mrm8488's picture
Initial commit from mrm8488
dda18f1
raw history blame
No virus
21.5 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.6920639267619961,
"global_step": 86500,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"learning_rate": 4.997283134350093e-05,
"loss": 1.6457,
"step": 500
},
{
"epoch": 0.02,
"learning_rate": 4.994566268700186e-05,
"loss": 1.6172,
"step": 1000
},
{
"epoch": 0.03,
"learning_rate": 4.991849403050279e-05,
"loss": 1.6618,
"step": 1500
},
{
"epoch": 0.04,
"learning_rate": 4.989132537400373e-05,
"loss": 1.6384,
"step": 2000
},
{
"epoch": 0.05,
"learning_rate": 4.986415671750466e-05,
"loss": 1.6445,
"step": 2500
},
{
"epoch": 0.06,
"learning_rate": 4.983698806100559e-05,
"loss": 1.644,
"step": 3000
},
{
"epoch": 0.07,
"learning_rate": 4.980981940450652e-05,
"loss": 1.6763,
"step": 3500
},
{
"epoch": 0.08,
"learning_rate": 4.9782650748007454e-05,
"loss": 1.6557,
"step": 4000
},
{
"epoch": 0.09,
"learning_rate": 4.9755482091508384e-05,
"loss": 1.6375,
"step": 4500
},
{
"epoch": 0.1,
"learning_rate": 4.972831343500931e-05,
"loss": 1.6605,
"step": 5000
},
{
"epoch": 0.11,
"learning_rate": 4.970114477851024e-05,
"loss": 1.6571,
"step": 5500
},
{
"epoch": 0.12,
"learning_rate": 4.967397612201118e-05,
"loss": 1.6556,
"step": 6000
},
{
"epoch": 0.13,
"learning_rate": 4.964680746551211e-05,
"loss": 1.6593,
"step": 6500
},
{
"epoch": 0.14,
"learning_rate": 4.961963880901304e-05,
"loss": 1.6578,
"step": 7000
},
{
"epoch": 0.15,
"learning_rate": 4.959247015251397e-05,
"loss": 1.6519,
"step": 7500
},
{
"epoch": 0.16,
"learning_rate": 4.9565301496014905e-05,
"loss": 1.6479,
"step": 8000
},
{
"epoch": 0.17,
"learning_rate": 4.9538132839515835e-05,
"loss": 1.6705,
"step": 8500
},
{
"epoch": 0.18,
"learning_rate": 4.9510964183016765e-05,
"loss": 1.6807,
"step": 9000
},
{
"epoch": 0.19,
"learning_rate": 4.9483795526517694e-05,
"loss": 1.6667,
"step": 9500
},
{
"epoch": 0.2,
"learning_rate": 4.945662687001863e-05,
"loss": 1.6604,
"step": 10000
},
{
"epoch": 0.21,
"learning_rate": 4.942945821351956e-05,
"loss": 1.6885,
"step": 10500
},
{
"epoch": 0.22,
"learning_rate": 4.940228955702049e-05,
"loss": 1.6648,
"step": 11000
},
{
"epoch": 0.22,
"learning_rate": 4.937512090052142e-05,
"loss": 1.6483,
"step": 11500
},
{
"epoch": 0.23,
"learning_rate": 4.9347952244022357e-05,
"loss": 1.6835,
"step": 12000
},
{
"epoch": 0.24,
"learning_rate": 4.9320783587523286e-05,
"loss": 1.6797,
"step": 12500
},
{
"epoch": 0.25,
"learning_rate": 4.9293614931024216e-05,
"loss": 1.6698,
"step": 13000
},
{
"epoch": 0.26,
"learning_rate": 4.9266446274525146e-05,
"loss": 1.663,
"step": 13500
},
{
"epoch": 0.27,
