finalMeat / last-checkpoint /trainer_state.json
dq158's picture
Training in progress, epoch 3, checkpoint
8492909
raw
history blame contribute delete
No virus
51.8 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 3.0,
"eval_steps": 500,
"global_step": 204657,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"learning_rate": 8e-05,
"loss": 2.273,
"step": 500
},
{
"epoch": 0.01,
"learning_rate": 7.999957460493864e-05,
"loss": 2.2497,
"step": 1000
},
{
"epoch": 0.02,
"learning_rate": 7.999829842880257e-05,
"loss": 2.3728,
"step": 1500
},
{
"epoch": 0.03,
"learning_rate": 7.999617149873574e-05,
"loss": 2.3576,
"step": 2000
},
{
"epoch": 0.04,
"learning_rate": 7.999319385997746e-05,
"loss": 2.3374,
"step": 2500
},
{
"epoch": 0.04,
"learning_rate": 7.998936557586135e-05,
"loss": 2.3538,
"step": 3000
},
{
"epoch": 0.05,
"learning_rate": 7.998468672781407e-05,
"loss": 2.366,
"step": 3500
},
{
"epoch": 0.06,
"learning_rate": 7.997915741535355e-05,
"loss": 2.3321,
"step": 4000
},
{
"epoch": 0.07,
"learning_rate": 7.997277775608694e-05,
"loss": 2.2838,
"step": 4500
},
{
"epoch": 0.07,
"learning_rate": 7.996554788570796e-05,
"loss": 2.2679,
"step": 5000
},
{
"epoch": 0.08,
"learning_rate": 7.995746795799422e-05,
"loss": 2.4041,
"step": 5500
},
{
"epoch": 0.09,
"learning_rate": 7.994853814480376e-05,
"loss": 2.3415,
"step": 6000
},
{
"epoch": 0.1,
"learning_rate": 7.99387586360715e-05,
"loss": 2.3589,
"step": 6500
},
{
"epoch": 0.1,
"learning_rate": 7.992812963980518e-05,
"loss": 2.3203,
"step": 7000
},
{
"epoch": 0.11,
"learning_rate": 7.991665138208094e-05,
"loss": 2.3517,
"step": 7500
},
{
"epoch": 0.12,
"learning_rate": 7.990432410703848e-05,
"loss": 2.4286,
"step": 8000
},
{
"epoch": 0.12,
"learning_rate": 7.989114807687589e-05,
"loss": 2.3227,
"step": 8500
},
{
"epoch": 0.13,
"learning_rate": 7.987712357184408e-05,
"loss": 2.3945,
"step": 9000
},
{
"epoch": 0.14,
"learning_rate": 7.98622508902408e-05,
"loss": 2.3775,
"step": 9500
},
{
"epoch": 0.15,
"learning_rate": 7.984653034840432e-05,
"loss": 2.3522,
"step": 10000
},
{
"epoch": 0.15,
"learning_rate": 7.982996228070671e-05,
"loss": 2.334,
"step": 10500
},
{
"epoch": 0.16,
"learning_rate": 7.981254703954664e-05,
"loss": 2.3742,
"step": 11000
},
{
"epoch": 0.17,
"learning_rate": 7.979428499534201e-05,
"loss": 2.2517,
"step": 11500
},
{
"epoch": 0.18,
"learning_rate": 7.977517653652199e-05,
"loss": 2.3994,
"step": 12000
},
{
"epoch": 0.18,
"learning_rate": 7.975522206951876e-05,
"loss": 2.3606,
"step": 12500
},
{
"epoch": 0.19,
"learning_rate": 7.973442201875895e-05,
"loss": 2.3634,
"step": 13000
},
{
"epoch": 0.2,
"learning_rate": 7.971277682665446e-05,
"loss": 2.3061,
"step": 13500
},
{
"epoch": 0.21,
"learning_rate": 7.969028695359319e-05,
"loss": 2.4985,
"step": 14000
},
{
"epoch": 0.21,
"learning_rate": 7.966695287792921e-05,
"loss": 2.3908,
"step": 14500
},
{
"epoch": 0.22,
"learning_rate": 7.96427750959725e-05,
"loss": 2.3605,
"step": 15000
},
{
"epoch": 0.23,
"learning_rate": 7.961775412197857e-05,
"loss": 2.347,
"step": 15500
},
{
"epoch": 0.23,
"learning_rate": 7.959189048813735e-05,
"loss": 2.3233,
"step": 16000
},
{
"epoch": 0.24,
"learning_rate": 7.95651847445619e-05,
"loss": 2.3415,
"step": 16500
},
{
"epoch": 0.25,
"learning_rate": 7.953763745927682e-05,
"loss": 2.4679,
"step": 17000
},
{
"epoch": 0.26,
"learning_rate": 7.950924921820606e-05,
"loss": 2.443,
"step": 17500
},
{
