llava-onevision-qwen2-7b-ov-chat / trainer_state.json
luodian's picture
be0f70caefe53dd92deb6041cda716d9ca689fb7882a9c006dec3d49976dbb51
859f01f verified
raw
history blame
93.8 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.9991503823279524,
"eval_steps": 500,
"global_step": 147,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"grad_norm": 48.577925295489194,
"learning_rate": 3.3333333333333334e-08,
"logps/chosen": -67.63638305664062,
"logps/rejected": -55.72713851928711,
"loss": 0.6931,
"losses/dpo": 0.6931471824645996,
"losses/sft": 0.34192484617233276,
"losses/total": 0.6931471824645996,
"ref_logps/chosen": -67.63638305664062,
"ref_logps/rejected": -55.72713851928711,
"rewards/accuracies": 0.0,
"rewards/chosen": 0.0,
"rewards/margins": 0.0,
"rewards/rejected": 0.0,
"step": 1
},
{
"epoch": 0.01,
"grad_norm": 31.47832995229055,
"learning_rate": 6.666666666666667e-08,
"logps/chosen": -68.23696899414062,
"logps/rejected": -57.0877799987793,
"loss": 0.6931,
"losses/dpo": 0.6931471824645996,
"losses/sft": 0.35053497552871704,
"losses/total": 0.6931471824645996,
"ref_logps/chosen": -68.23696899414062,
"ref_logps/rejected": -57.0877799987793,
"rewards/accuracies": 0.0,
"rewards/chosen": 0.0,
"rewards/margins": 0.0,
"rewards/rejected": 0.0,
"step": 2
},
{
"epoch": 0.02,
"grad_norm": 50.46042718530731,
"learning_rate": 1e-07,
"logps/chosen": -63.95356750488281,
"logps/rejected": -56.19622802734375,
"loss": 0.6913,
"losses/dpo": 0.7034662961959839,
"losses/sft": 0.34674179553985596,
"losses/total": 0.7034662961959839,
"ref_logps/chosen": -64.00633239746094,
"ref_logps/rejected": -56.198883056640625,
"rewards/accuracies": 0.5,
"rewards/chosen": 0.005276843905448914,
"rewards/margins": 0.005011209286749363,
"rewards/rejected": 0.0002656341530382633,
"step": 3
},
{
"epoch": 0.03,
"grad_norm": 32.81062658932443,
"learning_rate": 1.3333333333333334e-07,
"logps/chosen": -61.69102096557617,
"logps/rejected": -54.53285217285156,
"loss": 0.6963,
"losses/dpo": 0.6837673783302307,
"losses/sft": 0.3563593626022339,
"losses/total": 0.6837673783302307,
"ref_logps/chosen": -61.65943145751953,
"ref_logps/rejected": -54.55246353149414,
"rewards/accuracies": 0.390625,
"rewards/chosen": -0.003159371204674244,
"rewards/margins": -0.005120169371366501,
"rewards/rejected": 0.0019607983995229006,
"step": 4
},
{
"epoch": 0.03,
"grad_norm": 33.80575482734628,
"learning_rate": 1.6666666666666665e-07,
"logps/chosen": -66.74890899658203,
"logps/rejected": -54.70701599121094,
"loss": 0.6925,
"losses/dpo": 0.6997189521789551,
"losses/sft": 0.3123997151851654,
"losses/total": 0.6997189521789551,
"ref_logps/chosen": -66.69550323486328,
"ref_logps/rejected": -54.629825592041016,
"rewards/accuracies": 0.53125,
"rewards/chosen": -0.005340744741261005,
"rewards/margins": 0.0023781214840710163,
"rewards/rejected": -0.00771886482834816,
"step": 5
},
{
"epoch": 0.04,
"grad_norm": 39.07641400536226,
"learning_rate": 2e-07,
"logps/chosen": -68.85008239746094,
"logps/rejected": -58.72641372680664,
"loss": 0.6966,
"losses/dpo": 0.6637319922447205,
"losses/sft": 0.3361469805240631,
"losses/total": 0.6637319922447205,
"ref_logps/chosen": -68.81605529785156,
"ref_logps/rejected": -58.74713134765625,
"rewards/accuracies": 0.453125,
"rewards/chosen": -0.003402549307793379,
"rewards/margins": -0.005474450532346964,
"rewards/rejected": 0.0020719014573842287,
"step": 6
},
{
"epoch": 0.05,
"grad_norm": 34.000425983037914,
"learning_rate": 2.3333333333333333e-07,
"logps/chosen": -61.8620491027832,
"logps/rejected": -51.837215423583984,
"loss": 0.694,
"losses/dpo": 0.6815477013587952,
"losses/sft": 0.34513574838638306,
"losses/total": 0.6815477013587952,
"ref_logps/chosen": -61.80104446411133,
"ref_logps/rejected": -51.780853271484375,
"rewards/accuracies": 0.515625,
"rewards/chosen": -0.006101294886320829,
"rewards/margins": -0.0004650282207876444,
"rewards/rejected": -0.005636268295347691,
"step": 7
},
{
"epoch": 0.05,
"grad_norm": 39.18548739683539,
"learning_rate": 2.6666666666666667e-07,
"logps/chosen": -65.5419921875,
"logps/rejected": -53.712867736816406,
"loss": 0.6864,
"losses/dpo": 0.6901150345802307,
"losses/sft": 0.3755946159362793,
"losses/total": 0.6901150345802307,
"ref_logps/chosen": -65.61614227294922,
"ref_logps/rejected": -53.64268493652344,
"rewards/accuracies": 0.65625,
"rewards/chosen": 0.007414784282445908,
"rewards/margins": 0.014433156698942184,
"rewards/rejected": -0.007018372416496277,
"step": 8
},
{
"epoch": 0.06,
"grad_norm": 34.05787792436153,
"learning_rate": 3e-07,
"logps/chosen": -66.96398162841797,
"logps/rejected": -52.923221588134766,
"loss": 0.7011,
"losses/dpo": 0.6919645071029663,
"losses/sft": 0.34879329800605774,
"losses/total": 0.6919645071029663,
"ref_logps/chosen": -66.94356536865234,
"ref_logps/rejected": -53.04799270629883,
"rewards/accuracies": 0.4375,
"rewards/chosen": -0.0020427405834198,
"rewards/margins": -0.014519838616251945,
"rewards/rejected": 0.012477097101509571,
"step": 9
},
{
"epoch": 0.07,
"grad_norm": 32.92454994636796,
"learning_rate": 3.333333333333333e-07,
"logps/chosen": -68.3951416015625,
"logps/rejected": -53.54661178588867,
"loss": 0.6903,
"losses/dpo": 0.6679682731628418,
"losses/sft": 0.3381645083427429,
"losses/total": 0.6679682731628418,
"ref_logps/chosen": -68.44535827636719,
"ref_logps/rejected": -53.52064895629883,
"rewards/accuracies": 0.515625,
"rewards/chosen": 0.005021924152970314,
"rewards/margins": 0.007618834264576435,
"rewards/rejected": -0.002596910111606121,
"step": 10
},
{
"epoch": 0.07,
"grad_norm": 28.62792358507012,
"learning_rate": 3.666666666666666e-07,
"logps/chosen": -58.64153289794922,
"logps/rejected": -49.53267288208008,
"loss": 0.6999,
"losses/dpo": 0.6928383111953735,
"losses/sft": 0.34639859199523926,
"losses/total": 0.6928383111953735,
"ref_logps/chosen": -58.6944465637207,
"ref_logps/rejected": -49.710044860839844,
"rewards/accuracies": 0.375,
"rewards/chosen": 0.005291242152452469,
"rewards/margins": -0.012445551343262196,
"rewards/rejected": 0.01773679442703724,
"step": 11
},
{
"epoch": 0.08,
"grad_norm": 34.95324550960433,
"learning_rate": 4e-07,
"logps/chosen": -63.45814514160156,
"logps/rejected": -51.141937255859375,
"loss": 0.6926,
"losses/dpo": 0.6734712719917297,
"losses/sft": 0.38542595505714417,
"losses/total": 0.6734712719917297,
"ref_logps/chosen": -63.53767395019531,
"ref_logps/rejected": -51.197601318359375,
"rewards/accuracies": 0.5,
"rewards/chosen": 0.007953131571412086,
"rewards/margins": 0.0023870535660535097,
"rewards/rejected": 0.00556607823818922,
"step": 12
},
{
"epoch": 0.09,
"grad_norm": 40.49931666997992,
"learning_rate": 4.3333333333333335e-07,
"logps/chosen": -72.46746826171875,
"logps/rejected": -58.34828186035156,
"loss": 0.6943,
"losses/dpo": 0.6939187049865723,
"losses/sft": 0.3672393262386322,
"losses/total": 0.6939187049865723,
"ref_logps/chosen": -72.57909393310547,
"ref_logps/rejected": -58.46982955932617,
"rewards/accuracies": 0.421875,
"rewards/chosen": 0.011162188835442066,
"rewards/margins": -0.000992664834484458,
"rewards/rejected": 0.012154852971434593,
"step": 13
},
{
"epoch": 0.1,
"grad_norm": 35.43728228028493,
"learning_rate": 4.6666666666666666e-07,
"logps/chosen": -60.627281188964844,
"logps/rejected": -52.48558807373047,
"loss": 0.6921,
"losses/dpo": 0.6789003014564514,
"losses/sft": 0.3384055495262146,
"losses/total": 0.6789003014564514,
"ref_logps/chosen": -60.64947509765625,
"ref_logps/rejected": -52.47667694091797,
"rewards/accuracies": 0.484375,
"rewards/chosen": 0.0022186655551195145,
"rewards/margins": 0.003109627403318882,
"rewards/rejected": -0.0008909617317840457,
"step": 14
},
{
"epoch": 0.1,
"grad_norm": 37.2571656085439,
"learning_rate": 5e-07,
"logps/chosen": -64.5811538696289,
"logps/rejected": -56.402488708496094,
"loss": 0.6889,
"losses/dpo": 0.6773965358734131,
"losses/sft": 0.32888567447662354,
"losses/total": 0.6773965358734131,
"ref_logps/chosen": -64.77919006347656,
"ref_logps/rejected": -56.505523681640625,
"rewards/accuracies": 0.578125,
"rewards/chosen": 0.019803302362561226,
"rewards/margins": 0.009500503540039062,
"rewards/rejected": 0.010302798822522163,
"step": 15
},
{
"epoch": 0.11,
"grad_norm": 37.083526947632535,
"learning_rate": 4.999291986732822e-07,
"logps/chosen": -67.24109649658203,
"logps/rejected": -54.730838775634766,
"loss": 0.6882,
"losses/dpo": 0.6811863780021667,
"losses/sft": 0.32620126008987427,
"losses/total": 0.6811863780021667,
"ref_logps/chosen": -67.33450317382812,
"ref_logps/rejected": -54.712337493896484,
"rewards/accuracies": 0.546875,
"rewards/chosen": 0.009341094642877579,
"rewards/margins": 0.011191939935088158,
"rewards/rejected": -0.0018508443608880043,
"step": 16
},
{
"epoch": 0.12,
"grad_norm": 31.937569243375922,
"learning_rate": 4.99716834795752e-07,
"logps/chosen": -63.45280456542969,
"logps/rejected": -52.51921463012695,
"loss": 0.6823,
"losses/dpo": 0.678799569606781,
"losses/sft": 0.3536127507686615,
"losses/total": 0.678799569606781,
"ref_logps/chosen": -63.669334411621094,
"ref_logps/rejected": -52.50813674926758,
"rewards/accuracies": 0.65625,
"rewards/chosen": 0.021653007715940475,
"rewards/margins": 0.022760074585676193,
"rewards/rejected": -0.0011070647742599249,
"step": 17
},
{
"epoch": 0.12,
"grad_norm": 40.33340375657743,
"learning_rate": 4.993630286525634e-07,
"logps/chosen": -67.25548553466797,
"logps/rejected": -56.92215347290039,
"loss": 0.6874,
"losses/dpo": 0.6505956053733826,
"losses/sft": 0.3500673770904541,
"losses/total": 0.6505956053733826,
"ref_logps/chosen": -67.41217041015625,
"ref_logps/rejected": -56.942283630371094,
"rewards/accuracies": 0.578125,
"rewards/chosen": 0.01566872000694275,
"rewards/margins": 0.013655520044267178,
"rewards/rejected": 0.0020132004283368587,
"step": 18
},
{
"epoch": 0.13,
"grad_norm": 38.0338752718148,
