bert-base-swedish-cased-kubhist2 / trainer_state.json
Felix
Upload final model
854ae2a
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 2.3943855786431816,
"eval_steps": 500,
"global_step": 87000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"learning_rate": 4.977065272235219e-05,
"loss": 3.2765,
"step": 500
},
{
"epoch": 0.03,
"learning_rate": 4.954130544470437e-05,
"loss": 2.8884,
"step": 1000
},
{
"epoch": 0.04,
"learning_rate": 4.931195816705656e-05,
"loss": 2.7582,
"step": 1500
},
{
"epoch": 0.06,
"learning_rate": 4.908261088940875e-05,
"loss": 2.6689,
"step": 2000
},
{
"epoch": 0.07,
"learning_rate": 4.885326361176093e-05,
"loss": 2.6055,
"step": 2500
},
{
"epoch": 0.08,
"learning_rate": 4.862391633411312e-05,
"loss": 2.5549,
"step": 3000
},
{
"epoch": 0.1,
"learning_rate": 4.83945690564653e-05,
"loss": 2.5199,
"step": 3500
},
{
"epoch": 0.11,
"learning_rate": 4.816522177881749e-05,
"loss": 2.4834,
"step": 4000
},
{
"epoch": 0.12,
"learning_rate": 4.7935874501169674e-05,
"loss": 2.4538,
"step": 4500
},
{
"epoch": 0.14,
"learning_rate": 4.770652722352186e-05,
"loss": 2.4273,
"step": 5000
},
{
"epoch": 0.15,
"learning_rate": 4.7477179945874044e-05,
"loss": 2.4007,
"step": 5500
},
{
"epoch": 0.17,
"learning_rate": 4.7247832668226236e-05,
"loss": 2.3784,
"step": 6000
},
{
"epoch": 0.18,
"learning_rate": 4.7018485390578414e-05,
"loss": 2.3597,
"step": 6500
},
{
"epoch": 0.19,
"learning_rate": 4.67891381129306e-05,
"loss": 2.3433,
"step": 7000
},
{
"epoch": 0.21,
"learning_rate": 4.6559790835282785e-05,
"loss": 2.3291,
"step": 7500
},
{
"epoch": 0.22,
"learning_rate": 4.633044355763497e-05,
"loss": 2.3127,
"step": 8000
},
{
"epoch": 0.23,
"learning_rate": 4.610109627998716e-05,
"loss": 2.3013,
"step": 8500
},
{
"epoch": 0.25,
"learning_rate": 4.587174900233934e-05,
"loss": 2.2863,
"step": 9000
},
{
"epoch": 0.26,
"learning_rate": 4.564240172469153e-05,
"loss": 2.2772,
"step": 9500
},
{
"epoch": 0.28,
"learning_rate": 4.5413054447043716e-05,
"loss": 2.2645,
"step": 10000
},
{
"epoch": 0.29,
"learning_rate": 4.51837071693959e-05,
"loss": 2.2523,
"step": 10500
},
{
"epoch": 0.3,
"learning_rate": 4.4954359891748087e-05,
"loss": 2.2424,
"step": 11000
},
{
"epoch": 0.32,
"learning_rate": 4.472501261410027e-05,
"loss": 2.2344,
"step": 11500
},
{
"epoch": 0.33,
"learning_rate": 4.4495665336452457e-05,
"loss": 2.2204,
"step": 12000
},
{
"epoch": 0.34,
"learning_rate": 4.426631805880465e-05,
"loss": 2.2142,
"step": 12500
},
{
"epoch": 0.36,
"learning_rate": 4.4036970781156827e-05,
"loss": 2.2074,
"step": 13000
},
{
"epoch": 0.37,
"learning_rate": 4.380762350350902e-05,
"loss": 2.1972,
"step": 13500
},
{
"epoch": 0.39,
"learning_rate": 4.3578276225861203e-05,