"epoch": 0.26,
"learning_rate": 7.948002062516052e-05,
"loss": 2.5141,
"step": 18000
},
{
"epoch": 0.27,
"learning_rate": 7.944995230182513e-05,
"loss": 2.4339,
"step": 18500
},
{
"epoch": 0.28,
"learning_rate": 7.941904488774571e-05,
"loss": 2.4308,
"step": 19000
},
{
"epoch": 0.29,
"learning_rate": 7.938729904031533e-05,
"loss": 2.4312,
"step": 19500
},
{
"epoch": 0.29,
"learning_rate": 7.93547154347603e-05,
"loss": 2.4756,
"step": 20000
},
{
"epoch": 0.3,
"learning_rate": 7.932129476412592e-05,
"loss": 2.4426,
"step": 20500
},
{
"epoch": 0.31,
"learning_rate": 7.928703773926155e-05,
"loss": 2.4779,
"step": 21000
},
{
"epoch": 0.32,
"learning_rate": 7.925194508880567e-05,
"loss": 2.4671,
"step": 21500
},
{
"epoch": 0.32,
"learning_rate": 7.921601755917029e-05,
"loss": 2.4473,
"step": 22000
},
{
"epoch": 0.33,
"learning_rate": 7.917925591452508e-05,
"loss": 2.3929,
"step": 22500
},
{
"epoch": 0.34,
"learning_rate": 7.914166093678117e-05,
"loss": 2.4158,
"step": 23000
},
{
"epoch": 0.34,
"learning_rate": 7.910323342557442e-05,
"loss": 2.4607,
"step": 23500
},
{
"epoch": 0.35,
"learning_rate": 7.906397419824855e-05,
"loss": 2.4866,
"step": 24000
},
{
"epoch": 0.36,
"learning_rate": 7.902388408983759e-05,
"loss": 2.3708,
"step": 24500
},
{
"epoch": 0.37,
"learning_rate": 7.898296395304824e-05,
"loss": 2.4718,
"step": 25000
},
{
"epoch": 0.37,
"learning_rate": 7.894121465824175e-05,
"loss": 2.4436,
"step": 25500
},
{
"epoch": 0.38,
"learning_rate": 7.889863709341528e-05,
"loss": 2.498,
"step": 26000
},
{
"epoch": 0.39,
"learning_rate": 7.885523216418312e-05,
"loss": 2.4418,
"step": 26500
},
{
"epoch": 0.4,
"learning_rate": 7.881100079375742e-05,
"loss": 2.3653,
"step": 27000
},
{
"epoch": 0.4,
"learning_rate": 7.876594392292848e-05,
"loss": 2.5256,
"step": 27500
},
{
"epoch": 0.41,
"learning_rate": 7.872006251004482e-05,
"loss": 2.549,
"step": 28000
},
{
"epoch": 0.42,
"learning_rate": 7.867335753099278e-05,
"loss": 2.4543,
"step": 28500
},
{
"epoch": 0.43,
"learning_rate": 7.86258299791757e-05,
"loss": 2.4647,
"step": 29000
},
{
"epoch": 0.43,
"learning_rate": 7.857748086549292e-05,
"loss": 2.5375,
"step": 29500
},
{
"epoch": 0.44,
"learning_rate": 7.852831121831812e-05,
"loss": 2.4895,
"step": 30000
},
{
"epoch": 0.45,
"learning_rate": 7.847832208347754e-05,
"loss": 2.4156,
"step": 30500
},
{
"epoch": 0.45,
"learning_rate": 7.842751452422775e-05,
"loss": 2.5006,
"step": 31000
},
{
"epoch": 0.46,
"learning_rate": 7.8375889621233e-05,
"loss": 2.459,
"step": 31500
},
{
"epoch": 0.47,
"learning_rate": 7.83234484725422e-05,
"loss": 2.469,
"step": 32000
},
{
"epoch": 0.48,
"learning_rate": 7.827019219356568e-05,
"loss": 2.4331,
"step": 32500
},
{
"epoch": 0.48,
"learning_rate": 7.821612191705128e-05,
"loss": 2.487,
"step": 33000
},
{
"epoch": 0.49,
"learning_rate": 7.816123879306048e-05,
"loss": 2.5139,
"step": 33500
},
{
"epoch": 0.5,
"learning_rate": 7.810554398894376e-05,
"loss": 2.5117,
"step": 34000
},
{
"epoch": 0.51,
"learning_rate": 7.804903868931584e-05,
"loss": 2.5537,
"step": 34500
},
{
"epoch": 0.51,
"learning_rate": 7.79917240960305e-05,
"loss": 2.4394,
"step": 35000
},
{
"epoch": 0.52,
"learning_rate": 7.7933601428155e-05,
"loss": 2.4285,
"step": 35500
},
{
"epoch": 0.53,
"learning_rate": 7.78746719219441e-05,
"loss": 2.5693,
"step": 36000
},
{
"epoch": 0.54,
"learning_rate": 7.781493683081388e-05,
"loss": 2.4932,
"step": 36500
},
{
"epoch": 0.54,
