396014096912182e-06,
"logits/chosen": -0.02692921832203865,
"logits/rejected": 0.01183820329606533,
"logps/chosen": -657.1026611328125,
"logps/rejected": -674.083251953125,
"loss": 217000.05,
"rewards/accuracies": 0.59375,
"rewards/chosen": -0.35720986127853394,
"rewards/margins": 0.06607996672391891,
"rewards/rejected": -0.42328983545303345,
"step": 190
},
{
"epoch": 0.32,
"learning_rate": 4.302012913171584e-06,
"logits/chosen": 0.016970274969935417,
"logits/rejected": 0.03241819143295288,
"logps/chosen": -740.4771728515625,
"logps/rejected": -757.164794921875,
"loss": 228484.45,
"rewards/accuracies": 0.5562499761581421,
"rewards/chosen": -0.3992530405521393,
"rewards/margins": 0.035844482481479645,
"rewards/rejected": -0.4350975453853607,
"step": 200
},
{
"epoch": 0.34,
"learning_rate": 4.202382202273702e-06,
"logits/chosen": 0.00015654190792702138,
"logits/rejected": 0.08966602385044098,
"logps/chosen": -675.322509765625,
"logps/rejected": -718.8179931640625,
"loss": 214730.75,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.3644675612449646,
"rewards/margins": 0.0624375119805336,
"rewards/rejected": -0.42690515518188477,
"step": 210
},
{
"epoch": 0.35,
"learning_rate": 4.097433212705492e-06,
"logits/chosen": -0.06447938829660416,
"logits/rejected": -0.055021267384290695,
"logps/chosen": -613.5958251953125,
"logps/rejected": -697.4694213867188,
"loss": 211407.05,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -0.3579150140285492,
"rewards/margins": 0.05921386927366257,
"rewards/rejected": -0.41712889075279236,
"step": 220
},
{
"epoch": 0.37,
"learning_rate": 3.987493807371033e-06,
"logits/chosen": -0.06768902391195297,
"logits/rejected": 0.0020426944829523563,
"logps/chosen": -682.5296020507812,
"logps/rejected": -710.4071655273438,
"loss": 207694.0,
"rewards/accuracies": 0.643750011920929,
"rewards/chosen": -0.3751247823238373,
"rewards/margins": 0.06657867878675461,
"rewards/rejected": -0.4417034983634949,
"step": 230
},
{
"epoch": 0.38,
"learning_rate": 3.872907439340758e-06,
"logits/chosen": 0.005040665622800589,
"logits/rejected": 0.02079974114894867,
"logps/chosen": -712.57763671875,
"logps/rejected": -737.4443969726562,
"loss": 220043.025,
"rewards/accuracies": 0.574999988079071,
"rewards/chosen": -0.404686838388443,
"rewards/margins": 0.051380474120378494,
"rewards/rejected": -0.4560673236846924,
"step": 240
},
{
"epoch": 0.4,
"learning_rate": 3.75403207889666e-06,
"logits/chosen": 0.003915914800018072,
"logits/rejected": 0.031082844361662865,
"logps/chosen": -672.8313598632812,
"logps/rejected": -687.6966552734375,
"loss": 207725.825,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.3670479953289032,
"rewards/margins": 0.065892294049263,
"rewards/rejected": -0.4329403042793274,
"step": 250
},
{
"epoch": 0.42,
"learning_rate": 3.631239095225417e-06,
"logits/chosen": 0.06261593103408813,
"logits/rejected": 0.13836300373077393,
"logps/chosen": -689.614501953125,
"logps/rejected": -736.5167236328125,
"loss": 216697.45,
"rewards/accuracies": 0.574999988079071,
"rewards/chosen": -0.4008885324001312,
"rewards/margins": 0.0602349154651165,
"rewards/rejected": -0.4611234664916992,
"step": 260
},
{
"epoch": 0.43,
"learning_rate": 3.5049120962530608e-06,
"logits/chosen": 0.11205191910266876,
"logits/rejected": 0.14602160453796387,
"logps/chosen": -735.8699951171875,
"logps/rejected": -832.4979248046875,
"loss": 206147.4,
"rewards/accuracies": 0.5874999761581421,
"rewards/chosen": -0.44775643944740295,
"rewards/margins": 0.07730598002672195,
"rewards/rejected": -0.5250624418258667,
"step": 270
},
{
"epoch": 0.45,
