phi-2-gpo-v18-i1 / trainer_state.json
lole25's picture
Model save
8214ff1 verified
raw
history blame
45.7 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.9994666666666666,
"eval_steps": 500,
"global_step": 937,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 5.319148936170213e-08,
"logits/chosen": 0.0328633114695549,
"logits/rejected": 0.22779616713523865,
"logps/chosen": -407.2765197753906,
"logps/rejected": -295.7368469238281,
"loss": 0.4178,
"rewards/accuracies": 0.0,
"rewards/chosen": 0.0,
"rewards/margins": 0.0,
"rewards/rejected": 0.0,
"step": 1
},
{
"epoch": 0.01,
"learning_rate": 5.319148936170213e-07,
"logits/chosen": 0.12644250690937042,
"logits/rejected": 0.25490710139274597,
"logps/chosen": -342.5318603515625,
"logps/rejected": -339.8170471191406,
"loss": 0.3561,
"rewards/accuracies": 0.4305555522441864,
"rewards/chosen": 2.7654659788822755e-05,
"rewards/margins": 8.612655801698565e-05,
"rewards/rejected": -5.84718982281629e-05,
"step": 10
},
{
"epoch": 0.02,
"learning_rate": 1.0638297872340427e-06,
"logits/chosen": 0.11099036037921906,
"logits/rejected": 0.24831095337867737,
"logps/chosen": -378.44439697265625,
"logps/rejected": -353.71014404296875,
"loss": 0.3639,
"rewards/accuracies": 0.4000000059604645,
"rewards/chosen": -4.667802568292245e-05,
"rewards/margins": -9.378789400216192e-05,
"rewards/rejected": 4.710986831923947e-05,
"step": 20
},
{
"epoch": 0.03,
"learning_rate": 1.595744680851064e-06,
"logits/chosen": 0.13326099514961243,
"logits/rejected": 0.1938765048980713,
"logps/chosen": -396.07440185546875,
"logps/rejected": -353.0478210449219,
"loss": 0.3768,
"rewards/accuracies": 0.518750011920929,
"rewards/chosen": 4.021604763693176e-05,
"rewards/margins": -1.0102905434905551e-05,
"rewards/rejected": 5.0318958528805524e-05,
"step": 30
},
{
"epoch": 0.04,
"learning_rate": 2.1276595744680853e-06,
"logits/chosen": 0.15499012172222137,
"logits/rejected": 0.24013516306877136,
"logps/chosen": -357.16259765625,
"logps/rejected": -330.24609375,
"loss": 0.3838,
"rewards/accuracies": 0.46875,
"rewards/chosen": 5.2471255912678316e-05,
"rewards/margins": 0.00012843578588217497,
"rewards/rejected": -7.596451177960262e-05,
"step": 40
},
{
"epoch": 0.05,
"learning_rate": 2.6595744680851065e-06,
"logits/chosen": 0.08975875377655029,
"logits/rejected": 0.1924462616443634,
"logps/chosen": -360.7787780761719,
"logps/rejected": -345.1267395019531,
"loss": 0.3709,
"rewards/accuracies": 0.5562499761581421,
"rewards/chosen": 0.00022406328935176134,
"rewards/margins": 0.0003413105441723019,
"rewards/rejected": -0.00011724724026862532,
"step": 50
},
{
"epoch": 0.06,
"learning_rate": 3.191489361702128e-06,
"logits/chosen": 0.11686824262142181,
"logits/rejected": 0.2060682326555252,
"logps/chosen": -301.4020690917969,
"logps/rejected": -306.5481872558594,
"loss": 0.379,
"rewards/accuracies": 0.6625000238418579,
"rewards/chosen": 0.0006540452013723552,
"rewards/margins": 0.001110218232497573,
"rewards/rejected": -0.00045617311843670905,
"step": 60
},
{
"epoch": 0.07,
"learning_rate": 3.723404255319149e-06,
"logits/chosen": 0.061485785990953445,
"logits/rejected": 0.14367519319057465,
"logps/chosen": -309.8575134277344,
"logps/rejected": -319.0173645019531,
"loss": 0.3678,
"rewards/accuracies": 0.65625,
"rewards/chosen": 0.000993106747046113,
"rewards/margins": 0.0025861565954983234,
"rewards/rejected": -0.001593050197698176,
"step": 70
},
{
"epoch": 0.09,
"learning_rate": 4.255319148936171e-06,
"logits/chosen": 0.08192186057567596,
"logits/rejected": 0.18210968375205994,
"logps/chosen": -313.7688293457031,
"logps/rejected": -354.7149658203125,
"loss": 0.3692,
"rewards/accuracies": 0.7124999761581421,
"rewards/chosen": 0.0009643015218898654,
"rewards/margins": 0.005697316490113735,
"rewards/rejected": -0.004733014386147261,
"step": 80
},
{
"epoch": 0.1,
"learning_rate": 4.787234042553192e-06,
"logits/chosen": 0.11128780990839005,
"logits/rejected": 0.16921164095401764,
"logps/chosen": -364.577392578125,
"logps/rejected": -352.7664489746094,
"loss": 0.372,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.002881403313949704,
"rewards/margins": 0.005687831435352564,
"rewards/rejected": -0.008569234982132912,
"step": 90
},
{
"epoch": 0.11,
"learning_rate": 4.999375059004058e-06,
"logits/chosen": 0.08404798805713654,
"logits/rejected": 0.17940422892570496,
"logps/chosen": -375.03900146484375,
"logps/rejected": -333.00775146484375,
"loss": 0.3432,
"rewards/accuracies": 0.65625,
"rewards/chosen": -0.0073157488368451595,
"rewards/margins": 0.015402473509311676,
"rewards/rejected": -0.022718224674463272,
"step": 100
},
{
"epoch": 0.12,
"learning_rate": 4.9955571065548795e-06,
"logits/chosen": 0.0811820849776268,
"logits/rejected": 0.12991295754909515,
