phi-2-dpo-ultrachat-lora / trainer_state.json
lole25's picture
Model save
c5be93f verified
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.9968602825745683,
"eval_steps": 100,
"global_step": 954,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 1.0416666666666667e-07,
"logits/chosen": 0.8531318306922913,
"logits/rejected": 0.7327959537506104,
"logps/chosen": -133.83494567871094,
"logps/rejected": -129.6299285888672,
"loss": 0.6931,
"rewards/accuracies": 0.0,
"rewards/chosen": 0.0,
"rewards/margins": 0.0,
"rewards/rejected": 0.0,
"step": 1
},
{
"epoch": 0.02,
"learning_rate": 1.0416666666666667e-06,
"logits/chosen": 0.8151362538337708,
"logits/rejected": 0.7788704037666321,
"logps/chosen": -90.5947036743164,
"logps/rejected": -92.77328491210938,
"loss": 0.6931,
"rewards/accuracies": 0.2361111044883728,
"rewards/chosen": 0.00018473717500455678,
"rewards/margins": 6.259369547478855e-05,
"rewards/rejected": 0.00012214347952976823,
"step": 10
},
{
"epoch": 0.04,
"learning_rate": 2.0833333333333334e-06,
"logits/chosen": 0.8342723846435547,
"logits/rejected": 0.8797974586486816,
"logps/chosen": -81.16204071044922,
"logps/rejected": -81.1548080444336,
"loss": 0.6932,
"rewards/accuracies": 0.17499999701976776,
"rewards/chosen": -0.00041853776201605797,
"rewards/margins": -0.0003428882628213614,
"rewards/rejected": -7.564939733128995e-05,
"step": 20
},
{
"epoch": 0.06,
"learning_rate": 3.125e-06,
"logits/chosen": 0.8050435185432434,
"logits/rejected": 0.8031445741653442,
"logps/chosen": -101.43102264404297,
"logps/rejected": -98.3962631225586,
"loss": 0.6932,
"rewards/accuracies": 0.23749999701976776,
"rewards/chosen": 0.00019087354303337634,
"rewards/margins": -0.00028236288926564157,
"rewards/rejected": 0.000473236374091357,
"step": 30
},
{
"epoch": 0.08,
"learning_rate": 4.166666666666667e-06,
"logits/chosen": 0.8352361917495728,
"logits/rejected": 0.8269574046134949,
"logps/chosen": -111.55462646484375,
"logps/rejected": -99.02640533447266,
"loss": 0.6931,
"rewards/accuracies": 0.23749999701976776,
"rewards/chosen": -5.0148169975727797e-05,
"rewards/margins": 7.251273200381547e-05,
"rewards/rejected": -0.00012266085832379758,
"step": 40
},
{
"epoch": 0.1,
"learning_rate": 4.999731868769027e-06,
"logits/chosen": 0.8326011896133423,
"logits/rejected": 0.8801982998847961,
"logps/chosen": -98.504638671875,
"logps/rejected": -83.65934753417969,
"loss": 0.6932,
"rewards/accuracies": 0.19374999403953552,
"rewards/chosen": -0.0006545094074681401,
"rewards/margins": -0.0007917654584161937,
"rewards/rejected": 0.00013725618191529065,
"step": 50
},
{
"epoch": 0.13,
"learning_rate": 4.9903533134293035e-06,
"logits/chosen": 0.8338711857795715,
"logits/rejected": 0.8361239433288574,
"logps/chosen": -105.47569274902344,
"logps/rejected": -103.5544662475586,
"loss": 0.6931,
"rewards/accuracies": 0.26249998807907104,
"rewards/chosen": -6.698282231809571e-05,
"rewards/margins": -0.0007280272548086941,
"rewards/rejected": 0.0006610444979742169,
"step": 60
},
{
"epoch": 0.15,
"learning_rate": 4.967625656594782e-06,
"logits/chosen": 0.8040585517883301,
"logits/rejected": 0.8523794412612915,
"logps/chosen": -83.11436462402344,
"logps/rejected": -77.95774841308594,
"loss": 0.6931,
"rewards/accuracies": 0.26875001192092896,
"rewards/chosen": -0.0002090830384986475,
"rewards/margins": -0.0001823374768719077,
"rewards/rejected": -2.6745503419078887e-05,
"step": 70
},
{
"epoch": 0.17,
"learning_rate": 4.93167072587771e-06,
"logits/chosen": 0.8447190523147583,
"logits/rejected": 0.9011589288711548,
"logps/chosen": -101.15557098388672,
"logps/rejected": -84.94987487792969,
"loss": 0.693,
"rewards/accuracies": 0.3375000059604645,
"rewards/chosen": 0.00010431509872432798,
"rewards/margins": 0.000598998973146081,
"rewards/rejected": -0.0004946838016621768,
"step": 80
},
{
"epoch": 0.19,
"learning_rate": 4.882681251368549e-06,
"logits/chosen": 0.8417055010795593,
"logits/rejected": 0.8667267560958862,
"logps/chosen": -88.30390930175781,
"logps/rejected": -110.51224517822266,
"loss": 0.6931,
"rewards/accuracies": 0.24375000596046448,
"rewards/chosen": -0.0014625315088778734,
"rewards/margins": -0.0006530345417559147,
"rewards/rejected": -0.0008094970253296196,
"step": 90
},
{
"epoch": 0.21,
"learning_rate": 4.8209198325401815e-06,
"logits/chosen": 0.8801371455192566,
"logits/rejected": 0.9127315282821655,
"logps/chosen": -86.3653793334961,
"logps/rejected": -69.10803985595703,
"loss": 0.693,
"rewards/accuracies": 0.22499999403953552,
"rewards/chosen": -0.0005041404510848224,
"rewards/margins": 0.0003193179436493665,
"rewards/rejected": -0.0008234584820456803,
"step": 100
},
{
"epoch": 0.21,
"eval_logits/chosen": 0.7997792959213257,
"eval_logits/rejected": 0.8176011443138123,
"eval_logps/chosen": -91.77481842041016,
"eval_logps/rejected": -94.28044891357422,
"eval_loss": 0.6930604577064514,
"eval_rewards/accuracies": 0.2680000066757202,
"eval_rewards/chosen": -0.00047452302533201873,
"eval_rewards/margins": 0.00035985803697258234,
"eval_rewards/rejected": -0.0008343810332007706,
"eval_runtime": 273.126,
"eval_samples_per_second": 7.323,
"eval_steps_per_second": 0.458,
"step": 100
},
{
"epoch": 0.23,
"learning_rate": 4.746717530629565e-06,
"logits/chosen": 0.8060985803604126,
"logits/rejected": 0.8291690945625305,
"logps/chosen": -110.20963287353516,
"logps/rejected": -96.49766540527344,
"loss": 0.6929,
"rewards/accuracies": 0.26249998807907104,
"rewards/chosen": -0.0005915694055147469,
"rewards/margins": 0.00022233165509533137,