"learning_rate": 3.3754457302455464e-06,
"logits/chosen": 0.06379252672195435,
"logits/rejected": 0.08125611394643784,
"logps/chosen": -791.1683349609375,
"logps/rejected": -842.2650146484375,
"loss": 217755.6,
"rewards/accuracies": 0.5625,
"rewards/chosen": -0.5111533403396606,
"rewards/margins": 0.05805770307779312,
"rewards/rejected": -0.5692110061645508,
"step": 280
},
{
"epoch": 0.46,
"learning_rate": 3.2432444529190714e-06,
"logits/chosen": 0.03794580325484276,
"logits/rejected": 0.10185732692480087,
"logps/chosen": -766.0865478515625,
"logps/rejected": -786.9296875,
"loss": 205712.7375,
"rewards/accuracies": 0.643750011920929,
"rewards/chosen": -0.4407348036766052,
"rewards/margins": 0.07172349095344543,
"rewards/rejected": -0.5124582648277283,
"step": 290
},
{
"epoch": 0.48,
"learning_rate": 3.1087212639117057e-06,
"logits/chosen": 0.0560905747115612,
"logits/rejected": 0.056962113827466965,
"logps/chosen": -691.1881103515625,
"logps/rejected": -736.6007690429688,
"loss": 199775.475,
"rewards/accuracies": 0.606249988079071,
"rewards/chosen": -0.3910200595855713,
"rewards/margins": 0.06142463535070419,
"rewards/rejected": -0.4524447023868561,
"step": 300
},
{
"epoch": 0.5,
"learning_rate": 2.9722964165636263e-06,
"logits/chosen": 0.0340690053999424,
"logits/rejected": 0.10146850347518921,
"logps/chosen": -720.965087890625,
"logps/rejected": -739.9038696289062,
"loss": 217950.85,
"rewards/accuracies": 0.5375000238418579,
"rewards/chosen": -0.421273797750473,
"rewards/margins": 0.03930564597249031,
"rewards/rejected": -0.46057945489883423,
"step": 310
},
{
"epoch": 0.51,
"learning_rate": 2.8343961050366275e-06,
"logits/chosen": 0.054887332022190094,
"logits/rejected": 0.044551193714141846,
"logps/chosen": -735.8577880859375,
"logps/rejected": -789.2850952148438,
"loss": 218271.95,
"rewards/accuracies": 0.5562499761581421,
"rewards/chosen": -0.4242521822452545,
"rewards/margins": 0.05404456704854965,
"rewards/rejected": -0.47829669713974,
"step": 320
},
{
"epoch": 0.53,
"learning_rate": 2.695451132874385e-06,
"logits/chosen": 0.020503774285316467,
"logits/rejected": 0.08877702057361603,
"logps/chosen": -685.0552978515625,
"logps/rejected": -749.4927978515625,
"loss": 208541.825,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -0.4152056574821472,
"rewards/margins": 0.06724479049444199,
"rewards/rejected": -0.4824504256248474,
"step": 330
},
{
"epoch": 0.54,
"learning_rate": 2.5558955671628964e-06,
"logits/chosen": 0.03993639722466469,
"logits/rejected": 0.09598435461521149,
"logps/chosen": -687.928466796875,
"logps/rejected": -717.9952392578125,
"loss": 211084.3,
"rewards/accuracies": 0.612500011920929,
"rewards/chosen": -0.401594877243042,
"rewards/margins": 0.05332701653242111,
"rewards/rejected": -0.45492181181907654,
"step": 340
},
{
"epoch": 0.56,
"learning_rate": 2.4161653824955654e-06,
"logits/chosen": 0.06816108524799347,
"logits/rejected": 0.08216299116611481,
"logps/chosen": -716.1536865234375,
"logps/rejected": -749.4075927734375,
"loss": 205341.7,
"rewards/accuracies": 0.6312500238418579,
"rewards/chosen": -0.42900824546813965,
"rewards/margins": 0.06950432807207108,
"rewards/rejected": -0.4985125958919525,
"step": 350
},
{
"epoch": 0.58,
"learning_rate": 2.2766970989791697e-06,
"logits/chosen": 0.012849223800003529,
"logits/rejected": 0.02745252475142479,
"logps/chosen": -773.5811157226562,
"logps/rejected": -870.6488037109375,
"loss": 204889.025,
"rewards/accuracies": 0.6312500238418579,
"rewards/chosen": -0.4782876968383789,
"rewards/margins": 0.08239278197288513,
"rewards/rejected": -0.5606805086135864,