"logps/chosen": -374.31103515625,
"logps/rejected": -350.5423278808594,
"loss": 0.3508,
"rewards/accuracies": 0.6312500238418579,
"rewards/chosen": -0.02395443804562092,
"rewards/margins": 0.01998896151781082,
"rewards/rejected": -0.04394339770078659,
"step": 110
},
{
"epoch": 0.13,
"learning_rate": 4.9882736864879e-06,
"logits/chosen": 0.07552331686019897,
"logits/rejected": 0.10529439151287079,
"logps/chosen": -385.13165283203125,
"logps/rejected": -430.96142578125,
"loss": 0.3218,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": -0.06972260773181915,
"rewards/margins": 0.05090062692761421,
"rewards/rejected": -0.12062324583530426,
"step": 120
},
{
"epoch": 0.14,
"learning_rate": 4.977534912960124e-06,
"logits/chosen": 0.1114310622215271,
"logits/rejected": 0.16347981989383698,
"logps/chosen": -484.64544677734375,
"logps/rejected": -562.412109375,
"loss": 0.3292,
"rewards/accuracies": 0.6875,
"rewards/chosen": -0.13593843579292297,
"rewards/margins": 0.09484948217868805,
"rewards/rejected": -0.23078791797161102,
"step": 130
},
{
"epoch": 0.15,
"learning_rate": 4.963355698422092e-06,
"logits/chosen": 0.12350474298000336,
"logits/rejected": 0.17375971376895905,
"logps/chosen": -554.6981201171875,
"logps/rejected": -652.0894775390625,
"loss": 0.3099,
"rewards/accuracies": 0.65625,
"rewards/chosen": -0.20147164165973663,
"rewards/margins": 0.13413165509700775,
"rewards/rejected": -0.3356032967567444,
"step": 140
},
{
"epoch": 0.16,
"learning_rate": 4.945755732909625e-06,
"logits/chosen": 0.0980851799249649,
"logits/rejected": 0.21910460293293,
"logps/chosen": -511.48046875,
"logps/rejected": -560.3663940429688,
"loss": 0.3019,
"rewards/accuracies": 0.581250011920929,
"rewards/chosen": -0.1676788479089737,
"rewards/margins": 0.08112357556819916,
"rewards/rejected": -0.24880242347717285,
"step": 150
},
{
"epoch": 0.17,
"learning_rate": 4.924759456701167e-06,
"logits/chosen": 0.06849007308483124,
"logits/rejected": 0.16746065020561218,
"logps/chosen": -522.7907104492188,
"logps/rejected": -604.654541015625,
"loss": 0.2982,
"rewards/accuracies": 0.675000011920929,
"rewards/chosen": -0.16378344595432281,
"rewards/margins": 0.10557951033115387,
"rewards/rejected": -0.2693629264831543,
"step": 160
},
{
"epoch": 0.18,
"learning_rate": 4.900396026378671e-06,
"logits/chosen": 0.09706642478704453,
"logits/rejected": 0.22275814414024353,
"logps/chosen": -541.6332397460938,
"logps/rejected": -655.224609375,
"loss": 0.2733,
"rewards/accuracies": 0.768750011920929,
"rewards/chosen": -0.1621033251285553,
"rewards/margins": 0.12445729970932007,
"rewards/rejected": -0.286560595035553,
"step": 170
},
{
"epoch": 0.19,
"learning_rate": 4.872699274339169e-06,
"logits/chosen": 0.14361754059791565,
"logits/rejected": 0.1780318021774292,
"logps/chosen": -505.50262451171875,
"logps/rejected": -618.1754760742188,
"loss": 0.2927,
"rewards/accuracies": 0.71875,
"rewards/chosen": -0.15032632648944855,
"rewards/margins": 0.1392049789428711,
"rewards/rejected": -0.28953129053115845,
"step": 180
},
{
"epoch": 0.2,
"learning_rate": 4.8417076618132434e-06,
"logits/chosen": 0.07448507845401764,
"logits/rejected": 0.11850146949291229,
"logps/chosen": -497.01702880859375,
"logps/rejected": -640.9612426757812,
"loss": 0.2788,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.1675926297903061,
"rewards/margins": 0.15780577063560486,
"rewards/rejected": -0.32539841532707214,
"step": 190
},
{
"epoch": 0.21,
"learning_rate": 4.807464225455655e-06,
"logits/chosen": 0.1496686488389969,
"logits/rejected": 0.15641534328460693,
"logps/chosen": -477.30438232421875,
"logps/rejected": -642.1568603515625,
"loss": 0.2886,
"rewards/accuracies": 0.6875,
"rewards/chosen": -0.18345065414905548,
"rewards/margins": 0.14235511422157288,
"rewards/rejected": -0.32580578327178955,
"step": 200
},
{
"epoch": 0.22,
"learning_rate": 4.770016517582283e-06,
"logits/chosen": 0.029586156830191612,
"logits/rejected": 0.10910207033157349,
"logps/chosen": -545.3736572265625,
"logps/rejected": -619.1397705078125,
"loss": 0.3316,
"rewards/accuracies": 0.6625000238418579,
"rewards/chosen": -0.2202235460281372,
"rewards/margins": 0.1027408093214035,
"rewards/rejected": -0.3229643702507019,
"step": 210
},
{
"epoch": 0.23,
"learning_rate": 4.7294165401363616e-06,
"logits/chosen": 0.07184920459985733,
"logits/rejected": 0.10738018900156021,
"logps/chosen": -551.0501708984375,
"logps/rejected": -633.4860229492188,
"loss": 0.2955,
"rewards/accuracies": 0.643750011920929,
"rewards/chosen": -0.21274058520793915,
"rewards/margins": 0.10448522865772247,
"rewards/rejected": -0.3172258138656616,
"step": 220
},
{
"epoch": 0.25,
"learning_rate": 4.68572067247573e-06,
"logits/chosen": 0.07210972160100937,
"logits/rejected": 0.15474052727222443,