"rewards/rejected": -0.0008139010751619935,
"step": 110
},
{
"epoch": 0.25,
"learning_rate": 4.660472094042121e-06,
"logits/chosen": 0.8455309867858887,
"logits/rejected": 0.9219174385070801,
"logps/chosen": -84.42987060546875,
"logps/rejected": -80.36955261230469,
"loss": 0.6931,
"rewards/accuracies": 0.28125,
"rewards/chosen": -0.0005250257672742009,
"rewards/margins": 0.00011121686839032918,
"rewards/rejected": -0.0006362426793202758,
"step": 120
},
{
"epoch": 0.27,
"learning_rate": 4.5626458262912745e-06,
"logits/chosen": 0.7771774530410767,
"logits/rejected": 0.8348148465156555,
"logps/chosen": -91.55365753173828,
"logps/rejected": -80.16468811035156,
"loss": 0.693,
"rewards/accuracies": 0.26249998807907104,
"rewards/chosen": -0.0003579836920835078,
"rewards/margins": 0.0003911118838004768,
"rewards/rejected": -0.0007490954594686627,
"step": 130
},
{
"epoch": 0.29,
"learning_rate": 4.453763107901676e-06,
"logits/chosen": 0.8080952763557434,
"logits/rejected": 0.8902841806411743,
"logps/chosen": -102.3683853149414,
"logps/rejected": -121.3236312866211,
"loss": 0.6929,
"rewards/accuracies": 0.34375,
"rewards/chosen": -0.0011437158100306988,
"rewards/margins": 0.00023913508630357683,
"rewards/rejected": -0.001382850925438106,
"step": 140
},
{
"epoch": 0.31,
"learning_rate": 4.33440758555951e-06,
"logits/chosen": 0.8286693692207336,
"logits/rejected": 0.8885159492492676,
"logps/chosen": -98.10747528076172,
"logps/rejected": -75.61839294433594,
"loss": 0.6929,
"rewards/accuracies": 0.26875001192092896,
"rewards/chosen": -0.001014741021208465,
"rewards/margins": 0.00025303030270151794,
"rewards/rejected": -0.0012677714694291353,
"step": 150
},
{
"epoch": 0.33,
"learning_rate": 4.205219043576955e-06,
"logits/chosen": 0.7178612947463989,
"logits/rejected": 0.7653802633285522,
"logps/chosen": -80.03684997558594,
"logps/rejected": -73.78407287597656,
"loss": 0.6928,
"rewards/accuracies": 0.2562499940395355,
"rewards/chosen": -0.001032956875860691,
"rewards/margins": 0.0007252781069837511,
"rewards/rejected": -0.0017582349246367812,
"step": 160
},
{
"epoch": 0.36,
"learning_rate": 4.066889974440757e-06,
"logits/chosen": 0.8685981631278992,
"logits/rejected": 0.9337660670280457,
"logps/chosen": -72.55715942382812,
"logps/rejected": -81.37494659423828,
"loss": 0.6929,
"rewards/accuracies": 0.26875001192092896,
"rewards/chosen": -0.0013817875878885388,
"rewards/margins": 0.0006504356861114502,
"rewards/rejected": -0.002032223390415311,
"step": 170
},
{
"epoch": 0.38,
"learning_rate": 3.92016186682789e-06,
"logits/chosen": 0.7872442007064819,
"logits/rejected": 0.8030352592468262,
"logps/chosen": -98.71012878417969,
"logps/rejected": -94.06107330322266,
"loss": 0.6928,
"rewards/accuracies": 0.26249998807907104,
"rewards/chosen": -0.0005101398564875126,
"rewards/margins": 0.0011010088492184877,
"rewards/rejected": -0.0016111487057060003,
"step": 180
},
{
"epoch": 0.4,
"learning_rate": 3.7658212309857576e-06,
"logits/chosen": 0.7398073077201843,
"logits/rejected": 0.8337036967277527,
"logps/chosen": -97.44853973388672,
"logps/rejected": -107.1574478149414,
"loss": 0.6926,
"rewards/accuracies": 0.3187499940395355,
"rewards/chosen": -0.0017274795100092888,
"rewards/margins": 0.0012662711087614298,
"rewards/rejected": -0.0029937506187707186,
"step": 190
},
{
"epoch": 0.42,
"learning_rate": 3.604695382782159e-06,
"logits/chosen": 0.7769767642021179,
"logits/rejected": 0.8381963968276978,
"logps/chosen": -77.53779602050781,
"logps/rejected": -86.9189224243164,
"loss": 0.6922,
"rewards/accuracies": 0.3125,
"rewards/chosen": -0.0016526943072676659,
"rewards/margins": 0.0023483093827962875,
"rewards/rejected": -0.004001003224402666,
"step": 200
},
{
"epoch": 0.42,
"eval_logits/chosen": 0.7941381931304932,
"eval_logits/rejected": 0.8121381998062134,
"eval_logps/chosen": -91.90680694580078,
"eval_logps/rejected": -94.51410675048828,
"eval_loss": 0.692441999912262,
"eval_rewards/accuracies": 0.3019999861717224,
"eval_rewards/chosen": -0.0017943703569471836,
"eval_rewards/margins": 0.001376640284433961,
"eval_rewards/rejected": -0.0031710106413811445,
"eval_runtime": 273.1548,
"eval_samples_per_second": 7.322,
"eval_steps_per_second": 0.458,
"step": 200
},
{
"epoch": 0.44,
"learning_rate": 3.437648009023905e-06,
"logits/chosen": 0.837793231010437,
"logits/rejected": 0.834112286567688,
"logps/chosen": -93.53672790527344,
"logps/rejected": -86.74687194824219,
"loss": 0.6926,
"rewards/accuracies": 0.2874999940395355,
"rewards/chosen": -0.0022572961170226336,
"rewards/margins": 0.0016887232195585966,
"rewards/rejected": -0.00394601933658123,
"step": 210
},
{
"epoch": 0.46,
"learning_rate": 3.265574537815398e-06,
"logits/chosen": 0.7990659475326538,
"logits/rejected": 0.8434357643127441,
"logps/chosen": -101.04875946044922,
"logps/rejected": -95.13258361816406,
"loss": 0.6927,
"rewards/accuracies": 0.3187499940395355,
"rewards/chosen": -0.002797973807901144,
"rewards/margins": 0.0009291036985814571,
"rewards/rejected": -0.003727077739313245,
"step": 220
},
{
"epoch": 0.48,
"learning_rate": 3.089397338773569e-06,
"logits/chosen": 0.8693816065788269,
"logits/rejected": 0.8417167663574219,
"logps/chosen": -88.75157165527344,
"logps/rejected": -93.32563018798828,
"loss": 0.6923,
"rewards/accuracies": 0.35624998807907104,
"rewards/chosen": -0.0035379533655941486,
"rewards/margins": 0.001493574702180922,
"rewards/rejected": -0.005031527951359749,
"step": 230
},
{
"epoch": 0.5,
"learning_rate": 2.9100607788275547e-06,
