AmelieSchreiber's picture
Rename trainer_state (6).json to trainer_state.json
434d14d
raw history blame
No virus
55.7 kB
{
"best_metric": 0.5088262320154422,
"best_model_checkpoint": "esm2_t12_35M-lora-binding-sites_2023-09-13_00-24-11/checkpoint-90066",
"epoch": 1.0,
"eval_steps": 500,
"global_step": 90066,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 0.0005701566148067436,
"loss": 0.4436,
"step": 200
},
{
"epoch": 0.0,
"learning_rate": 0.0005701560386498644,
"loss": 0.1877,
"step": 400
},
{
"epoch": 0.01,
"learning_rate": 0.0005701550771046327,
"loss": 0.1187,
"step": 600
},
{
"epoch": 0.01,
"learning_rate": 0.0005701537301723485,
"loss": 0.1312,
"step": 800
},
{
"epoch": 0.01,
"learning_rate": 0.0005701519978548325,
"loss": 0.1103,
"step": 1000
},
{
"epoch": 0.01,
"learning_rate": 0.0005701498801544267,
"loss": 0.1073,
"step": 1200
},
{
"epoch": 0.02,
"learning_rate": 0.0005701473770739939,
"loss": 0.0926,
"step": 1400
},
{
"epoch": 0.02,
"learning_rate": 0.000570144488616918,
"loss": 0.0842,
"step": 1600
},
{
"epoch": 0.02,
"learning_rate": 0.0005701412147871038,
"loss": 0.0846,
"step": 1800
},
{
"epoch": 0.02,
"learning_rate": 0.0005701375555889771,
"loss": 0.073,
"step": 2000
},
{
"epoch": 0.02,
"learning_rate": 0.0005701335322088747,
"loss": 0.0787,
"step": 2200
},
{
"epoch": 0.03,
"learning_rate": 0.0005701291042162594,
"loss": 0.084,
"step": 2400
},
{
"epoch": 0.03,
"learning_rate": 0.0005701242908717034,
"loss": 0.0826,
"step": 2600
},
{
"epoch": 0.03,
"learning_rate": 0.0005701190921817139,
"loss": 0.0625,
"step": 2800
},
{
"epoch": 0.03,
"learning_rate": 0.0005701135081533188,
"loss": 0.0601,
"step": 3000
},
{
"epoch": 0.04,
"learning_rate": 0.0005701075387940671,
"loss": 0.0628,
"step": 3200
},
{
"epoch": 0.04,
"learning_rate": 0.0005701012168439152,
"loss": 0.0897,
"step": 3400
},
{
"epoch": 0.04,
"learning_rate": 0.0005700944787742291,
"loss": 0.0656,
"step": 3600
},
{
"epoch": 0.04,
"learning_rate": 0.0005700873553994116,
"loss": 0.0627,
"step": 3800
},
{
"epoch": 0.04,
"learning_rate": 0.0005700798467290926,
"loss": 0.0645,
"step": 4000
},
{
"epoch": 0.05,
"learning_rate": 0.0005700719527734229,
"loss": 0.0674,
"step": 4200
},
{
"epoch": 0.05,
"learning_rate": 0.0005700636735430738,
"loss": 0.0641,
"step": 4400
},
{
"epoch": 0.05,
"learning_rate": 0.0005700550090492382,
"loss": 0.058,
"step": 4600
},
{
"epoch": 0.05,
"learning_rate": 0.0005700459593036292,
"loss": 0.0581,
"step": 4800
},
{
"epoch": 0.06,
"learning_rate": 0.0005700365243184809,
"loss": 0.0476,
"step": 5000
},
{
"epoch": 0.06,
"learning_rate": 0.0005700267041065481,
"loss": 0.0527,
"step": 5200
},
{
"epoch": 0.06,
"learning_rate": 0.0005700164986811065,
"loss": 0.0612,
"step": 5400
},
{
"epoch": 0.06,
"learning_rate": 0.0005700059080559526,
"loss": 0.0528,
"step": 5600
},
{
"epoch": 0.06,
"learning_rate": 0.0005699949322454036,
"loss": 0.065,
"step": 5800
},
{
"epoch": 0.07,
"learning_rate": 0.0005699835712642972,
"loss": 0.0473,
"step": 6000
},
{
"epoch": 0.07,
"learning_rate": 0.0005699718251279921,
"loss": 0.0607,
"step": 6200
},
{
"epoch": 0.07,
"learning_rate": 0.0005699596938523677,
"loss": 0.0503,
"step": 6400
},
{
"epoch": 0.07,
"learning_rate": 0.0005699471774538237,
"loss": 0.0633,
"step": 6600
},
{
"epoch": 0.08,
"learning_rate": 0.0005699342759492807,
"loss": 0.052,
"step": 6800
},
{
"epoch": 0.08,
"learning_rate": 0.0005699209893561799,
"loss": 0.048,
"step": 7000
},
{
"epoch": 0.08,
"learning_rate": 0.0005699073176924832,
"loss": 0.0506,
"step": 7200
},
{
"epoch": 0.08,
"learning_rate": 0.0005698933322180377,
"loss": 0.0727,
"step": 7400
},
{
"epoch": 0.08,
"learning_rate": 0.000569878892394234,
"loss": 0.0464,
"step": 7600
},
{
"epoch": 0.09,
"learning_rate": 0.000569864067556744,
"loss": 0.0436,
"step": 7800
},
{
"epoch": 0.09,
"learning_rate": 0.0005698489347324025,
"loss": 0.0564,
"step": 8000
},
{
"epoch": 0.09,
"learning_rate": 0.0005698333418529975,
"loss": 0.0503,
"step": 8200
},
{
"epoch": 0.09,
"learning_rate": 0.0005698173640214845,
"loss": 0.054,
"step": 8400
},
{
"epoch": 0.1,
"learning_rate": 0.0005698010012594639,
"loss": 0.0548,
"step": 8600
},
{
"epoch": 0.1,
"learning_rate": 0.0005697842535890557,
"loss": 0.0625,
"step": 8800
},
{
"epoch": 0.1,
"learning_rate": 0.0005697671210329006,
"loss": 0.0547,
"step": 9000
},
{
"epoch": 0.1,
"learning_rate": 0.0005697496036141599,
"loss": 0.0458,
"step": 9200
},
{
"epoch": 0.1,
"learning_rate": 0.0005697317013565146,
"loss": 0.0381,
"step": 9400
},
{
"epoch": 0.11,
"learning_rate": 0.0005697134142841663,
"loss": 0.0493,
"step": 9600
},
{
"epoch": 0.11,
"learning_rate": 0.000569694742421837,
"loss": 0.0441,
"step": 9800
},
{
"epoch": 0.11,
"learning_rate": 0.0005696756857947685,
"loss": 0.0496,
"step": 10000
},
{
"epoch": 0.11,
"learning_rate": 0.0005696562444287228,
"loss": 0.0453,
"step": 10200
},
{
"epoch": 0.12,
"learning_rate": 0.0005696364183499823,
"loss": 0.0599,
"step": 10400
},
{
"epoch": 0.12,
"learning_rate": 0.0005696162075853492,
"loss": 0.0482,
"step": 10600
},
{
"epoch": 0.12,
"learning_rate": 0.0005695957160960544,
"loss": 0.056,
"step": 10800
},
{
"epoch": 0.12,