"logps/chosen": -530.9338989257812,
"logps/rejected": -641.7281494140625,
"loss": 0.2748,
"rewards/accuracies": 0.731249988079071,
"rewards/chosen": -0.20435886085033417,
"rewards/margins": 0.1295766979455948,
"rewards/rejected": -0.33393555879592896,
"step": 230
},
{
"epoch": 0.26,
"learning_rate": 4.638989593081364e-06,
"logits/chosen": 0.1024290919303894,
"logits/rejected": 0.1865706890821457,
"logps/chosen": -584.1736450195312,
"logps/rejected": -625.4171142578125,
"loss": 0.309,
"rewards/accuracies": 0.643750011920929,
"rewards/chosen": -0.20201639831066132,
"rewards/margins": 0.09904564172029495,
"rewards/rejected": -0.30106204748153687,
"step": 240
},
{
"epoch": 0.27,
"learning_rate": 4.5892881952959015e-06,
"logits/chosen": 0.02927372418344021,
"logits/rejected": 0.09811622649431229,
"logps/chosen": -533.7764892578125,
"logps/rejected": -629.5748901367188,
"loss": 0.2861,
"rewards/accuracies": 0.6875,
"rewards/chosen": -0.18896381556987762,
"rewards/margins": 0.11862628161907196,
"rewards/rejected": -0.3075900971889496,
"step": 250
},
{
"epoch": 0.28,
"learning_rate": 4.536685497209182e-06,
"logits/chosen": -0.005825365893542767,
"logits/rejected": 0.09718175232410431,
"logps/chosen": -493.98675537109375,
"logps/rejected": -583.2804565429688,
"loss": 0.2896,
"rewards/accuracies": 0.65625,
"rewards/chosen": -0.20344126224517822,
"rewards/margins": 0.09909246116876602,
"rewards/rejected": -0.30253371596336365,
"step": 260
},
{
"epoch": 0.29,
"learning_rate": 4.481254545815943e-06,
"logits/chosen": 0.044513873755931854,
"logits/rejected": 0.025778833776712418,
"logps/chosen": -558.865966796875,
"logps/rejected": -615.7289428710938,
"loss": 0.2943,
"rewards/accuracies": 0.6937500238418579,
"rewards/chosen": -0.2074216902256012,
"rewards/margins": 0.11120656877756119,
"rewards/rejected": -0.3186282515525818,
"step": 270
},
{
"epoch": 0.3,
"learning_rate": 4.42307231557875e-06,
"logits/chosen": -0.028864026069641113,
"logits/rejected": 0.07541537284851074,
"logps/chosen": -592.0741577148438,
"logps/rejected": -668.5736083984375,
"loss": 0.3057,
"rewards/accuracies": 0.65625,
"rewards/chosen": -0.21832111477851868,
"rewards/margins": 0.11492238938808441,
"rewards/rejected": -0.3332434892654419,
"step": 280
},
{
"epoch": 0.31,
"learning_rate": 4.3622196015370305e-06,
"logits/chosen": 0.0018409714102745056,
"logits/rejected": 0.11235576868057251,
"logps/chosen": -553.7425537109375,
"logps/rejected": -655.2655639648438,
"loss": 0.3027,
"rewards/accuracies": 0.6625000238418579,
"rewards/chosen": -0.20998337864875793,
"rewards/margins": 0.11225197464227676,
"rewards/rejected": -0.3222353756427765,
"step": 290
},
{
"epoch": 0.32,
"learning_rate": 4.298780907110648e-06,
"logits/chosen": 0.02845548465847969,
"logits/rejected": 0.060444425791502,
"logps/chosen": -538.5903930664062,
"logps/rejected": -638.1737060546875,
"loss": 0.2801,
"rewards/accuracies": 0.6625000238418579,
"rewards/chosen": -0.21026530861854553,
"rewards/margins": 0.11142082512378693,
"rewards/rejected": -0.32168611884117126,
"step": 300
},
{
"epoch": 0.33,
"learning_rate": 4.23284432675381e-06,
"logits/chosen": 0.03200574219226837,
"logits/rejected": 0.09657691419124603,
"logps/chosen": -566.3619995117188,
"logps/rejected": -676.0072631835938,
"loss": 0.2852,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": -0.209668830037117,
"rewards/margins": 0.10983745008707047,
"rewards/rejected": -0.3195062577724457,
"step": 310
},
{
"epoch": 0.34,
"learning_rate": 4.164501423622277e-06,
"logits/chosen": 0.009218290448188782,
"logits/rejected": 0.07068648189306259,
"logps/chosen": -493.89532470703125,
"logps/rejected": -595.8672485351562,
"loss": 0.2807,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": -0.19013245403766632,
"rewards/margins": 0.11394868791103363,
"rewards/rejected": -0.30408114194869995,
"step": 320
},
{
"epoch": 0.35,
"learning_rate": 4.0938471024237355e-06,
"logits/chosen": 0.040950141847133636,
"logits/rejected": 0.15327873826026917,
"logps/chosen": -557.1869506835938,
"logps/rejected": -644.422119140625,
"loss": 0.286,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": -0.21056047081947327,
"rewards/margins": 0.11874810606241226,
"rewards/rejected": -0.32930856943130493,
"step": 330
},
{
"epoch": 0.36,
"learning_rate": 4.020979477627907e-06,
"logits/chosen": 0.05679699778556824,
"logits/rejected": 0.10968853533267975,
"logps/chosen": -576.4144287109375,
"logps/rejected": -666.2213745117188,
"loss": 0.2967,
"rewards/accuracies": 0.6937500238418579,
"rewards/chosen": -0.23255427181720734,
"rewards/margins": 0.12515048682689667,
"rewards/rejected": -0.3577048182487488,
"step": 340
},
{
"epoch": 0.37,
"learning_rate": 3.9459997372194105e-06,
"logits/chosen": 0.06001776456832886,