"learning_rate": 4.923927761802608e-05,
"loss": 1.6552,
"step": 14000
},
{
"epoch": 0.28,
"learning_rate": 4.921210896152701e-05,
"loss": 1.6769,
"step": 14500
},
{
"epoch": 0.29,
"learning_rate": 4.918494030502794e-05,
"loss": 1.6656,
"step": 15000
},
{
"epoch": 0.3,
"learning_rate": 4.915777164852887e-05,
"loss": 1.6515,
"step": 15500
},
{
"epoch": 0.31,
"learning_rate": 4.913060299202981e-05,
"loss": 1.6606,
"step": 16000
},
{
"epoch": 0.32,
"learning_rate": 4.910343433553074e-05,
"loss": 1.6586,
"step": 16500
},
{
"epoch": 0.33,
"learning_rate": 4.907626567903167e-05,
"loss": 1.6945,
"step": 17000
},
{
"epoch": 0.34,
"learning_rate": 4.90490970225326e-05,
"loss": 1.7064,
"step": 17500
},
{
"epoch": 0.35,
"learning_rate": 4.9021928366033534e-05,
"loss": 1.6779,
"step": 18000
},
{
"epoch": 0.36,
"learning_rate": 4.899475970953446e-05,
"loss": 1.6548,
"step": 18500
},
{
"epoch": 0.37,
"learning_rate": 4.896759105303539e-05,
"loss": 1.6983,
"step": 19000
},
{
"epoch": 0.38,
"learning_rate": 4.894042239653632e-05,
"loss": 1.6528,
"step": 19500
},
{
"epoch": 0.39,
"learning_rate": 4.891325374003726e-05,
"loss": 1.6845,
"step": 20000
},
{
"epoch": 0.4,
"learning_rate": 4.888608508353818e-05,
"loss": 1.672,
"step": 20500
},
{
"epoch": 0.41,
"learning_rate": 4.885891642703912e-05,
"loss": 1.6565,
"step": 21000
},
{
"epoch": 0.42,
"learning_rate": 4.883174777054005e-05,
"loss": 1.6549,
"step": 21500
},
{
"epoch": 0.43,
"learning_rate": 4.8804579114040985e-05,
"loss": 1.6702,
"step": 22000
},
{
"epoch": 0.44,
"learning_rate": 4.877741045754191e-05,
"loss": 1.6959,
"step": 22500
},
{
"epoch": 0.45,
"learning_rate": 4.8750241801042845e-05,
"loss": 1.6585,
"step": 23000
},
{
"epoch": 0.46,
"learning_rate": 4.8723073144543774e-05,
"loss": 1.678,
"step": 23500
},
{
"epoch": 0.47,
"learning_rate": 4.869590448804471e-05,
"loss": 1.6927,
"step": 24000
},
{
"epoch": 0.48,
"learning_rate": 4.8668735831545634e-05,
"loss": 1.6876,
"step": 24500
},
{
"epoch": 0.49,
"learning_rate": 4.864156717504657e-05,
"loss": 1.6907,
"step": 25000
},
{
"epoch": 0.5,
"learning_rate": 4.86143985185475e-05,
"loss": 1.6822,
"step": 25500
},
{
"epoch": 0.51,
"learning_rate": 4.8587229862048436e-05,
"loss": 1.7038,
"step": 26000
},
{
"epoch": 0.52,
"learning_rate": 4.856006120554936e-05,
"loss": 1.6671,
"step": 26500
},
{
"epoch": 0.53,
"learning_rate": 4.8532892549050296e-05,
"loss": 1.6546,
"step": 27000
},
{
"epoch": 0.54,
"learning_rate": 4.8505723892551226e-05,
"loss": 1.6896,
"step": 27500
},
{
"epoch": 0.55,
"learning_rate": 4.847855523605216e-05,
"loss": 1.6644,
"step": 28000
},
{
"epoch": 0.56,
"learning_rate": 4.8451386579553085e-05,