"learning_rate": 0.0005695747379652068,
"loss": 0.0673,
"step": 11000
},
{
"epoch": 0.12,
"learning_rate": 0.0005695533752318507,
"loss": 0.0419,
"step": 11200
},
{
"epoch": 0.13,
"learning_rate": 0.0005695316279248656,
"loss": 0.0457,
"step": 11400
},
{
"epoch": 0.13,
"learning_rate": 0.0005695094960736512,
"loss": 0.0658,
"step": 11600
},
{
"epoch": 0.13,
"learning_rate": 0.0005694869797081268,
"loss": 0.0485,
"step": 11800
},
{
"epoch": 0.13,
"learning_rate": 0.0005694640788587318,
"loss": 0.0581,
"step": 12000
},
{
"epoch": 0.14,
"learning_rate": 0.0005694407935564252,
"loss": 0.0523,
"step": 12200
},
{
"epoch": 0.14,
"learning_rate": 0.0005694171238326854,
"loss": 0.0488,
"step": 12400
},
{
"epoch": 0.14,
"learning_rate": 0.0005693930697195113,
"loss": 0.0434,
"step": 12600
},
{
"epoch": 0.14,
"learning_rate": 0.0005693686312494207,
"loss": 0.0441,
"step": 12800
},
{
"epoch": 0.14,
"learning_rate": 0.0005693439335253714,
"loss": 0.0474,
"step": 13000
},
{
"epoch": 0.15,
"learning_rate": 0.0005693187283624477,
"loss": 0.0466,
"step": 13200
},
{
"epoch": 0.15,
"learning_rate": 0.0005692931389431076,
"loss": 0.0504,
"step": 13400
},
{
"epoch": 0.15,
"learning_rate": 0.0005692671653019445,
"loss": 0.0574,
"step": 13600
},
{
"epoch": 0.15,
"learning_rate": 0.0005692408074740716,
"loss": 0.0455,
"step": 13800
},
{
"epoch": 0.16,
"learning_rate": 0.0005692140654951213,
"loss": 0.047,
"step": 14000
},
{
"epoch": 0.16,
"learning_rate": 0.0005691869394012452,
"loss": 0.0399,
"step": 14200
},
{
"epoch": 0.16,
"learning_rate": 0.0005691594292291145,
"loss": 0.046,
"step": 14400
},
{
"epoch": 0.16,
"learning_rate": 0.0005691315350159191,
"loss": 0.0509,
"step": 14600
},
{
"epoch": 0.16,
"learning_rate": 0.0005691032567993687,
"loss": 0.0493,
"step": 14800
},
{
"epoch": 0.17,
"learning_rate": 0.0005690745946176916,
"loss": 0.0433,
"step": 15000
},
{
"epoch": 0.17,
"learning_rate": 0.0005690455485096356,
"loss": 0.0525,
"step": 15200
},
{
"epoch": 0.17,
"learning_rate": 0.0005690161185144672,
"loss": 0.0487,
"step": 15400
},
{
"epoch": 0.17,
"learning_rate": 0.000568986304671972,
"loss": 0.0493,
"step": 15600
},
{
"epoch": 0.18,
"learning_rate": 0.0005689562589653545,
"loss": 0.0596,
"step": 15800
},
{
"epoch": 0.18,
"learning_rate": 0.0005689256794683665,
"loss": 0.0381,
"step": 16000
},
{
"epoch": 0.18,
"learning_rate": 0.0005688947162463136,
"loss": 0.0463,
"step": 16200
},
{
"epoch": 0.18,
"learning_rate": 0.0005688633693410542,
"loss": 0.0411,
"step": 16400
},
{
"epoch": 0.18,
"learning_rate": 0.0005688316387949652,
"loss": 0.0532,
"step": 16600
},
{
"epoch": 0.19,
"learning_rate": 0.0005687995246509422,
"loss": 0.0407,
"step": 16800
},
{
"epoch": 0.19,
"learning_rate": 0.0005687670269523996,
"loss": 0.0426,
"step": 17000
},
{
"epoch": 0.19,
"learning_rate": 0.0005687341457432698,
"loss": 0.0415,
"step": 17200
},
{
"epoch": 0.19,
"learning_rate": 0.0005687008810680043,
"loss": 0.0449,
"step": 17400
},
{
"epoch": 0.2,
"learning_rate": 0.0005686672329715724,
"loss": 0.0411,
"step": 17600
},
{
"epoch": 0.2,
"learning_rate": 0.0005686332014994621,
"loss": 0.0465,
"step": 17800
},
{
"epoch": 0.2,
"learning_rate": 0.0005685987866976797,
"loss": 0.0386,
"step": 18000
},
{
"epoch": 0.2,
"learning_rate": 0.0005685639886127493,
"loss": 0.0379,
"step": 18200
},
{
"epoch": 0.2,
"learning_rate": 0.0005685288072917138,
"loss": 0.05,
"step": 18400
},
{
"epoch": 0.21,
"learning_rate": 0.0005684934215577838,
"loss": 0.0409,
"step": 18600
},
{
"epoch": 0.21,
"learning_rate": 0.0005684574758233195,
"loss": 0.0417,
"step": 18800
},
{
"epoch": 0.21,
"learning_rate": 0.0005684211469967415,
"loss": 0.0418,
"step": 19000
},
{
"epoch": 0.21,
"learning_rate": 0.0005683844351271619,
"loss": 0.0478,
"step": 19200
},
{
"epoch": 0.22,
"learning_rate": 0.0005683473402642101,
"loss": 0.04,
"step": 19400
},
{
"epoch": 0.22,
"learning_rate": 0.0005683098624580339,
"loss": 0.0448,
"step": 19600
},
{
"epoch": 0.22,
"learning_rate": 0.0005682720017592983,
"loss": 0.0374,
"step": 19800
},
{
"epoch": 0.22,
"learning_rate": 0.0005682337582191861,
"loss": 0.0382,
"step": 20000
},
{
"epoch": 0.22,
"learning_rate": 0.0005681951318893979,
"loss": 0.0431,
"step": 20200
},
{
"epoch": 0.23,
"learning_rate": 0.0005681561228221512,
"loss": 0.0437,
"step": 20400
},
{
"epoch": 0.23,
"learning_rate": 0.0005681167310701815,
"loss": 0.0421,
"step": 20600
},
{
"epoch": 0.23,
"learning_rate": 0.0005680769566867412,
"loss": 0.0441,
"step": 20800
},
{
"epoch": 0.23,
"learning_rate": 0.0005680370014619785,
"loss": 0.0625,
"step": 21000
},
{
"epoch": 0.24,
"learning_rate": 0.0005679964638899052,
"loss": 0.0418,
"step": 21200
},
{
"epoch": 0.24,
"learning_rate": 0.0005679555438489471,
"loss": 0.0454,
"step": 21400
},
{
"epoch": 0.24,
"learning_rate": 0.0005679142413944229,
"loss": 0.0416,
"step": 21600
},
{
"epoch": 0.24,
"learning_rate": 0.0005678725565821682,
"loss": 0.0462,
"step": 21800
},
{
"epoch": 0.24,
"learning_rate": 0.0005678304894685357,
"loss": 0.0456,
"step": 22000
},
{
"epoch": 0.25,
"learning_rate": 0.0005677880401103944,
"loss": 0.0454,
"step": 22200
},
{
"epoch": 0.25,
"learning_rate": 0.0005677452085651307,