"logits/chosen": 0.7980669736862183,
"logits/rejected": 0.8121516108512878,
"logps/chosen": -115.3040771484375,
"logps/rejected": -101.95747375488281,
"loss": 0.6917,
"rewards/accuracies": 0.36250001192092896,
"rewards/chosen": -0.0025563673116266727,
"rewards/margins": 0.0030108934734016657,
"rewards/rejected": -0.005567261017858982,
"step": 240
},
{
"epoch": 0.52,
"learning_rate": 2.72852616010567e-06,
"logits/chosen": 0.7765026688575745,
"logits/rejected": 0.8406999707221985,
"logps/chosen": -116.62477111816406,
"logps/rejected": -107.66845703125,
"loss": 0.692,
"rewards/accuracies": 0.36250001192092896,
"rewards/chosen": -0.0028718970715999603,
"rewards/margins": 0.0028597547207027674,
"rewards/rejected": -0.005731652025133371,
"step": 250
},
{
"epoch": 0.54,
"learning_rate": 2.5457665670441937e-06,
"logits/chosen": 0.7957627177238464,
"logits/rejected": 0.8751128911972046,
"logps/chosen": -99.27378845214844,
"logps/rejected": -88.92988586425781,
"loss": 0.6915,
"rewards/accuracies": 0.33125001192092896,
"rewards/chosen": -0.0027769659645855427,
"rewards/margins": 0.003127423347905278,
"rewards/rejected": -0.005904389079660177,
"step": 260
},
{
"epoch": 0.57,
"learning_rate": 2.3627616503391813e-06,
"logits/chosen": 0.8911062479019165,
"logits/rejected": 0.9384480714797974,
"logps/chosen": -114.53426361083984,
"logps/rejected": -83.79386901855469,
"loss": 0.6917,
"rewards/accuracies": 0.4124999940395355,
"rewards/chosen": -0.0030242924112826586,
"rewards/margins": 0.0038346436340361834,
"rewards/rejected": -0.006858936045318842,
"step": 270
},
{
"epoch": 0.59,
"learning_rate": 2.1804923757009885e-06,
"logits/chosen": 0.8408538699150085,
"logits/rejected": 0.839997410774231,
"logps/chosen": -121.15773010253906,
"logps/rejected": -109.73968505859375,
"loss": 0.6916,
"rewards/accuracies": 0.38749998807907104,
"rewards/chosen": -0.0034516516607254744,
"rewards/margins": 0.0034923548810184,
"rewards/rejected": -0.006944006774574518,
"step": 280
},
{
"epoch": 0.61,
"learning_rate": 1.9999357655598894e-06,
"logits/chosen": 0.825127124786377,
"logits/rejected": 0.8463200330734253,
"logps/chosen": -94.99199676513672,
"logps/rejected": -92.08692932128906,
"loss": 0.6919,
"rewards/accuracies": 0.2750000059604645,
"rewards/chosen": -0.005209151655435562,
"rewards/margins": 0.0012942428002133965,
"rewards/rejected": -0.006503394804894924,
"step": 290
},
{
"epoch": 0.63,
"learning_rate": 1.8220596619089576e-06,
"logits/chosen": 0.7896022796630859,
"logits/rejected": 0.8404830694198608,
"logps/chosen": -79.6166763305664,
"logps/rejected": -87.97811126708984,
"loss": 0.6917,
"rewards/accuracies": 0.3125,
"rewards/chosen": -0.002977000316604972,
"rewards/margins": 0.004023983143270016,
"rewards/rejected": -0.007000982761383057,
"step": 300
},
{
"epoch": 0.63,
"eval_logits/chosen": 0.787030041217804,
"eval_logits/rejected": 0.8057210445404053,
"eval_logps/chosen": -92.21893310546875,
"eval_logps/rejected": -94.9659194946289,
"eval_loss": 0.6917064189910889,
"eval_rewards/accuracies": 0.3100000023841858,
"eval_rewards/chosen": -0.004915657918900251,
"eval_rewards/margins": 0.0027734539471566677,
"eval_rewards/rejected": -0.0076891109347343445,
"eval_runtime": 273.1314,
"eval_samples_per_second": 7.322,
"eval_steps_per_second": 0.458,
"step": 300
},
{
"epoch": 0.65,
"learning_rate": 1.647817538357072e-06,
"logits/chosen": 0.797222912311554,
"logits/rejected": 0.8298647999763489,
"logps/chosen": -85.97596740722656,
"logps/rejected": -87.59102630615234,
"loss": 0.6915,
"rewards/accuracies": 0.33125001192092896,
"rewards/chosen": -0.005664575379341841,
"rewards/margins": 0.002779053058475256,
"rewards/rejected": -0.008443629369139671,
"step": 310
},
{
"epoch": 0.67,
"learning_rate": 1.4781433892011132e-06,
"logits/chosen": 0.7955067753791809,
"logits/rejected": 0.8460105657577515,
"logps/chosen": -88.00541687011719,
"logps/rejected": -91.3482437133789,
"loss": 0.6916,
"rewards/accuracies": 0.3687500059604645,
"rewards/chosen": -0.005240852013230324,
"rewards/margins": 0.003471267642453313,
"rewards/rejected": -0.008712120354175568,
"step": 320
},
{
"epoch": 0.69,
"learning_rate": 1.3139467229135999e-06,
"logits/chosen": 0.7453655004501343,
"logits/rejected": 0.7835742235183716,
"logps/chosen": -101.49515533447266,
"logps/rejected": -102.06614685058594,
"loss": 0.6914,
"rewards/accuracies": 0.33125001192092896,
"rewards/chosen": -0.00533133652061224,
"rewards/margins": 0.004351903218775988,
"rewards/rejected": -0.00968323927372694,
"step": 330
},
{
"epoch": 0.71,
"learning_rate": 1.1561076868822756e-06,
"logits/chosen": 0.7904757857322693,
"logits/rejected": 0.8207653760910034,
"logps/chosen": -83.08015441894531,
"logps/rejected": -73.13031768798828,
"loss": 0.6908,
"rewards/accuracies": 0.29374998807907104,
"rewards/chosen": -0.002843918278813362,
"rewards/margins": 0.005075609777122736,
"rewards/rejected": -0.007919528521597385,
"step": 340
},
{
"epoch": 0.73,
"learning_rate": 1.0054723495346484e-06,
"logits/chosen": 0.8648473620414734,
"logits/rejected": 0.88224858045578,
"logps/chosen": -88.79496002197266,
"logps/rejected": -82.97408294677734,
"loss": 0.6911,
"rewards/accuracies": 0.36250001192092896,
"rewards/chosen": -0.0072964876890182495,
"rewards/margins": 0.003556302282959223,
"rewards/rejected": -0.01085279043763876,
"step": 350
},
{
"epoch": 0.75,
"learning_rate": 8.628481651367876e-07,
"logits/chosen": 0.7940279841423035,
"logits/rejected": 0.808300793170929,