"learning_rate": 4.988679806432711e-07,
"logps/chosen": -68.8526611328125,
"logps/rejected": -57.075347900390625,
"loss": 0.6723,
"losses/dpo": 0.6660802960395813,
"losses/sft": 0.34118345379829407,
"losses/total": 0.6660802960395813,
"ref_logps/chosen": -69.11880493164062,
"ref_logps/rejected": -56.906070709228516,
"rewards/accuracies": 0.640625,
"rewards/chosen": 0.026613827794790268,
"rewards/margins": 0.04354187101125717,
"rewards/rejected": -0.016928039491176605,
"step": 19
},
{
"epoch": 0.14,
"grad_norm": 35.61816166775489,
"learning_rate": 4.982319711683221e-07,
"logps/chosen": -72.06625366210938,
"logps/rejected": -59.52400207519531,
"loss": 0.6721,
"losses/dpo": 0.6615892052650452,
"losses/sft": 0.34661561250686646,
"losses/total": 0.6615892052650452,
"ref_logps/chosen": -72.32228088378906,
"ref_logps/rejected": -59.32331848144531,
"rewards/accuracies": 0.65625,
"rewards/chosen": 0.02560277283191681,
"rewards/margins": 0.04567151144146919,
"rewards/rejected": -0.02006874606013298,
"step": 20
},
{
"epoch": 0.14,
"grad_norm": 35.39098813520557,
"learning_rate": 4.974553604702332e-07,
"logps/chosen": -69.02694702148438,
"logps/rejected": -56.37590789794922,
"loss": 0.6788,
"losses/dpo": 0.6626015901565552,
"losses/sft": 0.3223264813423157,
"losses/total": 0.6626015901565552,
"ref_logps/chosen": -69.218994140625,
"ref_logps/rejected": -56.257652282714844,
"rewards/accuracies": 0.625,
"rewards/chosen": 0.01920466497540474,
"rewards/margins": 0.031030891463160515,
"rewards/rejected": -0.011826226487755775,
"step": 21
},
{
"epoch": 0.15,
"grad_norm": 33.9907845041531,
"learning_rate": 4.965385884295466e-07,
"logps/chosen": -58.412986755371094,
"logps/rejected": -53.20189666748047,
"loss": 0.6719,
"losses/dpo": 0.6767945289611816,
"losses/sft": 0.3387165367603302,
"losses/total": 0.6767945289611816,
"ref_logps/chosen": -58.631317138671875,
"ref_logps/rejected": -52.960777282714844,
"rewards/accuracies": 0.640625,
"rewards/chosen": 0.021833162754774094,
"rewards/margins": 0.045944713056087494,
"rewards/rejected": -0.024111557751893997,
"step": 22
},
{
"epoch": 0.16,
"grad_norm": 33.80480054954672,
"learning_rate": 4.954821743156767e-07,
"logps/chosen": -67.24698638916016,
"logps/rejected": -56.11846160888672,
"loss": 0.6741,
"losses/dpo": 0.6613680124282837,
"losses/sft": 0.34880080819129944,
"losses/total": 0.6613680124282837,
"ref_logps/chosen": -67.44197082519531,
"ref_logps/rejected": -55.9046630859375,
"rewards/accuracies": 0.609375,
"rewards/chosen": 0.019499044865369797,
"rewards/margins": 0.040879033505916595,
"rewards/rejected": -0.0213799886405468,
"step": 23
},
{
"epoch": 0.16,
"grad_norm": 33.73597666461483,
"learning_rate": 4.942867164927899e-07,
"logps/chosen": -62.87995910644531,
"logps/rejected": -55.92186737060547,
"loss": 0.6554,
"losses/dpo": 0.6621065139770508,
"losses/sft": 0.350619375705719,
"losses/total": 0.6621065139770508,
"ref_logps/chosen": -63.19189453125,
"ref_logps/rejected": -55.41385269165039,
"rewards/accuracies": 0.6875,
"rewards/chosen": 0.031193653121590614,
"rewards/margins": 0.08199521154165268,
"rewards/rejected": -0.05080154910683632,
"step": 24
},
{
"epoch": 0.17,
"grad_norm": 31.252590331214783,
"learning_rate": 4.929528920808854e-07,
"logps/chosen": -70.07772064208984,
"logps/rejected": -58.75320816040039,
"loss": 0.6566,
"losses/dpo": 0.646221935749054,
"losses/sft": 0.3677119016647339,
"losses/total": 0.646221935749054,
"ref_logps/chosen": -70.34449005126953,
"ref_logps/rejected": -58.24370574951172,
"rewards/accuracies": 0.75,
"rewards/chosen": 0.026676896959543228,
"rewards/margins": 0.07762698084115982,
"rewards/rejected": -0.05095008388161659,
"step": 25
},
{
"epoch": 0.18,
"grad_norm": 31.628164542125273,
"learning_rate": 4.91481456572267e-07,
"logps/chosen": -63.15786361694336,
"logps/rejected": -51.286197662353516,
"loss": 0.6378,
"losses/dpo": 0.6170041561126709,
"losses/sft": 0.36194464564323425,
"losses/total": 0.6170041561126709,
"ref_logps/chosen": -63.57653045654297,
"ref_logps/rejected": -50.502952575683594,
"rewards/accuracies": 0.796875,
"rewards/chosen": 0.04186610132455826,
"rewards/margins": 0.12019069492816925,
"rewards/rejected": -0.07832459360361099,
"step": 26
},
{
"epoch": 0.18,
"grad_norm": 29.472445320476258,
"learning_rate": 4.898732434036243e-07,
"logps/chosen": -60.576107025146484,
"logps/rejected": -51.86144256591797,
"loss": 0.6304,
"losses/dpo": 0.6577066779136658,
"losses/sft": 0.31096988916397095,
"losses/total": 0.6577066779136658,
"ref_logps/chosen": -61.04180145263672,
"ref_logps/rejected": -50.955745697021484,
"rewards/accuracies": 0.828125,
"rewards/chosen": 0.04656900465488434,
"rewards/margins": 0.13713900744915009,
"rewards/rejected": -0.09057000279426575,
"step": 27
},
{
"epoch": 0.19,
"grad_norm": 34.12892875093247,
"learning_rate": 4.881291634839651e-07,
"logps/chosen": -71.571044921875,
"logps/rejected": -59.838462829589844,
"loss": 0.6348,
"losses/dpo": 0.6427409052848816,
"losses/sft": 0.370533287525177,
"losses/total": 0.6427409052848816,
"ref_logps/chosen": -71.80653381347656,
"ref_logps/rejected": -58.735103607177734,
"rewards/accuracies": 0.640625,
"rewards/chosen": 0.023549124598503113,
"rewards/margins": 0.1338854283094406,
"rewards/rejected": -0.1103362962603569,
"step": 28
},
{
"epoch": 0.2,
"grad_norm": 32.265156441586456,
"learning_rate": 4.86250204678667e-07,
"logps/chosen": -65.21573638916016,
"logps/rejected": -56.004364013671875,
"loss": 0.6235,
"losses/dpo": 0.6029931902885437,
"losses/sft": 0.3526380658149719,
"losses/total": 0.6029931902885437,
"ref_logps/chosen": -65.50960540771484,
"ref_logps/rejected": -54.726261138916016,
"rewards/accuracies": 0.8125,
"rewards/chosen": 0.029387805610895157,
"rewards/margins": 0.15719717741012573,
"rewards/rejected": -0.12780936062335968,
"step": 29
},
{
"epoch": 0.2,
"grad_norm": 37.3823888174941,
"learning_rate": 4.842374312499405e-07,
"logps/chosen": -73.87957763671875,
"logps/rejected": -60.16520309448242,
"loss": 0.5862,
"losses/dpo": 0.5859757661819458,
"losses/sft": 0.34089845418930054,
"losses/total": 0.5859757661819458,
"ref_logps/chosen": -74.58474731445312,
"ref_logps/rejected": -58.38753890991211,
"rewards/accuracies": 0.796875,
"rewards/chosen": 0.0705181285738945,
"rewards/margins": 0.24828468263149261,
"rewards/rejected": -0.1777665764093399,
"step": 30
},
{
"epoch": 0.21,
"grad_norm": 32.144180117174194,
"learning_rate": 4.820919832540181e-07,
"logps/chosen": -65.9237060546875,
"logps/rejected": -57.733497619628906,
"loss": 0.6071,
"losses/dpo": 0.6222472190856934,
"losses/sft": 0.33616167306900024,
"losses/total": 0.6222472190856934,
"ref_logps/chosen": -65.80138397216797,
"ref_logps/rejected": -55.510379791259766,
"rewards/accuracies": 0.71875,
"rewards/chosen": -0.012232396751642227,
"rewards/margins": 0.21007922291755676,
"rewards/rejected": -0.22231163084506989,
"step": 31
},
{
"epoch": 0.22,
"grad_norm": 35.501453167718694,
"learning_rate": 4.798150758954164e-07,
"logps/chosen": -68.75838470458984,
"logps/rejected": -60.68508529663086,
"loss": 0.5798,
"losses/dpo": 0.5868023037910461,
"losses/sft": 0.3538406789302826,
"losses/total": 0.5868023037910461,
"ref_logps/chosen": -68.71626281738281,
"ref_logps/rejected": -57.89375305175781,
"rewards/accuracies": 0.765625,
"rewards/chosen": -0.004212833940982819,
"rewards/margins": 0.2749200761318207,
"rewards/rejected": -0.2791329324245453,
"step": 32
},
{
"epoch": 0.22,
"grad_norm": 32.699996982131026,
"learning_rate": 4.774079988386296e-07,
"logps/chosen": -70.96333312988281,
"logps/rejected": -61.67507553100586,
"loss": 0.5944,
"losses/dpo": 0.5362982749938965,
"losses/sft": 0.37232011556625366,
"losses/total": 0.5362982749938965,
"ref_logps/chosen": -70.30509948730469,
"ref_logps/rejected": -58.60476303100586,
"rewards/accuracies": 0.765625,
"rewards/chosen": -0.06582287698984146,
"rewards/margins": 0.24120771884918213,
"rewards/rejected": -0.3070306181907654,
"step": 33
},
{
"epoch": 0.23,
"grad_norm": 37.488816567711076,
"learning_rate": 4.748721154776508e-07,
"logps/chosen": -65.87406921386719,
"logps/rejected": -59.7636604309082,
"loss": 0.5652,
"losses/dpo": 0.500527024269104,
"losses/sft": 0.3647816479206085,
"losses/total": 0.500527024269104,
"ref_logps/chosen": -65.68878173828125,
"ref_logps/rejected": -56.097267150878906,
"rewards/accuracies": 0.734375,
"rewards/chosen": -0.018529564142227173,
"rewards/margins": 0.3481099605560303,
"rewards/rejected": -0.36663955450057983,
"step": 34
},
{
"epoch": 0.24,
"grad_norm": 33.919463466183984,
"learning_rate": 4.7220886216373085e-07,
"logps/chosen": -67.95443725585938,
"logps/rejected": -64.25914001464844,
"loss": 0.5351,
"losses/dpo": 0.5877989530563354,
"losses/sft": 0.3324495255947113,
"losses/total": 0.5877989530563354,
"ref_logps/chosen": -67.34272766113281,
"ref_logps/rejected": -59.69349670410156,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.06117025017738342,
"rewards/margins": 0.3953939974308014,
"rewards/rejected": -0.4565642774105072,
"step": 35
},
{
"epoch": 0.24,
"grad_norm": 36.85611869447123,
"learning_rate": 4.694197473918139e-07,
"logps/chosen": -73.41097259521484,
"logps/rejected": -63.35462188720703,
"loss": 0.5958,
"losses/dpo": 0.5920328497886658,
"losses/sft": 0.34827345609664917,
"losses/total": 0.5920328497886658,
"ref_logps/chosen": -72.06607055664062,
"ref_logps/rejected": -59.26370620727539,
"rewards/accuracies": 0.671875,
"rewards/chosen": -0.1344902515411377,
"rewards/margins": 0.2746003270149231,
"rewards/rejected": -0.4090905785560608,
"step": 36
},
{
"epoch": 0.25,
"grad_norm": 29.396807219224034,
"learning_rate": 4.6650635094610966e-07,
"logps/chosen": -69.74252319335938,
"logps/rejected": -59.216861724853516,
"loss": 0.511,
"losses/dpo": 0.4569849967956543,
"losses/sft": 0.330870658159256,
"losses/total": 0.4569849967956543,