"logits/rejected": 0.1731962114572525,
"logps/chosen": -578.7145385742188,
"logps/rejected": -695.6481323242188,
"loss": 0.2869,
"rewards/accuracies": 0.6937500238418579,
"rewards/chosen": -0.23581092059612274,
"rewards/margins": 0.13799193501472473,
"rewards/rejected": -0.37380287051200867,
"step": 350
},
{
"epoch": 0.38,
"learning_rate": 3.869012002182573e-06,
"logits/chosen": 0.12720148265361786,
"logits/rejected": 0.11129485070705414,
"logps/chosen": -585.1661376953125,
"logps/rejected": -674.7999877929688,
"loss": 0.2751,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": -0.24305827915668488,
"rewards/margins": 0.1303601711988449,
"rewards/rejected": -0.3734184205532074,
"step": 360
},
{
"epoch": 0.39,
"learning_rate": 3.7901231819133104e-06,
"logits/chosen": 0.046785108745098114,
"logits/rejected": 0.12497253715991974,
"logps/chosen": -613.0939331054688,
"logps/rejected": -778.9556274414062,
"loss": 0.2816,
"rewards/accuracies": 0.6937500238418579,
"rewards/chosen": -0.26795250177383423,
"rewards/margins": 0.17298182845115662,
"rewards/rejected": -0.44093433022499084,
"step": 370
},
{
"epoch": 0.41,
"learning_rate": 3.709442825758875e-06,
"logits/chosen": -0.016536608338356018,
"logits/rejected": 0.08203241974115372,
"logps/chosen": -619.8163452148438,
"logps/rejected": -756.1831665039062,
"loss": 0.2809,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": -0.28122490644454956,
"rewards/margins": 0.1667798012495041,
"rewards/rejected": -0.44800466299057007,
"step": 380
},
{
"epoch": 0.42,
"learning_rate": 3.6270829708916113e-06,
"logits/chosen": -0.013438811525702477,
"logits/rejected": 0.056597866117954254,
"logps/chosen": -660.5250244140625,
"logps/rejected": -809.8643798828125,
"loss": 0.2683,
"rewards/accuracies": 0.7437499761581421,
"rewards/chosen": -0.30733537673950195,
"rewards/margins": 0.1618378460407257,
"rewards/rejected": -0.46917325258255005,
"step": 390
},
{
"epoch": 0.43,
"learning_rate": 3.543157986727991e-06,
"logits/chosen": 0.033776335418224335,
"logits/rejected": 0.08476515114307404,
"logps/chosen": -612.4915161132812,
"logps/rejected": -791.068603515625,
"loss": 0.2687,
"rewards/accuracies": 0.6812499761581421,
"rewards/chosen": -0.298635333776474,
"rewards/margins": 0.14923743903636932,
"rewards/rejected": -0.4478727877140045,
"step": 400
},
{
"epoch": 0.44,
"learning_rate": 3.4577844161089614e-06,
"logits/chosen": 0.04248402267694473,
"logits/rejected": 0.13372144103050232,
"logps/chosen": -653.1471557617188,
"logps/rejected": -754.6251831054688,
"loss": 0.2659,
"rewards/accuracies": 0.706250011920929,
"rewards/chosen": -0.30667322874069214,
"rewards/margins": 0.1425372064113617,
"rewards/rejected": -0.4492104649543762,
"step": 410
},
{
"epoch": 0.45,
"learning_rate": 3.3710808134621577e-06,
"logits/chosen": 0.016178572550415993,
"logits/rejected": 0.09873557835817337,
"logps/chosen": -632.523193359375,
"logps/rejected": -767.3197021484375,
"loss": 0.2715,
"rewards/accuracies": 0.7124999761581421,
"rewards/chosen": -0.28947144746780396,
"rewards/margins": 0.1752663552761078,
"rewards/rejected": -0.46473783254623413,
"step": 420
},
{
"epoch": 0.46,
"learning_rate": 3.2831675801707126e-06,
"logits/chosen": 0.010262581519782543,
"logits/rejected": 0.09065081179141998,
"logps/chosen": -607.0426025390625,
"logps/rejected": -816.6497192382812,
"loss": 0.2543,
"rewards/accuracies": 0.7437499761581421,
"rewards/chosen": -0.3082737326622009,
"rewards/margins": 0.19562499225139618,
"rewards/rejected": -0.5038987398147583,
"step": 430
},
{
"epoch": 0.47,
"learning_rate": 3.194166797377289e-06,
"logits/chosen": 0.03939592093229294,
"logits/rejected": 0.051005780696868896,
"logps/chosen": -630.1431884765625,
"logps/rejected": -726.6351318359375,
"loss": 0.2793,
"rewards/accuracies": 0.65625,
"rewards/chosen": -0.28141292929649353,
"rewards/margins": 0.13616052269935608,
"rewards/rejected": -0.4175735116004944,
"step": 440
},
{
"epoch": 0.48,
"learning_rate": 3.104202056455501e-06,
"logits/chosen": 0.03237801790237427,
"logits/rejected": 0.128599613904953,
"logps/chosen": -699.2102661132812,
"logps/rejected": -776.4459228515625,
"loss": 0.2784,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": -0.2957838177680969,
"rewards/margins": 0.1545422077178955,
"rewards/rejected": -0.4503260552883148,
"step": 450
},
{
"epoch": 0.49,
"learning_rate": 3.013398287384144e-06,
"logits/chosen": -0.011108310893177986,
"logits/rejected": 0.0545300655066967,
"logps/chosen": -605.3892822265625,
"logps/rejected": -729.7080078125,
"loss": 0.3022,
"rewards/accuracies": 0.6875,
"rewards/chosen": -0.27824196219444275,
"rewards/margins": 0.1415918618440628,
"rewards/rejected": -0.41983380913734436,
"step": 460
},
{
"epoch": 0.5,
"learning_rate": 2.9218815852625717e-06,