"logps/chosen": -95.09690856933594,
"logps/rejected": -98.6402816772461,
"loss": 0.6911,
"rewards/accuracies": 0.375,
"rewards/chosen": -0.007194930221885443,
"rewards/margins": 0.004204220604151487,
"rewards/rejected": -0.01139915082603693,
"step": 360
},
{
"epoch": 0.77,
"learning_rate": 7.289996455765749e-07,
"logits/chosen": 0.8749688863754272,
"logits/rejected": 0.8549054861068726,
"logps/chosen": -107.79057312011719,
"logps/rejected": -104.4919662475586,
"loss": 0.6913,
"rewards/accuracies": 0.33125001192092896,
"rewards/chosen": -0.005484831985086203,
"rewards/margins": 0.0049090199172496796,
"rewards/rejected": -0.01039385236799717,
"step": 370
},
{
"epoch": 0.8,
"learning_rate": 6.046442623320145e-07,
"logits/chosen": 0.8186184167861938,
"logits/rejected": 0.8605905771255493,
"logps/chosen": -102.05201721191406,
"logps/rejected": -88.45350646972656,
"loss": 0.6905,
"rewards/accuracies": 0.35624998807907104,
"rewards/chosen": -0.006084176246076822,
"rewards/margins": 0.004532010294497013,
"rewards/rejected": -0.010616186074912548,
"step": 380
},
{
"epoch": 0.82,
"learning_rate": 4.904486005914027e-07,
"logits/chosen": 0.8054735064506531,
"logits/rejected": 0.8569159507751465,
"logps/chosen": -110.53129577636719,
"logps/rejected": -99.7970199584961,
"loss": 0.6905,
"rewards/accuracies": 0.35624998807907104,
"rewards/chosen": -0.0049868240021169186,
"rewards/margins": 0.005900658201426268,
"rewards/rejected": -0.01088748313486576,
"step": 390
},
{
"epoch": 0.84,
"learning_rate": 3.8702478614051353e-07,
"logits/chosen": 0.8212454915046692,
"logits/rejected": 0.8515514135360718,
"logps/chosen": -100.35879516601562,
"logps/rejected": -103.1037368774414,
"loss": 0.6905,
"rewards/accuracies": 0.3375000059604645,
"rewards/chosen": -0.006732915993779898,
"rewards/margins": 0.004675927106291056,
"rewards/rejected": -0.011408843100070953,
"step": 400
},
{
"epoch": 0.84,
"eval_logits/chosen": 0.7826990485191345,
"eval_logits/rejected": 0.8012421727180481,
"eval_logps/chosen": -92.42469787597656,
"eval_logps/rejected": -95.25089263916016,
"eval_loss": 0.6913294792175293,
"eval_rewards/accuracies": 0.328000009059906,
"eval_rewards/chosen": -0.00697335647419095,
"eval_rewards/margins": 0.0035653903614729643,
"eval_rewards/rejected": -0.010538745671510696,
"eval_runtime": 273.6535,
"eval_samples_per_second": 7.309,
"eval_steps_per_second": 0.457,
"step": 400
},
{
"epoch": 0.86,
"learning_rate": 3.5218566107988872e-06,
"logits/chosen": 0.8185564279556274,
"logits/rejected": 0.8151994943618774,
"logps/chosen": -95.8418960571289,
"logps/rejected": -80.64591979980469,
"loss": 0.6916,
"rewards/accuracies": 0.33125001192092896,
"rewards/chosen": -0.008152286522090435,
"rewards/margins": 0.003440212458372116,
"rewards/rejected": -0.011592499911785126,
"step": 410
},
{
"epoch": 0.88,
"learning_rate": 3.437648009023905e-06,
"logits/chosen": 0.7549771070480347,
"logits/rejected": 0.8158624768257141,
"logps/chosen": -95.72089385986328,
"logps/rejected": -97.96694946289062,
"loss": 0.6914,
"rewards/accuracies": 0.34375,
"rewards/chosen": -0.00923125259578228,
"rewards/margins": 0.0032367813400924206,
"rewards/rejected": -0.012468034401535988,
"step": 420
},
{
"epoch": 0.9,
"learning_rate": 3.352182461642929e-06,
"logits/chosen": 0.862383246421814,
"logits/rejected": 0.8848444819450378,
"logps/chosen": -83.51128387451172,
"logps/rejected": -90.3624038696289,
"loss": 0.6909,
"rewards/accuracies": 0.41874998807907104,
"rewards/chosen": -0.0076131029054522514,
"rewards/margins": 0.005502562969923019,
"rewards/rejected": -0.013115664944052696,
"step": 430
},
{
"epoch": 0.92,
"learning_rate": 3.265574537815398e-06,
"logits/chosen": 0.7784366011619568,
"logits/rejected": 0.8296224474906921,
"logps/chosen": -109.9617919921875,
"logps/rejected": -82.34262084960938,
"loss": 0.6907,
"rewards/accuracies": 0.3499999940395355,
"rewards/chosen": -0.006960420869290829,
"rewards/margins": 0.007026113569736481,
"rewards/rejected": -0.013986535370349884,
"step": 440
},
{
"epoch": 0.94,
"learning_rate": 3.177940338091043e-06,
"logits/chosen": 0.769826352596283,
"logits/rejected": 0.8450337648391724,
"logps/chosen": -81.3624496459961,
"logps/rejected": -72.23616027832031,
"loss": 0.6903,
"rewards/accuracies": 0.30000001192092896,
"rewards/chosen": -0.00711588840931654,
"rewards/margins": 0.005811682902276516,
"rewards/rejected": -0.012927571311593056,
"step": 450
},
{
"epoch": 0.96,
"learning_rate": 3.089397338773569e-06,
"logits/chosen": 0.7976155877113342,
"logits/rejected": 0.8512013554573059,
"logps/chosen": -106.02879333496094,
"logps/rejected": -88.0615234375,
"loss": 0.6899,
"rewards/accuracies": 0.32499998807907104,
"rewards/chosen": -0.007455066777765751,
"rewards/margins": 0.00965816993266344,
"rewards/rejected": -0.017113234847784042,
"step": 460
},
{
"epoch": 0.98,
"learning_rate": 3.0000642344401115e-06,
"logits/chosen": 0.8498729467391968,
"logits/rejected": 0.8459364771842957,
"logps/chosen": -116.07499694824219,
"logps/rejected": -111.56224060058594,
"loss": 0.6892,
"rewards/accuracies": 0.375,
"rewards/chosen": -0.010130222886800766,
"rewards/margins": 0.009535903111100197,
"rewards/rejected": -0.019666125997900963,
"step": 470
},
{
"epoch": 1.0,
"learning_rate": 2.9100607788275547e-06,
"logits/chosen": 0.8643083572387695,
"logits/rejected": 0.8725606799125671,
"logps/chosen": -76.1094741821289,
"logps/rejected": -66.09565734863281,
"loss": 0.6903,
"rewards/accuracies": 0.26875001192092896,
"rewards/chosen": -0.010615186765789986,
