phi-2-dpo-ultrachat-lora / trainer_state.json
lole25's picture
Model save
b80f3e8 verified
raw
history blame
26.3 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.9984301412872841,
"eval_steps": 100,
"global_step": 477,
"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": 2.9492720416985004e-07,
"logits/chosen": 0.8186850547790527,
"logits/rejected": 0.8151271939277649,
"logps/chosen": -95.82102966308594,
"logps/rejected": -80.586669921875,
"loss": 0.6916,
"rewards/accuracies": 0.3375000059604645,
"rewards/chosen": -0.00794359389692545,
"rewards/margins": 0.003056485904380679,
"rewards/rejected": -0.011000080034136772,
"step": 410
},
{
"epoch": 0.88,
"learning_rate": 2.1464952759020857e-07,
"logits/chosen": 0.7571959495544434,
"logits/rejected": 0.8163139224052429,
"logps/chosen": -95.65677642822266,
"logps/rejected": -97.86299896240234,
"loss": 0.6915,
"rewards/accuracies": 0.3125,
"rewards/chosen": -0.00859010498970747,
"rewards/margins": 0.0028384437318891287,
"rewards/rejected": -0.011428548023104668,
"step": 420
},
{
"epoch": 0.9,
"learning_rate": 1.4662207078575685e-07,
"logits/chosen": 0.8645572662353516,
"logits/rejected": 0.887597918510437,
"logps/chosen": -83.41182708740234,
"logps/rejected": -90.17640686035156,
"loss": 0.6913,
"rewards/accuracies": 0.39375001192092896,
"rewards/chosen": -0.006618577986955643,
"rewards/margins": 0.004637080244719982,
"rewards/rejected": -0.011255658231675625,
"step": 430
},
{
"epoch": 0.92,
"learning_rate": 9.120948298936422e-08,
"logits/chosen": 0.7825466394424438,
"logits/rejected": 0.8328782916069031,
"logps/chosen": -109.79942321777344,
"logps/rejected": -82.07637023925781,
"loss": 0.6911,
"rewards/accuracies": 0.34375,
"rewards/chosen": -0.005336672533303499,
"rewards/margins": 0.005987245589494705,
"rewards/rejected": -0.011323917657136917,
"step": 440
},
{
"epoch": 0.94,
"learning_rate": 4.870879364444109e-08,
"logits/chosen": 0.774901270866394,
"logits/rejected": 0.8499285578727722,
"logps/chosen": -81.24304962158203,
"logps/rejected": -71.88755798339844,
"loss": 0.6911,
"rewards/accuracies": 0.29374998807907104,
"rewards/chosen": -0.005921828560531139,
"rewards/margins": 0.0035197760444134474,
"rewards/rejected": -0.00944160483777523,
"step": 450
},
{
"epoch": 0.96,
"learning_rate": 1.93478202307823e-08,
"logits/chosen": 0.8049672842025757,
"logits/rejected": 0.8570533990859985,
"logps/chosen": -105.7918930053711,
"logps/rejected": -87.53938293457031,
"loss": 0.6907,
"rewards/accuracies": 0.3187499940395355,
"rewards/chosen": -0.005085950251668692,
"rewards/margins": 0.0068057505413889885,
"rewards/rejected": -0.011891700327396393,
"step": 460
},
{
"epoch": 0.98,
"learning_rate": 3.283947088983663e-09,
"logits/chosen": 0.8561135530471802,
"logits/rejected": 0.8520036935806274,
"logps/chosen": -115.6650390625,
"logps/rejected": -110.88753509521484,
"loss": 0.6903,
"rewards/accuracies": 0.39375001192092896,
"rewards/chosen": -0.006030657794326544,
"rewards/margins": 0.00688832625746727,
"rewards/rejected": -0.012918984517455101,
"step": 470
},
{
"epoch": 1.0,
"step": 477,
"total_flos": 0.0,
"train_loss": 0.6920521804121805,
"train_runtime": 6813.2628,
"train_samples_per_second": 4.486,
"train_steps_per_second": 0.07
}
],
"logging_steps": 10,
"max_steps": 477,
"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
}