"ref_logps/chosen": -68.98128509521484,
"ref_logps/rejected": -53.205997467041016,
"rewards/accuracies": 0.84375,
"rewards/chosen": -0.07612372934818268,
"rewards/margins": 0.5249630212783813,
"rewards/rejected": -0.6010866761207581,
"step": 37
},
{
"epoch": 0.26,
"grad_norm": 32.158672339711444,
"learning_rate": 4.63470323005287e-07,
"logps/chosen": -74.17395782470703,
"logps/rejected": -65.09367370605469,
"loss": 0.4616,
"losses/dpo": 0.4395897388458252,
"losses/sft": 0.3282167613506317,
"losses/total": 0.4395897388458252,
"ref_logps/chosen": -72.66117858886719,
"ref_logps/rejected": -56.74953079223633,
"rewards/accuracies": 0.859375,
"rewards/chosen": -0.15127676725387573,
"rewards/margins": 0.6831383109092712,
"rewards/rejected": -0.8344150185585022,
"step": 38
},
{
"epoch": 0.27,
"grad_norm": 30.39214935377121,
"learning_rate": 4.603133832077953e-07,
"logps/chosen": -72.17027282714844,
"logps/rejected": -66.1070785522461,
"loss": 0.5353,
"losses/dpo": 0.47761982679367065,
"losses/sft": 0.38627445697784424,
"losses/total": 0.47761982679367065,
"ref_logps/chosen": -69.07978820800781,
"ref_logps/rejected": -58.215328216552734,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.3090481758117676,
"rewards/margins": 0.4801270365715027,
"rewards/rejected": -0.7891752123832703,
"step": 39
},
{
"epoch": 0.27,
"grad_norm": 29.86640568721065,
"learning_rate": 4.5703731967784265e-07,
"logps/chosen": -61.24689483642578,
"logps/rejected": -59.919918060302734,
"loss": 0.5273,
"losses/dpo": 0.7497649192810059,
"losses/sft": 0.35406211018562317,
"losses/total": 0.7497649192810059,
"ref_logps/chosen": -57.38946533203125,
"ref_logps/rejected": -50.80248260498047,
"rewards/accuracies": 0.734375,
"rewards/chosen": -0.3857432007789612,
"rewards/margins": 0.5260001420974731,
"rewards/rejected": -0.9117434024810791,
"step": 40
},
{
"epoch": 0.28,
"grad_norm": 29.500472780356045,
"learning_rate": 4.5364398801258394e-07,
"logps/chosen": -63.82752990722656,
"logps/rejected": -60.21675491333008,
"loss": 0.5103,
"losses/dpo": 0.5127725601196289,
"losses/sft": 0.3456054925918579,
"losses/total": 0.5127725601196289,
"ref_logps/chosen": -60.51768493652344,
"ref_logps/rejected": -50.69208526611328,
"rewards/accuracies": 0.84375,
"rewards/chosen": -0.3309844732284546,
"rewards/margins": 0.6214821934700012,
"rewards/rejected": -0.952466607093811,
"step": 41
},
{
"epoch": 0.29,
"grad_norm": 28.303717346192418,
"learning_rate": 4.501353102310901e-07,
"logps/chosen": -64.48641967773438,
"logps/rejected": -60.165767669677734,
"loss": 0.5197,
"losses/dpo": 0.4626217782497406,
"losses/sft": 0.3867911994457245,
"losses/total": 0.4626217782497406,
"ref_logps/chosen": -60.505496978759766,
"ref_logps/rejected": -50.563045501708984,
"rewards/accuracies": 0.765625,
"rewards/chosen": -0.3980926275253296,
"rewards/margins": 0.562179684638977,
"rewards/rejected": -0.9602723121643066,
"step": 42
},
{
"epoch": 0.29,
"grad_norm": 26.546916183591545,
"learning_rate": 4.4651327368569684e-07,
"logps/chosen": -74.27953338623047,
"logps/rejected": -66.18184661865234,
"loss": 0.478,
"losses/dpo": 0.5494570136070251,
"losses/sft": 0.3961746394634247,
"losses/total": 0.5494570136070251,
"ref_logps/chosen": -68.87181854248047,
"ref_logps/rejected": -54.055946350097656,
"rewards/accuracies": 0.84375,
"rewards/chosen": -0.5407715439796448,
"rewards/margins": 0.671818733215332,
"rewards/rejected": -1.212590217590332,
"step": 43
},
{
"epoch": 0.3,
"grad_norm": 26.197389360540555,
"learning_rate": 4.4277992993634686e-07,
"logps/chosen": -79.60818481445312,
"logps/rejected": -79.91624450683594,
"loss": 0.4052,
"losses/dpo": 0.4185926914215088,
"losses/sft": 0.3548857569694519,
"losses/total": 0.4185926914215088,
"ref_logps/chosen": -72.74970245361328,
"ref_logps/rejected": -62.91980743408203,
"rewards/accuracies": 0.84375,
"rewards/chosen": -0.6858479976654053,
"rewards/margins": 1.0137956142425537,
"rewards/rejected": -1.699643611907959,
"step": 44
},
{
"epoch": 0.31,
"grad_norm": 27.708445312624427,
"learning_rate": 4.3893739358856455e-07,
"logps/chosen": -79.15251159667969,
"logps/rejected": -72.04096221923828,
"loss": 0.4729,
"losses/dpo": 0.4918556213378906,
"losses/sft": 0.37935709953308105,
"losses/total": 0.4918556213378906,
"ref_logps/chosen": -72.23287963867188,
"ref_logps/rejected": -56.32197570800781,
"rewards/accuracies": 0.71875,
"rewards/chosen": -0.6919631958007812,
"rewards/margins": 0.8799355626106262,
"rewards/rejected": -1.5718988180160522,
"step": 45
},
{
"epoch": 0.31,
"grad_norm": 30.962017558121154,
"learning_rate": 4.34987841095721e-07,
"logps/chosen": -75.75450134277344,
"logps/rejected": -75.48016357421875,
"loss": 0.4757,
"losses/dpo": 0.47784677147865295,
"losses/sft": 0.37939363718032837,
"losses/total": 0.47784677147865295,
"ref_logps/chosen": -67.21387481689453,
"ref_logps/rejected": -58.56239318847656,
"rewards/accuracies": 0.84375,
"rewards/chosen": -0.8540633320808411,
"rewards/margins": 0.8377144932746887,
"rewards/rejected": -1.6917778253555298,
"step": 46
},
{
"epoch": 0.32,
"grad_norm": 27.327055413855135,
"learning_rate": 4.309335095262675e-07,
"logps/chosen": -84.39332580566406,
"logps/rejected": -78.34506225585938,
"loss": 0.4763,
"losses/dpo": 0.3084137439727783,
"losses/sft": 0.3708834648132324,
"losses/total": 0.3084137439727783,
"ref_logps/chosen": -74.23382568359375,
"ref_logps/rejected": -59.50791931152344,
"rewards/accuracies": 0.671875,
"rewards/chosen": -1.0159505605697632,
"rewards/margins": 0.8677631616592407,
"rewards/rejected": -1.8837138414382935,
"step": 47
},
{
"epoch": 0.33,
"grad_norm": 37.60224677954145,
"learning_rate": 4.2677669529663686e-07,
"logps/chosen": -85.45326232910156,
"logps/rejected": -80.37303161621094,
"loss": 0.498,
"losses/dpo": 0.5172345638275146,
"losses/sft": 0.46193623542785645,
"losses/total": 0.5172345638275146,
"ref_logps/chosen": -76.32438659667969,
"ref_logps/rejected": -63.45566940307617,
"rewards/accuracies": 0.734375,
"rewards/chosen": -0.9128875732421875,
"rewards/margins": 0.7788479328155518,
"rewards/rejected": -1.6917355060577393,
"step": 48
},
{
"epoch": 0.33,
"grad_norm": 26.012407404622028,
"learning_rate": 4.22519752870528e-07,
"logps/chosen": -86.257080078125,
"logps/rejected": -79.96172332763672,
"loss": 0.4015,
"losses/dpo": 0.3850885331630707,
"losses/sft": 0.4463062882423401,
"losses/total": 0.3850885331630707,
"ref_logps/chosen": -77.25395965576172,
"ref_logps/rejected": -60.78544998168945,
"rewards/accuracies": 0.890625,
"rewards/chosen": -0.9003111124038696,
"rewards/margins": 1.0173161029815674,
"rewards/rejected": -1.9176270961761475,
"step": 49
},
{
"epoch": 0.34,
"grad_norm": 29.5339729693226,
"learning_rate": 4.1816509342531317e-07,
"logps/chosen": -72.80648803710938,
"logps/rejected": -73.68412780761719,
"loss": 0.4609,
"losses/dpo": 0.5073346495628357,
"losses/sft": 0.3690868318080902,
"losses/total": 0.5073346495628357,
"ref_logps/chosen": -61.75213623046875,
"ref_logps/rejected": -53.534000396728516,
"rewards/accuracies": 0.765625,
"rewards/chosen": -1.1054348945617676,
"rewards/margins": 0.9095778465270996,
"rewards/rejected": -2.015012741088867,
"step": 50
},
{
"epoch": 0.35,
"grad_norm": 24.67832174233356,
"learning_rate": 4.137151834863213e-07,
"logps/chosen": -78.76303100585938,
"logps/rejected": -81.20369720458984,
"loss": 0.3581,
"losses/dpo": 0.31153589487075806,
"losses/sft": 0.34659337997436523,
"losses/total": 0.31153589487075806,
"ref_logps/chosen": -67.64138793945312,
"ref_logps/rejected": -56.615203857421875,
"rewards/accuracies": 0.84375,
"rewards/chosen": -1.1121646165847778,
"rewards/margins": 1.3466840982437134,
"rewards/rejected": -2.458848714828491,
"step": 51
},
{
"epoch": 0.35,
"grad_norm": 25.652161578301136,
"learning_rate": 4.0917254352977206e-07,
"logps/chosen": -80.27860260009766,
"logps/rejected": -78.06929016113281,
"loss": 0.4625,
"losses/dpo": 0.5235142707824707,
"losses/sft": 0.3862667977809906,
"losses/total": 0.5235142707824707,
"ref_logps/chosen": -65.9322280883789,
"ref_logps/rejected": -54.390098571777344,
"rewards/accuracies": 0.8125,
"rewards/chosen": -1.4346369504928589,
"rewards/margins": 0.9332831501960754,
"rewards/rejected": -2.367920160293579,
"step": 52
},
{
"epoch": 0.36,
"grad_norm": 31.613726978757875,
"learning_rate": 4.045397465551513e-07,
"logps/chosen": -84.57078552246094,
"logps/rejected": -76.68920135498047,
"loss": 0.5117,
"losses/dpo": 0.4083520770072937,
"losses/sft": 0.32504379749298096,
"losses/total": 0.4083520770072937,
"ref_logps/chosen": -72.13128662109375,
"ref_logps/rejected": -53.99201965332031,
"rewards/accuracies": 0.78125,
"rewards/chosen": -1.2439489364624023,
"rewards/margins": 1.0257703065872192,
"rewards/rejected": -2.269719123840332,
"step": 53
},
{
"epoch": 0.37,
"grad_norm": 29.84992723438259,
"learning_rate": 3.998194166278367e-07,
"logps/chosen": -76.96097564697266,
"logps/rejected": -74.58714294433594,
"loss": 0.4777,
"losses/dpo": 0.4091704785823822,
"losses/sft": 0.4479122757911682,
"losses/total": 0.4091704785823822,
"ref_logps/chosen": -61.228057861328125,
"ref_logps/rejected": -50.48232650756836,
"rewards/accuracies": 0.71875,
"rewards/chosen": -1.573291540145874,
"rewards/margins": 0.8371899127960205,
"rewards/rejected": -2.4104812145233154,
"step": 54
},
{
"epoch": 0.37,
"grad_norm": 36.88672265696095,
"learning_rate": 3.9501422739279953e-07,
"logps/chosen": -81.66922760009766,
"logps/rejected": -75.91566467285156,
"loss": 0.6061,
"losses/dpo": 0.5785080194473267,
"losses/sft": 0.37718018889427185,
"losses/total": 0.5785080194473267,
"ref_logps/chosen": -67.54790496826172,
"ref_logps/rejected": -57.624385833740234,
"rewards/accuracies": 0.640625,
"rewards/chosen": -1.4121325016021729,
"rewards/margins": 0.4169955253601074,
"rewards/rejected": -1.8291279077529907,
"step": 55
},
{
"epoch": 0.38,