"rewards/margins": 0.004579311236739159,
"rewards/rejected": -0.015194499865174294,
"step": 480
},
{
"epoch": 1.03,
"learning_rate": 2.8195076242990124e-06,
"logits/chosen": 0.8051859736442566,
"logits/rejected": 0.8318646550178528,
"logps/chosen": -93.38316345214844,
"logps/rejected": -100.01469421386719,
"loss": 0.6903,
"rewards/accuracies": 0.35624998807907104,
"rewards/chosen": -0.014997744932770729,
"rewards/margins": 0.004119081888347864,
"rewards/rejected": -0.019116824492812157,
"step": 490
},
{
"epoch": 1.05,
"learning_rate": 2.72852616010567e-06,
"logits/chosen": 0.8241451978683472,
"logits/rejected": 0.8089090585708618,
"logps/chosen": -102.48905181884766,
"logps/rejected": -102.73881530761719,
"loss": 0.6898,
"rewards/accuracies": 0.39375001192092896,
"rewards/chosen": -0.013014930300414562,
"rewards/margins": 0.011042198166251183,
"rewards/rejected": -0.02405713126063347,
"step": 500
},
{
"epoch": 1.05,
"eval_logits/chosen": 0.7711244225502014,
"eval_logits/rejected": 0.7902986407279968,
"eval_logps/chosen": -93.14293670654297,
"eval_logps/rejected": -96.24901580810547,
"eval_loss": 0.6899796724319458,
"eval_rewards/accuracies": 0.335999995470047,
"eval_rewards/chosen": -0.014155692420899868,
"eval_rewards/margins": 0.006364365108311176,
"eval_rewards/rejected": -0.020520057529211044,
"eval_runtime": 272.796,
"eval_samples_per_second": 7.331,
"eval_steps_per_second": 0.458,
"step": 500
},
{
"epoch": 1.07,
"learning_rate": 2.637238349660819e-06,
"logits/chosen": 0.8259177207946777,
"logits/rejected": 0.8255017995834351,
"logps/chosen": -92.70307159423828,
"logps/rejected": -90.75215148925781,
"loss": 0.6893,
"rewards/accuracies": 0.35624998807907104,
"rewards/chosen": -0.016096794977784157,
"rewards/margins": 0.006517867557704449,
"rewards/rejected": -0.02261466160416603,
"step": 510
},
{
"epoch": 1.09,
"learning_rate": 2.5457665670441937e-06,
"logits/chosen": 0.8241807222366333,
"logits/rejected": 0.8674284219741821,
"logps/chosen": -103.48587799072266,
"logps/rejected": -95.00794219970703,
"loss": 0.6887,
"rewards/accuracies": 0.375,
"rewards/chosen": -0.012511089444160461,
"rewards/margins": 0.011780844070017338,
"rewards/rejected": -0.024291934445500374,
"step": 520
},
{
"epoch": 1.11,
"learning_rate": 2.4542334329558075e-06,
"logits/chosen": 0.8090038299560547,
"logits/rejected": 0.8909789323806763,
"logps/chosen": -108.34500885009766,
"logps/rejected": -90.74520111083984,
"loss": 0.6897,
"rewards/accuracies": 0.38749998807907104,
"rewards/chosen": -0.013912905938923359,
"rewards/margins": 0.01008325070142746,
"rewards/rejected": -0.023996157571673393,
"step": 530
},
{
"epoch": 1.13,
"learning_rate": 2.3627616503391813e-06,
"logits/chosen": 0.7575622797012329,
"logits/rejected": 0.7894734144210815,
"logps/chosen": -103.4853515625,
"logps/rejected": -89.14204406738281,
"loss": 0.6886,
"rewards/accuracies": 0.36250001192092896,
"rewards/chosen": -0.01601695641875267,
"rewards/margins": 0.009688997641205788,
"rewards/rejected": -0.025705954059958458,
"step": 540
},
{
"epoch": 1.15,
"learning_rate": 2.271473839894331e-06,
"logits/chosen": 0.7494063377380371,
"logits/rejected": 0.7767230272293091,
"logps/chosen": -93.18793487548828,
"logps/rejected": -85.53865814208984,
"loss": 0.689,
"rewards/accuracies": 0.34375,
"rewards/chosen": -0.015046611428260803,
"rewards/margins": 0.009437174536287785,
"rewards/rejected": -0.024483786895871162,
"step": 550
},
{
"epoch": 1.17,
"learning_rate": 2.1804923757009885e-06,
"logits/chosen": 0.7773897051811218,
"logits/rejected": 0.8226049542427063,
"logps/chosen": -118.14359283447266,
"logps/rejected": -118.2892837524414,
"loss": 0.6883,
"rewards/accuracies": 0.39375001192092896,
"rewards/chosen": -0.01944814994931221,
"rewards/margins": 0.012502019293606281,
"rewards/rejected": -0.03195016831159592,
"step": 560
},
{
"epoch": 1.19,
"learning_rate": 2.089939221172446e-06,
"logits/chosen": 0.7639147639274597,
"logits/rejected": 0.8038470149040222,
"logps/chosen": -96.55006408691406,
"logps/rejected": -94.61703491210938,
"loss": 0.6893,
"rewards/accuracies": 0.36250001192092896,
"rewards/chosen": -0.01976608671247959,
"rewards/margins": 0.01228870265185833,
"rewards/rejected": -0.03205478936433792,
"step": 570
},
{
"epoch": 1.21,
"learning_rate": 1.9999357655598894e-06,
"logits/chosen": 0.799372673034668,
"logits/rejected": 0.7758508920669556,
"logps/chosen": -107.0300521850586,
"logps/rejected": -101.98541259765625,
"loss": 0.6898,
"rewards/accuracies": 0.32499998807907104,
"rewards/chosen": -0.023156706243753433,
"rewards/margins": 0.0033106685150414705,
"rewards/rejected": -0.026467373594641685,
"step": 580
},
{
"epoch": 1.23,
"learning_rate": 1.9106026612264316e-06,
"logits/chosen": 0.8342978358268738,
"logits/rejected": 0.818748950958252,
"logps/chosen": -82.849609375,
"logps/rejected": -82.92151641845703,
"loss": 0.6875,
"rewards/accuracies": 0.34375,
"rewards/chosen": -0.021987412124872208,
"rewards/margins": 0.00925590842962265,
"rewards/rejected": -0.03124331869184971,
"step": 590
},
{
"epoch": 1.26,
"learning_rate": 1.8220596619089576e-06,
"logits/chosen": 0.7792515158653259,
"logits/rejected": 0.7631284594535828,
"logps/chosen": -108.453369140625,
"logps/rejected": -109.84110260009766,
"loss": 0.6882,
"rewards/accuracies": 0.3812499940395355,
"rewards/chosen": -0.024591121822595596,
"rewards/margins": 0.010857349261641502,
"rewards/rejected": -0.03544846922159195,
"step": 600
},
{
"epoch": 1.26,
"eval_logits/chosen": 0.7526758909225464,