"learning_rate": 7.775439742531495e-05,
"loss": 2.4468,
"step": 37000
},
{
"epoch": 0.55,
"learning_rate": 7.769305499310553e-05,
"loss": 2.4614,
"step": 37500
},
{
"epoch": 0.56,
"learning_rate": 7.763091083892402e-05,
"loss": 2.6549,
"step": 38000
},
{
"epoch": 0.56,
"learning_rate": 7.756796628456121e-05,
"loss": 2.4871,
"step": 38500
},
{
"epoch": 0.57,
"learning_rate": 7.750422266883222e-05,
"loss": 2.413,
"step": 39000
},
{
"epoch": 0.58,
"learning_rate": 7.743968134754806e-05,
"loss": 2.3566,
"step": 39500
},
{
"epoch": 0.59,
"learning_rate": 7.737434369348664e-05,
"loss": 2.4936,
"step": 40000
},
{
"epoch": 0.59,
"learning_rate": 7.730821109636379e-05,
"loss": 2.453,
"step": 40500
},
{
"epoch": 0.6,
"learning_rate": 7.724128496280346e-05,
"loss": 2.4851,
"step": 41000
},
{
"epoch": 0.61,
"learning_rate": 7.717356671630802e-05,
"loss": 2.4564,
"step": 41500
},
{
"epoch": 0.62,
"learning_rate": 7.710505779722786e-05,
"loss": 2.5133,
"step": 42000
},
{
"epoch": 0.62,
"learning_rate": 7.703575966273073e-05,
"loss": 2.4546,
"step": 42500
},
{
"epoch": 0.63,
"learning_rate": 7.696567378677089e-05,
"loss": 2.5157,
"step": 43000
},
{
"epoch": 0.64,
"learning_rate": 7.689480166005756e-05,
"loss": 2.4248,
"step": 43500
},
{
"epoch": 0.64,
"learning_rate": 7.682314479002344e-05,
"loss": 2.4853,
"step": 44000
},
{
"epoch": 0.65,
"learning_rate": 7.67507047007924e-05,
"loss": 2.4615,
"step": 44500
},
{
"epoch": 0.66,
"learning_rate": 7.667748293314729e-05,
"loss": 2.5391,
"step": 45000
},
{
"epoch": 0.67,
"learning_rate": 7.6603481044497e-05,
"loss": 2.4464,
"step": 45500
},
{
"epoch": 0.67,
"learning_rate": 7.652870060884345e-05,
"loss": 2.4941,
"step": 46000
},
{
"epoch": 0.68,
"learning_rate": 7.645314321674803e-05,
"loss": 2.4708,
"step": 46500
},
{
"epoch": 0.69,
"learning_rate": 7.637681047529781e-05,
"loss": 2.5972,
"step": 47000
},
{
"epoch": 0.7,
"learning_rate": 7.629970400807136e-05,
"loss": 2.5369,
"step": 47500
},
{
"epoch": 0.7,
"learning_rate": 7.622182545510419e-05,
"loss": 2.4348,
"step": 48000
},
{
"epoch": 0.71,
"learning_rate": 7.61431764728539e-05,
"loss": 2.4546,
"step": 48500
},
{
"epoch": 0.72,
"learning_rate": 7.606375873416491e-05,
"loss": 2.4378,
"step": 49000
},
{
"epoch": 0.73,
"learning_rate": 7.598357392823292e-05,
"loss": 2.471,
"step": 49500
},
{
"epoch": 0.73,
"learning_rate": 7.590262376056896e-05,
"loss": 2.4677,
"step": 50000
},
{
"epoch": 0.74,
"learning_rate": 7.58209099529631e-05,
"loss": 2.4245,
"step": 50500
},
{
"epoch": 0.75,
"learning_rate": 7.573843424344783e-05,
"loss": 2.4734,
"step": 51000
},
{
"epoch": 0.75,
"learning_rate": 7.565519838626113e-05,
"loss": 2.4158,
"step": 51500
},
{
"epoch": 0.76,
"learning_rate": 7.557120415180916e-05,
"loss": 2.5098,
"step": 52000
},
{
"epoch": 0.77,
"learning_rate": 7.548645332662853e-05,
"loss": 2.5478,
"step": 52500
},
{
"epoch": 0.78,
"learning_rate": 7.540094771334835e-05,
"loss": 2.4502,
"step": 53000
},
{
"epoch": 0.78,
"learning_rate": 7.531468913065192e-05,
"loss": 2.4264,
"step": 53500
},
{
"epoch": 0.79,
"learning_rate": 7.522767941323798e-05,
"loss": 2.4903,
"step": 54000
},
{
"epoch": 0.8,
"learning_rate": 7.513992041178174e-05,
"loss": 2.4486,
"step": 54500
},
{
"epoch": 0.81,
"learning_rate": 7.505141399289549e-05,
"loss": 2.5171,
"step": 55000
},
{
"epoch": 0.81,
"learning_rate": 7.496216203908891e-05,
"loss": 2.5396,
"step": 55500
},
{
"epoch": 0.82,