"loss": 2.1857,
"step": 14000
},
{
"epoch": 0.4,
"learning_rate": 4.334892894821339e-05,
"loss": 2.1819,
"step": 14500
},
{
"epoch": 0.41,
"learning_rate": 4.3119581670565573e-05,
"loss": 2.1736,
"step": 15000
},
{
"epoch": 0.43,
"learning_rate": 4.289023439291776e-05,
"loss": 2.165,
"step": 15500
},
{
"epoch": 0.44,
"learning_rate": 4.2660887115269944e-05,
"loss": 2.1599,
"step": 16000
},
{
"epoch": 0.45,
"learning_rate": 4.243153983762213e-05,
"loss": 2.1566,
"step": 16500
},
{
"epoch": 0.47,
"learning_rate": 4.2202192559974314e-05,
"loss": 2.1496,
"step": 17000
},
{
"epoch": 0.48,
"learning_rate": 4.19728452823265e-05,
"loss": 2.1421,
"step": 17500
},
{
"epoch": 0.5,
"learning_rate": 4.174349800467869e-05,
"loss": 2.138,
"step": 18000
},
{
"epoch": 0.51,
"learning_rate": 4.151415072703087e-05,
"loss": 2.1319,
"step": 18500
},
{
"epoch": 0.52,
"learning_rate": 4.128480344938306e-05,
"loss": 2.1262,
"step": 19000
},
{
"epoch": 0.54,
"learning_rate": 4.1055456171735245e-05,
"loss": 2.1194,
"step": 19500
},
{
"epoch": 0.55,
"learning_rate": 4.082610889408743e-05,
"loss": 2.1162,
"step": 20000
},
{
"epoch": 0.56,
"learning_rate": 4.0596761616439616e-05,
"loss": 2.1084,
"step": 20500
},
{
"epoch": 0.58,
"learning_rate": 4.03674143387918e-05,
"loss": 2.1045,
"step": 21000
},
{
"epoch": 0.59,
"learning_rate": 4.0138067061143986e-05,
"loss": 2.0962,
"step": 21500
},
{
"epoch": 0.61,
"learning_rate": 3.990871978349617e-05,
"loss": 2.0967,
"step": 22000
},
{
"epoch": 0.62,
"learning_rate": 3.9679372505848356e-05,
"loss": 2.0875,
"step": 22500
},
{
"epoch": 0.63,
"learning_rate": 3.945002522820054e-05,
"loss": 2.0823,
"step": 23000
},
{
"epoch": 0.65,
"learning_rate": 3.922067795055273e-05,
"loss": 2.0831,
"step": 23500
},
{
"epoch": 0.66,
"learning_rate": 3.899133067290491e-05,
"loss": 2.0766,
"step": 24000
},
{
"epoch": 0.67,
"learning_rate": 3.87619833952571e-05,
"loss": 2.0725,
"step": 24500
},
{
"epoch": 0.69,
"learning_rate": 3.853263611760928e-05,
"loss": 2.0637,
"step": 25000
},
{
"epoch": 0.7,
"learning_rate": 3.830328883996147e-05,
"loss": 2.064,
"step": 25500
},
{
"epoch": 0.72,
"learning_rate": 3.807394156231366e-05,
"loss": 2.0582,
"step": 26000
},
{
"epoch": 0.73,
"learning_rate": 3.784459428466584e-05,
"loss": 2.0575,
"step": 26500
},
{
"epoch": 0.74,
"learning_rate": 3.761524700701803e-05,
"loss": 2.0531,
"step": 27000
},
{
"epoch": 0.76,
"learning_rate": 3.738589972937022e-05,
"loss": 2.0473,
"step": 27500
},
{
"epoch": 0.77,
"learning_rate": 3.71565524517224e-05,
"loss": 2.0474,
"step": 28000
},
{
"epoch": 0.78,
"learning_rate": 3.692720517407459e-05,
"loss": 2.0426,
"step": 28500
},
{
"epoch": 0.8,
"learning_rate": 3.6697857896426775e-05,
"loss": 2.0421,