"grad_norm": 26.409540611725895,
"learning_rate": 3.9012690056022345e-07,
"logps/chosen": -88.31976318359375,
"logps/rejected": -80.9065170288086,
"loss": 0.3836,
"losses/dpo": 0.4441046714782715,
"losses/sft": 0.4260564148426056,
"losses/total": 0.4441046714782715,
"ref_logps/chosen": -74.53655242919922,
"ref_logps/rejected": -56.45116424560547,
"rewards/accuracies": 0.875,
"rewards/chosen": -1.3783206939697266,
"rewards/margins": 1.0672156810760498,
"rewards/rejected": -2.4455361366271973,
"step": 56
},
{
"epoch": 0.39,
"grad_norm": 26.34257247806165,
"learning_rate": 3.851602043638994e-07,
"logps/chosen": -80.94815826416016,
"logps/rejected": -77.87812805175781,
"loss": 0.4378,
"losses/dpo": 0.38587698340415955,
"losses/sft": 0.39670416712760925,
"losses/total": 0.38587698340415955,
"ref_logps/chosen": -64.29881286621094,
"ref_logps/rejected": -51.452911376953125,
"rewards/accuracies": 0.8125,
"rewards/chosen": -1.6649348735809326,
"rewards/margins": 0.9775880575180054,
"rewards/rejected": -2.6425228118896484,
"step": 57
},
{
"epoch": 0.39,
"grad_norm": 27.08511242764537,
"learning_rate": 3.8011695199326805e-07,
"logps/chosen": -80.61515808105469,
"logps/rejected": -81.2110366821289,
"loss": 0.4141,
"losses/dpo": 0.3337501883506775,
"losses/sft": 0.3477111756801605,
"losses/total": 0.3337501883506775,
"ref_logps/chosen": -66.6812744140625,
"ref_logps/rejected": -56.64939880371094,
"rewards/accuracies": 0.84375,
"rewards/chosen": -1.393389344215393,
"rewards/margins": 1.062773585319519,
"rewards/rejected": -2.456162929534912,
"step": 58
},
{
"epoch": 0.4,
"grad_norm": 29.443828368623393,
"learning_rate": 3.75e-07,
"logps/chosen": -79.15809631347656,
"logps/rejected": -77.99955749511719,
"loss": 0.4174,
"losses/dpo": 0.2030327320098877,
"losses/sft": 0.36094900965690613,
"losses/total": 0.2030327320098877,
"ref_logps/chosen": -64.46220397949219,
"ref_logps/rejected": -51.404056549072266,
"rewards/accuracies": 0.75,
"rewards/chosen": -1.4695887565612793,
"rewards/margins": 1.1899616718292236,
"rewards/rejected": -2.659550428390503,
"step": 59
},
{
"epoch": 0.41,
"grad_norm": 31.360451766133608,
"learning_rate": 3.698122466800142e-07,
"logps/chosen": -83.48946380615234,
"logps/rejected": -80.81828308105469,
"loss": 0.5084,
"losses/dpo": 0.6314846873283386,
"losses/sft": 0.4423762559890747,
"losses/total": 0.6314846873283386,
"ref_logps/chosen": -69.77751159667969,
"ref_logps/rejected": -59.079166412353516,
"rewards/accuracies": 0.734375,
"rewards/chosen": -1.3711949586868286,
"rewards/margins": 0.802715539932251,
"rewards/rejected": -2.173910617828369,
"step": 60
},
{
"epoch": 0.41,
"grad_norm": 32.77515833972762,
"learning_rate": 3.645566304318526e-07,
"logps/chosen": -88.54704284667969,
"logps/rejected": -83.86685180664062,
"loss": 0.4477,
"losses/dpo": 0.4280022382736206,
"losses/sft": 0.3504616618156433,
"losses/total": 0.4280022382736206,
"ref_logps/chosen": -75.23544311523438,
"ref_logps/rejected": -60.14040756225586,
"rewards/accuracies": 0.84375,
"rewards/chosen": -1.331160068511963,
"rewards/margins": 1.04148530960083,
"rewards/rejected": -2.372645378112793,
"step": 61
},
{
"epoch": 0.42,
"grad_norm": 27.946308305497556,
"learning_rate": 3.5923612809233984e-07,
"logps/chosen": -78.10906982421875,
"logps/rejected": -79.12262725830078,
"loss": 0.4413,
"losses/dpo": 0.4408959746360779,
"losses/sft": 0.3510644733905792,
"losses/total": 0.4408959746360779,
"ref_logps/chosen": -64.80330657958984,
"ref_logps/rejected": -56.04015350341797,
"rewards/accuracies": 0.828125,
"rewards/chosen": -1.3305763006210327,
"rewards/margins": 0.9776719212532043,
"rewards/rejected": -2.308248281478882,
"step": 62
},
{
"epoch": 0.43,
"grad_norm": 41.17153515866478,
"learning_rate": 3.5385375325047163e-07,
"logps/chosen": -80.082275390625,
"logps/rejected": -79.17640686035156,
"loss": 0.4871,
"losses/dpo": 0.45268240571022034,
"losses/sft": 0.3942197263240814,
"losses/total": 0.45268240571022034,
"ref_logps/chosen": -65.72177124023438,
"ref_logps/rejected": -55.820762634277344,
"rewards/accuracies": 0.765625,
"rewards/chosen": -1.4360499382019043,
"rewards/margins": 0.8995141983032227,
"rewards/rejected": -2.335564136505127,
"step": 63
},
{
"epoch": 0.44,
"grad_norm": 26.987470660431804,
"learning_rate": 3.4841255454048535e-07,
"logps/chosen": -78.40692138671875,
"logps/rejected": -81.3203125,
"loss": 0.4319,
"losses/dpo": 0.3416772782802582,
"losses/sft": 0.351028174161911,
"losses/total": 0.3416772782802582,
"ref_logps/chosen": -66.26089477539062,
"ref_logps/rejected": -58.60002136230469,
"rewards/accuracies": 0.84375,
"rewards/chosen": -1.214601755142212,
"rewards/margins": 1.057426929473877,
"rewards/rejected": -2.272028923034668,
"step": 64
},
{
"epoch": 0.44,
"grad_norm": 33.31694146367439,
"learning_rate": 3.4291561391508185e-07,
"logps/chosen": -82.72046661376953,
"logps/rejected": -85.01447296142578,
"loss": 0.494,
"losses/dpo": 0.4605589509010315,
"losses/sft": 0.3883049190044403,
"losses/total": 0.4605589509010315,
"ref_logps/chosen": -67.7024917602539,
"ref_logps/rejected": -59.51292419433594,
"rewards/accuracies": 0.703125,
"rewards/chosen": -1.5017977952957153,
"rewards/margins": 1.0483572483062744,
"rewards/rejected": -2.5501549243927,
"step": 65
},
{
"epoch": 0.45,
"grad_norm": 30.20641981592105,
"learning_rate": 3.373660448997746e-07,
"logps/chosen": -78.78214263916016,
"logps/rejected": -76.67687225341797,
"loss": 0.461,
"losses/dpo": 0.3649926483631134,
"losses/sft": 0.3876701891422272,
"losses/total": 0.3649926483631134,
"ref_logps/chosen": -65.60338592529297,
"ref_logps/rejected": -53.363441467285156,
"rewards/accuracies": 0.765625,
"rewards/chosen": -1.3178752660751343,
"rewards/margins": 1.0134679079055786,
"rewards/rejected": -2.331343173980713,
"step": 66
},
{
"epoch": 0.46,
"grad_norm": 28.499338933696293,
"learning_rate": 3.317669908293554e-07,
"logps/chosen": -80.8163070678711,
"logps/rejected": -78.53517150878906,
"loss": 0.4977,
"losses/dpo": 0.5431786775588989,
"losses/sft": 0.3635212779045105,
"losses/total": 0.5431786775588989,
"ref_logps/chosen": -67.38165283203125,
"ref_logps/rejected": -56.12194061279297,
"rewards/accuracies": 0.765625,
"rewards/chosen": -1.3434655666351318,
"rewards/margins": 0.8978573679924011,
"rewards/rejected": -2.2413227558135986,
"step": 67
},
{
"epoch": 0.46,
"grad_norm": 30.8426766786845,
"learning_rate": 3.261216230674768e-07,
"logps/chosen": -80.0118408203125,
"logps/rejected": -78.16852569580078,
"loss": 0.4656,
"losses/dpo": 0.22820590436458588,
"losses/sft": 0.41188591718673706,
"losses/total": 0.22820590436458588,
"ref_logps/chosen": -67.98802185058594,
"ref_logps/rejected": -55.998008728027344,
"rewards/accuracies": 0.765625,
"rewards/chosen": -1.2023829221725464,
"rewards/margins": 1.0146691799163818,
"rewards/rejected": -2.2170519828796387,
"step": 68
},
{
"epoch": 0.47,
"grad_norm": 26.8942337262588,
"learning_rate": 3.204331392103574e-07,
"logps/chosen": -76.27379608154297,
"logps/rejected": -72.5108413696289,
"loss": 0.3977,
"losses/dpo": 0.349028617143631,
"losses/sft": 0.36189785599708557,
"losses/total": 0.349028617143631,
"ref_logps/chosen": -65.81362915039062,
"ref_logps/rejected": -51.00428009033203,
"rewards/accuracies": 0.890625,
"rewards/chosen": -1.0460165739059448,
"rewards/margins": 1.1046404838562012,
"rewards/rejected": -2.1506569385528564,
"step": 69
},
{
"epoch": 0.48,
"grad_norm": 26.047792222109905,
"learning_rate": 3.147047612756302e-07,
"logps/chosen": -86.95272827148438,
"logps/rejected": -83.82672119140625,
"loss": 0.3953,
"losses/dpo": 0.46198731660842896,
"losses/sft": 0.4407782554626465,
"losses/total": 0.46198731660842896,
"ref_logps/chosen": -75.7898178100586,
"ref_logps/rejected": -60.74498748779297,
"rewards/accuracies": 0.8125,
"rewards/chosen": -1.1162903308868408,
"rewards/margins": 1.1918830871582031,
"rewards/rejected": -2.308173418045044,
"step": 70
},
{
"epoch": 0.48,
"grad_norm": 24.670894739162968,
"learning_rate": 3.0893973387735683e-07,
"logps/chosen": -70.71044921875,
"logps/rejected": -71.96138763427734,
"loss": 0.3944,
"losses/dpo": 0.29774680733680725,
"losses/sft": 0.3790546953678131,
"losses/total": 0.29774680733680725,
"ref_logps/chosen": -61.44975280761719,
"ref_logps/rejected": -51.039268493652344,
"rewards/accuracies": 0.859375,
"rewards/chosen": -0.9260695576667786,
"rewards/margins": 1.166142225265503,
"rewards/rejected": -2.092211961746216,
"step": 71
},
{
"epoch": 0.49,
"grad_norm": 28.239155495003125,
"learning_rate": 3.0314132238824415e-07,
"logps/chosen": -81.26232147216797,
"logps/rejected": -78.04046630859375,
"loss": 0.3912,
"losses/dpo": 0.5779224038124084,
"losses/sft": 0.3678759038448334,
"losses/total": 0.5779224038124084,
"ref_logps/chosen": -72.41909790039062,
"ref_logps/rejected": -57.58437728881836,
"rewards/accuracies": 0.859375,
"rewards/chosen": -0.8843216300010681,
"rewards/margins": 1.1612873077392578,
"rewards/rejected": -2.0456089973449707,
"step": 72
},
{
"epoch": 0.5,
"grad_norm": 23.65680821014391,
"learning_rate": 2.9731281109010253e-07,
"logps/chosen": -78.29800415039062,
"logps/rejected": -76.63243103027344,
"loss": 0.3589,
"losses/dpo": 0.39184096455574036,
"losses/sft": 0.35749632120132446,
"losses/total": 0.39184096455574036,
"ref_logps/chosen": -70.57292175292969,
"ref_logps/rejected": -56.80116653442383,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.7725088000297546,
"rewards/margins": 1.210618257522583,
"rewards/rejected": -1.9831271171569824,
"step": 73
},
{
"epoch": 0.5,
"grad_norm": 57.763902094880514,
"learning_rate": 2.9145750131359355e-07,
"logps/chosen": -71.69022369384766,
"logps/rejected": -74.53846740722656,
"loss": 0.4711,
"losses/dpo": 0.3447693884372711,
"losses/sft": 0.3251613676548004,
"losses/total": 0.3447693884372711,
"ref_logps/chosen": -61.327369689941406,