"eval_logits/rejected": 0.7722000479698181,
"eval_logps/chosen": -93.8980941772461,
"eval_logps/rejected": -97.25941467285156,
"eval_loss": 0.6886565089225769,
"eval_rewards/accuracies": 0.33399999141693115,
"eval_rewards/chosen": -0.02170729637145996,
"eval_rewards/margins": 0.008916829712688923,
"eval_rewards/rejected": -0.03062412701547146,
"eval_runtime": 272.575,
"eval_samples_per_second": 7.337,
"eval_steps_per_second": 0.459,
"step": 600
},
{
"epoch": 1.28,
"learning_rate": 1.7344254621846018e-06,
"logits/chosen": 0.7705439329147339,
"logits/rejected": 0.7898679375648499,
"logps/chosen": -114.97818756103516,
"logps/rejected": -105.07425689697266,
"loss": 0.6871,
"rewards/accuracies": 0.35624998807907104,
"rewards/chosen": -0.020355116575956345,
"rewards/margins": 0.01100640743970871,
"rewards/rejected": -0.031361524015665054,
"step": 610
},
{
"epoch": 1.3,
"learning_rate": 1.647817538357072e-06,
"logits/chosen": 0.8015000224113464,
"logits/rejected": 0.8592731356620789,
"logps/chosen": -105.63825988769531,
"logps/rejected": -101.30821228027344,
"loss": 0.6877,
"rewards/accuracies": 0.38749998807907104,
"rewards/chosen": -0.025122905150055885,
"rewards/margins": 0.012796345166862011,
"rewards/rejected": -0.03791924566030502,
"step": 620
},
{
"epoch": 1.32,
"learning_rate": 1.5623519909760953e-06,
"logits/chosen": 0.72571861743927,
"logits/rejected": 0.7672609090805054,
"logps/chosen": -95.39134216308594,
"logps/rejected": -91.43064880371094,
"loss": 0.6875,
"rewards/accuracies": 0.3125,
"rewards/chosen": -0.023522889241576195,
"rewards/margins": 0.008042030036449432,
"rewards/rejected": -0.031564921140670776,
"step": 630
},
{
"epoch": 1.34,
"learning_rate": 1.4781433892011132e-06,
"logits/chosen": 0.7983066439628601,
"logits/rejected": 0.828285813331604,
"logps/chosen": -91.58604431152344,
"logps/rejected": -98.34529876708984,
"loss": 0.6877,
"rewards/accuracies": 0.3375000059604645,
"rewards/chosen": -0.02216607704758644,
"rewards/margins": 0.009402362629771233,
"rewards/rejected": -0.03156844526529312,
"step": 640
},
{
"epoch": 1.36,
"learning_rate": 1.3953046172178413e-06,
"logits/chosen": 0.8052563667297363,
"logits/rejected": 0.861487090587616,
"logps/chosen": -102.22875213623047,
"logps/rejected": -115.90311431884766,
"loss": 0.688,
"rewards/accuracies": 0.39375001192092896,
"rewards/chosen": -0.02779082953929901,
"rewards/margins": 0.011276346631348133,
"rewards/rejected": -0.03906718268990517,
"step": 650
},
{
"epoch": 1.38,
"learning_rate": 1.3139467229135999e-06,
"logits/chosen": 0.7473502159118652,
"logits/rejected": 0.7619670629501343,
"logps/chosen": -81.72496032714844,
"logps/rejected": -88.32991027832031,
"loss": 0.6867,
"rewards/accuracies": 0.32499998807907104,
"rewards/chosen": -0.023514125496149063,
"rewards/margins": 0.01081976480782032,
"rewards/rejected": -0.034333888441324234,
"step": 660
},
{
"epoch": 1.4,
"learning_rate": 1.2341787690142436e-06,
"logits/chosen": 0.677151083946228,
"logits/rejected": 0.7038652300834656,
"logps/chosen": -116.34004211425781,
"logps/rejected": -98.69480895996094,
"loss": 0.6883,
"rewards/accuracies": 0.3687500059604645,
"rewards/chosen": -0.024621013551950455,
"rewards/margins": 0.016095632687211037,
"rewards/rejected": -0.040716640651226044,
"step": 670
},
{
"epoch": 1.42,
"learning_rate": 1.1561076868822756e-06,
"logits/chosen": 0.7057438492774963,
"logits/rejected": 0.734573483467102,
"logps/chosen": -77.70423889160156,
"logps/rejected": -83.44449615478516,
"loss": 0.6875,
"rewards/accuracies": 0.32499998807907104,
"rewards/chosen": -0.029654154554009438,
"rewards/margins": 0.00593325262889266,
"rewards/rejected": -0.035587407648563385,
"step": 680
},
{
"epoch": 1.44,
"learning_rate": 1.079838133172111e-06,
"logits/chosen": 0.7637497186660767,
"logits/rejected": 0.8190714716911316,
"logps/chosen": -95.7881088256836,
"logps/rejected": -82.04200744628906,
"loss": 0.6882,
"rewards/accuracies": 0.35624998807907104,
"rewards/chosen": -0.026545196771621704,
"rewards/margins": 0.014608954079449177,
"rewards/rejected": -0.041154149919748306,
"step": 690
},
{
"epoch": 1.47,
"learning_rate": 1.0054723495346484e-06,
"logits/chosen": 0.8055804371833801,
"logits/rejected": 0.8074037432670593,
"logps/chosen": -106.52473449707031,
"logps/rejected": -99.84661865234375,
"loss": 0.6858,
"rewards/accuracies": 0.375,
"rewards/chosen": -0.03077777661383152,
"rewards/margins": 0.016430744901299477,
"rewards/rejected": -0.0472085177898407,
"step": 700
},
{
"epoch": 1.47,
"eval_logits/chosen": 0.7395281791687012,
"eval_logits/rejected": 0.7599726319313049,
"eval_logps/chosen": -94.4716567993164,
"eval_logps/rejected": -98.02493286132812,
"eval_loss": 0.6878523826599121,
"eval_rewards/accuracies": 0.328000009059906,
"eval_rewards/chosen": -0.02744293212890625,
"eval_rewards/margins": 0.010836233384907246,
"eval_rewards/rejected": -0.03827916085720062,
"eval_runtime": 272.6941,
"eval_samples_per_second": 7.334,
"eval_steps_per_second": 0.458,
"step": 700
},
{
"epoch": 1.49,
"learning_rate": 9.331100255592437e-07,
"logits/chosen": 0.7606201767921448,
"logits/rejected": 0.8386918306350708,
"logps/chosen": -89.20939636230469,
"logps/rejected": -82.81990814208984,
"loss": 0.6861,
"rewards/accuracies": 0.3062500059604645,
"rewards/chosen": -0.022489020600914955,
"rewards/margins": 0.012643699534237385,
"rewards/rejected": -0.03513271361589432,
"step": 710
},
{
"epoch": 1.51,
"learning_rate": 8.628481651367876e-07,