"step": 29000
},
{
"epoch": 0.81,
"learning_rate": 3.646851061877895e-05,
"loss": 2.0366,
"step": 29500
},
{
"epoch": 0.83,
"learning_rate": 3.6239163341131145e-05,
"loss": 2.0312,
"step": 30000
},
{
"epoch": 0.84,
"learning_rate": 3.600981606348332e-05,
"loss": 2.0291,
"step": 30500
},
{
"epoch": 0.85,
"learning_rate": 3.5780468785835515e-05,
"loss": 2.0279,
"step": 31000
},
{
"epoch": 0.87,
"learning_rate": 3.55511215081877e-05,
"loss": 2.028,
"step": 31500
},
{
"epoch": 0.88,
"learning_rate": 3.5321774230539885e-05,
"loss": 2.023,
"step": 32000
},
{
"epoch": 0.89,
"learning_rate": 3.509242695289207e-05,
"loss": 2.0178,
"step": 32500
},
{
"epoch": 0.91,
"learning_rate": 3.486307967524426e-05,
"loss": 2.015,
"step": 33000
},
{
"epoch": 0.92,
"learning_rate": 3.463373239759644e-05,
"loss": 2.0145,
"step": 33500
},
{
"epoch": 0.94,
"learning_rate": 3.440438511994863e-05,
"loss": 2.0159,
"step": 34000
},
{
"epoch": 0.95,
"learning_rate": 3.417503784230081e-05,
"loss": 2.0089,
"step": 34500
},
{
"epoch": 0.96,
"learning_rate": 3.3945690564653e-05,
"loss": 2.004,
"step": 35000
},
{
"epoch": 0.98,
"learning_rate": 3.371634328700519e-05,
"loss": 2.0045,
"step": 35500
},
{
"epoch": 0.99,
"learning_rate": 3.348699600935737e-05,
"loss": 2.0026,
"step": 36000
},
{
"epoch": 1.0,
"learning_rate": 3.325764873170956e-05,
"loss": 1.9939,
"step": 36500
},
{
"epoch": 1.02,
"learning_rate": 3.302830145406174e-05,
"loss": 1.9928,
"step": 37000
},
{
"epoch": 1.03,
"learning_rate": 3.279895417641393e-05,
"loss": 1.9924,
"step": 37500
},
{
"epoch": 1.05,
"learning_rate": 3.256960689876611e-05,
"loss": 1.9886,
"step": 38000
},
{
"epoch": 1.06,
"learning_rate": 3.23402596211183e-05,
"loss": 1.9853,
"step": 38500
},
{
"epoch": 1.07,
"learning_rate": 3.211091234347048e-05,
"loss": 1.9814,
"step": 39000
},
{
"epoch": 1.09,
"learning_rate": 3.1881565065822674e-05,
"loss": 1.9826,
"step": 39500
},
{
"epoch": 1.1,
"learning_rate": 3.165221778817485e-05,
"loss": 1.98,
"step": 40000
},
{
"epoch": 1.11,
"learning_rate": 3.1422870510527044e-05,
"loss": 1.9785,
"step": 40500
},
{
"epoch": 1.13,
"learning_rate": 3.119352323287923e-05,
"loss": 1.9761,
"step": 41000
},
{
"epoch": 1.14,
"learning_rate": 3.0964175955231414e-05,
"loss": 1.9724,
"step": 41500
},
{
"epoch": 1.16,
"learning_rate": 3.07348286775836e-05,
"loss": 1.9746,
"step": 42000
},
{
"epoch": 1.17,
"learning_rate": 3.0505481399935787e-05,
"loss": 1.972,
"step": 42500
},
{
"epoch": 1.18,
"learning_rate": 3.027613412228797e-05,
"loss": 1.9707,
"step": 43000
},
{
"epoch": 1.2,
"learning_rate": 3.0046786844640157e-05,
"loss": 1.9685,
"step": 43500
},
{
"epoch": 1.21,
"learning_rate": 2.981743956699234e-05,
"loss": 1.9605,
"step": 44000
},
{