"loss": 1.6728,
"step": 28500
},
{
"epoch": 0.57,
"learning_rate": 4.842421792305402e-05,
"loss": 1.687,
"step": 29000
},
{
"epoch": 0.58,
"learning_rate": 4.839704926655495e-05,
"loss": 1.6557,
"step": 29500
},
{
"epoch": 0.59,
"learning_rate": 4.836988061005588e-05,
"loss": 1.7109,
"step": 30000
},
{
"epoch": 0.6,
"learning_rate": 4.834271195355681e-05,
"loss": 1.7037,
"step": 30500
},
{
"epoch": 0.61,
"learning_rate": 4.831554329705775e-05,
"loss": 1.6809,
"step": 31000
},
{
"epoch": 0.62,
"learning_rate": 4.828837464055868e-05,
"loss": 1.6785,
"step": 31500
},
{
"epoch": 0.63,
"learning_rate": 4.826120598405961e-05,
"loss": 1.6842,
"step": 32000
},
{
"epoch": 0.64,
"learning_rate": 4.8234037327560536e-05,
"loss": 1.6569,
"step": 32500
},
{
"epoch": 0.65,
"learning_rate": 4.820686867106147e-05,
"loss": 1.662,
"step": 33000
},
{
"epoch": 0.66,
"learning_rate": 4.81797000145624e-05,
"loss": 1.6857,
"step": 33500
},
{
"epoch": 0.67,
"learning_rate": 4.815253135806333e-05,
"loss": 1.7066,
"step": 34000
},
{
"epoch": 0.67,
"learning_rate": 4.812536270156426e-05,
"loss": 1.6824,
"step": 34500
},
{
"epoch": 0.68,
"learning_rate": 4.80981940450652e-05,
"loss": 1.6847,
"step": 35000
},
{
"epoch": 0.69,
"learning_rate": 4.807102538856613e-05,
"loss": 1.6629,
"step": 35500
},
{
"epoch": 0.7,
"learning_rate": 4.804385673206706e-05,
"loss": 1.6498,
"step": 36000
},
{
"epoch": 0.71,
"learning_rate": 4.801668807556799e-05,
"loss": 1.6833,
"step": 36500
},
{
"epoch": 0.72,
"learning_rate": 4.7989519419068924e-05,
"loss": 1.6693,
"step": 37000
},
{
"epoch": 0.73,
"learning_rate": 4.7962350762569854e-05,
"loss": 1.6912,
"step": 37500
},
{
"epoch": 0.74,
"learning_rate": 4.7935182106070784e-05,
"loss": 1.6591,
"step": 38000
},
{
"epoch": 0.75,
"learning_rate": 4.7908013449571714e-05,
"loss": 1.6655,
"step": 38500
},
{
"epoch": 0.76,
"learning_rate": 4.788084479307265e-05,
"loss": 1.6699,
"step": 39000
},
{
"epoch": 0.77,
"learning_rate": 4.785367613657358e-05,
"loss": 1.6544,
"step": 39500
},
{
"epoch": 0.78,
"learning_rate": 4.782650748007451e-05,
"loss": 1.6876,
"step": 40000
},
{
"epoch": 0.79,
"learning_rate": 4.779933882357544e-05,
"loss": 1.6916,
"step": 40500
},
{
"epoch": 0.8,
"learning_rate": 4.7772170167076376e-05,
"loss": 1.6976,
"step": 41000
},
{
"epoch": 0.81,
"learning_rate": 4.7745001510577305e-05,
"loss": 1.6905,
"step": 41500
},
{
"epoch": 0.82,
"learning_rate": 4.7717832854078235e-05,
"loss": 1.6886,
"step": 42000
},
{
"epoch": 0.83,
"learning_rate": 4.7690664197579165e-05,
"loss": 1.6585,
"step": 42500
},
{
"epoch": 0.84,
"learning_rate": 4.76634955410801e-05,
"loss": 1.6877,
"step": 43000
},
{