"learning_rate": 7.487216644872901e-05,
"loss": 2.4514,
"step": 56000
},
{
"epoch": 0.83,
"learning_rate": 7.478142913599978e-05,
"loss": 2.4017,
"step": 56500
},
{
"epoch": 0.84,
"learning_rate": 7.468995203086146e-05,
"loss": 2.4591,
"step": 57000
},
{
"epoch": 0.84,
"learning_rate": 7.459773707900946e-05,
"loss": 2.5764,
"step": 57500
},
{
"epoch": 0.85,
"learning_rate": 7.450478624183306e-05,
"loss": 2.5013,
"step": 58000
},
{
"epoch": 0.86,
"learning_rate": 7.441110149637363e-05,
"loss": 2.51,
"step": 58500
},
{
"epoch": 0.86,
"learning_rate": 7.431668483528254e-05,
"loss": 2.3992,
"step": 59000
},
{
"epoch": 0.87,
"learning_rate": 7.422153826677887e-05,
"loss": 2.4671,
"step": 59500
},
{
"epoch": 0.88,
"learning_rate": 7.412566381460662e-05,
"loss": 2.5362,
"step": 60000
},
{
"epoch": 0.89,
"learning_rate": 7.402906351799175e-05,
"loss": 2.4981,
"step": 60500
},
{
"epoch": 0.89,
"learning_rate": 7.39317394315987e-05,
"loss": 2.4574,
"step": 61000
},
{
"epoch": 0.9,
"learning_rate": 7.383369362548674e-05,
"loss": 2.4777,
"step": 61500
},
{
"epoch": 0.91,
"learning_rate": 7.373492818506597e-05,
"loss": 2.4628,
"step": 62000
},
{
"epoch": 0.92,
"learning_rate": 7.363544521105292e-05,
"loss": 2.4506,
"step": 62500
},
{
"epoch": 0.92,
"learning_rate": 7.353524681942585e-05,
"loss": 2.5644,
"step": 63000
},
{
"epoch": 0.93,
"learning_rate": 7.343433514137987e-05,
"loss": 2.5131,
"step": 63500
},
{
"epoch": 0.94,
"learning_rate": 7.333271232328141e-05,
"loss": 2.5575,
"step": 64000
},
{
"epoch": 0.95,
"learning_rate": 7.32303805266227e-05,
"loss": 2.5901,
"step": 64500
},
{
"epoch": 0.95,
"learning_rate": 7.312734192797583e-05,
"loss": 2.5701,
"step": 65000
},
{
"epoch": 0.96,
"learning_rate": 7.302359871894635e-05,
"loss": 2.5201,
"step": 65500
},
{
"epoch": 0.97,
"learning_rate": 7.291915310612666e-05,
"loss": 2.5402,
"step": 66000
},
{
"epoch": 0.97,
"learning_rate": 7.281400731104918e-05,
"loss": 2.4797,
"step": 66500
},
{
"epoch": 0.98,
"learning_rate": 7.2708163570139e-05,
"loss": 2.4836,
"step": 67000
},
{
"epoch": 0.99,
"learning_rate": 7.260162413466636e-05,
"loss": 2.4707,
"step": 67500
},
{
"epoch": 1.0,
"learning_rate": 7.249439127069873e-05,
"loss": 2.4334,
"step": 68000
},
{
"epoch": 1.0,
"eval_bleu": 1.0,
"eval_brevity_penalty": 1.0,
"eval_length_ratio": 1.0,
"eval_loss": 2.243044853210449,
"eval_precisions": [
1.0,
1.0,
1.0,
1.0
],
"eval_reference_length": 7761920,
"eval_runtime": 17988.3779,
"eval_samples_per_second": 0.843,
"eval_steps_per_second": 0.421,
"eval_translation_length": 7761920,
"step": 68219
},
{
"epoch": 1.0,
"learning_rate": 7.238646725905267e-05,
"loss": 2.3238,
"step": 68500
},
{
"epoch": 1.01,
"learning_rate": 7.227785439524523e-05,
"loss": 2.1194,
"step": 69000
},
{
"epoch": 1.02,
"learning_rate": 7.216855498944522e-05,
"loss": 2.1879,
"step": 69500
},
{
"epoch": 1.03,
"learning_rate": 7.205857136642402e-05,
"loss": 2.0581,
"step": 70000
},
{
"epoch": 1.03,
"learning_rate": 7.194790586550611e-05,
"loss": 2.1256,
"step": 70500
},
{
"epoch": 1.04,
"learning_rate": 7.183656084051938e-05,
"loss": 2.1933,
"step": 71000
},
{
"epoch": 1.05,
"learning_rate": 7.172453865974503e-05,
"loss": 2.1403,
"step": 71500
},
{
"epoch": 1.06,
"learning_rate": 7.161184170586717e-05,
"loss": 2.1402,
"step": 72000
},
{
"epoch": 1.06,
"learning_rate": 7.149847237592218e-05,
"loss": 2.0527,
"step": 72500
},
{
"epoch": 1.07,