"loss": 0.0422,
"step": 22400
},
{
"epoch": 0.25,
"learning_rate": 0.0005677019948906472,
"loss": 0.039,
"step": 22600
},
{
"epoch": 0.25,
"learning_rate": 0.0005676583991453631,
"loss": 0.0356,
"step": 22800
},
{
"epoch": 0.26,
"learning_rate": 0.0005676144213882145,
"loss": 0.0386,
"step": 23000
},
{
"epoch": 0.26,
"learning_rate": 0.000567570284427209,
"loss": 0.0431,
"step": 23200
},
{
"epoch": 0.26,
"learning_rate": 0.0005675255447345165,
"loss": 0.0377,
"step": 23400
},
{
"epoch": 0.26,
"learning_rate": 0.0005674804232095615,
"loss": 0.0419,
"step": 23600
},
{
"epoch": 0.26,
"learning_rate": 0.0005674349199133426,
"loss": 0.04,
"step": 23800
},
{
"epoch": 0.27,
"learning_rate": 0.0005673890349073741,
"loss": 0.05,
"step": 24000
},
{
"epoch": 0.27,
"learning_rate": 0.0005673427682536871,
"loss": 0.0306,
"step": 24200
},
{
"epoch": 0.27,
"learning_rate": 0.0005672961200148277,
"loss": 0.0366,
"step": 24400
},
{
"epoch": 0.27,
"learning_rate": 0.0005672490902538587,
"loss": 0.0463,
"step": 24600
},
{
"epoch": 0.28,
"learning_rate": 0.0005672016790343581,
"loss": 0.0421,
"step": 24800
},
{
"epoch": 0.28,
"learning_rate": 0.0005671538864204198,
"loss": 0.0478,
"step": 25000
},
{
"epoch": 0.28,
"learning_rate": 0.0005671059542948227,
"loss": 0.0521,
"step": 25200
},
{
"epoch": 0.28,
"learning_rate": 0.0005670574009925137,
"loss": 0.0449,
"step": 25400
},
{
"epoch": 0.28,
"learning_rate": 0.0005670084664908124,
"loss": 0.0485,
"step": 25600
},
{
"epoch": 0.29,
"learning_rate": 0.0005669591508558719,
"loss": 0.0308,
"step": 25800
},
{
"epoch": 0.29,
"learning_rate": 0.0005669094541543607,
"loss": 0.0354,
"step": 26000
},
{
"epoch": 0.29,
"learning_rate": 0.0005668596277895911,
"loss": 0.0415,
"step": 26200
},
{
"epoch": 0.29,
"learning_rate": 0.0005668091710614934,
"loss": 0.0555,
"step": 26400
},
{
"epoch": 0.3,
"learning_rate": 0.0005667583334695783,
"loss": 0.0399,
"step": 26600
},
{
"epoch": 0.3,
"learning_rate": 0.000566707115082572,
"loss": 0.052,
"step": 26800
},
{
"epoch": 0.3,
"learning_rate": 0.0005666555159697149,
"loss": 0.0398,
"step": 27000
},
{
"epoch": 0.3,
"learning_rate": 0.0005666035362007624,
"loss": 0.053,
"step": 27200
},
{
"epoch": 0.3,
"learning_rate": 0.0005665511758459846,
"loss": 0.0454,
"step": 27400
},
{
"epoch": 0.31,
"learning_rate": 0.0005664984349761658,
"loss": 0.0322,
"step": 27600
},
{
"epoch": 0.31,
"learning_rate": 0.0005664453136626051,
"loss": 0.0508,
"step": 27800
},
{
"epoch": 0.31,
"learning_rate": 0.0005663918119771154,
"loss": 0.0366,
"step": 28000
},
{
"epoch": 0.31,
"learning_rate": 0.0005663379299920243,
"loss": 0.0429,
"step": 28200
},
{
"epoch": 0.32,
"learning_rate": 0.0005662836677801735,
"loss": 0.0375,
"step": 28400
},
{
"epoch": 0.32,
"learning_rate": 0.0005662290254149182,
"loss": 0.0408,
"step": 28600
},
{
"epoch": 0.32,
"learning_rate": 0.0005661740029701281,
"loss": 0.0417,
"step": 28800
},
{
"epoch": 0.32,
"learning_rate": 0.0005661186005201862,
"loss": 0.0442,
"step": 29000
},
{
"epoch": 0.32,
"learning_rate": 0.0005660628181399899,
"loss": 0.0393,
"step": 29200
},
{
"epoch": 0.33,
"learning_rate": 0.0005660066559049495,
"loss": 0.0413,
"step": 29400
},
{
"epoch": 0.33,
"learning_rate": 0.0005659501138909893,
"loss": 0.036,
"step": 29600
},
{
"epoch": 0.33,
"learning_rate": 0.0005658931921745466,
"loss": 0.0391,
"step": 29800
},
{
"epoch": 0.33,
"learning_rate": 0.0005658358908325724,
"loss": 0.035,
"step": 30000
},
{
"epoch": 0.34,
"learning_rate": 0.000565778209942531,
"loss": 0.0397,
"step": 30200
},
{
"epoch": 0.34,
"learning_rate": 0.0005657201495823991,
"loss": 0.0363,
"step": 30400
},
{
"epoch": 0.34,
"learning_rate": 0.0005656620029730321,
"loss": 0.0536,
"step": 30600
},
{
"epoch": 0.34,
"learning_rate": 0.0005656031858050686,
"loss": 0.0525,
"step": 30800
},
{
"epoch": 0.34,
"learning_rate": 0.0005655439894036252,
"loss": 0.0343,
"step": 31000
},
{
"epoch": 0.35,
"learning_rate": 0.0005654844138487275,
"loss": 0.0463,
"step": 31200
},
{
"epoch": 0.35,
"learning_rate": 0.0005654244592209141,
"loss": 0.0399,
"step": 31400
},
{
"epoch": 0.35,
"learning_rate": 0.0005653641256012361,
"loss": 0.0584,
"step": 31600
},
{
"epoch": 0.35,
"learning_rate": 0.0005653034130712565,
"loss": 0.0398,
"step": 31800
},
{
"epoch": 0.36,
"learning_rate": 0.0005652423217130511,
"loss": 0.0473,
"step": 32000
},
{
"epoch": 0.36,
"learning_rate": 0.0005651808516092075,
"loss": 0.043,
"step": 32200
},
{
"epoch": 0.36,
"learning_rate": 0.0005651190028428255,
"loss": 0.0478,
"step": 32400
},
{
"epoch": 0.36,
"learning_rate": 0.000565057087575819,
"loss": 0.0531,
"step": 32600
},
{
"epoch": 0.36,
"learning_rate": 0.0005649944836279704,
"loss": 0.0346,
"step": 32800
},
{
"epoch": 0.37,
"learning_rate": 0.0005649315012695288,
"loss": 0.0452,
"step": 33000
},
{
"epoch": 0.37,
"learning_rate": 0.0005648681405856382,
"loss": 0.0385,
"step": 33200
},
{
"epoch": 0.37,
"learning_rate": 0.0005648044016619544,
"loss": 0.0418,
"step": 33400
},
{
"epoch": 0.37,
"learning_rate": 0.0005647402845846439,
"loss": 0.0386,
"step": 33600
},
{
"epoch": 0.38,
"learning_rate": 0.0005646757894403851,
"loss": 0.0363,
"step": 33800
},
{