"epoch": 1.22,
"learning_rate": 2.9588092289344527e-05,
"loss": 1.9596,
"step": 44500
},
{
"epoch": 1.24,
"learning_rate": 2.9358745011696716e-05,
"loss": 1.9608,
"step": 45000
},
{
"epoch": 1.25,
"learning_rate": 2.9129397734048897e-05,
"loss": 1.9598,
"step": 45500
},
{
"epoch": 1.27,
"learning_rate": 2.8900050456401086e-05,
"loss": 1.9564,
"step": 46000
},
{
"epoch": 1.28,
"learning_rate": 2.867070317875327e-05,
"loss": 1.9557,
"step": 46500
},
{
"epoch": 1.29,
"learning_rate": 2.8441355901105456e-05,
"loss": 1.9526,
"step": 47000
},
{
"epoch": 1.31,
"learning_rate": 2.821200862345764e-05,
"loss": 1.9472,
"step": 47500
},
{
"epoch": 1.32,
"learning_rate": 2.7982661345809826e-05,
"loss": 1.9466,
"step": 48000
},
{
"epoch": 1.33,
"learning_rate": 2.775331406816201e-05,
"loss": 1.9479,
"step": 48500
},
{
"epoch": 1.35,
"learning_rate": 2.75239667905142e-05,
"loss": 1.9448,
"step": 49000
},
{
"epoch": 1.36,
"learning_rate": 2.729461951286638e-05,
"loss": 1.9451,
"step": 49500
},
{
"epoch": 1.38,
"learning_rate": 2.706527223521857e-05,
"loss": 1.9458,
"step": 50000
},
{
"epoch": 1.39,
"learning_rate": 2.6835924957570758e-05,
"loss": 1.9373,
"step": 50500
},
{
"epoch": 1.4,
"learning_rate": 2.660657767992294e-05,
"loss": 1.9364,
"step": 51000
},
{
"epoch": 1.42,
"learning_rate": 2.6377230402275128e-05,
"loss": 1.9345,
"step": 51500
},
{
"epoch": 1.43,
"learning_rate": 2.614788312462731e-05,
"loss": 1.9366,
"step": 52000
},
{
"epoch": 1.44,
"learning_rate": 2.5918535846979498e-05,
"loss": 1.9348,
"step": 52500
},
{
"epoch": 1.46,
"learning_rate": 2.5689188569331686e-05,
"loss": 1.9336,
"step": 53000
},
{
"epoch": 1.47,
"learning_rate": 2.5459841291683868e-05,
"loss": 1.932,
"step": 53500
},
{
"epoch": 1.49,
"learning_rate": 2.5230494014036056e-05,
"loss": 1.9333,
"step": 54000
},
{
"epoch": 1.5,
"learning_rate": 2.500114673638824e-05,
"loss": 1.9303,
"step": 54500
},
{
"epoch": 1.51,
"learning_rate": 2.4771799458740426e-05,
"loss": 1.9276,
"step": 55000
},
{
"epoch": 1.53,
"learning_rate": 2.454245218109261e-05,
"loss": 1.9268,
"step": 55500
},
{
"epoch": 1.54,
"learning_rate": 2.4313104903444796e-05,
"loss": 1.9244,
"step": 56000
},
{
"epoch": 1.55,
"learning_rate": 2.408375762579698e-05,
"loss": 1.9256,
"step": 56500
},
{
"epoch": 1.57,
"learning_rate": 2.385441034814917e-05,
"loss": 1.9191,
"step": 57000
},
{
"epoch": 1.58,
"learning_rate": 2.3625063070501355e-05,
"loss": 1.9186,
"step": 57500
},
{
"epoch": 1.6,
"learning_rate": 2.339571579285354e-05,
"loss": 1.9188,
"step": 58000
},
{
"epoch": 1.61,
"learning_rate": 2.3166368515205725e-05,
"loss": 1.9179,
"step": 58500
},
{
"epoch": 1.62,
"learning_rate": 2.2937021237557913e-05,
"loss": 1.9176,
"step": 59000
},
{