"learning_rate": 7.138443308124771e-05,
"loss": 2.169,
"step": 73000
},
{
"epoch": 1.08,
"learning_rate": 7.12697262474314e-05,
"loss": 2.1674,
"step": 73500
},
{
"epoch": 1.08,
"learning_rate": 7.115435431425928e-05,
"loss": 2.0606,
"step": 74000
},
{
"epoch": 1.09,
"learning_rate": 7.103831973566391e-05,
"loss": 2.1421,
"step": 74500
},
{
"epoch": 1.1,
"learning_rate": 7.092162497967207e-05,
"loss": 2.1357,
"step": 75000
},
{
"epoch": 1.11,
"learning_rate": 7.080427252835243e-05,
"loss": 2.1281,
"step": 75500
},
{
"epoch": 1.11,
"learning_rate": 7.068626487776266e-05,
"loss": 2.095,
"step": 76000
},
{
"epoch": 1.12,
"learning_rate": 7.056760453789635e-05,
"loss": 2.1362,
"step": 76500
},
{
"epoch": 1.13,
"learning_rate": 7.044829403262961e-05,
"loss": 2.1443,
"step": 77000
},
{
"epoch": 1.14,
"learning_rate": 7.032833589966745e-05,
"loss": 2.2343,
"step": 77500
},
{
"epoch": 1.14,
"learning_rate": 7.02077326904897e-05,
"loss": 2.0832,
"step": 78000
},
{
"epoch": 1.15,
"learning_rate": 7.008648697029687e-05,
"loss": 2.087,
"step": 78500
},
{
"epoch": 1.16,
"learning_rate": 6.996460131795548e-05,
"loss": 2.146,
"step": 79000
},
{
"epoch": 1.17,
"learning_rate": 6.984207832594325e-05,
"loss": 2.1542,
"step": 79500
},
{
"epoch": 1.17,
"learning_rate": 6.971892060029398e-05,
"loss": 2.0636,
"step": 80000
},
{
"epoch": 1.18,
"learning_rate": 6.959513076054207e-05,
"loss": 2.1275,
"step": 80500
},
{
"epoch": 1.19,
"learning_rate": 6.947071143966686e-05,
"loss": 2.1221,
"step": 81000
},
{
"epoch": 1.19,
"learning_rate": 6.934566528403655e-05,
"loss": 2.097,
"step": 81500
},
{
"epoch": 1.2,
"learning_rate": 6.921999495335204e-05,
"loss": 2.1852,
"step": 82000
},
{
"epoch": 1.21,
"learning_rate": 6.909370312059018e-05,
"loss": 2.125,
"step": 82500
},
{
"epoch": 1.22,
"learning_rate": 6.896679247194712e-05,
"loss": 2.1584,
"step": 83000
},
{
"epoch": 1.22,
"learning_rate": 6.883926570678098e-05,
"loss": 2.203,
"step": 83500
},
{
"epoch": 1.23,
"learning_rate": 6.871112553755457e-05,
"loss": 2.1556,
"step": 84000
},
{
"epoch": 1.24,
"learning_rate": 6.858237468977765e-05,
"loss": 2.081,
"step": 84500
},
{
"epoch": 1.25,
"learning_rate": 6.845301590194898e-05,
"loss": 2.2032,
"step": 85000
},
{
"epoch": 1.25,
"learning_rate": 6.8323051925498e-05,
"loss": 2.169,
"step": 85500
},
{
"epoch": 1.26,
"learning_rate": 6.819248552472642e-05,
"loss": 2.1197,
"step": 86000
},
{
"epoch": 1.27,
"learning_rate": 6.806131947674935e-05,
"loss": 2.1572,
"step": 86500
},
{
"epoch": 1.28,
"learning_rate": 6.792955657143622e-05,
"loss": 2.1894,
"step": 87000
},
{
"epoch": 1.28,
"learning_rate": 6.779719961135148e-05,
"loss": 2.2259,
"step": 87500
},
{
"epoch": 1.29,
"learning_rate": 6.766425141169503e-05,
"loss": 2.0972,
"step": 88000
},
{
"epoch": 1.3,
"learning_rate": 6.753071480024223e-05,
"loss": 2.1572,
"step": 88500
},
{
"epoch": 1.3,
"learning_rate": 6.739659261728381e-05,
"loss": 2.1574,
"step": 89000
},
{
"epoch": 1.31,
"learning_rate": 6.726188771556552e-05,
"loss": 2.1689,
"step": 89500
},
{
"epoch": 1.32,
"learning_rate": 6.712660296022731e-05,
"loss": 2.1866,
"step": 90000
},
{
"epoch": 1.33,
"learning_rate": 6.699074122874255e-05,
"loss": 2.1624,
"step": 90500
},
{
"epoch": 1.33,
"learning_rate": 6.685430541085673e-05,
"loss": 2.147,
"step": 91000
},
{
"epoch": 1.34,
"learning_rate": 6.671729840852598e-05,
"loss": 2.2042,
"step": 91500
},
{
"epoch": 1.35,