"epoch": 0.38,
"learning_rate": 0.0005646109163163668,
"loss": 0.0398,
"step": 34000
},
{
"epoch": 0.38,
"learning_rate": 0.0005645456653002893,
"loss": 0.0469,
"step": 34200
},
{
"epoch": 0.38,
"learning_rate": 0.0005644800364803637,
"loss": 0.0393,
"step": 34400
},
{
"epoch": 0.38,
"learning_rate": 0.0005644143609174058,
"loss": 0.0453,
"step": 34600
},
{
"epoch": 0.39,
"learning_rate": 0.0005643479786443663,
"loss": 0.0347,
"step": 34800
},
{
"epoch": 0.39,
"learning_rate": 0.0005642812188347254,
"loss": 0.037,
"step": 35000
},
{
"epoch": 0.39,
"learning_rate": 0.0005642140815787342,
"loss": 0.036,
"step": 35200
},
{
"epoch": 0.39,
"learning_rate": 0.0005641465669671531,
"loss": 0.057,
"step": 35400
},
{
"epoch": 0.4,
"learning_rate": 0.0005640786750912536,
"loss": 0.0368,
"step": 35600
},
{
"epoch": 0.4,
"learning_rate": 0.0005640104060428169,
"loss": 0.0402,
"step": 35800
},
{
"epoch": 0.4,
"learning_rate": 0.0005639417599141336,
"loss": 0.0456,
"step": 36000
},
{
"epoch": 0.4,
"learning_rate": 0.0005638727367980045,
"loss": 0.0489,
"step": 36200
},
{
"epoch": 0.4,
"learning_rate": 0.0005638033367877403,
"loss": 0.041,
"step": 36400
},
{
"epoch": 0.41,
"learning_rate": 0.0005637335599771606,
"loss": 0.0665,
"step": 36600
},
{
"epoch": 0.41,
"learning_rate": 0.0005636634064605948,
"loss": 0.0418,
"step": 36800
},
{
"epoch": 0.41,
"learning_rate": 0.0005635928763328815,
"loss": 0.0341,
"step": 37000
},
{
"epoch": 0.41,
"learning_rate": 0.0005635223251590105,
"loss": 0.0583,
"step": 37200
},
{
"epoch": 0.42,
"learning_rate": 0.0005634510439774146,
"loss": 0.0392,
"step": 37400
},
{
"epoch": 0.42,
"learning_rate": 0.0005633793864717577,
"loss": 0.0599,
"step": 37600
},
{
"epoch": 0.42,
"learning_rate": 0.0005633073527389121,
"loss": 0.0436,
"step": 37800
},
{
"epoch": 0.42,
"learning_rate": 0.0005632349428762577,
"loss": 0.037,
"step": 38000
},
{
"epoch": 0.42,
"learning_rate": 0.0005631621569816834,
"loss": 0.0318,
"step": 38200
},
{
"epoch": 0.43,
"learning_rate": 0.0005630889951535865,
"loss": 0.0625,
"step": 38400
},
{
"epoch": 0.43,
"learning_rate": 0.0005630154574908725,
"loss": 0.0433,
"step": 38600
},
{
"epoch": 0.43,
"learning_rate": 0.0005629415440929546,
"loss": 0.0314,
"step": 38800
},
{
"epoch": 0.43,
"learning_rate": 0.0005628672550597543,
"loss": 0.0363,
"step": 39000
},
{
"epoch": 0.44,
"learning_rate": 0.0005627925904917011,
"loss": 0.05,
"step": 39200
},
{
"epoch": 0.44,
"learning_rate": 0.0005627175504897316,
"loss": 0.0477,
"step": 39400
},
{
"epoch": 0.44,
"learning_rate": 0.0005626425131654336,
"loss": 0.0529,
"step": 39600
},
{
"epoch": 0.44,
"learning_rate": 0.0005625667244763705,
"loss": 0.0282,
"step": 39800
},
{
"epoch": 0.44,
"learning_rate": 0.0005624905606587331,
"loss": 0.0447,
"step": 40000
},
{
"epoch": 0.45,
"learning_rate": 0.0005624140218154851,
"loss": 0.0393,
"step": 40200
},
{
"epoch": 0.45,
"learning_rate": 0.0005623371080500971,
"loss": 0.0411,
"step": 40400
},
{
"epoch": 0.45,
"learning_rate": 0.0005622598194665467,
"loss": 0.0455,
"step": 40600
},
{
"epoch": 0.45,
"learning_rate": 0.000562182156169318,
"loss": 0.0378,
"step": 40800
},
{
"epoch": 0.46,
"learning_rate": 0.0005621041182634019,
"loss": 0.0384,
"step": 41000
},
{
"epoch": 0.46,
"learning_rate": 0.0005620257058542957,
"loss": 0.0497,
"step": 41200
},
{
"epoch": 0.46,
"learning_rate": 0.0005619469190480028,
"loss": 0.0352,
"step": 41400
},
{
"epoch": 0.46,
"learning_rate": 0.0005618677579510328,
"loss": 0.0429,
"step": 41600
},
{
"epoch": 0.46,
"learning_rate": 0.0005617886212774089,
"loss": 0.0566,
"step": 41800
},
{
"epoch": 0.47,
"learning_rate": 0.0005617087137907502,
"loss": 0.0357,
"step": 42000
},
{
"epoch": 0.47,
"learning_rate": 0.0005616284323354373,
"loss": 0.0534,
"step": 42200
},
{
"epoch": 0.47,
"learning_rate": 0.0005615477770200006,
"loss": 0.0653,
"step": 42400
},
{
"epoch": 0.47,
"learning_rate": 0.0005614667479534755,
"loss": 0.057,
"step": 42600
},
{
"epoch": 0.48,
"learning_rate": 0.0005613853452454032,
"loss": 0.0451,
"step": 42800
},
{
"epoch": 0.48,
"learning_rate": 0.0005613035690058294,
"loss": 0.0394,
"step": 43000
},
{
"epoch": 0.48,
"learning_rate": 0.0005612214193453053,
"loss": 0.044,
"step": 43200
},
{
"epoch": 0.48,
"learning_rate": 0.0005611388963748866,
"loss": 0.0376,
"step": 43400
},
{
"epoch": 0.48,
"learning_rate": 0.0005610560002061337,
"loss": 0.0443,
"step": 43600
},
{
"epoch": 0.49,
"learning_rate": 0.0005609727309511118,
"loss": 0.0435,
"step": 43800
},
{
"epoch": 0.49,
"learning_rate": 0.0005608895078611188,
"loss": 0.0332,
"step": 44000
},
{
"epoch": 0.49,
"learning_rate": 0.0005608054946357917,
"loss": 0.0486,
"step": 44200
},
{
"epoch": 0.49,
"learning_rate": 0.0005607211086628469,
"loss": 0.0341,
"step": 44400
},
{
"epoch": 0.5,
"learning_rate": 0.0005606363500563632,
"loss": 0.0381,
"step": 44600
},
{
"epoch": 0.5,
"learning_rate": 0.0005605512189309235,
"loss": 0.041,
"step": 44800
},
{
"epoch": 0.5,
"learning_rate": 0.0005604657154016139,
"loss": 0.0353,
"step": 45000
},
{
"epoch": 0.5,
"learning_rate": 0.0005603798395840247,
"loss": 0.0489,
"step": 45200
},
{
"epoch": 0.5,
"learning_rate": 0.0005602935915942485,