"logits/chosen": -0.010739525780081749,
"logits/rejected": 0.1061888337135315,
"logps/chosen": -675.0782470703125,
"logps/rejected": -811.0116577148438,
"loss": 0.2543,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.32911890745162964,
"rewards/margins": 0.14946062862873077,
"rewards/rejected": -0.4785795211791992,
"step": 470
},
{
"epoch": 0.51,
"learning_rate": 2.829779035208113e-06,
"logits/chosen": -0.007434132508933544,
"logits/rejected": 0.005777590908110142,
"logps/chosen": -675.082275390625,
"logps/rejected": -852.5006103515625,
"loss": 0.2821,
"rewards/accuracies": 0.7437499761581421,
"rewards/chosen": -0.3303678631782532,
"rewards/margins": 0.17629772424697876,
"rewards/rejected": -0.5066655874252319,
"step": 480
},
{
"epoch": 0.52,
"learning_rate": 2.737218535878705e-06,
"logits/chosen": -0.010230760090053082,
"logits/rejected": 0.06991882622241974,
"logps/chosen": -657.92919921875,
"logps/rejected": -769.2183227539062,
"loss": 0.2732,
"rewards/accuracies": 0.675000011920929,
"rewards/chosen": -0.29096969962120056,
"rewards/margins": 0.1650645136833191,
"rewards/rejected": -0.45603418350219727,
"step": 490
},
{
"epoch": 0.53,
"learning_rate": 2.64432862186579e-06,
"logits/chosen": 0.010941145941615105,
"logits/rejected": 0.06464840471744537,
"logps/chosen": -660.7554931640625,
"logps/rejected": -808.1925048828125,
"loss": 0.2525,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": -0.3170284628868103,
"rewards/margins": 0.16747084259986877,
"rewards/rejected": -0.4844993054866791,
"step": 500
},
{
"epoch": 0.54,
"learning_rate": 2.551238285204126e-06,
"logits/chosen": 0.01938098482787609,
"logits/rejected": 0.10472504049539566,
"logps/chosen": -665.8427734375,
"logps/rejected": -784.5371704101562,
"loss": 0.2657,
"rewards/accuracies": 0.675000011920929,
"rewards/chosen": -0.29728442430496216,
"rewards/margins": 0.14796006679534912,
"rewards/rejected": -0.4452444911003113,
"step": 510
},
{
"epoch": 0.55,
"learning_rate": 2.4580767962463688e-06,
"logits/chosen": 0.014335828833281994,
"logits/rejected": 0.10104580968618393,
"logps/chosen": -667.510986328125,
"logps/rejected": -799.0164794921875,
"loss": 0.2633,
"rewards/accuracies": 0.706250011920929,
"rewards/chosen": -0.30110082030296326,
"rewards/margins": 0.16439731419086456,
"rewards/rejected": -0.4654981195926666,
"step": 520
},
{
"epoch": 0.57,
"learning_rate": 2.3649735241511546e-06,
"logits/chosen": 0.028812021017074585,
"logits/rejected": 0.09060634672641754,
"logps/chosen": -620.7320556640625,
"logps/rejected": -778.6414184570312,
"loss": 0.2401,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": -0.2825697362422943,
"rewards/margins": 0.1802894026041031,
"rewards/rejected": -0.4628591537475586,
"step": 530
},
{
"epoch": 0.58,
"learning_rate": 2.2720577572339914e-06,
"logits/chosen": -0.015088727697730064,
"logits/rejected": 0.08517689257860184,
"logps/chosen": -646.2489013671875,
"logps/rejected": -777.7031860351562,
"loss": 0.261,
"rewards/accuracies": 0.71875,
"rewards/chosen": -0.2979923486709595,
"rewards/margins": 0.16907855868339539,
"rewards/rejected": -0.46707090735435486,
"step": 540
},
{
"epoch": 0.59,
"learning_rate": 2.1794585234303995e-06,
"logits/chosen": 0.0029529170133173466,
"logits/rejected": 0.08159308135509491,
"logps/chosen": -626.0859985351562,
"logps/rejected": -760.864990234375,
"loss": 0.2519,
"rewards/accuracies": 0.6937500238418579,
"rewards/chosen": -0.2840486168861389,
"rewards/margins": 0.16967108845710754,
"rewards/rejected": -0.4537196755409241,
"step": 550
},
{
"epoch": 0.6,
"learning_rate": 2.0873044111206407e-06,
"logits/chosen": 0.04737887904047966,
"logits/rejected": 0.09217114746570587,
"logps/chosen": -670.5003662109375,
"logps/rejected": -781.8992309570312,
"loss": 0.2568,
"rewards/accuracies": 0.668749988079071,
"rewards/chosen": -0.3173016309738159,
"rewards/margins": 0.15325884521007538,
"rewards/rejected": -0.4705604612827301,
"step": 560
},
{
"epoch": 0.61,
"learning_rate": 1.9957233905648293e-06,
"logits/chosen": 0.05551145225763321,
"logits/rejected": 0.07084596157073975,
"logps/chosen": -708.875732421875,
"logps/rejected": -875.9669189453125,
"loss": 0.2604,
"rewards/accuracies": 0.7437499761581421,
"rewards/chosen": -0.3540639579296112,
"rewards/margins": 0.1768617331981659,
"rewards/rejected": -0.5309256315231323,
"step": 570
},
{
"epoch": 0.62,
"learning_rate": 1.904842636196402e-06,
"logits/chosen": 0.05049177259206772,
"logits/rejected": 0.09144983440637589,
"logps/chosen": -662.368896484375,
"logps/rejected": -762.068359375,
"loss": 0.2818,
"rewards/accuracies": 0.699999988079071,
"rewards/chosen": -0.2937859892845154,
"rewards/margins": 0.15120849013328552,
"rewards/rejected": -0.4449945092201233,
"step": 580