"learning_rate": 6.657972313585541e-05,
"loss": 2.1646,
"step": 92000
},
{
"epoch": 1.36,
"learning_rate": 6.644158251903713e-05,
"loss": 2.151,
"step": 92500
},
{
"epoch": 1.36,
"learning_rate": 6.630287949628791e-05,
"loss": 2.1218,
"step": 93000
},
{
"epoch": 1.37,
"learning_rate": 6.616361701778681e-05,
"loss": 2.1922,
"step": 93500
},
{
"epoch": 1.38,
"learning_rate": 6.602379804561237e-05,
"loss": 2.1459,
"step": 94000
},
{
"epoch": 1.39,
"learning_rate": 6.588342555367958e-05,
"loss": 2.1345,
"step": 94500
},
{
"epoch": 1.39,
"learning_rate": 6.57425025276767e-05,
"loss": 2.2256,
"step": 95000
},
{
"epoch": 1.4,
"learning_rate": 6.560103196500169e-05,
"loss": 2.163,
"step": 95500
},
{
"epoch": 1.41,
"learning_rate": 6.545901687469848e-05,
"loss": 2.2915,
"step": 96000
},
{
"epoch": 1.41,
"learning_rate": 6.531646027739298e-05,
"loss": 2.1037,
"step": 96500
},
{
"epoch": 1.42,
"learning_rate": 6.517336520522879e-05,
"loss": 2.2256,
"step": 97000
},
{
"epoch": 1.43,
"learning_rate": 6.50297347018028e-05,
"loss": 2.1624,
"step": 97500
},
{
"epoch": 1.44,
"learning_rate": 6.48855718221003e-05,
"loss": 2.2168,
"step": 98000
},
{
"epoch": 1.44,
"learning_rate": 6.47408796324302e-05,
"loss": 2.2301,
"step": 98500
},
{
"epoch": 1.45,
"learning_rate": 6.45956612103596e-05,
"loss": 2.1579,
"step": 99000
},
{
"epoch": 1.46,
"learning_rate": 6.444991964464851e-05,
"loss": 2.1901,
"step": 99500
},
{
"epoch": 1.47,
"learning_rate": 6.430365803518404e-05,
"loss": 2.179,
"step": 100000
},
{
"epoch": 1.47,
"learning_rate": 6.415687949291448e-05,
"loss": 2.2105,
"step": 100500
},
{
"epoch": 1.48,
"learning_rate": 6.40095871397832e-05,
"loss": 2.1975,
"step": 101000
},
{
"epoch": 1.49,
"learning_rate": 6.386178410866217e-05,
"loss": 2.1788,
"step": 101500
},
{
"epoch": 1.5,
"learning_rate": 6.371347354328537e-05,
"loss": 2.2073,
"step": 102000
},
{
"epoch": 1.5,
"learning_rate": 6.356465859818192e-05,
"loss": 2.1706,
"step": 102500
},
{
"epoch": 1.51,
"learning_rate": 6.341534243860894e-05,
"loss": 2.1783,
"step": 103000
},
{
"epoch": 1.52,
"learning_rate": 6.326552824048426e-05,
"loss": 2.1604,
"step": 103500
},
{
"epoch": 1.52,
"learning_rate": 6.31152191903189e-05,
"loss": 2.1701,
"step": 104000
},
{
"epoch": 1.53,
"learning_rate": 6.296441848514925e-05,
"loss": 2.1284,
"step": 104500
},
{
"epoch": 1.54,
"learning_rate": 6.281312933246902e-05,
"loss": 2.158,
"step": 105000
},
{
"epoch": 1.55,
"learning_rate": 6.26613549501612e-05,
"loss": 2.1539,
"step": 105500
},
{
"epoch": 1.55,
"learning_rate": 6.250909856642937e-05,
"loss": 2.2176,
"step": 106000
},
{
"epoch": 1.56,
"learning_rate": 6.235636341972923e-05,
"loss": 2.227,
"step": 106500
},
{
"epoch": 1.57,
"learning_rate": 6.220315275869965e-05,
"loss": 2.1622,
"step": 107000
},
{
"epoch": 1.58,
"learning_rate": 6.204946984209355e-05,
"loss": 2.2304,
"step": 107500
},
{
"epoch": 1.58,
"learning_rate": 6.189531793870862e-05,
"loss": 2.1883,
"step": 108000
},
{
"epoch": 1.59,
"learning_rate": 6.174070032731776e-05,
"loss": 2.1939,
"step": 108500
},
{
"epoch": 1.6,
"learning_rate": 6.15856202965994e-05,
"loss": 2.225,
"step": 109000
},
{
"epoch": 1.61,
"learning_rate": 6.14300811450675e-05,
"loss": 2.1847,
"step": 109500
},
{
"epoch": 1.61,
"learning_rate": 6.12740861810014e-05,
"loss": 2.2027,
"step": 110000
},
{
"epoch": 1.62,
"learning_rate": 6.111763872237548e-05,
"loss": 2.1629,
"step": 110500
},
{
"epoch": 1.63,