"epoch": 0.85,
"learning_rate": 4.763632688458103e-05,
"loss": 1.6786,
"step": 43500
},
{
"epoch": 0.86,
"learning_rate": 4.760915822808196e-05,
"loss": 1.676,
"step": 44000
},
{
"epoch": 0.87,
"learning_rate": 4.758198957158289e-05,
"loss": 1.6702,
"step": 44500
},
{
"epoch": 0.88,
"learning_rate": 4.755482091508383e-05,
"loss": 1.6733,
"step": 45000
},
{
"epoch": 0.89,
"learning_rate": 4.752765225858476e-05,
"loss": 1.6813,
"step": 45500
},
{
"epoch": 0.9,
"learning_rate": 4.750048360208569e-05,
"loss": 1.6726,
"step": 46000
},
{
"epoch": 0.91,
"learning_rate": 4.7473314945586616e-05,
"loss": 1.6844,
"step": 46500
},
{
"epoch": 0.92,
"learning_rate": 4.744614628908755e-05,
"loss": 1.6694,
"step": 47000
},
{
"epoch": 0.93,
"learning_rate": 4.741897763258848e-05,
"loss": 1.6795,
"step": 47500
},
{
"epoch": 0.94,
"learning_rate": 4.739180897608941e-05,
"loss": 1.6512,
"step": 48000
},
{
"epoch": 0.95,
"learning_rate": 4.736464031959034e-05,
"loss": 1.7015,
"step": 48500
},
{
"epoch": 0.96,
"learning_rate": 4.733747166309128e-05,
"loss": 1.6633,
"step": 49000
},
{
"epoch": 0.97,
"learning_rate": 4.73103030065922e-05,
"loss": 1.648,
"step": 49500
},
{
"epoch": 0.98,
"learning_rate": 4.728313435009314e-05,
"loss": 1.6985,
"step": 50000
},
{
"epoch": 0.99,
"learning_rate": 4.725596569359407e-05,
"loss": 1.6694,
"step": 50500
},
{
"epoch": 1.0,
"learning_rate": 4.7228797037095004e-05,
"loss": 1.709,
"step": 51000
},
{
"epoch": 1.01,
"learning_rate": 4.720162838059593e-05,
"loss": 1.5724,
"step": 51500
},
{
"epoch": 1.02,
"learning_rate": 4.7174459724096864e-05,
"loss": 1.5297,
"step": 52000
},
{
"epoch": 1.03,
"learning_rate": 4.7147291067597793e-05,
"loss": 1.5361,
"step": 52500
},
{
"epoch": 1.04,
"learning_rate": 4.712012241109873e-05,
"loss": 1.5141,
"step": 53000
},
{
"epoch": 1.05,
"learning_rate": 4.709295375459965e-05,
"loss": 1.5395,
"step": 53500
},
{
"epoch": 1.06,
"learning_rate": 4.706578509810059e-05,
"loss": 1.5198,
"step": 54000
},
{
"epoch": 1.07,
"learning_rate": 4.703861644160152e-05,
"loss": 1.5742,
"step": 54500
},
{
"epoch": 1.08,
"learning_rate": 4.701144778510245e-05,
"loss": 1.5446,
"step": 55000
},
{
"epoch": 1.09,
"learning_rate": 4.698427912860338e-05,
"loss": 1.5587,
"step": 55500
},
{
"epoch": 1.1,
"learning_rate": 4.695711047210431e-05,
"loss": 1.5487,
"step": 56000
},
{
"epoch": 1.11,
"learning_rate": 4.6929941815605245e-05,
"loss": 1.5434,
"step": 56500
},
{
"epoch": 1.12,
"learning_rate": 4.6902773159106175e-05,
"loss": 1.5424,
"step": 57000
},
{
"epoch": 1.12,
"learning_rate": 4.6875604502607104e-05,
"loss": 1.5343,
"step": 57500
},
{
"epoch": 1.13,
"learning_rate": 4.6848435846108034e-05,