},
{
"epoch": 0.63,
"learning_rate": 1.814788350020726e-06,
"logits/chosen": 0.013850146904587746,
"logits/rejected": 0.08181043714284897,
"logps/chosen": -655.6898193359375,
"logps/rejected": -799.6488037109375,
"loss": 0.2732,
"rewards/accuracies": 0.6875,
"rewards/chosen": -0.2904633581638336,
"rewards/margins": 0.16650694608688354,
"rewards/rejected": -0.45697030425071716,
"step": 590
},
{
"epoch": 0.64,
"learning_rate": 1.725685586364051e-06,
"logits/chosen": 0.026665305718779564,
"logits/rejected": 0.10492020845413208,
"logps/chosen": -667.7979736328125,
"logps/rejected": -860.6505737304688,
"loss": 0.2529,
"rewards/accuracies": 0.7124999761581421,
"rewards/chosen": -0.2826516032218933,
"rewards/margins": 0.18477080762386322,
"rewards/rejected": -0.4674224257469177,
"step": 600
},
{
"epoch": 0.65,
"learning_rate": 1.6376580782162172e-06,
"logits/chosen": 0.04557771235704422,
"logits/rejected": 0.06558578461408615,
"logps/chosen": -630.7008056640625,
"logps/rejected": -797.5521240234375,
"loss": 0.2553,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.29132452607154846,
"rewards/margins": 0.18406057357788086,
"rewards/rejected": -0.4753850996494293,
"step": 610
},
{
"epoch": 0.66,
"learning_rate": 1.550828065408227e-06,
"logits/chosen": 0.0122049106284976,
"logits/rejected": 0.05510292202234268,
"logps/chosen": -702.2847900390625,
"logps/rejected": -777.2984619140625,
"loss": 0.2793,
"rewards/accuracies": 0.6875,
"rewards/chosen": -0.32928770780563354,
"rewards/margins": 0.15212389826774597,
"rewards/rejected": -0.4814116060733795,
"step": 620
},
{
"epoch": 0.67,
"learning_rate": 1.4653161248633053e-06,
"logits/chosen": 0.027809197083115578,
"logits/rejected": 0.12329945713281631,
"logps/chosen": -693.8055419921875,
"logps/rejected": -849.8357543945312,
"loss": 0.2712,
"rewards/accuracies": 0.6875,
"rewards/chosen": -0.3381686210632324,
"rewards/margins": 0.15602417290210724,
"rewards/rejected": -0.4941927492618561,
"step": 630
},
{
"epoch": 0.68,
"learning_rate": 1.381241003157162e-06,
"logits/chosen": 0.018910501152276993,
"logits/rejected": 0.06395964324474335,
"logps/chosen": -668.9143676757812,
"logps/rejected": -784.6621704101562,
"loss": 0.304,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": -0.3322431445121765,
"rewards/margins": 0.13410837948322296,
"rewards/rejected": -0.4663515090942383,
"step": 640
},
{
"epoch": 0.69,
"learning_rate": 1.298719451619979e-06,
"logits/chosen": 0.032313134521245956,
"logits/rejected": 0.09517451375722885,
"logps/chosen": -650.8678588867188,
"logps/rejected": -775.7073974609375,
"loss": 0.2671,
"rewards/accuracies": 0.6875,
"rewards/chosen": -0.302242636680603,
"rewards/margins": 0.15537433326244354,
"rewards/rejected": -0.45761698484420776,
"step": 650
},
{
"epoch": 0.7,
"learning_rate": 1.2178660642091036e-06,
"logits/chosen": -0.0034284330904483795,
"logits/rejected": 0.08462049067020416,
"logps/chosen": -646.4454345703125,
"logps/rejected": -803.33056640625,
"loss": 0.2708,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": -0.3094111979007721,
"rewards/margins": 0.16766026616096497,
"rewards/rejected": -0.47707146406173706,
"step": 660
},
{
"epoch": 0.71,
"learning_rate": 1.1387931183775821e-06,
"logits/chosen": 0.01865808106958866,
"logits/rejected": 0.07240469753742218,
"logps/chosen": -656.0631103515625,
"logps/rejected": -819.2371826171875,
"loss": 0.2438,
"rewards/accuracies": 0.706250011920929,
"rewards/chosen": -0.31997841596603394,
"rewards/margins": 0.1749737411737442,
"rewards/rejected": -0.49495214223861694,
"step": 670
},
{
"epoch": 0.73,
"learning_rate": 1.061610419159532e-06,
"logits/chosen": 0.007644282188266516,
"logits/rejected": 0.12147297710180283,
"logps/chosen": -666.0787353515625,
"logps/rejected": -814.1128540039062,
"loss": 0.27,
"rewards/accuracies": 0.7250000238418579,
"rewards/chosen": -0.31804323196411133,
"rewards/margins": 0.1811751127243042,
"rewards/rejected": -0.49921831488609314,
"step": 680
},
{
"epoch": 0.74,
"learning_rate": 9.864251466888364e-07,
"logits/chosen": 0.002938181161880493,
"logits/rejected": 0.05325765162706375,
"logps/chosen": -659.8150634765625,
"logps/rejected": -826.27294921875,
"loss": 0.2679,
"rewards/accuracies": 0.706250011920929,
"rewards/chosen": -0.33443090319633484,
"rewards/margins": 0.17140033841133118,
"rewards/rejected": -0.505831241607666,
"step": 690
},
{
"epoch": 0.75,
"learning_rate": 9.133417073629288e-07,
"logits/chosen": 0.009498851373791695,
"logits/rejected": 0.10675854980945587,
"logps/chosen": -641.385986328125,
"logps/rejected": -772.0265502929688,
"loss": 0.2745,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.33288660645484924,
"rewards/margins": 0.14625254273414612,
"rewards/rejected": -0.47913914918899536,