"step": 360
},
{
"epoch": 0.59,
"learning_rate": 2.1379264185356545e-06,
"logits/chosen": -0.01748507469892502,
"logits/rejected": 0.026378531008958817,
"logps/chosen": -870.5720825195312,
"logps/rejected": -918.7122192382812,
"loss": 221285.175,
"rewards/accuracies": 0.5625,
"rewards/chosen": -0.5346072316169739,
"rewards/margins": 0.046777307987213135,
"rewards/rejected": -0.581384539604187,
"step": 370
},
{
"epoch": 0.61,
"learning_rate": 2.000286863759934e-06,
"logits/chosen": 0.007047905120998621,
"logits/rejected": 0.08078267425298691,
"logps/chosen": -755.0501098632812,
"logps/rejected": -820.9991455078125,
"loss": 201255.875,
"rewards/accuracies": 0.5874999761581421,
"rewards/chosen": -0.45793017745018005,
"rewards/margins": 0.08488230407238007,
"rewards/rejected": -0.5428124666213989,
"step": 380
},
{
"epoch": 0.62,
"learning_rate": 1.8642084235859764e-06,
"logits/chosen": 0.011273056268692017,
"logits/rejected": -0.012196743860840797,
"logps/chosen": -764.2548828125,
"logps/rejected": -848.6171875,
"loss": 214622.65,
"rewards/accuracies": 0.574999988079071,
"rewards/chosen": -0.4602470397949219,
"rewards/margins": 0.07196761667728424,
"rewards/rejected": -0.5322146415710449,
"step": 390
},
{
"epoch": 0.64,
"learning_rate": 1.7301162099921013e-06,
"logits/chosen": -0.0019163116812705994,
"logits/rejected": 0.05834005028009415,
"logps/chosen": -858.6873168945312,
"logps/rejected": -904.6046142578125,
"loss": 206360.4125,
"rewards/accuracies": 0.675000011920929,
"rewards/chosen": -0.5314964652061462,
"rewards/margins": 0.08296509087085724,
"rewards/rejected": -0.6144616603851318,
"step": 400
},
{
"epoch": 0.66,
"learning_rate": 1.5984291299420117e-06,
"logits/chosen": 0.012182576581835747,
"logits/rejected": 0.0708267092704773,
"logps/chosen": -813.0255737304688,
"logps/rejected": -832.9349365234375,
"loss": 211531.0,
"rewards/accuracies": 0.6312500238418579,
"rewards/chosen": -0.49179133772850037,
"rewards/margins": 0.08578650653362274,
"rewards/rejected": -0.5775778293609619,
"step": 410
},
{
"epoch": 0.67,
"learning_rate": 1.4695585767104092e-06,
"logits/chosen": 0.0914839580655098,
"logits/rejected": 0.04479590803384781,
"logps/chosen": -832.50439453125,
"logps/rejected": -880.9533081054688,
"loss": 205329.4,
"rewards/accuracies": 0.606249988079071,
"rewards/chosen": -0.48841309547424316,
"rewards/margins": 0.07340176403522491,
"rewards/rejected": -0.5618147850036621,
"step": 420
},
{
"epoch": 0.69,
"learning_rate": 1.3439071446815452e-06,
"logits/chosen": 0.08182860910892487,
"logits/rejected": 0.100291907787323,
"logps/chosen": -752.081298828125,
"logps/rejected": -833.1486206054688,
"loss": 201124.725,
"rewards/accuracies": 0.6625000238418579,
"rewards/chosen": -0.45724090933799744,
"rewards/margins": 0.08672616630792618,
"rewards/rejected": -0.543967068195343,
"step": 430
},
{
"epoch": 0.7,
"learning_rate": 1.2218673716356919e-06,
"logits/chosen": 0.056249700486660004,
"logits/rejected": 0.04042553901672363,
"logps/chosen": -763.6502075195312,
"logps/rejected": -836.1224365234375,
"loss": 206556.925,
"rewards/accuracies": 0.581250011920929,
"rewards/chosen": -0.4586957097053528,
"rewards/margins": 0.05668143555521965,
"rewards/rejected": -0.5153771638870239,
"step": 440
},
{
"epoch": 0.72,
"learning_rate": 1.103820512452661e-06,
"logits/chosen": 0.058336324989795685,
"logits/rejected": 0.09516619145870209,
"logps/chosen": -799.2554931640625,
"logps/rejected": -850.5153198242188,
"loss": 200567.1625,
"rewards/accuracies": 0.6625000238418579,
"rewards/chosen": -0.45398440957069397,