"loss": 0.0508,
"step": 45400
},
{
"epoch": 0.51,
"learning_rate": 0.0005602069715488817,
"loss": 0.0339,
"step": 45600
},
{
"epoch": 0.51,
"learning_rate": 0.0005601199795650235,
"loss": 0.0382,
"step": 45800
},
{
"epoch": 0.51,
"learning_rate": 0.0005600326157602758,
"loss": 0.0422,
"step": 46000
},
{
"epoch": 0.51,
"learning_rate": 0.0005599448802527431,
"loss": 0.0317,
"step": 46200
},
{
"epoch": 0.52,
"learning_rate": 0.000559856773161033,
"loss": 0.0398,
"step": 46400
},
{
"epoch": 0.52,
"learning_rate": 0.0005597682946042545,
"loss": 0.0389,
"step": 46600
},
{
"epoch": 0.52,
"learning_rate": 0.0005596794447020196,
"loss": 0.0422,
"step": 46800
},
{
"epoch": 0.52,
"learning_rate": 0.0005595902235744417,
"loss": 0.0353,
"step": 47000
},
{
"epoch": 0.52,
"learning_rate": 0.0005595010802262207,
"loss": 0.0427,
"step": 47200
},
{
"epoch": 0.53,
"learning_rate": 0.0005594111188649213,
"loss": 0.0357,
"step": 47400
},
{
"epoch": 0.53,
"learning_rate": 0.000559320786641021,
"loss": 0.0519,
"step": 47600
},
{
"epoch": 0.53,
"learning_rate": 0.0005592300836766373,
"loss": 0.0521,
"step": 47800
},
{
"epoch": 0.53,
"learning_rate": 0.0005591390100943891,
"loss": 0.0474,
"step": 48000
},
{
"epoch": 0.54,
"learning_rate": 0.0005590475660173964,
"loss": 0.043,
"step": 48200
},
{
"epoch": 0.54,
"learning_rate": 0.0005589557515692793,
"loss": 0.0485,
"step": 48400
},
{
"epoch": 0.54,
"learning_rate": 0.0005588635668741596,
"loss": 0.0355,
"step": 48600
},
{
"epoch": 0.54,
"learning_rate": 0.0005587710120566592,
"loss": 0.0333,
"step": 48800
},
{
"epoch": 0.54,
"learning_rate": 0.0005586780872419001,
"loss": 0.0435,
"step": 49000
},
{
"epoch": 0.55,
"learning_rate": 0.0005585847925555047,
"loss": 0.0405,
"step": 49200
},
{
"epoch": 0.55,
"learning_rate": 0.0005584911281235955,
"loss": 0.0409,
"step": 49400
},
{
"epoch": 0.55,
"learning_rate": 0.0005583975651622658,
"loss": 0.0394,
"step": 49600
},
{
"epoch": 0.55,
"learning_rate": 0.0005583031634668374,
"loss": 0.0541,
"step": 49800
},
{
"epoch": 0.56,
"learning_rate": 0.0005582083924066216,
"loss": 0.0369,
"step": 50000
},
{
"epoch": 0.56,
"learning_rate": 0.0005581132521097367,
"loss": 0.0335,
"step": 50200
},
{
"epoch": 0.56,
"learning_rate": 0.0005580177427048001,
"loss": 0.0558,
"step": 50400
},
{
"epoch": 0.56,
"learning_rate": 0.0005579218643209284,
"loss": 0.0388,
"step": 50600
},
{
"epoch": 0.56,
"learning_rate": 0.0005578256170877366,
"loss": 0.0408,
"step": 50800
},
{
"epoch": 0.57,
"learning_rate": 0.0005577290011353388,
"loss": 0.0396,
"step": 51000
},
{
"epoch": 0.57,
"learning_rate": 0.0005576320165943473,
"loss": 0.0371,
"step": 51200
},
{
"epoch": 0.57,
"learning_rate": 0.0005575346635958727,
"loss": 0.0431,
"step": 51400
},
{
"epoch": 0.57,
"learning_rate": 0.000557437431794138,
"loss": 0.0415,
"step": 51600
},
{
"epoch": 0.58,
"learning_rate": 0.000557339344116661,
"loss": 0.0284,
"step": 51800
},
{
"epoch": 0.58,
"learning_rate": 0.0005572408883773566,
"loss": 0.0447,
"step": 52000
},
{
"epoch": 0.58,
"learning_rate": 0.0005571420647093241,
"loss": 0.0466,
"step": 52200
},
{
"epoch": 0.58,
"learning_rate": 0.0005570428732461607,
"loss": 0.035,
"step": 52400
},
{
"epoch": 0.58,
"learning_rate": 0.0005569433141219605,
"loss": 0.0378,
"step": 52600
},
{
"epoch": 0.59,
"learning_rate": 0.0005568433874713144,
"loss": 0.0318,
"step": 52800
},
{
"epoch": 0.59,
"learning_rate": 0.0005567430934293107,
"loss": 0.0422,
"step": 53000
},
{
"epoch": 0.59,
"learning_rate": 0.0005566424321315337,
"loss": 0.0415,
"step": 53200
},
{
"epoch": 0.59,
"learning_rate": 0.0005565414037140649,
"loss": 0.0501,
"step": 53400
},
{
"epoch": 0.6,
"learning_rate": 0.0005564400083134816,
"loss": 0.0399,
"step": 53600
},
{
"epoch": 0.6,
"learning_rate": 0.0005563382460668574,
"loss": 0.0345,
"step": 53800
},
{
"epoch": 0.6,
"learning_rate": 0.0005562361171117617,
"loss": 0.0315,
"step": 54000
},
{
"epoch": 0.6,
"learning_rate": 0.0005561336215862598,
"loss": 0.0425,
"step": 54200
},
{
"epoch": 0.6,
"learning_rate": 0.0005560307596289125,
"loss": 0.0402,
"step": 54400
},
{
"epoch": 0.61,
"learning_rate": 0.0005559280484309493,
"loss": 0.0552,
"step": 54600
},
{
"epoch": 0.61,
"learning_rate": 0.000555824455857993,
"loss": 0.0539,
"step": 54800
},
{
"epoch": 0.61,
"learning_rate": 0.0005557204972711436,
"loss": 0.046,
"step": 55000
},
{
"epoch": 0.61,
"learning_rate": 0.0005556161728109399,
"loss": 0.0483,
"step": 55200
},
{
"epoch": 0.62,
"learning_rate": 0.0005555114826184149,
"loss": 0.0355,
"step": 55400
},
{
"epoch": 0.62,
"learning_rate": 0.0005554064268350967,
"loss": 0.0371,
"step": 55600
},
{
"epoch": 0.62,
"learning_rate": 0.0005553010056030069,
"loss": 0.0488,
"step": 55800
},
{
"epoch": 0.62,
"learning_rate": 0.0005551952190646615,
"loss": 0.0288,
"step": 56000
},
{
"epoch": 0.62,
"learning_rate": 0.0005550890673630703,
"loss": 0.034,
"step": 56200
},
{
"epoch": 0.63,
"learning_rate": 0.0005549830841330922,
"loss": 0.0469,
"step": 56400
},
{
"epoch": 0.63,
"learning_rate": 0.0005548762043600333,
"loss": 0.0463,
"step": 56600
},
{
"epoch": 0.63,
"learning_rate": 0.0005547689598549955,
"loss": 0.0567,