"ref_logps/rejected": -53.874820709228516,
"rewards/accuracies": 0.75,
"rewards/chosen": -1.036285638809204,
"rewards/margins": 1.0300801992416382,
"rewards/rejected": -2.0663657188415527,
"step": 74
},
{
"epoch": 0.51,
"grad_norm": 37.61505605094831,
"learning_rate": 2.8557870956832133e-07,
"logps/chosen": -78.79322052001953,
"logps/rejected": -73.63280487060547,
"loss": 0.4466,
"losses/dpo": 0.4499085545539856,
"losses/sft": 0.3389120101928711,
"losses/total": 0.4499085545539856,
"ref_logps/chosen": -70.82493591308594,
"ref_logps/rejected": -56.25859832763672,
"rewards/accuracies": 0.8125,
"rewards/chosen": -0.7968282699584961,
"rewards/margins": 0.9405927658081055,
"rewards/rejected": -1.7374211549758911,
"step": 75
},
{
"epoch": 0.52,
"grad_norm": 25.268092842517042,
"learning_rate": 2.7967976566432627e-07,
"logps/chosen": -79.96376037597656,
"logps/rejected": -75.98302459716797,
"loss": 0.3581,
"losses/dpo": 0.3253936469554901,
"losses/sft": 0.3978617191314697,
"losses/total": 0.3253936469554901,
"ref_logps/chosen": -70.92337036132812,
"ref_logps/rejected": -53.73899841308594,
"rewards/accuracies": 0.84375,
"rewards/chosen": -0.9040391445159912,
"rewards/margins": 1.320363163948059,
"rewards/rejected": -2.2244021892547607,
"step": 76
},
{
"epoch": 0.52,
"grad_norm": 24.79898935884772,
"learning_rate": 2.737640108260456e-07,
"logps/chosen": -74.60429382324219,
"logps/rejected": -74.33484649658203,
"loss": 0.3639,
"losses/dpo": 0.19668929278850555,
"losses/sft": 0.34319061040878296,
"losses/total": 0.19668929278850555,
"ref_logps/chosen": -67.25056457519531,
"ref_logps/rejected": -54.8765869140625,
"rewards/accuracies": 0.921875,
"rewards/chosen": -0.7353739142417908,
"rewards/margins": 1.2104520797729492,
"rewards/rejected": -1.9458260536193848,
"step": 77
},
{
"epoch": 0.53,
"grad_norm": 32.80096075705349,
"learning_rate": 2.6783479579980807e-07,
"logps/chosen": -73.52947998046875,
"logps/rejected": -75.42818450927734,
"loss": 0.4477,
"losses/dpo": 0.4471086859703064,
"losses/sft": 0.3155350685119629,
"losses/total": 0.4471086859703064,
"ref_logps/chosen": -64.02719116210938,
"ref_logps/rejected": -56.57443618774414,
"rewards/accuracies": 0.828125,
"rewards/chosen": -0.9502288103103638,
"rewards/margins": 0.9351459741592407,
"rewards/rejected": -1.885374903678894,
"step": 78
},
{
"epoch": 0.54,
"grad_norm": 30.01631729835111,
"learning_rate": 2.618954789559356e-07,
"logps/chosen": -78.71662902832031,
"logps/rejected": -70.70683288574219,
"loss": 0.4583,
"losses/dpo": 0.4225349426269531,
"losses/sft": 0.37885919213294983,
"losses/total": 0.4225349426269531,
"ref_logps/chosen": -70.06804656982422,
"ref_logps/rejected": -53.02828598022461,
"rewards/accuracies": 0.78125,
"rewards/chosen": -0.8648585677146912,
"rewards/margins": 0.9029962420463562,
"rewards/rejected": -1.7678548097610474,
"step": 79
},
{
"epoch": 0.54,
"grad_norm": 27.778753710279016,
"learning_rate": 2.5594942438652685e-07,
"logps/chosen": -74.57829284667969,
"logps/rejected": -75.41488647460938,
"loss": 0.4252,
"losses/dpo": 0.4811527132987976,
"losses/sft": 0.3381648361682892,
"losses/total": 0.4811527132987976,
"ref_logps/chosen": -65.43878173828125,
"ref_logps/rejected": -56.21765899658203,
"rewards/accuracies": 0.78125,
"rewards/chosen": -0.9139504432678223,
"rewards/margins": 1.005772352218628,
"rewards/rejected": -1.9197226762771606,
"step": 80
},
{
"epoch": 0.55,
"grad_norm": 27.418542939543443,
"learning_rate": 2.5e-07,
"logps/chosen": -84.89808654785156,
"logps/rejected": -77.45452117919922,
"loss": 0.457,
"losses/dpo": 0.4078216552734375,
"losses/sft": 0.3400785028934479,
"losses/total": 0.4078216552734375,
"ref_logps/chosen": -78.10498046875,
"ref_logps/rejected": -61.716976165771484,
"rewards/accuracies": 0.78125,
"rewards/chosen": -0.6793106198310852,
"rewards/margins": 0.8944438695907593,
"rewards/rejected": -1.5737544298171997,
"step": 81
},
{
"epoch": 0.56,
"grad_norm": 29.24936699437702,
"learning_rate": 2.4405057561347313e-07,
"logps/chosen": -71.50326538085938,
"logps/rejected": -74.53978729248047,
"loss": 0.496,
"losses/dpo": 0.3848089277744293,
"losses/sft": 0.3731555938720703,
"losses/total": 0.3848089277744293,
"ref_logps/chosen": -60.67548370361328,
"ref_logps/rejected": -55.07884216308594,
"rewards/accuracies": 0.75,
"rewards/chosen": -1.0827784538269043,
"rewards/margins": 0.8633174896240234,
"rewards/rejected": -1.9460958242416382,
"step": 82
},
{
"epoch": 0.56,
"grad_norm": 48.23269343517185,
"learning_rate": 2.381045210440644e-07,
"logps/chosen": -78.87052917480469,
"logps/rejected": -73.31388092041016,
"loss": 0.4528,
"losses/dpo": 0.43188297748565674,
"losses/sft": 0.36860212683677673,
"losses/total": 0.43188297748565674,
"ref_logps/chosen": -70.29788970947266,
"ref_logps/rejected": -54.075828552246094,
"rewards/accuracies": 0.796875,
"rewards/chosen": -0.8572635650634766,
"rewards/margins": 1.0665409564971924,
"rewards/rejected": -1.923804521560669,
"step": 83
},
{
"epoch": 0.57,
"grad_norm": 36.11750167575466,
"learning_rate": 2.321652042001919e-07,
"logps/chosen": -75.86955261230469,
"logps/rejected": -77.73924255371094,
"loss": 0.5125,
"losses/dpo": 0.49389997124671936,
"losses/sft": 0.4043150544166565,
"losses/total": 0.49389997124671936,
"ref_logps/chosen": -67.49108123779297,
"ref_logps/rejected": -60.198638916015625,
"rewards/accuracies": 0.734375,
"rewards/chosen": -0.8378474712371826,
"rewards/margins": 0.9162120819091797,
"rewards/rejected": -1.7540594339370728,
"step": 84
},
{
"epoch": 0.58,
"grad_norm": 44.25422567833337,
"learning_rate": 2.2623598917395436e-07,
"logps/chosen": -78.2550048828125,
"logps/rejected": -69.63221740722656,
"loss": 0.4812,
"losses/dpo": 0.24741792678833008,
"losses/sft": 0.38957494497299194,
"losses/total": 0.24741792678833008,
"ref_logps/chosen": -69.76954650878906,
"ref_logps/rejected": -52.76110076904297,
"rewards/accuracies": 0.765625,
"rewards/chosen": -0.8485451936721802,
"rewards/margins": 0.8385663628578186,
"rewards/rejected": -1.6871116161346436,
"step": 85
},
{
"epoch": 0.58,
"grad_norm": 29.36328595493774,
"learning_rate": 2.2032023433567376e-07,
"logps/chosen": -72.15760040283203,
"logps/rejected": -76.89938354492188,
"loss": 0.3683,
"losses/dpo": 0.3277527093887329,
"losses/sft": 0.34014564752578735,
"losses/total": 0.3277527093887329,
"ref_logps/chosen": -63.951847076416016,
"ref_logps/rejected": -56.66627883911133,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.8205756545066833,
"rewards/margins": 1.2027339935302734,
"rewards/rejected": -2.0233094692230225,
"step": 86
},
{
"epoch": 0.59,
"grad_norm": 24.453783435169065,
"learning_rate": 2.1442129043167873e-07,
"logps/chosen": -75.3466796875,
"logps/rejected": -76.63810729980469,
"loss": 0.352,
"losses/dpo": 0.4673713743686676,
"losses/sft": 0.3854469656944275,
"losses/total": 0.4673713743686676,
"ref_logps/chosen": -68.66751861572266,
"ref_logps/rejected": -56.724952697753906,
"rewards/accuracies": 0.84375,
"rewards/chosen": -0.6679168939590454,
"rewards/margins": 1.3233978748321533,
"rewards/rejected": -1.9913148880004883,
"step": 87
},
{
"epoch": 0.6,
"grad_norm": 28.45352429062705,
"learning_rate": 2.085424986864065e-07,
"logps/chosen": -80.1172103881836,
"logps/rejected": -72.56062316894531,
"loss": 0.4422,
"losses/dpo": 0.3042376637458801,
"losses/sft": 0.28263095021247864,
"losses/total": 0.3042376637458801,
"ref_logps/chosen": -74.51995849609375,
"ref_logps/rejected": -56.53531265258789,
"rewards/accuracies": 0.8125,
"rewards/chosen": -0.5597256422042847,
"rewards/margins": 1.0428049564361572,
"rewards/rejected": -1.602530598640442,
"step": 88
},
{
"epoch": 0.6,
"grad_norm": 29.271005488942123,
"learning_rate": 2.0268718890989752e-07,
"logps/chosen": -71.47293853759766,
"logps/rejected": -76.03471374511719,
"loss": 0.4194,
"losses/dpo": 0.4656490683555603,
"losses/sft": 0.3225964307785034,
"losses/total": 0.4656490683555603,
"ref_logps/chosen": -63.167877197265625,
"ref_logps/rejected": -56.3370361328125,
"rewards/accuracies": 0.859375,
"rewards/chosen": -0.8305058479309082,
"rewards/margins": 1.1392616033554077,
"rewards/rejected": -1.9697675704956055,
"step": 89
},
{
"epoch": 0.61,
"grad_norm": 35.681125753631115,
"learning_rate": 1.968586776117558e-07,
"logps/chosen": -79.76475524902344,
"logps/rejected": -73.98800659179688,
"loss": 0.4659,
"losses/dpo": 0.35364383459091187,
"losses/sft": 0.3435375690460205,
"losses/total": 0.35364383459091187,
"ref_logps/chosen": -72.63055419921875,
"ref_logps/rejected": -56.81049346923828,
"rewards/accuracies": 0.734375,
"rewards/chosen": -0.7134197950363159,
"rewards/margins": 1.004331350326538,
"rewards/rejected": -1.7177510261535645,
"step": 90
},
{
"epoch": 0.62,
"grad_norm": 36.49093259706875,
"learning_rate": 1.9106026612264315e-07,
"logps/chosen": -78.425537109375,
"logps/rejected": -74.1988525390625,
"loss": 0.3985,
"losses/dpo": 0.19245147705078125,
"losses/sft": 0.39025193452835083,
"losses/total": 0.19245147705078125,
"ref_logps/chosen": -69.4635009765625,
"ref_logps/rejected": -53.60189437866211,
"rewards/accuracies": 0.8125,
"rewards/chosen": -0.8962031006813049,
"rewards/margins": 1.1634931564331055,
"rewards/rejected": -2.0596961975097656,
"step": 91
},
{
"epoch": 0.63,
"grad_norm": 27.105142818170325,
"learning_rate": 1.8529523872436977e-07,
"logps/chosen": -75.47040557861328,
"logps/rejected": -75.96112060546875,
"loss": 0.4417,
"losses/dpo": 0.3523982763290405,
"losses/sft": 0.3631323575973511,
"losses/total": 0.3523982763290405,
"ref_logps/chosen": -67.11583709716797,
"ref_logps/rejected": -58.03672790527344,
"rewards/accuracies": 0.796875,
"rewards/chosen": -0.8354578018188477,
"rewards/margins": 0.9569823145866394,
"rewards/rejected": -1.7924400568008423,
"step": 92
},
{
"epoch": 0.63,
"grad_norm": 38.458971249921234,
"learning_rate": 1.7956686078964255e-07,