"rewards/margins": 0.09574355185031891,
"rewards/rejected": -0.5497279763221741,
"step": 450
},
{
"epoch": 0.74,
"learning_rate": 9.901353480633468e-07,
"logits/chosen": 0.06983263790607452,
"logits/rejected": 0.10929594188928604,
"logps/chosen": -735.544189453125,
"logps/rejected": -794.2623291015625,
"loss": 208963.425,
"rewards/accuracies": 0.59375,
"rewards/chosen": -0.4254637360572815,
"rewards/margins": 0.08059998601675034,
"rewards/rejected": -0.5060637593269348,
"step": 460
},
{
"epoch": 0.75,
"learning_rate": 8.811670333701544e-07,
"logits/chosen": 0.035215824842453,
"logits/rejected": 0.10739400237798691,
"logps/chosen": -752.6171875,
"logps/rejected": -820.1688232421875,
"loss": 208260.5375,
"rewards/accuracies": 0.606249988079071,
"rewards/chosen": -0.4630301594734192,
"rewards/margins": 0.055133186280727386,
"rewards/rejected": -0.5181634426116943,
"step": 470
},
{
"epoch": 0.77,
"learning_rate": 7.772559877354341e-07,
"logits/chosen": 0.10807295143604279,
"logits/rejected": 0.0317668542265892,
"logps/chosen": -756.6256103515625,
"logps/rejected": -816.7699584960938,
"loss": 216200.35,
"rewards/accuracies": 0.612500011920929,
"rewards/chosen": -0.4670717716217041,
"rewards/margins": 0.07191431522369385,
"rewards/rejected": -0.538986086845398,
"step": 480
},
{
"epoch": 0.78,
"learning_rate": 6.787268315040604e-07,
"logits/chosen": 0.04728057235479355,
"logits/rejected": 0.15024212002754211,
"logps/chosen": -767.487060546875,
"logps/rejected": -839.1052856445312,
"loss": 199459.25,
"rewards/accuracies": 0.6625000238418579,
"rewards/chosen": -0.44638481736183167,
"rewards/margins": 0.08024169504642487,
"rewards/rejected": -0.526626467704773,
"step": 490
},
{
"epoch": 0.8,
"learning_rate": 5.858873718824829e-07,
"logits/chosen": 0.09359405189752579,
"logits/rejected": 0.10223015397787094,
"logps/chosen": -759.4922485351562,
"logps/rejected": -814.4105224609375,
"loss": 206686.275,
"rewards/accuracies": 0.612500011920929,
"rewards/chosen": -0.46266812086105347,
"rewards/margins": 0.0593070387840271,
"rewards/rejected": -0.5219752192497253,
"step": 500
},
{
"epoch": 0.82,
"learning_rate": 4.990276413423817e-07,
"logits/chosen": 0.01493439543992281,
"logits/rejected": 0.0643661767244339,
"logps/chosen": -735.6563720703125,
"logps/rejected": -793.5201416015625,
"loss": 198584.575,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.42836418747901917,
"rewards/margins": 0.08260954916477203,
"rewards/rejected": -0.5109737515449524,
"step": 510
},
{
"epoch": 0.83,
"learning_rate": 4.184189915529796e-07,
"logits/chosen": 0.09607640653848648,
"logits/rejected": 0.12576672434806824,
"logps/chosen": -765.7574462890625,
"logps/rejected": -835.0860595703125,
"loss": 198521.6875,
"rewards/accuracies": 0.6312500238418579,
"rewards/chosen": -0.45677047967910767,
"rewards/margins": 0.08099553734064102,
"rewards/rejected": -0.5377659797668457,
"step": 520
},
{
"epoch": 0.85,
"learning_rate": 3.4431324567258176e-07,
"logits/chosen": 0.03705473989248276,
"logits/rejected": 0.07135260850191116,
"logps/chosen": -806.9525146484375,
"logps/rejected": -851.9264526367188,
"loss": 200813.6625,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.4773276448249817,
"rewards/margins": 0.0656534880399704,
"rewards/rejected": -0.5429811477661133,
"step": 530
},
{
"epoch": 0.86,
"learning_rate": 2.769419116476052e-07,
"logits/chosen": 0.013349669054150581,
"logits/rejected": 0.04136626794934273,
"logps/chosen": -748.1133422851562,
"logps/rejected": -810.5701293945312,
"loss": 198242.825,
"rewards/accuracies": 0.6187499761581421,
"rewards/chosen": -0.4668392241001129,