"epoch": 1.64,
"learning_rate": 2.27076739599101e-05,
"loss": 1.9154,
"step": 59500
},
{
"epoch": 1.65,
"learning_rate": 2.2478326682262283e-05,
"loss": 1.9119,
"step": 60000
},
{
"epoch": 1.67,
"learning_rate": 2.224897940461447e-05,
"loss": 1.9138,
"step": 60500
},
{
"epoch": 1.68,
"learning_rate": 2.2019632126966653e-05,
"loss": 1.9129,
"step": 61000
},
{
"epoch": 1.69,
"learning_rate": 2.179028484931884e-05,
"loss": 1.9114,
"step": 61500
},
{
"epoch": 1.71,
"learning_rate": 2.1560937571671023e-05,
"loss": 1.909,
"step": 62000
},
{
"epoch": 1.72,
"learning_rate": 2.133159029402321e-05,
"loss": 1.9054,
"step": 62500
},
{
"epoch": 1.73,
"learning_rate": 2.1102243016375397e-05,
"loss": 1.905,
"step": 63000
},
{
"epoch": 1.75,
"learning_rate": 2.0872895738727582e-05,
"loss": 1.9037,
"step": 63500
},
{
"epoch": 1.76,
"learning_rate": 2.0643548461079767e-05,
"loss": 1.9061,
"step": 64000
},
{
"epoch": 1.78,
"learning_rate": 2.0414201183431952e-05,
"loss": 1.902,
"step": 64500
},
{
"epoch": 1.79,
"learning_rate": 2.018485390578414e-05,
"loss": 1.8995,
"step": 65000
},
{
"epoch": 1.8,
"learning_rate": 1.9955506628136325e-05,
"loss": 1.8987,
"step": 65500
},
{
"epoch": 1.82,
"learning_rate": 1.972615935048851e-05,
"loss": 1.8988,
"step": 66000
},
{
"epoch": 1.83,
"learning_rate": 1.9496812072840695e-05,
"loss": 1.9013,
"step": 66500
},
{
"epoch": 1.84,
"learning_rate": 1.9267464795192884e-05,
"loss": 1.8992,
"step": 67000
},
{
"epoch": 1.86,
"learning_rate": 1.903811751754507e-05,
"loss": 1.8974,
"step": 67500
},
{
"epoch": 1.87,
"learning_rate": 1.8808770239897254e-05,
"loss": 1.8943,
"step": 68000
},
{
"epoch": 1.89,
"learning_rate": 1.857942296224944e-05,
"loss": 1.8954,
"step": 68500
},
{
"epoch": 1.9,
"learning_rate": 1.8350075684601624e-05,
"loss": 1.8942,
"step": 69000
},
{
"epoch": 1.91,
"learning_rate": 1.812072840695381e-05,
"loss": 1.8904,
"step": 69500
},
{
"epoch": 1.93,
"learning_rate": 1.7891381129305994e-05,
"loss": 1.8915,
"step": 70000
},
{
"epoch": 1.94,
"learning_rate": 1.7662033851658182e-05,
"loss": 1.8871,
"step": 70500
},
{
"epoch": 1.95,
"learning_rate": 1.7432686574010367e-05,
"loss": 1.8872,
"step": 71000
},
{
"epoch": 1.97,
"learning_rate": 1.7203339296362552e-05,
"loss": 1.8885,
"step": 71500
},
{
"epoch": 1.98,
"learning_rate": 1.6973992018714737e-05,
"loss": 1.8835,
"step": 72000
},
{
"epoch": 2.0,
"learning_rate": 1.6744644741066926e-05,
"loss": 1.8854,
"step": 72500
},
{
"epoch": 2.01,
"learning_rate": 1.651529746341911e-05,
"loss": 1.8856,
"step": 73000
},
{
"epoch": 2.02,
"learning_rate": 1.6285950185771296e-05,
"loss": 1.8836,
"step": 73500
},
{
"epoch": 2.04,
"learning_rate": 1.605660290812348e-05,
"loss": 1.8834,
"step": 74000
},