"step": 56800
},
{
"epoch": 0.63,
"learning_rate": 0.0005546613507629596,
"loss": 0.0432,
"step": 57000
},
{
"epoch": 0.64,
"learning_rate": 0.0005545533772293995,
"loss": 0.0486,
"step": 57200
},
{
"epoch": 0.64,
"learning_rate": 0.0005544450394002813,
"loss": 0.0362,
"step": 57400
},
{
"epoch": 0.64,
"learning_rate": 0.0005543363374220639,
"loss": 0.0489,
"step": 57600
},
{
"epoch": 0.64,
"learning_rate": 0.0005542272714416986,
"loss": 0.0479,
"step": 57800
},
{
"epoch": 0.64,
"learning_rate": 0.0005541178416066283,
"loss": 0.035,
"step": 58000
},
{
"epoch": 0.65,
"learning_rate": 0.0005540080480647882,
"loss": 0.0363,
"step": 58200
},
{
"epoch": 0.65,
"learning_rate": 0.0005538978909646053,
"loss": 0.0395,
"step": 58400
},
{
"epoch": 0.65,
"learning_rate": 0.0005537873704549974,
"loss": 0.0286,
"step": 58600
},
{
"epoch": 0.65,
"learning_rate": 0.0005536764866853745,
"loss": 0.0426,
"step": 58800
},
{
"epoch": 0.66,
"learning_rate": 0.0005535652398056369,
"loss": 0.0327,
"step": 59000
},
{
"epoch": 0.66,
"learning_rate": 0.0005534536299661762,
"loss": 0.0343,
"step": 59200
},
{
"epoch": 0.66,
"learning_rate": 0.0005533416573178747,
"loss": 0.0426,
"step": 59400
},
{
"epoch": 0.66,
"learning_rate": 0.0005532293220121047,
"loss": 0.0326,
"step": 59600
},
{
"epoch": 0.66,
"learning_rate": 0.0005531166242007295,
"loss": 0.0401,
"step": 59800
},
{
"epoch": 0.67,
"learning_rate": 0.0005530035640361017,
"loss": 0.0547,
"step": 60000
},
{
"epoch": 0.67,
"learning_rate": 0.0005528901416710644,
"loss": 0.0406,
"step": 60200
},
{
"epoch": 0.67,
"learning_rate": 0.00055277635725895,
"loss": 0.0392,
"step": 60400
},
{
"epoch": 0.67,
"learning_rate": 0.0005526622109535803,
"loss": 0.0331,
"step": 60600
},
{
"epoch": 0.68,
"learning_rate": 0.000552548276349058,
"loss": 0.0497,
"step": 60800
},
{
"epoch": 0.68,
"learning_rate": 0.0005524334085281352,
"loss": 0.0493,
"step": 61000
},
{
"epoch": 0.68,
"learning_rate": 0.0005523181792775799,
"loss": 0.0509,
"step": 61200
},
{
"epoch": 0.68,
"learning_rate": 0.0005522025887531671,
"loss": 0.0278,
"step": 61400
},
{
"epoch": 0.68,
"learning_rate": 0.0005520866371111605,
"loss": 0.0386,
"step": 61600
},
{
"epoch": 0.69,
"learning_rate": 0.0005519703245083118,
"loss": 0.0357,
"step": 61800
},
{
"epoch": 0.69,
"learning_rate": 0.0005518536511018605,
"loss": 0.0352,
"step": 62000
},
{
"epoch": 0.69,
"learning_rate": 0.0005517366170495342,
"loss": 0.0404,
"step": 62200
},
{
"epoch": 0.69,
"learning_rate": 0.000551619222509548,
"loss": 0.0434,
"step": 62400
},
{
"epoch": 0.7,
"learning_rate": 0.0005515014676406042,
"loss": 0.0357,
"step": 62600
},
{
"epoch": 0.7,
"learning_rate": 0.0005513833526018923,
"loss": 0.0374,
"step": 62800
},
{
"epoch": 0.7,
"learning_rate": 0.0005512648775530885,
"loss": 0.0444,
"step": 63000
},
{
"epoch": 0.7,
"learning_rate": 0.0005511460426543559,
"loss": 0.038,
"step": 63200
},
{
"epoch": 0.7,
"learning_rate": 0.0005510268480663441,
"loss": 0.0444,
"step": 63400
},
{
"epoch": 0.71,
"learning_rate": 0.0005509072939501887,
"loss": 0.0357,
"step": 63600
},
{
"epoch": 0.71,
"learning_rate": 0.0005507879809285815,
"loss": 0.0418,
"step": 63800
},
{
"epoch": 0.71,
"learning_rate": 0.0005506677100371086,
"loss": 0.0314,
"step": 64000
},
{
"epoch": 0.71,
"learning_rate": 0.0005505470801030006,
"loss": 0.0377,
"step": 64200
},
{
"epoch": 0.72,
"learning_rate": 0.0005504260912893339,
"loss": 0.0358,
"step": 64400
},
{
"epoch": 0.72,
"learning_rate": 0.0005503047437596697,
"loss": 0.0405,
"step": 64600
},
{
"epoch": 0.72,
"learning_rate": 0.0005501830376780543,
"loss": 0.0348,
"step": 64800
},
{
"epoch": 0.72,
"learning_rate": 0.0005500609732090188,
"loss": 0.0349,
"step": 65000
},
{
"epoch": 0.72,
"learning_rate": 0.0005499385505175786,
"loss": 0.0395,
"step": 65200
},
{
"epoch": 0.73,
"learning_rate": 0.0005498157697692334,
"loss": 0.0415,
"step": 65400
},
{
"epoch": 0.73,
"learning_rate": 0.000549692631129967,
"loss": 0.0302,
"step": 65600
},
{
"epoch": 0.73,
"learning_rate": 0.0005495691347662473,
"loss": 0.0423,
"step": 65800
},
{
"epoch": 0.73,
"learning_rate": 0.0005494452808450252,
"loss": 0.0412,
"step": 66000
},
{
"epoch": 0.74,
"learning_rate": 0.000549321691479022,
"loss": 0.0338,
"step": 66200
},
{
"epoch": 0.74,
"learning_rate": 0.0005491977484533271,
"loss": 0.0386,
"step": 66400
},
{
"epoch": 0.74,
"learning_rate": 0.0005490728264358403,
"loss": 0.0319,
"step": 66600
},
{
"epoch": 0.74,
"learning_rate": 0.0005489475475317964,
"loss": 0.0425,
"step": 66800
},
{
"epoch": 0.74,
"learning_rate": 0.0005488219119105563,
"loss": 0.0395,
"step": 67000
},
{
"epoch": 0.75,
"learning_rate": 0.0005486959197419632,
"loss": 0.0537,
"step": 67200
},
{
"epoch": 0.75,
"learning_rate": 0.0005485695711963426,
"loss": 0.0456,
"step": 67400
},
{
"epoch": 0.75,
"learning_rate": 0.000548442866444501,
"loss": 0.0408,
"step": 67600
},
{
"epoch": 0.75,
"learning_rate": 0.0005483158056577273,
"loss": 0.0371,
"step": 67800
},
{
"epoch": 0.76,
"learning_rate": 0.0005481883890077914,
"loss": 0.0415,
"step": 68000
},
{
"epoch": 0.76,
"learning_rate": 0.0005480606166669441,
"loss": 0.0458,
"step": 68200
},
{
"epoch": 0.76,