"logps/chosen": -72.20645141601562,
"logps/rejected": -72.76055145263672,
"loss": 0.4654,
"losses/dpo": 0.48378482460975647,
"losses/sft": 0.2976914644241333,
"losses/total": 0.48378482460975647,
"ref_logps/chosen": -64.7495346069336,
"ref_logps/rejected": -56.04286193847656,
"rewards/accuracies": 0.8125,
"rewards/chosen": -0.7456916570663452,
"rewards/margins": 0.9260772466659546,
"rewards/rejected": -1.671769142150879,
"step": 93
},
{
"epoch": 0.64,
"grad_norm": 21.64298997501433,
"learning_rate": 1.7387837693252327e-07,
"logps/chosen": -80.37052917480469,
"logps/rejected": -79.18669891357422,
"loss": 0.3604,
"losses/dpo": 0.3508569896221161,
"losses/sft": 0.39343804121017456,
"losses/total": 0.3508569896221161,
"ref_logps/chosen": -72.72434997558594,
"ref_logps/rejected": -58.45079803466797,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.764617919921875,
"rewards/margins": 1.3089728355407715,
"rewards/rejected": -2.0735907554626465,
"step": 94
},
{
"epoch": 0.65,
"grad_norm": 28.31625318441289,
"learning_rate": 1.6823300917064458e-07,
"logps/chosen": -74.7480697631836,
"logps/rejected": -71.78422546386719,
"loss": 0.4383,
"losses/dpo": 0.17416918277740479,
"losses/sft": 0.3635005056858063,
"losses/total": 0.17416918277740479,
"ref_logps/chosen": -66.6783447265625,
"ref_logps/rejected": -53.81765365600586,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.8069729804992676,
"rewards/margins": 0.9896839261054993,
"rewards/rejected": -1.796656847000122,
"step": 95
},
{
"epoch": 0.65,
"grad_norm": 31.43003387376896,
"learning_rate": 1.6263395510022543e-07,
"logps/chosen": -82.76495361328125,
"logps/rejected": -77.74391174316406,
"loss": 0.3885,
"losses/dpo": 0.40873783826828003,
"losses/sft": 0.3334735929965973,
"losses/total": 0.40873783826828003,
"ref_logps/chosen": -74.447998046875,
"ref_logps/rejected": -57.47111511230469,
"rewards/accuracies": 0.84375,
"rewards/chosen": -0.8316947221755981,
"rewards/margins": 1.1955851316452026,
"rewards/rejected": -2.0272796154022217,
"step": 96
},
{
"epoch": 0.66,
"grad_norm": 29.307347493310413,
"learning_rate": 1.5708438608491815e-07,
"logps/chosen": -73.7046890258789,
"logps/rejected": -74.98145294189453,
"loss": 0.4818,
"losses/dpo": 0.3637702465057373,
"losses/sft": 0.3260781168937683,
"losses/total": 0.3637702465057373,
"ref_logps/chosen": -67.44843292236328,
"ref_logps/rejected": -59.09331130981445,
"rewards/accuracies": 0.765625,
"rewards/chosen": -0.625625491142273,
"rewards/margins": 0.9631882905960083,
"rewards/rejected": -1.5888137817382812,
"step": 97
},
{
"epoch": 0.67,
"grad_norm": 51.123126271441144,
"learning_rate": 1.5158744545951465e-07,
"logps/chosen": -75.0676040649414,
"logps/rejected": -72.10035705566406,
"loss": 0.4318,
"losses/dpo": 0.29737696051597595,
"losses/sft": 0.3478523790836334,
"losses/total": 0.29737696051597595,
"ref_logps/chosen": -67.30294799804688,
"ref_logps/rejected": -53.60346603393555,
"rewards/accuracies": 0.8125,
"rewards/chosen": -0.7764655947685242,
"rewards/margins": 1.073223352432251,
"rewards/rejected": -1.84968900680542,
"step": 98
},
{
"epoch": 0.67,
"grad_norm": 26.667202752541407,
"learning_rate": 1.461462467495284e-07,
"logps/chosen": -75.2646484375,
"logps/rejected": -74.6376953125,
"loss": 0.3632,
"losses/dpo": 0.4617415964603424,
"losses/sft": 0.3319079577922821,
"losses/total": 0.4617415964603424,
"ref_logps/chosen": -67.51861572265625,
"ref_logps/rejected": -55.421478271484375,
"rewards/accuracies": 0.890625,
"rewards/chosen": -0.7746031284332275,
"rewards/margins": 1.1470186710357666,
"rewards/rejected": -1.9216216802597046,
"step": 99
},
{
"epoch": 0.68,
"grad_norm": 32.0757645021365,
"learning_rate": 1.4076387190766014e-07,
"logps/chosen": -67.63773345947266,
"logps/rejected": -70.43733215332031,
"loss": 0.4539,
"losses/dpo": 0.4691275358200073,
"losses/sft": 0.3476483225822449,
"losses/total": 0.4691275358200073,
"ref_logps/chosen": -57.368900299072266,
"ref_logps/rejected": -50.622802734375,
"rewards/accuracies": 0.75,
"rewards/chosen": -1.0268828868865967,
"rewards/margins": 0.9545705914497375,
"rewards/rejected": -1.9814532995224,
"step": 100
},
{
"epoch": 0.69,
"grad_norm": 26.664220237307592,
"learning_rate": 1.354433695681474e-07,
"logps/chosen": -80.13877868652344,
"logps/rejected": -81.61981201171875,
"loss": 0.4337,
"losses/dpo": 0.4848173260688782,
"losses/sft": 0.36751094460487366,
"losses/total": 0.4848173260688782,
"ref_logps/chosen": -72.32598876953125,
"ref_logps/rejected": -63.25422668457031,
"rewards/accuracies": 0.765625,
"rewards/chosen": -0.7812784910202026,
"rewards/margins": 1.0552804470062256,
"rewards/rejected": -1.8365588188171387,
"step": 101
},
{
"epoch": 0.69,
"grad_norm": 121.65683713646945,
"learning_rate": 1.301877533199859e-07,
"logps/chosen": -72.0447006225586,
"logps/rejected": -72.38668823242188,
"loss": 0.4732,
"losses/dpo": 0.3546653389930725,
"losses/sft": 0.34984883666038513,
"losses/total": 0.3546653389930725,
"ref_logps/chosen": -63.54951095581055,
"ref_logps/rejected": -53.592952728271484,
"rewards/accuracies": 0.734375,
"rewards/chosen": -0.8495190739631653,
"rewards/margins": 1.0298542976379395,
"rewards/rejected": -1.87937331199646,
"step": 102
},
{
"epoch": 0.7,
"grad_norm": 32.59677272775599,
"learning_rate": 1.2500000000000005e-07,
"logps/chosen": -66.81201934814453,
"logps/rejected": -67.73169708251953,
"loss": 0.4377,
"losses/dpo": 0.4282957911491394,
"losses/sft": 0.3223702311515808,
"losses/total": 0.4282957911491394,
"ref_logps/chosen": -59.40269088745117,
"ref_logps/rejected": -50.77289581298828,
"rewards/accuracies": 0.84375,
"rewards/chosen": -0.7409323453903198,
"rewards/margins": 0.9549475908279419,
"rewards/rejected": -1.6958800554275513,
"step": 103
},
{
"epoch": 0.71,
"grad_norm": 24.94130531155071,
"learning_rate": 1.1988304800673196e-07,
"logps/chosen": -69.9290542602539,
"logps/rejected": -70.86064147949219,
"loss": 0.4243,
"losses/dpo": 0.4698595404624939,
"losses/sft": 0.37894606590270996,
"losses/total": 0.4698595404624939,
"ref_logps/chosen": -62.15961456298828,
"ref_logps/rejected": -51.80973815917969,
"rewards/accuracies": 0.78125,
"rewards/chosen": -0.7769437432289124,
"rewards/margins": 1.1281468868255615,
"rewards/rejected": -1.9050908088684082,
"step": 104
},
{
"epoch": 0.71,
"grad_norm": 29.429373971218872,
"learning_rate": 1.1483979563610069e-07,
"logps/chosen": -78.253173828125,
"logps/rejected": -75.22711181640625,
"loss": 0.4335,
"losses/dpo": 0.2583315968513489,
"losses/sft": 0.35668855905532837,
"losses/total": 0.2583315968513489,
"ref_logps/chosen": -68.73954010009766,
"ref_logps/rejected": -54.715335845947266,
"rewards/accuracies": 0.765625,
"rewards/chosen": -0.9513629674911499,
"rewards/margins": 1.0998142957687378,
"rewards/rejected": -2.0511772632598877,
"step": 105
},
{
"epoch": 0.72,
"grad_norm": 27.54291684365646,
"learning_rate": 1.0987309943977646e-07,
"logps/chosen": -79.83787536621094,
"logps/rejected": -78.37857055664062,
"loss": 0.3904,
"losses/dpo": 0.4721170663833618,
"losses/sft": 0.3833865225315094,
"losses/total": 0.4721170663833618,
"ref_logps/chosen": -71.21440887451172,
"ref_logps/rejected": -57.57903289794922,
"rewards/accuracies": 0.8125,
"rewards/chosen": -0.8623473048210144,
"rewards/margins": 1.2176061868667603,
"rewards/rejected": -2.07995343208313,
"step": 106
},
{
"epoch": 0.73,
"grad_norm": 28.492473761434397,
"learning_rate": 1.0498577260720048e-07,
"logps/chosen": -78.24122619628906,
"logps/rejected": -77.19696044921875,
"loss": 0.4256,
"losses/dpo": 0.5148700475692749,
"losses/sft": 0.3965683579444885,
"losses/total": 0.5148700475692749,
"ref_logps/chosen": -72.94232177734375,
"ref_logps/rejected": -62.735538482666016,
"rewards/accuracies": 0.8125,
"rewards/chosen": -0.5298899412155151,
"rewards/margins": 0.9162522554397583,
"rewards/rejected": -1.4461421966552734,
"step": 107
},
{
"epoch": 0.73,
"grad_norm": 20.982493953158375,
"learning_rate": 1.0018058337216326e-07,
"logps/chosen": -76.80966186523438,
"logps/rejected": -72.25653839111328,
"loss": 0.3378,
"losses/dpo": 0.3177239000797272,
"losses/sft": 0.333109587430954,
"losses/total": 0.3177239000797272,
"ref_logps/chosen": -69.32987976074219,
"ref_logps/rejected": -51.46967697143555,
"rewards/accuracies": 0.890625,
"rewards/chosen": -0.7479775547981262,
"rewards/margins": 1.3307081460952759,
"rewards/rejected": -2.078685760498047,
"step": 108
},
{
"epoch": 0.74,
"grad_norm": 64.19848206989985,
"learning_rate": 9.546025344484868e-08,
"logps/chosen": -80.88861083984375,
"logps/rejected": -78.88575744628906,
"loss": 0.4023,
"losses/dpo": 0.39368999004364014,
"losses/sft": 0.34988299012184143,
"losses/total": 0.39368999004364014,
"ref_logps/chosen": -73.32632446289062,
"ref_logps/rejected": -61.007301330566406,
"rewards/accuracies": 0.890625,
"rewards/chosen": -0.7562285661697388,
"rewards/margins": 1.0316171646118164,
"rewards/rejected": -1.7878458499908447,
"step": 109
},
{
"epoch": 0.75,
"grad_norm": 23.38411036651133,
"learning_rate": 9.082745647022797e-08,
"logps/chosen": -81.5500717163086,
"logps/rejected": -78.32415771484375,
"loss": 0.3158,
"losses/dpo": 0.32109877467155457,
"losses/sft": 0.4066527783870697,
"losses/total": 0.32109877467155457,
"ref_logps/chosen": -74.28262329101562,
"ref_logps/rejected": -55.555335998535156,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.7267451882362366,
"rewards/margins": 1.5501360893249512,
"rewards/rejected": -2.276881217956543,
"step": 110
},
{
"epoch": 0.75,
"grad_norm": 40.260701911712964,
"learning_rate": 8.628481651367875e-08,
"logps/chosen": -74.5044174194336,
"logps/rejected": -76.19696044921875,
"loss": 0.3743,
"losses/dpo": 0.31068116426467896,
"losses/sft": 0.34140655398368835,
"losses/total": 0.31068116426467896,
"ref_logps/chosen": -64.63957214355469,
"ref_logps/rejected": -53.67190170288086,