"loss": 1.5479,
"step": 58000
},
{
"epoch": 1.14,
"learning_rate": 4.682126718960897e-05,
"loss": 1.5453,
"step": 58500
},
{
"epoch": 1.15,
"learning_rate": 4.67940985331099e-05,
"loss": 1.5283,
"step": 59000
},
{
"epoch": 1.16,
"learning_rate": 4.676692987661083e-05,
"loss": 1.5689,
"step": 59500
},
{
"epoch": 1.17,
"learning_rate": 4.673976122011176e-05,
"loss": 1.5636,
"step": 60000
},
{
"epoch": 1.18,
"learning_rate": 4.6712592563612696e-05,
"loss": 1.5483,
"step": 60500
},
{
"epoch": 1.19,
"learning_rate": 4.6685423907113626e-05,
"loss": 1.5425,
"step": 61000
},
{
"epoch": 1.2,
"learning_rate": 4.6658255250614556e-05,
"loss": 1.5662,
"step": 61500
},
{
"epoch": 1.21,
"learning_rate": 4.6631086594115485e-05,
"loss": 1.576,
"step": 62000
},
{
"epoch": 1.22,
"learning_rate": 4.660391793761642e-05,
"loss": 1.5613,
"step": 62500
},
{
"epoch": 1.23,
"learning_rate": 4.657674928111735e-05,
"loss": 1.5404,
"step": 63000
},
{
"epoch": 1.24,
"learning_rate": 4.654958062461828e-05,
"loss": 1.5614,
"step": 63500
},
{
"epoch": 1.25,
"learning_rate": 4.652241196811921e-05,
"loss": 1.551,
"step": 64000
},
{
"epoch": 1.26,
"learning_rate": 4.649524331162015e-05,
"loss": 1.5603,
"step": 64500
},
{
"epoch": 1.27,
"learning_rate": 4.646807465512108e-05,
"loss": 1.5794,
"step": 65000
},
{
"epoch": 1.28,
"learning_rate": 4.644090599862201e-05,
"loss": 1.5551,
"step": 65500
},
{
"epoch": 1.29,
"learning_rate": 4.641373734212294e-05,
"loss": 1.5965,
"step": 66000
},
{
"epoch": 1.3,
"learning_rate": 4.638656868562387e-05,
"loss": 1.5662,
"step": 66500
},
{
"epoch": 1.31,
"learning_rate": 4.63594000291248e-05,
"loss": 1.6079,
"step": 67000
},
{
"epoch": 1.32,
"learning_rate": 4.633223137262573e-05,
"loss": 1.5807,
"step": 67500
},
{
"epoch": 1.33,
"learning_rate": 4.630506271612666e-05,
"loss": 1.5592,
"step": 68000
},
{
"epoch": 1.34,
"learning_rate": 4.62778940596276e-05,
"loss": 1.5741,
"step": 68500
},
{
"epoch": 1.35,
"learning_rate": 4.625072540312852e-05,
"loss": 1.5827,
"step": 69000
},
{
"epoch": 1.36,
"learning_rate": 4.622355674662946e-05,
"loss": 1.5656,
"step": 69500
},
{
"epoch": 1.37,
"learning_rate": 4.619638809013039e-05,
"loss": 1.5764,
"step": 70000
},
{
"epoch": 1.38,
"learning_rate": 4.6169219433631325e-05,
"loss": 1.556,
"step": 70500
},
{
"epoch": 1.39,
"learning_rate": 4.614205077713225e-05,
"loss": 1.582,
"step": 71000
},
{
"epoch": 1.4,
"learning_rate": 4.6114882120633184e-05,
"loss": 1.5789,
"step": 71500
},
{
"epoch": 1.41,
"learning_rate": 4.6087713464134114e-05,
"loss": 1.5719,
"step": 72000
},
{
"epoch": 1.42,
"learning_rate": 4.606054480763505e-05,
"loss": 1.581,
"step": 72500