"learning_rate": 0.0005479324888079173,
"loss": 0.0385,
"step": 68400
},
{
"epoch": 0.76,
"learning_rate": 0.0005478046489035769,
"loss": 0.0437,
"step": 68600
},
{
"epoch": 0.76,
"learning_rate": 0.0005476758123037319,
"loss": 0.0331,
"step": 68800
},
{
"epoch": 0.77,
"learning_rate": 0.0005475466207059137,
"loss": 0.042,
"step": 69000
},
{
"epoch": 0.77,
"learning_rate": 0.0005474170742847724,
"loss": 0.0377,
"step": 69200
},
{
"epoch": 0.77,
"learning_rate": 0.0005472871732154382,
"loss": 0.0399,
"step": 69400
},
{
"epoch": 0.77,
"learning_rate": 0.0005471569176735208,
"loss": 0.0568,
"step": 69600
},
{
"epoch": 0.77,
"learning_rate": 0.0005470263078351086,
"loss": 0.031,
"step": 69800
},
{
"epoch": 0.78,
"learning_rate": 0.0005468953438767697,
"loss": 0.0308,
"step": 70000
},
{
"epoch": 0.78,
"learning_rate": 0.0005467640259755503,
"loss": 0.0417,
"step": 70200
},
{
"epoch": 0.78,
"learning_rate": 0.0005466323543089754,
"loss": 0.0467,
"step": 70400
},
{
"epoch": 0.78,
"learning_rate": 0.0005465003290550484,
"loss": 0.0316,
"step": 70600
},
{
"epoch": 0.79,
"learning_rate": 0.0005463679503922502,
"loss": 0.0411,
"step": 70800
},
{
"epoch": 0.79,
"learning_rate": 0.0005462352184995399,
"loss": 0.0365,
"step": 71000
},
{
"epoch": 0.79,
"learning_rate": 0.0005461021335563539,
"loss": 0.0358,
"step": 71200
},
{
"epoch": 0.79,
"learning_rate": 0.0005459686957426061,
"loss": 0.0497,
"step": 71400
},
{
"epoch": 0.79,
"learning_rate": 0.0005458349052386872,
"loss": 0.0415,
"step": 71600
},
{
"epoch": 0.8,
"learning_rate": 0.0005457007622254647,
"loss": 0.0367,
"step": 71800
},
{
"epoch": 0.8,
"learning_rate": 0.0005455662668842827,
"loss": 0.0343,
"step": 72000
},
{
"epoch": 0.8,
"learning_rate": 0.0005454314193969619,
"loss": 0.0434,
"step": 72200
},
{
"epoch": 0.8,
"learning_rate": 0.0005452962199457985,
"loss": 0.0465,
"step": 72400
},
{
"epoch": 0.81,
"learning_rate": 0.0005451606687135648,
"loss": 0.036,
"step": 72600
},
{
"epoch": 0.81,
"learning_rate": 0.0005450247658835089,
"loss": 0.0375,
"step": 72800
},
{
"epoch": 0.81,
"learning_rate": 0.0005448885116393537,
"loss": 0.0358,
"step": 73000
},
{
"epoch": 0.81,
"learning_rate": 0.0005447519061652976,
"loss": 0.0473,
"step": 73200
},
{
"epoch": 0.81,
"learning_rate": 0.0005446149496460137,
"loss": 0.0347,
"step": 73400
},
{
"epoch": 0.82,
"learning_rate": 0.000544478329676004,
"loss": 0.0562,
"step": 73600
},
{
"epoch": 0.82,
"learning_rate": 0.0005443406733750912,
"loss": 0.0354,
"step": 73800
},
{
"epoch": 0.82,
"learning_rate": 0.0005442026665848845,
"loss": 0.052,
"step": 74000
},
{
"epoch": 0.82,
"learning_rate": 0.0005440643094919513,
"loss": 0.0347,
"step": 74200
},
{
"epoch": 0.83,
"learning_rate": 0.0005439256022833326,
"loss": 0.0345,
"step": 74400
},
{
"epoch": 0.83,
"learning_rate": 0.0005437865451465428,
"loss": 0.0502,
"step": 74600
},
{
"epoch": 0.83,
"learning_rate": 0.0005436471382695693,
"loss": 0.0384,
"step": 74800
},
{
"epoch": 0.83,
"learning_rate": 0.0005435073818408721,
"loss": 0.0327,
"step": 75000
},
{
"epoch": 0.83,
"learning_rate": 0.0005433672760493841,
"loss": 0.0468,
"step": 75200
},
{
"epoch": 0.84,
"learning_rate": 0.0005432268210845104,
"loss": 0.0531,
"step": 75400
},
{
"epoch": 0.84,
"learning_rate": 0.000543086017136128,
"loss": 0.0342,
"step": 75600
},
{
"epoch": 0.84,
"learning_rate": 0.0005429448643945857,
"loss": 0.041,
"step": 75800
},
{
"epoch": 0.84,
"learning_rate": 0.0005428040714242554,
"loss": 0.0525,
"step": 76000
},
{
"epoch": 0.85,
"learning_rate": 0.0005426622234109044,
"loss": 0.0412,
"step": 76200
},
{
"epoch": 0.85,
"learning_rate": 0.0005425200271773083,
"loss": 0.0394,
"step": 76400
},
{
"epoch": 0.85,
"learning_rate": 0.000542377482915698,
"loss": 0.031,
"step": 76600
},
{
"epoch": 0.85,
"learning_rate": 0.0005422345908187753,
"loss": 0.0324,
"step": 76800
},
{
"epoch": 0.85,
"learning_rate": 0.0005420913510797115,
"loss": 0.0365,
"step": 77000
},
{
"epoch": 0.86,
"learning_rate": 0.0005419477638921487,
"loss": 0.0373,
"step": 77200
},
{
"epoch": 0.86,
"learning_rate": 0.0005418038294501984,
"loss": 0.0402,
"step": 77400
},
{
"epoch": 0.86,
"learning_rate": 0.0005416595479484413,
"loss": 0.0352,
"step": 77600
},
{
"epoch": 0.86,
"learning_rate": 0.0005415149195819273,
"loss": 0.0357,
"step": 77800
},
{
"epoch": 0.87,
"learning_rate": 0.0005413699445461759,
"loss": 0.0294,
"step": 78000
},
{
"epoch": 0.87,
"learning_rate": 0.0005412246230371741,
"loss": 0.0369,
"step": 78200
},
{
"epoch": 0.87,
"learning_rate": 0.0005410789552513784,
"loss": 0.0463,
"step": 78400
},
{
"epoch": 0.87,
"learning_rate": 0.0005409329413857127,
"loss": 0.0447,
"step": 78600
},
{
"epoch": 0.87,
"learning_rate": 0.000540786581637569,
"loss": 0.0476,
"step": 78800
},
{
"epoch": 0.88,
"learning_rate": 0.0005406398762048069,
"loss": 0.032,
"step": 79000
},
{
"epoch": 0.88,
"learning_rate": 0.0005404928252857535,
"loss": 0.0512,
"step": 79200
},
{
"epoch": 0.88,
"learning_rate": 0.0005403454290792025,
"loss": 0.0407,
"step": 79400
},
{
"epoch": 0.88,
"learning_rate": 0.0005401984273489655,
"loss": 0.0435,
"step": 79600
},
{
"epoch": 0.89,
"learning_rate": 0.0005400503428896132,