"step": 700
},
{
"epoch": 0.76,
"learning_rate": 8.424615888583332e-07,
"logits/chosen": 0.01779480278491974,
"logits/rejected": 0.06852717697620392,
"logps/chosen": -658.0755615234375,
"logps/rejected": -790.7129516601562,
"loss": 0.2472,
"rewards/accuracies": 0.6937500238418579,
"rewards/chosen": -0.3279511630535126,
"rewards/margins": 0.15394431352615356,
"rewards/rejected": -0.48189544677734375,
"step": 710
},
{
"epoch": 0.77,
"learning_rate": 7.738832191993092e-07,
"logits/chosen": -0.008415929041802883,
"logits/rejected": 0.025492772459983826,
"logps/chosen": -639.2633056640625,
"logps/rejected": -759.4827880859375,
"loss": 0.2888,
"rewards/accuracies": 0.675000011920929,
"rewards/chosen": -0.31683486700057983,
"rewards/margins": 0.16635778546333313,
"rewards/rejected": -0.48319268226623535,
"step": 720
},
{
"epoch": 0.78,
"learning_rate": 7.077018300752917e-07,
"logits/chosen": 0.036400362849235535,
"logits/rejected": 0.06037065386772156,
"logps/chosen": -635.8643798828125,
"logps/rejected": -806.2174072265625,
"loss": 0.2624,
"rewards/accuracies": 0.6937500238418579,
"rewards/chosen": -0.3112700879573822,
"rewards/margins": 0.17294862866401672,
"rewards/rejected": -0.48421868681907654,
"step": 730
},
{
"epoch": 0.79,
"learning_rate": 6.440093245969342e-07,
"logits/chosen": 0.006747276522219181,
"logits/rejected": 0.036448314785957336,
"logps/chosen": -682.7285766601562,
"logps/rejected": -770.6795654296875,
"loss": 0.2916,
"rewards/accuracies": 0.643750011920929,
"rewards/chosen": -0.3282105326652527,
"rewards/margins": 0.13396592438220978,
"rewards/rejected": -0.46217647194862366,
"step": 740
},
{
"epoch": 0.8,
"learning_rate": 5.828941496744075e-07,
"logits/chosen": -0.008215249516069889,
"logits/rejected": 0.039237380027770996,
"logps/chosen": -615.4769897460938,
"logps/rejected": -744.2301025390625,
"loss": 0.2858,
"rewards/accuracies": 0.6312500238418579,
"rewards/chosen": -0.31000182032585144,
"rewards/margins": 0.12073155492544174,
"rewards/rejected": -0.43073344230651855,
"step": 750
},
{
"epoch": 0.81,
"learning_rate": 5.244411731951671e-07,
"logits/chosen": -0.018334541469812393,
"logits/rejected": 0.0751979798078537,
"logps/chosen": -653.4116821289062,
"logps/rejected": -775.2245483398438,
"loss": 0.2744,
"rewards/accuracies": 0.699999988079071,
"rewards/chosen": -0.3128724694252014,
"rewards/margins": 0.16823306679725647,
"rewards/rejected": -0.4811055064201355,
"step": 760
},
{
"epoch": 0.82,
"learning_rate": 4.6873156617173594e-07,
"logits/chosen": -0.01749856024980545,
"logits/rejected": 0.0061824568547308445,
"logps/chosen": -671.5619506835938,
"logps/rejected": -798.6920776367188,
"loss": 0.2781,
"rewards/accuracies": 0.675000011920929,
"rewards/chosen": -0.32307571172714233,
"rewards/margins": 0.15197516977787018,
"rewards/rejected": -0.4750508666038513,
"step": 770
},
{
"epoch": 0.83,
"learning_rate": 4.1584269002318653e-07,
"logits/chosen": -0.031299591064453125,
"logits/rejected": 0.11445782333612442,
"logps/chosen": -633.4989624023438,
"logps/rejected": -759.5948486328125,
"loss": 0.276,
"rewards/accuracies": 0.6812499761581421,
"rewards/chosen": -0.2988068461418152,
"rewards/margins": 0.14791473746299744,
"rewards/rejected": -0.446721613407135,
"step": 780
},
{
"epoch": 0.84,
"learning_rate": 3.658479891468258e-07,
"logits/chosen": 0.01681383140385151,
"logits/rejected": 0.06748757511377335,
"logps/chosen": -703.5291748046875,
"logps/rejected": -790.0008544921875,
"loss": 0.2645,
"rewards/accuracies": 0.71875,
"rewards/chosen": -0.3278062641620636,
"rewards/margins": 0.1466275155544281,
"rewards/rejected": -0.4744337499141693,
"step": 790
},
{
"epoch": 0.85,
"learning_rate": 3.18816888929272e-07,
"logits/chosen": -0.006874733604490757,
"logits/rejected": 0.09821389615535736,
"logps/chosen": -696.3887939453125,
"logps/rejected": -831.1228637695312,
"loss": 0.2456,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": -0.31410273909568787,
"rewards/margins": 0.19106796383857727,
"rewards/rejected": -0.5051707029342651,
"step": 800
},
{
"epoch": 0.86,
"learning_rate": 2.748146993385484e-07,
"logits/chosen": 0.02232988551259041,
"logits/rejected": 0.11882144212722778,
"logps/chosen": -737.1534423828125,
"logps/rejected": -799.8753662109375,
"loss": 0.2901,
"rewards/accuracies": 0.643750011920929,
"rewards/chosen": -0.3397114872932434,
"rewards/margins": 0.11829358339309692,
"rewards/rejected": -0.45800504088401794,
"step": 810
},
{
"epoch": 0.87,
"learning_rate": 2.3390252423108077e-07,
"logits/chosen": -0.005960514303296804,
"logits/rejected": 0.08128391206264496,
"logps/chosen": -600.1956787109375,
"logps/rejected": -725.2703857421875,
"loss": 0.2618,
"rewards/accuracies": 0.6875,
"rewards/chosen": -0.28923314809799194,