"learning_rate": 6.096074209678854e-05,
"loss": 2.2492,
"step": 111000
},
{
"epoch": 1.63,
"learning_rate": 6.080339964139306e-05,
"loss": 2.1836,
"step": 111500
},
{
"epoch": 1.64,
"learning_rate": 6.0645614702824225e-05,
"loss": 2.2321,
"step": 112000
},
{
"epoch": 1.65,
"learning_rate": 6.0487390637128715e-05,
"loss": 2.1377,
"step": 112500
},
{
"epoch": 1.66,
"learning_rate": 6.032873080969333e-05,
"loss": 2.1757,
"step": 113000
},
{
"epoch": 1.66,
"learning_rate": 6.016963859517342e-05,
"loss": 2.2636,
"step": 113500
},
{
"epoch": 1.67,
"learning_rate": 6.00101173774211e-05,
"loss": 2.1175,
"step": 114000
},
{
"epoch": 1.68,
"learning_rate": 5.9850170549413294e-05,
"loss": 2.2032,
"step": 114500
},
{
"epoch": 1.69,
"learning_rate": 5.968980151317953e-05,
"loss": 2.1685,
"step": 115000
},
{
"epoch": 1.69,
"learning_rate": 5.9529013679729616e-05,
"loss": 2.1367,
"step": 115500
},
{
"epoch": 1.7,
"learning_rate": 5.936781046898106e-05,
"loss": 2.144,
"step": 116000
},
{
"epoch": 1.71,
"learning_rate": 5.920619530968634e-05,
"loss": 2.2216,
"step": 116500
},
{
"epoch": 1.72,
"learning_rate": 5.904417163936001e-05,
"loss": 2.1873,
"step": 117000
},
{
"epoch": 1.72,
"learning_rate": 5.88817429042055e-05,
"loss": 2.2138,
"step": 117500
},
{
"epoch": 1.73,
"learning_rate": 5.871891255904191e-05,
"loss": 2.1854,
"step": 118000
},
{
"epoch": 1.74,
"learning_rate": 5.855568406723049e-05,
"loss": 2.2012,
"step": 118500
},
{
"epoch": 1.74,
"learning_rate": 5.839206090060094e-05,
"loss": 2.2523,
"step": 119000
},
{
"epoch": 1.75,
"learning_rate": 5.82280465393776e-05,
"loss": 2.2806,
"step": 119500
},
{
"epoch": 1.76,
"learning_rate": 5.806364447210545e-05,
"loss": 2.2008,
"step": 120000
},
{
"epoch": 1.77,
"learning_rate": 5.789885819557585e-05,
"loss": 2.2525,
"step": 120500
},
{
"epoch": 1.77,
"learning_rate": 5.773369121475223e-05,
"loss": 2.2509,
"step": 121000
},
{
"epoch": 1.78,
"learning_rate": 5.756814704269547e-05,
"loss": 2.2109,
"step": 121500
},
{
"epoch": 1.79,
"learning_rate": 5.7402229200489246e-05,
"loss": 2.1961,
"step": 122000
},
{
"epoch": 1.8,
"learning_rate": 5.7235941217165076e-05,
"loss": 2.2703,
"step": 122500
},
{
"epoch": 1.8,
"learning_rate": 5.706928662962732e-05,
"loss": 2.2202,
"step": 123000
},
{
"epoch": 1.81,
"learning_rate": 5.690226898257789e-05,
"loss": 2.2061,
"step": 123500
},
{
"epoch": 1.82,
"learning_rate": 5.6734891828440906e-05,
"loss": 2.2972,
"step": 124000
},
{
"epoch": 1.83,
"learning_rate": 5.6567158727287094e-05,
"loss": 2.2,
"step": 124500
},
{
"epoch": 1.83,
"learning_rate": 5.639907324675809e-05,
"loss": 2.2358,
"step": 125000
},
{
"epoch": 1.84,
"learning_rate": 5.623063896199059e-05,
"loss": 2.2434,
"step": 125500
},
{
"epoch": 1.85,
"learning_rate": 5.606185945554023e-05,
"loss": 2.1901,
"step": 126000
},
{
"epoch": 1.85,
"learning_rate": 5.589273831730541e-05,
"loss": 2.1735,
"step": 126500
},
{
"epoch": 1.86,
"learning_rate": 5.572327914445101e-05,
"loss": 2.2449,
"step": 127000
},
{
"epoch": 1.87,
"learning_rate": 5.555348554133178e-05,
"loss": 2.1991,
"step": 127500
},
{
"epoch": 1.88,
"learning_rate": 5.538336111941571e-05,
"loss": 2.1521,
"step": 128000
},
{
"epoch": 1.88,
"learning_rate": 5.521290949720728e-05,
"loss": 2.2191,
"step": 128500
},
{
"epoch": 1.89,
"learning_rate": 5.504213430017038e-05,
"loss": 2.237,
"step": 129000
},
{
"epoch": 1.9,
"learning_rate": 5.4871039160651295e-05,