"logits/chosen": 0.771535336971283,
"logits/rejected": 0.8548757433891296,
"logps/chosen": -113.86335754394531,
"logps/rejected": -101.86293029785156,
"loss": 0.6856,
"rewards/accuracies": 0.35624998807907104,
"rewards/chosen": -0.02375701069831848,
"rewards/margins": 0.01785343512892723,
"rewards/rejected": -0.041610442101955414,
"step": 720
},
{
"epoch": 1.53,
"learning_rate": 7.947809564230446e-07,
"logits/chosen": 0.7982994914054871,
"logits/rejected": 0.8568431735038757,
"logps/chosen": -98.13008880615234,
"logps/rejected": -97.87364196777344,
"loss": 0.6874,
"rewards/accuracies": 0.375,
"rewards/chosen": -0.03274186700582504,
"rewards/margins": 0.010785898193717003,
"rewards/rejected": -0.0435277596116066,
"step": 730
},
{
"epoch": 1.55,
"learning_rate": 7.289996455765749e-07,
"logits/chosen": 0.7194357514381409,
"logits/rejected": 0.7669156789779663,
"logps/chosen": -88.61851501464844,
"logps/rejected": -82.62269592285156,
"loss": 0.6889,
"rewards/accuracies": 0.2562499940395355,
"rewards/chosen": -0.02295432984828949,
"rewards/margins": 0.008415495045483112,
"rewards/rejected": -0.03136982396245003,
"step": 740
},
{
"epoch": 1.57,
"learning_rate": 6.655924144404907e-07,
"logits/chosen": 0.7268679141998291,
"logits/rejected": 0.7419520616531372,
"logps/chosen": -90.82257843017578,
"logps/rejected": -87.84976196289062,
"loss": 0.687,
"rewards/accuracies": 0.35624998807907104,
"rewards/chosen": -0.024226779118180275,
"rewards/margins": 0.014533626846969128,
"rewards/rejected": -0.03876040503382683,
"step": 750
},
{
"epoch": 1.59,
"learning_rate": 6.046442623320145e-07,
"logits/chosen": 0.775734543800354,
"logits/rejected": 0.8094171285629272,
"logps/chosen": -111.5390853881836,
"logps/rejected": -107.2262954711914,
"loss": 0.6868,
"rewards/accuracies": 0.36250001192092896,
"rewards/chosen": -0.035627130419015884,
"rewards/margins": 0.011030396446585655,
"rewards/rejected": -0.04665752500295639,
"step": 760
},
{
"epoch": 1.61,
"learning_rate": 5.462368920983249e-07,
"logits/chosen": 0.7696752548217773,
"logits/rejected": 0.7860345840454102,
"logps/chosen": -86.94595336914062,
"logps/rejected": -74.9552230834961,
"loss": 0.6867,
"rewards/accuracies": 0.3125,
"rewards/chosen": -0.02532133087515831,
"rewards/margins": 0.010699031874537468,
"rewards/rejected": -0.03602036461234093,
"step": 770
},
{
"epoch": 1.63,
"learning_rate": 4.904486005914027e-07,
"logits/chosen": 0.7500615119934082,
"logits/rejected": 0.7661712765693665,
"logps/chosen": -119.89558410644531,
"logps/rejected": -111.26094055175781,
"loss": 0.6851,
"rewards/accuracies": 0.3812499940395355,
"rewards/chosen": -0.02393215522170067,
"rewards/margins": 0.01878645084798336,
"rewards/rejected": -0.04271860793232918,
"step": 780
},
{
"epoch": 1.65,
"learning_rate": 4.373541737087264e-07,
"logits/chosen": 0.7226775288581848,
"logits/rejected": 0.7507297992706299,
"logps/chosen": -103.21110534667969,
"logps/rejected": -99.74224853515625,
"loss": 0.6876,
"rewards/accuracies": 0.29374998807907104,
"rewards/chosen": -0.028694171458482742,
"rewards/margins": 0.013696588575839996,
"rewards/rejected": -0.04239075630903244,
"step": 790
},
{
"epoch": 1.67,
"learning_rate": 3.8702478614051353e-07,
"logits/chosen": 0.7573332786560059,
"logits/rejected": 0.7167325019836426,
"logps/chosen": -101.85111236572266,
"logps/rejected": -95.10759735107422,
"loss": 0.6857,
"rewards/accuracies": 0.4124999940395355,
"rewards/chosen": -0.02707928977906704,
"rewards/margins": 0.022238872945308685,
"rewards/rejected": -0.049318164587020874,
"step": 800
},
{
"epoch": 1.67,
"eval_logits/chosen": 0.7340908646583557,
"eval_logits/rejected": 0.7547872066497803,
"eval_logps/chosen": -94.76182556152344,
"eval_logps/rejected": -98.42695617675781,
"eval_loss": 0.687368631362915,
"eval_rewards/accuracies": 0.33399999141693115,
"eval_rewards/chosen": -0.030344627797603607,
"eval_rewards/margins": 0.011954776011407375,
"eval_rewards/rejected": -0.04229940101504326,
"eval_runtime": 272.7101,
"eval_samples_per_second": 7.334,
"eval_steps_per_second": 0.458,
"step": 800
},
{
"epoch": 1.7,
"learning_rate": 3.3952790595787986e-07,
"logits/chosen": 0.8084260821342468,
"logits/rejected": 0.8206745982170105,
"logps/chosen": -118.43504333496094,
"logps/rejected": -105.46468353271484,
"loss": 0.6876,
"rewards/accuracies": 0.3187499940395355,
"rewards/chosen": -0.03346724063158035,
"rewards/margins": 0.010504155419766903,
"rewards/rejected": -0.04397139698266983,
"step": 810
},
{
"epoch": 1.72,
"learning_rate": 2.9492720416985004e-07,
"logits/chosen": 0.6800391674041748,
"logits/rejected": 0.6965945363044739,
"logps/chosen": -96.83241271972656,
"logps/rejected": -101.45925903320312,
"loss": 0.6869,
"rewards/accuracies": 0.3499999940395355,
"rewards/chosen": -0.027126986533403397,
"rewards/margins": 0.01599222794175148,
"rewards/rejected": -0.04311921447515488,
"step": 820
},
{
"epoch": 1.74,
"learning_rate": 2.5328246937043526e-07,
"logits/chosen": 0.7581857442855835,
"logits/rejected": 0.7956913113594055,
"logps/chosen": -98.04803466796875,
"logps/rejected": -102.8603515625,
"loss": 0.6858,
"rewards/accuracies": 0.38749998807907104,
"rewards/chosen": -0.034231387078762054,
"rewards/margins": 0.015285378322005272,
"rewards/rejected": -0.049516770988702774,
"step": 830
},
{
"epoch": 1.76,
"learning_rate": 2.1464952759020857e-07,
"logits/chosen": 0.6740394830703735,
"logits/rejected": 0.7218228578567505,
"logps/chosen": -96.54594421386719,
"logps/rejected": -99.02603912353516,
"loss": 0.6878,