"loss": 0.037,
"step": 79800
},
{
"epoch": 0.89,
"learning_rate": 0.0005399019137409428,
"loss": 0.0396,
"step": 80000
},
{
"epoch": 0.89,
"learning_rate": 0.0005397531401036115,
"loss": 0.0398,
"step": 80200
},
{
"epoch": 0.89,
"learning_rate": 0.0005396040221787419,
"loss": 0.0346,
"step": 80400
},
{
"epoch": 0.89,
"learning_rate": 0.0005394545601679225,
"loss": 0.033,
"step": 80600
},
{
"epoch": 0.9,
"learning_rate": 0.0005393047542732067,
"loss": 0.0449,
"step": 80800
},
{
"epoch": 0.9,
"learning_rate": 0.0005391546046971128,
"loss": 0.0292,
"step": 81000
},
{
"epoch": 0.9,
"learning_rate": 0.0005390041116426239,
"loss": 0.0419,
"step": 81200
},
{
"epoch": 0.9,
"learning_rate": 0.0005388532753131871,
"loss": 0.0417,
"step": 81400
},
{
"epoch": 0.91,
"learning_rate": 0.0005387020959127137,
"loss": 0.0393,
"step": 81600
},
{
"epoch": 0.91,
"learning_rate": 0.0005385505736455791,
"loss": 0.0408,
"step": 81800
},
{
"epoch": 0.91,
"learning_rate": 0.0005383994688932983,
"loss": 0.0354,
"step": 82000
},
{
"epoch": 0.91,
"learning_rate": 0.0005382472632195911,
"loss": 0.0335,
"step": 82200
},
{
"epoch": 0.91,
"learning_rate": 0.0005380947152940981,
"loss": 0.0378,
"step": 82400
},
{
"epoch": 0.92,
"learning_rate": 0.0005379418253230443,
"loss": 0.0355,
"step": 82600
},
{
"epoch": 0.92,
"learning_rate": 0.0005377885935131172,
"loss": 0.0325,
"step": 82800
},
{
"epoch": 0.92,
"learning_rate": 0.0005376350200714668,
"loss": 0.0506,
"step": 83000
},
{
"epoch": 0.92,
"learning_rate": 0.0005374811052057044,
"loss": 0.0378,
"step": 83200
},
{
"epoch": 0.93,
"learning_rate": 0.0005373268491239034,
"loss": 0.0337,
"step": 83400
},
{
"epoch": 0.93,
"learning_rate": 0.000537172252034598,
"loss": 0.0499,
"step": 83600
},
{
"epoch": 0.93,
"learning_rate": 0.0005370173141467838,
"loss": 0.0344,
"step": 83800
},
{
"epoch": 0.93,
"learning_rate": 0.0005368620356699166,
"loss": 0.0406,
"step": 84000
},
{
"epoch": 0.93,
"learning_rate": 0.0005367071957545386,
"loss": 0.0563,
"step": 84200
},
{
"epoch": 0.94,
"learning_rate": 0.0005365512384300956,
"loss": 0.0382,
"step": 84400
},
{
"epoch": 0.94,
"learning_rate": 0.0005363949411466742,
"loss": 0.0396,
"step": 84600
},
{
"epoch": 0.94,
"learning_rate": 0.0005362383041155683,
"loss": 0.0408,
"step": 84800
},
{
"epoch": 0.94,
"learning_rate": 0.0005360813275485313,
"loss": 0.0319,
"step": 85000
},
{
"epoch": 0.95,
"learning_rate": 0.0005359240116577751,
"loss": 0.0491,
"step": 85200
},
{
"epoch": 0.95,
"learning_rate": 0.0005357663566559707,
"loss": 0.0347,
"step": 85400
},
{
"epoch": 0.95,
"learning_rate": 0.0005356083627562474,
"loss": 0.0417,
"step": 85600
},
{
"epoch": 0.95,
"learning_rate": 0.0005354500301721926,
"loss": 0.0595,
"step": 85800
},
{
"epoch": 0.95,
"learning_rate": 0.0005352913591178517,
"loss": 0.033,
"step": 86000
},
{
"epoch": 0.96,
"learning_rate": 0.0005351331456953341,
"loss": 0.0459,
"step": 86200
},
{
"epoch": 0.96,
"learning_rate": 0.0005349738000340555,
"loss": 0.0398,
"step": 86400
},
{
"epoch": 0.96,
"learning_rate": 0.0005348141165462928,
"loss": 0.0376,
"step": 86600
},
{
"epoch": 0.96,
"learning_rate": 0.0005346540954479174,
"loss": 0.0344,
"step": 86800
},
{
"epoch": 0.97,
"learning_rate": 0.0005344937369552576,
"loss": 0.0353,
"step": 87000
},
{
"epoch": 0.97,
"learning_rate": 0.0005343330412850971,
"loss": 0.0447,
"step": 87200
},
{
"epoch": 0.97,
"learning_rate": 0.0005341720086546761,
"loss": 0.0326,
"step": 87400
},
{
"epoch": 0.97,
"learning_rate": 0.0005340106392816899,
"loss": 0.0356,
"step": 87600
},
{
"epoch": 0.97,
"learning_rate": 0.0005338489333842892,
"loss": 0.0323,
"step": 87800
},
{
"epoch": 0.98,
"learning_rate": 0.0005336868911810796,
"loss": 0.0381,
"step": 88000
},
{
"epoch": 0.98,
"learning_rate": 0.0005335245128911213,
"loss": 0.0434,
"step": 88200
},
{
"epoch": 0.98,
"learning_rate": 0.000533361798733929,
"loss": 0.0331,
"step": 88400
},
{
"epoch": 0.98,
"learning_rate": 0.0005331987489294713,
"loss": 0.0513,
"step": 88600
},
{
"epoch": 0.99,
"learning_rate": 0.0005330353636981706,
"loss": 0.0333,
"step": 88800
},
{
"epoch": 0.99,
"learning_rate": 0.0005328716432609027,
"loss": 0.0448,
"step": 89000
},
{
"epoch": 0.99,
"learning_rate": 0.0005327084089490138,
"loss": 0.0479,
"step": 89200
},
{
"epoch": 0.99,
"learning_rate": 0.0005325440204375134,
"loss": 0.0394,
"step": 89400
},
{
"epoch": 0.99,
"learning_rate": 0.0005323792973842787,
"loss": 0.042,
"step": 89600
},
{
"epoch": 1.0,
"learning_rate": 0.0005322142400119943,
"loss": 0.0421,
"step": 89800
},
{
"epoch": 1.0,
"learning_rate": 0.0005320488485437965,
"loss": 0.0374,
"step": 90000
},
{
"epoch": 1.0,
"eval_accuracy": 0.9394310337232581,
"eval_auc": 0.8882998322705576,
"eval_f1": 0.5088262320154422,
"eval_loss": 0.30141326785087585,
"eval_mcc": 0.5283107198885248,
"eval_precision": 0.3662816185295454,
"eval_recall": 0.8330034442739891,
"eval_runtime": 4672.3874,
"eval_samples_per_second": 24.286,
"eval_steps_per_second": 4.857,
"step": 90066
}
],
"logging_steps": 200,
"max_steps": 540396,
"num_train_epochs": 6,
"save_steps": 500,
"total_flos": 9.00427404555e+16,
"trial_name": null,
"trial_params": null
}