},
{
"epoch": 1.43,
"learning_rate": 4.603337615113597e-05,
"loss": 1.5885,
"step": 73000
},
{
"epoch": 1.44,
"learning_rate": 4.600620749463691e-05,
"loss": 1.5516,
"step": 73500
},
{
"epoch": 1.45,
"learning_rate": 4.597903883813784e-05,
"loss": 1.59,
"step": 74000
},
{
"epoch": 1.46,
"learning_rate": 4.5951870181638776e-05,
"loss": 1.5968,
"step": 74500
},
{
"epoch": 1.47,
"learning_rate": 4.59247015251397e-05,
"loss": 1.5865,
"step": 75000
},
{
"epoch": 1.48,
"learning_rate": 4.5897532868640636e-05,
"loss": 1.5682,
"step": 75500
},
{
"epoch": 1.49,
"learning_rate": 4.5870364212141565e-05,
"loss": 1.5711,
"step": 76000
},
{
"epoch": 1.5,
"learning_rate": 4.58431955556425e-05,
"loss": 1.5595,
"step": 76500
},
{
"epoch": 1.51,
"learning_rate": 4.5816026899143425e-05,
"loss": 1.5898,
"step": 77000
},
{
"epoch": 1.52,
"learning_rate": 4.578885824264436e-05,
"loss": 1.6049,
"step": 77500
},
{
"epoch": 1.53,
"learning_rate": 4.576168958614529e-05,
"loss": 1.613,
"step": 78000
},
{
"epoch": 1.54,
"learning_rate": 4.573452092964622e-05,
"loss": 1.5761,
"step": 78500
},
{
"epoch": 1.55,
"learning_rate": 4.570735227314715e-05,
"loss": 1.5901,
"step": 79000
},
{
"epoch": 1.56,
"learning_rate": 4.568018361664809e-05,
"loss": 1.6102,
"step": 79500
},
{
"epoch": 1.56,
"learning_rate": 4.565301496014902e-05,
"loss": 1.5958,
"step": 80000
},
{
"epoch": 1.57,
"learning_rate": 4.5625846303649946e-05,
"loss": 1.5903,
"step": 80500
},
{
"epoch": 1.58,
"learning_rate": 4.5598677647150876e-05,
"loss": 1.5798,
"step": 81000
},
{
"epoch": 1.59,
"learning_rate": 4.557150899065181e-05,
"loss": 1.5832,
"step": 81500
},
{
"epoch": 1.6,
"learning_rate": 4.554434033415274e-05,
"loss": 1.6358,
"step": 82000
},
{
"epoch": 1.61,
"learning_rate": 4.551717167765367e-05,
"loss": 1.5969,
"step": 82500
},
{
"epoch": 1.62,
"learning_rate": 4.54900030211546e-05,
"loss": 1.6018,
"step": 83000
},
{
"epoch": 1.63,
"learning_rate": 4.546283436465554e-05,
"loss": 1.5919,
"step": 83500
},
{
"epoch": 1.64,
"learning_rate": 4.543566570815647e-05,
"loss": 1.5925,
"step": 84000
},
{
"epoch": 1.65,
"learning_rate": 4.54084970516574e-05,
"loss": 1.6047,
"step": 84500
},
{
"epoch": 1.66,
"learning_rate": 4.538132839515833e-05,
"loss": 1.6178,
"step": 85000
},
{
"epoch": 1.67,
"learning_rate": 4.5354159738659264e-05,
"loss": 1.5892,
"step": 85500
},
{
"epoch": 1.68,
"learning_rate": 4.5326991082160194e-05,
"loss": 1.6217,
"step": 86000
},
{
"epoch": 1.69,
"learning_rate": 4.5299822425661123e-05,
"loss": 1.5807,
"step": 86500
}
],
"max_steps": 920178,
"num_train_epochs": 18,
"total_flos": 1930009185792000.0,
"trial_name": null,
"trial_params": null
}