"rewards/accuracies": 0.30000001192092896,
"rewards/chosen": -0.029307430610060692,
"rewards/margins": 0.010790064930915833,
"rewards/rejected": -0.040097493678331375,
"step": 840
},
{
"epoch": 1.78,
"learning_rate": 1.790801674598186e-07,
"logits/chosen": 0.7388015985488892,
"logits/rejected": 0.7736684083938599,
"logps/chosen": -126.44425201416016,
"logps/rejected": -102.01686096191406,
"loss": 0.686,
"rewards/accuracies": 0.39375001192092896,
"rewards/chosen": -0.030984923243522644,
"rewards/margins": 0.01829494535923004,
"rewards/rejected": -0.049279868602752686,
"step": 850
},
{
"epoch": 1.8,
"learning_rate": 1.4662207078575685e-07,
"logits/chosen": 0.7661498785018921,
"logits/rejected": 0.7773095965385437,
"logps/chosen": -92.13005065917969,
"logps/rejected": -91.37489318847656,
"loss": 0.6844,
"rewards/accuracies": 0.33125001192092896,
"rewards/chosen": -0.028016680851578712,
"rewards/margins": 0.021260341629385948,
"rewards/rejected": -0.04927702248096466,
"step": 860
},
{
"epoch": 1.82,
"learning_rate": 1.1731874863145143e-07,
"logits/chosen": 0.7595205903053284,
"logits/rejected": 0.8001340627670288,
"logps/chosen": -94.4384765625,
"logps/rejected": -92.37102508544922,
"loss": 0.687,
"rewards/accuracies": 0.3812499940395355,
"rewards/chosen": -0.026580199599266052,
"rewards/margins": 0.01631156913936138,
"rewards/rejected": -0.042891766875982285,
"step": 870
},
{
"epoch": 1.84,
"learning_rate": 9.120948298936422e-08,
"logits/chosen": 0.7684468030929565,
"logits/rejected": 0.8293699026107788,
"logps/chosen": -89.8713150024414,
"logps/rejected": -87.5567855834961,
"loss": 0.6844,
"rewards/accuracies": 0.34375,
"rewards/chosen": -0.026115071028470993,
"rewards/margins": 0.01522884052246809,
"rewards/rejected": -0.04134391248226166,
"step": 880
},
{
"epoch": 1.86,
"learning_rate": 6.832927412229017e-08,
"logits/chosen": 0.7296298742294312,
"logits/rejected": 0.769476056098938,
"logps/chosen": -93.165283203125,
"logps/rejected": -96.9769287109375,
"loss": 0.6851,
"rewards/accuracies": 0.34375,
"rewards/chosen": -0.029177119955420494,
"rewards/margins": 0.016743745654821396,
"rewards/rejected": -0.04592086747288704,
"step": 890
},
{
"epoch": 1.88,
"learning_rate": 4.870879364444109e-08,
"logits/chosen": 0.7372065782546997,
"logits/rejected": 0.7822612524032593,
"logps/chosen": -89.78585052490234,
"logps/rejected": -89.9983901977539,
"loss": 0.6866,
"rewards/accuracies": 0.3499999940395355,
"rewards/chosen": -0.028756320476531982,
"rewards/margins": 0.017704127356410027,
"rewards/rejected": -0.04646044969558716,
"step": 900
},
{
"epoch": 1.88,
"eval_logits/chosen": 0.732122004032135,
"eval_logits/rejected": 0.7528373599052429,
"eval_logps/chosen": -94.85501861572266,
"eval_logps/rejected": -98.56551361083984,
"eval_loss": 0.6872289776802063,
"eval_rewards/accuracies": 0.34200000762939453,
"eval_rewards/chosen": -0.03127633407711983,
"eval_rewards/margins": 0.012408819980919361,
"eval_rewards/rejected": -0.043685153126716614,
"eval_runtime": 272.6053,
"eval_samples_per_second": 7.337,
"eval_steps_per_second": 0.459,
"step": 900
},
{
"epoch": 1.9,
"learning_rate": 3.237434340521789e-08,
"logits/chosen": 0.710226833820343,
"logits/rejected": 0.7852008938789368,
"logps/chosen": -103.32108306884766,
"logps/rejected": -105.41300964355469,
"loss": 0.6851,
"rewards/accuracies": 0.35624998807907104,
"rewards/chosen": -0.030542368069291115,
"rewards/margins": 0.015573601238429546,
"rewards/rejected": -0.04611596092581749,
"step": 910
},
{
"epoch": 1.93,
"learning_rate": 1.93478202307823e-08,
"logits/chosen": 0.7026702165603638,
"logits/rejected": 0.7310872673988342,
"logps/chosen": -98.82904815673828,
"logps/rejected": -96.06184387207031,
"loss": 0.6866,
"rewards/accuracies": 0.33125001192092896,
"rewards/chosen": -0.02804415300488472,
"rewards/margins": 0.011173558421432972,
"rewards/rejected": -0.03921770304441452,
"step": 920
},
{
"epoch": 1.95,
"learning_rate": 9.646686570697062e-09,
"logits/chosen": 0.7164516448974609,
"logits/rejected": 0.8162258863449097,
"logps/chosen": -104.56522369384766,
"logps/rejected": -101.3372802734375,
"loss": 0.6871,
"rewards/accuracies": 0.3812499940395355,
"rewards/chosen": -0.024641428142786026,
"rewards/margins": 0.01806877739727497,
"rewards/rejected": -0.042710207402706146,
"step": 930
},
{
"epoch": 1.97,
"learning_rate": 3.283947088983663e-09,
"logits/chosen": 0.7736842632293701,
"logits/rejected": 0.7780998945236206,
"logps/chosen": -100.5300064086914,
"logps/rejected": -111.0368881225586,
"loss": 0.6875,
"rewards/accuracies": 0.38749998807907104,
"rewards/chosen": -0.0371762290596962,
"rewards/margins": 0.015147706493735313,
"rewards/rejected": -0.05232393741607666,
"step": 940
},
{
"epoch": 1.99,
"learning_rate": 2.681312309735229e-10,
"logits/chosen": 0.8036483526229858,
"logits/rejected": 0.8175506591796875,
"logps/chosen": -94.41046905517578,
"logps/rejected": -104.76774597167969,
"loss": 0.6875,
"rewards/accuracies": 0.3062500059604645,
"rewards/chosen": -0.03001193329691887,
"rewards/margins": 0.011342789977788925,
"rewards/rejected": -0.04135472699999809,
"step": 950
},
{
"epoch": 2.0,
"step": 954,
"total_flos": 0.0,
"train_loss": 0.3994182998029441,
"train_runtime": 8026.3391,
"train_samples_per_second": 7.617,
"train_steps_per_second": 0.119
}
],
"logging_steps": 10,
"max_steps": 954,
"num_input_tokens_seen": 0,
"num_train_epochs": 2,
"save_steps": 100,
"total_flos": 0.0,
"train_batch_size": 4,
"trial_name": null,
"trial_params": null
}