"rewards/margins": 0.14518693089485168,
"rewards/rejected": -0.43442004919052124,
"step": 820
},
{
"epoch": 0.89,
"learning_rate": 1.961371764995243e-07,
"logits/chosen": -0.027128243818879128,
"logits/rejected": 0.08068221807479858,
"logps/chosen": -648.6610107421875,
"logps/rejected": -814.8963012695312,
"loss": 0.2565,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -0.31799566745758057,
"rewards/margins": 0.18419261276721954,
"rewards/rejected": -0.5021882653236389,
"step": 830
},
{
"epoch": 0.9,
"learning_rate": 1.61571099179261e-07,
"logits/chosen": 0.04195733740925789,
"logits/rejected": 0.05591679736971855,
"logps/chosen": -618.9046630859375,
"logps/rejected": -764.7545776367188,
"loss": 0.2554,
"rewards/accuracies": 0.7124999761581421,
"rewards/chosen": -0.3056134879589081,
"rewards/margins": 0.1550132781267166,
"rewards/rejected": -0.4606267511844635,
"step": 840
},
{
"epoch": 0.91,
"learning_rate": 1.3025229262312367e-07,
"logits/chosen": 0.007567937020212412,
"logits/rejected": 0.08190431445837021,
"logps/chosen": -702.777587890625,
"logps/rejected": -844.9808349609375,
"loss": 0.2738,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.33942800760269165,
"rewards/margins": 0.16622349619865417,
"rewards/rejected": -0.5056515336036682,
"step": 850
},
{
"epoch": 0.92,
"learning_rate": 1.0222424784546853e-07,
"logits/chosen": -0.011982535012066364,
"logits/rejected": 0.029620468616485596,
"logps/chosen": -671.661865234375,
"logps/rejected": -762.5552978515625,
"loss": 0.297,
"rewards/accuracies": 0.6499999761581421,
"rewards/chosen": -0.355450838804245,
"rewards/margins": 0.11624264717102051,
"rewards/rejected": -0.4716935157775879,
"step": 860
},
{
"epoch": 0.93,
"learning_rate": 7.752588612816553e-08,
"logits/chosen": -0.010390078648924828,
"logits/rejected": 0.059226177632808685,
"logps/chosen": -709.0919189453125,
"logps/rejected": -846.7449340820312,
"loss": 0.296,
"rewards/accuracies": 0.71875,
"rewards/chosen": -0.3520728349685669,
"rewards/margins": 0.14244066178798676,
"rewards/rejected": -0.49451351165771484,
"step": 870
},
{
"epoch": 0.94,
"learning_rate": 5.619150497236991e-08,
"logits/chosen": 0.030358577147126198,
"logits/rejected": 0.1358805149793625,
"logps/chosen": -658.2210693359375,
"logps/rejected": -764.5792236328125,
"loss": 0.2756,
"rewards/accuracies": 0.668749988079071,
"rewards/chosen": -0.31784769892692566,
"rewards/margins": 0.1385163813829422,
"rewards/rejected": -0.45636409521102905,
"step": 880
},
{
"epoch": 0.95,
"learning_rate": 3.825073047112743e-08,
"logits/chosen": 0.07536911964416504,
"logits/rejected": 0.11495663970708847,
"logps/chosen": -674.4959716796875,
"logps/rejected": -830.2818603515625,
"loss": 0.2589,
"rewards/accuracies": 0.59375,
"rewards/chosen": -0.32807761430740356,
"rewards/margins": 0.15293636918067932,
"rewards/rejected": -0.48101386427879333,
"step": 890
},
{
"epoch": 0.96,
"learning_rate": 2.372847616895685e-08,
"logits/chosen": 0.015253638848662376,
"logits/rejected": 0.014745418913662434,
"logps/chosen": -697.5753173828125,
"logps/rejected": -841.8138427734375,
"loss": 0.2527,
"rewards/accuracies": 0.699999988079071,
"rewards/chosen": -0.33767208456993103,
"rewards/margins": 0.15997497737407684,
"rewards/rejected": -0.4976470470428467,
"step": 900
},
{
"epoch": 0.97,
"learning_rate": 1.264490846553279e-08,
"logits/chosen": 0.04343446344137192,
"logits/rejected": 0.05051248520612717,
"logps/chosen": -659.9283447265625,
"logps/rejected": -793.960205078125,
"loss": 0.2637,
"rewards/accuracies": 0.675000011920929,
"rewards/chosen": -0.3193858563899994,
"rewards/margins": 0.15680603682994843,
"rewards/rejected": -0.4761919379234314,
"step": 910
},
{
"epoch": 0.98,
"learning_rate": 5.015418611516165e-09,
"logits/chosen": -0.0020548023749142885,
"logits/rejected": 0.06288515031337738,
"logps/chosen": -699.4730224609375,
"logps/rejected": -794.502197265625,
"loss": 0.268,
"rewards/accuracies": 0.668749988079071,
"rewards/chosen": -0.34633034467697144,
"rewards/margins": 0.136570543050766,
"rewards/rejected": -0.48290085792541504,
"step": 920
},
{
"epoch": 0.99,
"learning_rate": 8.506013354186993e-10,
"logits/chosen": 0.013004904612898827,
"logits/rejected": 0.07795099914073944,
"logps/chosen": -636.2744750976562,
"logps/rejected": -769.828125,
"loss": 0.264,
"rewards/accuracies": 0.7124999761581421,
"rewards/chosen": -0.2970486283302307,
"rewards/margins": 0.1587102711200714,
"rewards/rejected": -0.4557588994503021,
"step": 930
},
{
"epoch": 1.0,
"step": 937,
"total_flos": 0.0,
"train_loss": 0.06840609829860919,
"train_runtime": 2864.3388,
"train_samples_per_second": 10.474,
"train_steps_per_second": 0.327
}
],
"logging_steps": 10,
"max_steps": 937,
"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
}