"rewards/accuracies": 0.84375,
"rewards/chosen": -0.986484169960022,
"rewards/margins": 1.266021728515625,
"rewards/rejected": -2.2525057792663574,
"step": 111
},
{
"epoch": 0.76,
"grad_norm": 31.80013523303015,
"learning_rate": 8.183490657468686e-08,
"logps/chosen": -71.22296142578125,
"logps/rejected": -71.00870513916016,
"loss": 0.5294,
"losses/dpo": 0.4347510039806366,
"losses/sft": 0.36885449290275574,
"losses/total": 0.4347510039806366,
"ref_logps/chosen": -60.00767517089844,
"ref_logps/rejected": -51.94877624511719,
"rewards/accuracies": 0.71875,
"rewards/chosen": -1.1215291023254395,
"rewards/margins": 0.7844638228416443,
"rewards/rejected": -1.905992865562439,
"step": 112
},
{
"epoch": 0.77,
"grad_norm": 26.736632589832187,
"learning_rate": 7.748024712947204e-08,
"logps/chosen": -79.0911636352539,
"logps/rejected": -78.3819580078125,
"loss": 0.3736,
"losses/dpo": 0.38549065589904785,
"losses/sft": 0.36682289838790894,
"losses/total": 0.38549065589904785,
"ref_logps/chosen": -71.06074523925781,
"ref_logps/rejected": -57.33882141113281,
"rewards/accuracies": 0.765625,
"rewards/chosen": -0.8030425906181335,
"rewards/margins": 1.301270842552185,
"rewards/rejected": -2.104313373565674,
"step": 113
},
{
"epoch": 0.77,
"grad_norm": 27.914463891065783,
"learning_rate": 7.322330470336313e-08,
"logps/chosen": -79.21043395996094,
"logps/rejected": -78.50171661376953,
"loss": 0.4552,
"losses/dpo": 0.4180038869380951,
"losses/sft": 0.3502272665500641,
"losses/total": 0.4180038869380951,
"ref_logps/chosen": -70.78549194335938,
"ref_logps/rejected": -59.83561706542969,
"rewards/accuracies": 0.78125,
"rewards/chosen": -0.8424935340881348,
"rewards/margins": 1.024115800857544,
"rewards/rejected": -1.8666093349456787,
"step": 114
},
{
"epoch": 0.78,
"grad_norm": 36.14683391187098,
"learning_rate": 6.906649047373245e-08,
"logps/chosen": -66.70146179199219,
"logps/rejected": -67.81217193603516,
"loss": 0.3941,
"losses/dpo": 0.3595428466796875,
"losses/sft": 0.4029202461242676,
"losses/total": 0.3595428466796875,
"ref_logps/chosen": -57.97339630126953,
"ref_logps/rejected": -47.349143981933594,
"rewards/accuracies": 0.765625,
"rewards/chosen": -0.8728064298629761,
"rewards/margins": 1.173496961593628,
"rewards/rejected": -2.0463035106658936,
"step": 115
},
{
"epoch": 0.79,
"grad_norm": 34.93282588073872,
"learning_rate": 6.501215890427908e-08,
"logps/chosen": -74.01496887207031,
"logps/rejected": -71.6580810546875,
"loss": 0.4737,
"losses/dpo": 0.47708314657211304,
"losses/sft": 0.3861905336380005,
"losses/total": 0.47708314657211304,
"ref_logps/chosen": -66.82437896728516,
"ref_logps/rejected": -56.22287368774414,
"rewards/accuracies": 0.765625,
"rewards/chosen": -0.7190604209899902,
"rewards/margins": 0.8244602680206299,
"rewards/rejected": -1.5435206890106201,
"step": 116
},
{
"epoch": 0.8,
"grad_norm": 33.61435457152851,
"learning_rate": 6.106260641143546e-08,
"logps/chosen": -79.60284423828125,
"logps/rejected": -77.92535400390625,
"loss": 0.4682,
"losses/dpo": 0.290701687335968,
"losses/sft": 0.3620028495788574,
"losses/total": 0.290701687335968,
"ref_logps/chosen": -70.14583587646484,
"ref_logps/rejected": -59.40486526489258,
"rewards/accuracies": 0.78125,
"rewards/chosen": -0.9457005262374878,
"rewards/margins": 0.9063488245010376,
"rewards/rejected": -1.8520495891571045,
"step": 117
},
{
"epoch": 0.8,
"grad_norm": 32.27039772238261,
"learning_rate": 5.722007006365309e-08,
"logps/chosen": -85.39405059814453,
"logps/rejected": -77.39488220214844,
"loss": 0.4029,
"losses/dpo": 0.29392948746681213,
"losses/sft": 0.4053553342819214,
"losses/total": 0.29392948746681213,
"ref_logps/chosen": -75.33457946777344,
"ref_logps/rejected": -56.223270416259766,
"rewards/accuracies": 0.84375,
"rewards/chosen": -1.0059471130371094,
"rewards/margins": 1.111214280128479,
"rewards/rejected": -2.117161273956299,
"step": 118
},
{
"epoch": 0.81,
"grad_norm": 30.691331571937976,
"learning_rate": 5.3486726314303175e-08,
"logps/chosen": -76.40483856201172,
"logps/rejected": -76.4593734741211,
"loss": 0.4526,
"losses/dpo": 0.48409292101860046,
"losses/sft": 0.3616543412208557,
"losses/total": 0.48409292101860046,
"ref_logps/chosen": -66.07306671142578,
"ref_logps/rejected": -56.153717041015625,
"rewards/accuracies": 0.828125,
"rewards/chosen": -1.0331764221191406,
"rewards/margins": 0.9973889589309692,
"rewards/rejected": -2.0305655002593994,
"step": 119
},
{
"epoch": 0.82,
"grad_norm": 29.441590057939685,
"learning_rate": 4.986468976890992e-08,
"logps/chosen": -75.17059326171875,
"logps/rejected": -73.5225830078125,
"loss": 0.4307,
"losses/dpo": 0.33158645033836365,
"losses/sft": 0.3496292531490326,
"losses/total": 0.33158645033836365,
"ref_logps/chosen": -68.26935577392578,
"ref_logps/rejected": -55.77709197998047,
"rewards/accuracies": 0.8125,
"rewards/chosen": -0.69012451171875,
"rewards/margins": 1.0844244956970215,
"rewards/rejected": -1.774549126625061,
"step": 120
},
{
"epoch": 0.82,
"grad_norm": 24.37008051594277,
"learning_rate": 4.635601198741607e-08,
"logps/chosen": -73.0021743774414,
"logps/rejected": -74.11600494384766,
"loss": 0.3949,
"losses/dpo": 0.5414682030677795,
"losses/sft": 0.37046751379966736,
"losses/total": 0.5414682030677795,
"ref_logps/chosen": -64.94493103027344,
"ref_logps/rejected": -54.75990676879883,
"rewards/accuracies": 0.8125,
"rewards/chosen": -0.8057246804237366,
"rewards/margins": 1.1298854351043701,
"rewards/rejected": -1.935610294342041,
"step": 121
},
{
"epoch": 0.83,
"grad_norm": 26.892127512970333,
"learning_rate": 4.296268032215733e-08,
"logps/chosen": -75.42649841308594,
"logps/rejected": -76.27777099609375,
"loss": 0.4499,
"losses/dpo": 0.5582671761512756,
"losses/sft": 0.3608691692352295,
"losses/total": 0.5582671761512756,
"ref_logps/chosen": -66.92321014404297,
"ref_logps/rejected": -57.275753021240234,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.8503296971321106,
"rewards/margins": 1.0498719215393066,
"rewards/rejected": -1.9002017974853516,
"step": 122
},
{
"epoch": 0.84,
"grad_norm": 41.688513331786744,
"learning_rate": 3.968661679220467e-08,
"logps/chosen": -72.4287338256836,
"logps/rejected": -75.88028717041016,
"loss": 0.4095,
"losses/dpo": 0.6144567131996155,
"losses/sft": 0.3270949125289917,
"losses/total": 0.6144567131996155,
"ref_logps/chosen": -63.542545318603516,
"ref_logps/rejected": -54.83192443847656,
"rewards/accuracies": 0.84375,
"rewards/chosen": -0.8886189460754395,
"rewards/margins": 1.2162187099456787,
"rewards/rejected": -2.104837656021118,
"step": 123
},
{
"epoch": 0.84,
"grad_norm": 25.469129701195566,
"learning_rate": 3.652967699471299e-08,
"logps/chosen": -74.50592803955078,
"logps/rejected": -72.23234558105469,
"loss": 0.3866,
"losses/dpo": 0.28718867897987366,
"losses/sft": 0.37371402978897095,
"losses/total": 0.28718867897987366,
"ref_logps/chosen": -65.68087768554688,
"ref_logps/rejected": -51.324337005615234,
"rewards/accuracies": 0.828125,
"rewards/chosen": -0.8825055360794067,
"rewards/margins": 1.2082949876785278,
"rewards/rejected": -2.0908007621765137,
"step": 124
},
{
"epoch": 0.85,
"grad_norm": 35.98399239694971,
"learning_rate": 3.349364905389032e-08,
"logps/chosen": -79.32814025878906,
"logps/rejected": -75.2787094116211,
"loss": 0.3999,
"losses/dpo": 0.5298404693603516,
"losses/sft": 0.36322808265686035,
"losses/total": 0.5298404693603516,
"ref_logps/chosen": -72.24817657470703,
"ref_logps/rejected": -56.569156646728516,
"rewards/accuracies": 0.859375,
"rewards/chosen": -0.7079958915710449,
"rewards/margins": 1.1629595756530762,
"rewards/rejected": -1.8709555864334106,
"step": 125
},
{
"epoch": 0.86,
"grad_norm": 29.48115104276339,
"learning_rate": 3.058025260818609e-08,
"logps/chosen": -79.290283203125,
"logps/rejected": -73.2989730834961,
"loss": 0.4999,
"losses/dpo": 0.6051620841026306,
"losses/sft": 0.35250741243362427,
"losses/total": 0.6051620841026306,
"ref_logps/chosen": -70.70803833007812,
"ref_logps/rejected": -55.791748046875,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.8582238554954529,
"rewards/margins": 0.8924983143806458,
"rewards/rejected": -1.7507224082946777,
"step": 126
},
{
"epoch": 0.86,
"grad_norm": 24.758070012254365,
"learning_rate": 2.7791137836269158e-08,
"logps/chosen": -78.37055969238281,
"logps/rejected": -74.77693939208984,
"loss": 0.3954,
"losses/dpo": 0.3199983537197113,
"losses/sft": 0.39879244565963745,
"losses/total": 0.3199983537197113,
"ref_logps/chosen": -69.10183715820312,
"ref_logps/rejected": -54.88367462158203,
"rewards/accuracies": 0.78125,
"rewards/chosen": -0.9268715381622314,
"rewards/margins": 1.062455177307129,
"rewards/rejected": -1.98932683467865,
"step": 127
},
{
"epoch": 0.87,
"grad_norm": 20.827088770452676,
"learning_rate": 2.512788452234921e-08,
"logps/chosen": -70.80789184570312,
"logps/rejected": -75.08622741699219,
"loss": 0.3325,
"losses/dpo": 0.3453621566295624,
"losses/sft": 0.3389894664287567,
"losses/total": 0.3453621566295624,
"ref_logps/chosen": -61.59326934814453,
"ref_logps/rejected": -52.41365432739258,
"rewards/accuracies": 0.859375,
"rewards/chosen": -0.9214622974395752,
"rewards/margins": 1.345794677734375,
"rewards/rejected": -2.26725697517395,
"step": 128
},
{
"epoch": 0.88,
"grad_norm": 31.45091455444117,
"learning_rate": 2.259200116137039e-08,
"logps/chosen": -76.39757537841797,
"logps/rejected": -77.06990051269531,
"loss": 0.4837,
"losses/dpo": 0.6226012706756592,
"losses/sft": 0.37185585498809814,
"losses/total": 0.6226012706756592,
"ref_logps/chosen": -66.43987274169922,
"ref_logps/rejected": -58.27755355834961,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.9957698583602905,
"rewards/margins": 0.8834652900695801,
"rewards/rejected": -1.8792351484298706,
"step": 129
},
{
"epoch": 0.88,
"grad_norm": 25.29491524946383,
"learning_rate": 2.0184924104583612e-08,