"rewards/margins": 0.06354349106550217,
"rewards/rejected": -0.5303827524185181,
"step": 540
},
{
"epoch": 0.88,
"learning_rate": 2.1651545897676512e-07,
"logits/chosen": 0.06408552080392838,
"logits/rejected": 0.11870439350605011,
"logps/chosen": -763.1972045898438,
"logps/rejected": -839.8497314453125,
"loss": 211745.225,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": -0.4748295843601227,
"rewards/margins": 0.0728473886847496,
"rewards/rejected": -0.5476769208908081,
"step": 550
},
{
"epoch": 0.9,
"learning_rate": 1.6322266119983222e-07,
"logits/chosen": 0.0492372028529644,
"logits/rejected": 0.09988833963871002,
"logps/chosen": -789.6986083984375,
"logps/rejected": -825.2615966796875,
"loss": 211827.35,
"rewards/accuracies": 0.581250011920929,
"rewards/chosen": -0.4770093560218811,
"rewards/margins": 0.06789447367191315,
"rewards/rejected": -0.5449038743972778,
"step": 560
},
{
"epoch": 0.91,
"learning_rate": 1.1723000616502167e-07,
"logits/chosen": 0.06743012368679047,
"logits/rejected": 0.01723661459982395,
"logps/chosen": -782.1122436523438,
"logps/rejected": -836.3314208984375,
"loss": 205469.55,
"rewards/accuracies": 0.606249988079071,
"rewards/chosen": -0.47319063544273376,
"rewards/margins": 0.07614056020975113,
"rewards/rejected": -0.5493311882019043,
"step": 570
},
{
"epoch": 0.93,
"learning_rate": 7.868117591737585e-08,
"logits/chosen": 0.0372871533036232,
"logits/rejected": 0.07801343500614166,
"logps/chosen": -741.191650390625,
"logps/rejected": -787.1198120117188,
"loss": 213403.7,
"rewards/accuracies": 0.59375,
"rewards/chosen": -0.44429221749305725,
"rewards/margins": 0.07286076247692108,
"rewards/rejected": -0.5171529650688171,
"step": 580
},
{
"epoch": 0.94,
"learning_rate": 4.769659783295383e-08,
"logits/chosen": 0.07339660078287125,
"logits/rejected": 0.138061061501503,
"logps/chosen": -771.2107543945312,
"logps/rejected": -813.8905029296875,
"loss": 210919.15,
"rewards/accuracies": 0.6312500238418579,
"rewards/chosen": -0.47977980971336365,
"rewards/margins": 0.08075427263975143,
"rewards/rejected": -0.5605341196060181,
"step": 590
},
{
"epoch": 0.96,
"learning_rate": 2.4373068401120358e-08,
"logits/chosen": 0.11042066663503647,
"logits/rejected": 0.07605646550655365,
"logps/chosen": -791.4373779296875,
"logps/rejected": -826.2615356445312,
"loss": 216413.4,
"rewards/accuracies": 0.543749988079071,
"rewards/chosen": -0.49039316177368164,
"rewards/margins": 0.056141942739486694,
"rewards/rejected": -0.5465351343154907,
"step": 600
},
{
"epoch": 0.98,
"learning_rate": 8.78345083022425e-09,
"logits/chosen": 0.07623746991157532,
"logits/rejected": 0.1269301176071167,
"logps/chosen": -798.3727416992188,
"logps/rejected": -851.33740234375,
"loss": 209334.025,
"rewards/accuracies": 0.581250011920929,
"rewards/chosen": -0.49323853850364685,
"rewards/margins": 0.07322792708873749,
"rewards/rejected": -0.5664664506912231,
"step": 610
},
{
"epoch": 0.99,
"learning_rate": 9.764474213677654e-10,
"logits/chosen": 0.060989219695329666,
"logits/rejected": 0.09394808858633041,
"logps/chosen": -785.9288330078125,
"logps/rejected": -820.1795043945312,
"loss": 210348.9,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": -0.49476709961891174,
"rewards/margins": 0.057541172951459885,
"rewards/rejected": -0.5523082613945007,
"step": 620
},
{
"epoch": 1.0,
"step": 625,
"total_flos": 0.0,
"train_loss": 217680.7002,
"train_runtime": 7796.5816,
"train_samples_per_second": 2.565,
"train_steps_per_second": 0.08
}
],
"logging_steps": 10,
"max_steps": 625,
"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
}