{
"epoch": 2.05,
"learning_rate": 1.582725563047567e-05,
"loss": 1.8814,
"step": 74500
},
{
"epoch": 2.06,
"learning_rate": 1.5597908352827854e-05,
"loss": 1.8795,
"step": 75000
},
{
"epoch": 2.08,
"learning_rate": 1.536856107518004e-05,
"loss": 1.8823,
"step": 75500
},
{
"epoch": 2.09,
"learning_rate": 1.5139213797532223e-05,
"loss": 1.8776,
"step": 76000
},
{
"epoch": 2.11,
"learning_rate": 1.4909866519884411e-05,
"loss": 1.8741,
"step": 76500
},
{
"epoch": 2.12,
"learning_rate": 1.4680519242236596e-05,
"loss": 1.8779,
"step": 77000
},
{
"epoch": 2.13,
"learning_rate": 1.4451171964588781e-05,
"loss": 1.8771,
"step": 77500
},
{
"epoch": 2.15,
"learning_rate": 1.4221824686940966e-05,
"loss": 1.8791,
"step": 78000
},
{
"epoch": 2.16,
"learning_rate": 1.3992477409293153e-05,
"loss": 1.873,
"step": 78500
},
{
"epoch": 2.17,
"learning_rate": 1.3763130131645338e-05,
"loss": 1.8713,
"step": 79000
},
{
"epoch": 2.19,
"learning_rate": 1.3533782853997523e-05,
"loss": 1.8717,
"step": 79500
},
{
"epoch": 2.2,
"learning_rate": 1.3304435576349708e-05,
"loss": 1.8699,
"step": 80000
},
{
"epoch": 2.22,
"learning_rate": 1.3075088298701896e-05,
"loss": 1.8729,
"step": 80500
},
{
"epoch": 2.23,
"learning_rate": 1.2845741021054082e-05,
"loss": 1.8708,
"step": 81000
},
{
"epoch": 2.24,
"learning_rate": 1.2616393743406265e-05,
"loss": 1.869,
"step": 81500
},
{
"epoch": 2.26,
"learning_rate": 1.2387046465758452e-05,
"loss": 1.8693,
"step": 82000
},
{
"epoch": 2.27,
"learning_rate": 1.2157699188110638e-05,
"loss": 1.8697,
"step": 82500
},
{
"epoch": 2.28,
"learning_rate": 1.1928351910462823e-05,
"loss": 1.8694,
"step": 83000
},
{
"epoch": 2.3,
"learning_rate": 1.169900463281501e-05,
"loss": 1.8692,
"step": 83500
},
{
"epoch": 2.31,
"learning_rate": 1.1469657355167195e-05,
"loss": 1.871,
"step": 84000
},
{
"epoch": 2.33,
"learning_rate": 1.1240310077519382e-05,
"loss": 1.8667,
"step": 84500
},
{
"epoch": 2.34,
"learning_rate": 1.1010962799871565e-05,
"loss": 1.8603,
"step": 85000
},
{
"epoch": 2.35,
"learning_rate": 1.0781615522223752e-05,
"loss": 1.8638,
"step": 85500
},
{
"epoch": 2.37,
"learning_rate": 1.0552268244575937e-05,
"loss": 1.8632,
"step": 86000
},
{
"epoch": 2.38,
"learning_rate": 1.0322920966928124e-05,
"loss": 1.8652,
"step": 86500
},
{
"epoch": 2.39,
"learning_rate": 1.0093573689280309e-05,
"loss": 1.8638,
"step": 87000
},
{
"epoch": 2.39,
"step": 87000,
"total_flos": 5.8634472383906e+18,
"train_loss": 0.0,
"train_runtime": 3.9321,
"train_samples_per_second": 7096699.346,
"train_steps_per_second": 13860.374
}
],
"logging_steps": 500,
"max_steps": 54501,
"num_train_epochs": 3,
"save_steps": 500,
"total_flos": 5.8634472383906e+18,
"trial_name": null,
"trial_params": null
}