"logps/chosen": -69.09542083740234,
"logps/rejected": -73.32018280029297,
"loss": 0.4023,
"losses/dpo": 0.36410456895828247,
"losses/sft": 0.33447620272636414,
"losses/total": 0.36410456895828247,
"ref_logps/chosen": -59.17249298095703,
"ref_logps/rejected": -51.49285125732422,
"rewards/accuracies": 0.828125,
"rewards/chosen": -0.992293655872345,
"rewards/margins": 1.190440058708191,
"rewards/rejected": -2.1827335357666016,
"step": 130
},
{
"epoch": 0.89,
"grad_norm": 27.582904411320246,
"learning_rate": 1.7908016745981856e-08,
"logps/chosen": -86.50100708007812,
"logps/rejected": -82.30694580078125,
"loss": 0.4445,
"losses/dpo": 0.4145382344722748,
"losses/sft": 0.346317321062088,
"losses/total": 0.4145382344722748,
"ref_logps/chosen": -77.55247497558594,
"ref_logps/rejected": -61.19115447998047,
"rewards/accuracies": 0.78125,
"rewards/chosen": -0.8948546648025513,
"rewards/margins": 1.2167249917984009,
"rewards/rejected": -2.111579418182373,
"step": 131
},
{
"epoch": 0.9,
"grad_norm": 28.77405060036572,
"learning_rate": 1.57625687500596e-08,
"logps/chosen": -75.88212585449219,
"logps/rejected": -74.88499450683594,
"loss": 0.5229,
"losses/dpo": 0.4454580843448639,
"losses/sft": 0.36321359872817993,
"losses/total": 0.4454580843448639,
"ref_logps/chosen": -67.06256866455078,
"ref_logps/rejected": -57.12347412109375,
"rewards/accuracies": 0.71875,
"rewards/chosen": -0.8819558024406433,
"rewards/margins": 0.8941954970359802,
"rewards/rejected": -1.776151418685913,
"step": 132
},
{
"epoch": 0.9,
"grad_norm": 27.7470456349191,
"learning_rate": 1.3749795321332885e-08,
"logps/chosen": -73.93744659423828,
"logps/rejected": -76.04200744628906,
"loss": 0.3594,
"losses/dpo": 0.17011409997940063,
"losses/sft": 0.40163373947143555,
"losses/total": 0.17011409997940063,
"ref_logps/chosen": -64.7332534790039,
"ref_logps/rejected": -53.972320556640625,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.9204192757606506,
"rewards/margins": 1.2865492105484009,
"rewards/rejected": -2.206968307495117,
"step": 133
},
{
"epoch": 0.91,
"grad_norm": 36.37444245491115,
"learning_rate": 1.1870836516034876e-08,
"logps/chosen": -82.49986267089844,
"logps/rejected": -78.48277282714844,
"loss": 0.481,
"losses/dpo": 0.45800262689590454,
"losses/sft": 0.4599795341491699,
"losses/total": 0.45800262689590454,
"ref_logps/chosen": -73.60972595214844,
"ref_logps/rejected": -60.565696716308594,
"rewards/accuracies": 0.828125,
"rewards/chosen": -0.8890135288238525,
"rewards/margins": 0.9026941657066345,
"rewards/rejected": -1.7917077541351318,
"step": 134
},
{
"epoch": 0.92,
"grad_norm": 85.09720359081324,
"learning_rate": 1.0126756596375685e-08,
"logps/chosen": -69.55728149414062,
"logps/rejected": -67.4758071899414,
"loss": 0.4461,
"losses/dpo": 0.3670344650745392,
"losses/sft": 0.37169790267944336,
"losses/total": 0.3670344650745392,
"ref_logps/chosen": -60.00959396362305,
"ref_logps/rejected": -48.21699905395508,
"rewards/accuracies": 0.796875,
"rewards/chosen": -0.9547686576843262,
"rewards/margins": 0.9711121916770935,
"rewards/rejected": -1.9258809089660645,
"step": 135
},
{
"epoch": 0.92,
"grad_norm": 25.805500137707952,
"learning_rate": 8.518543427732949e-09,
"logps/chosen": -77.5588607788086,
"logps/rejected": -72.98957824707031,
"loss": 0.3857,
"losses/dpo": 0.34388694167137146,
"losses/sft": 0.39463409781455994,
"losses/total": 0.34388694167137146,
"ref_logps/chosen": -71.34376525878906,
"ref_logps/rejected": -55.052791595458984,
"rewards/accuracies": 0.84375,
"rewards/chosen": -0.6215101480484009,
"rewards/margins": 1.1721692085266113,
"rewards/rejected": -1.7936793565750122,
"step": 136
},
{
"epoch": 0.93,
"grad_norm": 28.925653233926095,
"learning_rate": 7.047107919114586e-09,
"logps/chosen": -76.60894775390625,
"logps/rejected": -67.70733642578125,
"loss": 0.5413,
"losses/dpo": 0.40706807374954224,
"losses/sft": 0.38619473576545715,
"losses/total": 0.40706807374954224,
"ref_logps/chosen": -65.6761245727539,
"ref_logps/rejected": -48.80441665649414,
"rewards/accuracies": 0.75,
"rewards/chosen": -1.0932811498641968,
"rewards/margins": 0.7970114350318909,
"rewards/rejected": -1.8902926445007324,
"step": 137
},
{
"epoch": 0.94,
"grad_norm": 28.194863492832845,
"learning_rate": 5.7132835072101486e-09,
"logps/chosen": -73.64652252197266,
"logps/rejected": -75.85942077636719,
"loss": 0.3847,
"losses/dpo": 0.4270258843898773,
"losses/sft": 0.4282100200653076,
"losses/total": 0.4270258843898773,
"ref_logps/chosen": -65.63545227050781,
"ref_logps/rejected": -55.20812225341797,
"rewards/accuracies": 0.8125,
"rewards/chosen": -0.8011073470115662,
"rewards/margins": 1.2640223503112793,
"rewards/rejected": -2.0651297569274902,
"step": 138
},
{
"epoch": 0.94,
"grad_norm": 24.471573256004906,
"learning_rate": 4.517825684323323e-09,
"logps/chosen": -76.01177215576172,
"logps/rejected": -72.83094787597656,
"loss": 0.3493,
"losses/dpo": 0.42532169818878174,
"losses/sft": 0.38909339904785156,
"losses/total": 0.42532169818878174,
"ref_logps/chosen": -69.16812896728516,
"ref_logps/rejected": -52.844810485839844,
"rewards/accuracies": 0.90625,
"rewards/chosen": -0.6843640208244324,
"rewards/margins": 1.3142496347427368,
"rewards/rejected": -1.9986138343811035,
"step": 139
},
{
"epoch": 0.95,
"grad_norm": 28.104692013897658,
"learning_rate": 3.4614115704533766e-09,
"logps/chosen": -80.1229019165039,
"logps/rejected": -77.62194061279297,
"loss": 0.3651,
"losses/dpo": 0.3558647334575653,
"losses/sft": 0.3259223401546478,
"losses/total": 0.3558647334575653,
"ref_logps/chosen": -71.98576354980469,
"ref_logps/rejected": -57.123634338378906,
"rewards/accuracies": 0.84375,
"rewards/chosen": -0.8137134313583374,
"rewards/margins": 1.2361174821853638,
"rewards/rejected": -2.049830913543701,
"step": 140
},
{
"epoch": 0.96,
"grad_norm": 30.06950212059891,
"learning_rate": 2.5446395297668287e-09,
"logps/chosen": -77.057373046875,
"logps/rejected": -77.57244873046875,
"loss": 0.3862,
"losses/dpo": 0.46156057715415955,
"losses/sft": 0.3272428512573242,
"losses/total": 0.46156057715415955,
"ref_logps/chosen": -69.33092498779297,
"ref_logps/rejected": -57.40209197998047,
"rewards/accuracies": 0.78125,
"rewards/chosen": -0.772644579410553,
"rewards/margins": 1.2443914413452148,
"rewards/rejected": -2.017035961151123,
"step": 141
},
{
"epoch": 0.97,
"grad_norm": 22.27867702599573,
"learning_rate": 1.7680288316779256e-09,
"logps/chosen": -76.52747344970703,
"logps/rejected": -76.79586029052734,
"loss": 0.3613,
"losses/dpo": 0.38265928626060486,
"losses/sft": 0.330098032951355,
"losses/total": 0.38265928626060486,
"ref_logps/chosen": -67.75516510009766,
"ref_logps/rejected": -55.386016845703125,
"rewards/accuracies": 0.859375,
"rewards/chosen": -0.8772311806678772,
"rewards/margins": 1.263753056526184,
"rewards/rejected": -2.140984058380127,
"step": 142
},
{
"epoch": 0.97,
"grad_norm": 33.24181116137987,
"learning_rate": 1.1320193567288527e-09,
"logps/chosen": -68.91748809814453,
"logps/rejected": -68.46461486816406,
"loss": 0.4453,
"losses/dpo": 0.6369032859802246,
"losses/sft": 0.38167503476142883,
"losses/total": 0.6369032859802246,
"ref_logps/chosen": -59.43241500854492,
"ref_logps/rejected": -47.996742248535156,
"rewards/accuracies": 0.78125,
"rewards/chosen": -0.9485068321228027,
"rewards/margins": 1.0982803106307983,
"rewards/rejected": -2.0467872619628906,
"step": 143
},
{
"epoch": 0.98,
"grad_norm": 27.26573554059411,
"learning_rate": 6.369713474366212e-10,
"logps/chosen": -79.28267669677734,
"logps/rejected": -79.29057312011719,
"loss": 0.4268,
"losses/dpo": 0.3871089816093445,
"losses/sft": 0.3575269877910614,
"losses/total": 0.3871089816093445,
"ref_logps/chosen": -70.6344985961914,
"ref_logps/rejected": -60.456382751464844,
"rewards/accuracies": 0.8125,
"rewards/chosen": -0.864818811416626,
"rewards/margins": 1.018599510192871,
"rewards/rejected": -1.883418321609497,
"step": 144
},
{
"epoch": 0.99,
"grad_norm": 28.26392966654158,
"learning_rate": 2.831652042480093e-10,
"logps/chosen": -71.9800796508789,
"logps/rejected": -73.54934692382812,
"loss": 0.4465,
"losses/dpo": 0.4069312512874603,
"losses/sft": 0.35074400901794434,
"losses/total": 0.4069312512874603,
"ref_logps/chosen": -62.18163299560547,
"ref_logps/rejected": -53.78350067138672,
"rewards/accuracies": 0.765625,
"rewards/chosen": -0.9798437356948853,
"rewards/margins": 0.996740460395813,
"rewards/rejected": -1.9765844345092773,
"step": 145
},
{
"epoch": 0.99,
"grad_norm": 23.91111528235258,
"learning_rate": 7.080132671774541e-11,
"logps/chosen": -75.3148422241211,
"logps/rejected": -74.4091567993164,
"loss": 0.3508,
"losses/dpo": 0.42828118801116943,
"losses/sft": 0.41488099098205566,
"losses/total": 0.42828118801116943,
"ref_logps/chosen": -68.02532196044922,
"ref_logps/rejected": -53.932071685791016,
"rewards/accuracies": 0.921875,
"rewards/chosen": -0.7289534211158752,
"rewards/margins": 1.3187549114227295,
"rewards/rejected": -2.047708511352539,
"step": 146
},
{
"epoch": 1.0,
"grad_norm": 25.683687754039113,
"learning_rate": 0.0,
"logps/chosen": -77.05089569091797,
"logps/rejected": -74.92003631591797,
"loss": 0.4269,
"losses/dpo": 0.3264519274234772,
"losses/sft": 0.35349562764167786,
"losses/total": 0.3264519274234772,
"ref_logps/chosen": -69.42140197753906,
"ref_logps/rejected": -56.2528190612793,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.7629485130310059,
"rewards/margins": 1.103774070739746,
"rewards/rejected": -1.8667224645614624,
"step": 147
},
{
"epoch": 1.0,
"step": 147,
"total_flos": 0.0,
"train_loss": 0.4901739539337807,
"train_runtime": 4483.0286,
"train_samples_per_second": 2.102,
"train_steps_per_second": 0.033
}
],
"logging_steps": 1.0,
"max_steps": 147,
"num_input_tokens_seen": 0,
"num_train_epochs": 1,
"save_steps": 10000,
"total_flos": 0.0,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}