"loss": 2.2709,
"step": 129500
},
{
"epoch": 1.91,
"learning_rate": 5.4699627717801396e-05,
"loss": 2.2217,
"step": 130000
},
{
"epoch": 1.91,
"learning_rate": 5.452790361749973e-05,
"loss": 2.2418,
"step": 130500
},
{
"epoch": 1.92,
"learning_rate": 5.4355870512275515e-05,
"loss": 2.2513,
"step": 131000
},
{
"epoch": 1.93,
"learning_rate": 5.418353206123042e-05,
"loss": 2.2671,
"step": 131500
},
{
"epoch": 1.93,
"learning_rate": 5.401089192996074e-05,
"loss": 2.2706,
"step": 132000
},
{
"epoch": 1.94,
"learning_rate": 5.383795379047944e-05,
"loss": 2.23,
"step": 132500
},
{
"epoch": 1.95,
"learning_rate": 5.3664721321138046e-05,
"loss": 2.2251,
"step": 133000
},
{
"epoch": 1.96,
"learning_rate": 5.349119820654839e-05,
"loss": 2.2968,
"step": 133500
},
{
"epoch": 1.96,
"learning_rate": 5.331738813750429e-05,
"loss": 2.2243,
"step": 134000
},
{
"epoch": 1.97,
"learning_rate": 5.314329481090299e-05,
"loss": 2.277,
"step": 134500
},
{
"epoch": 1.98,
"learning_rate": 5.296892192966653e-05,
"loss": 2.2377,
"step": 135000
},
{
"epoch": 1.99,
"learning_rate": 5.2794273202663085e-05,
"loss": 2.2582,
"step": 135500
},
{
"epoch": 1.99,
"learning_rate": 5.261935234462791e-05,
"loss": 2.2732,
"step": 136000
},
{
"epoch": 2.0,
"eval_bleu": 1.0,
"eval_brevity_penalty": 1.0,
"eval_length_ratio": 1.0,
"eval_loss": 2.259059429168701,
"eval_precisions": [
1.0,
1.0,
1.0,
1.0
],
"eval_reference_length": 7761920,
"eval_runtime": 17938.9236,
"eval_samples_per_second": 0.845,
"eval_steps_per_second": 0.423,
"eval_translation_length": 7761920,
"step": 136438
},
{
"epoch": 2.0,
"learning_rate": 5.2444163076084484e-05,
"loss": 2.1805,
"step": 136500
},
{
"epoch": 2.01,
"learning_rate": 5.226870912326529e-05,
"loss": 1.8318,
"step": 137000
},
{
"epoch": 2.02,
"learning_rate": 5.209299421803256e-05,
"loss": 1.7868,
"step": 137500
},
{
"epoch": 2.02,
"learning_rate": 5.1917022097798964e-05,
"loss": 1.8616,
"step": 138000
},
{
"epoch": 2.03,
"learning_rate": 5.174079650544804e-05,
"loss": 1.862,
"step": 138500
},
{
"epoch": 2.04,
"learning_rate": 5.1564321189254605e-05,
"loss": 1.7721,
"step": 139000
},
{
"epoch": 2.04,
"learning_rate": 5.138759990280507e-05,
"loss": 1.8478,
"step": 139500
},
{
"epoch": 2.05,
"learning_rate": 5.1210636404917554e-05,
"loss": 1.935,
"step": 140000
},
{
"epoch": 2.06,
"learning_rate": 5.1033434459561965e-05,
"loss": 1.8431,
"step": 140500
},
{
"epoch": 2.07,
"learning_rate": 5.0855997835779926e-05,
"loss": 1.8654,
"step": 141000
},
{
"epoch": 2.07,
"learning_rate": 5.067833030760462e-05,
"loss": 1.7853,
"step": 141500
},
{
"epoch": 2.08,
"learning_rate": 5.0500435653980455e-05,
"loss": 1.8631,
"step": 142000
},
{
"epoch": 2.09,
"learning_rate": 5.032231765868284e-05,
"loss": 1.8427,
"step": 142500
},
{
"epoch": 2.1,
"learning_rate": 5.014398011023752e-05,
"loss": 1.8528,
"step": 143000
},
{
"epoch": 2.1,
"learning_rate": 4.9965426801840127e-05,
"loss": 1.8048,
"step": 143500
},
{
"epoch": 2.11,
"learning_rate": 4.9786661531275425e-05,
"loss": 1.8064,
"step": 144000
},
{
"epoch": 2.12,
"learning_rate": 4.96076881008366e-05,
"loss": 1.9335,
"step": 144500
},
{
"epoch": 2.13,
"learning_rate": 4.9428510317244295e-05,
"loss": 1.9127,
"step": 145000
},
{
"epoch": 2.13,
"learning_rate": 4.924913199156575e-05,
"loss": 1.8177,
"step": 145500
},
{
"epoch": 2.14,
"learning_rate": 4.906955693913363e-05,
"loss": 1.8531,
"step": 146000
},
{
"epoch": 2.15,
"learning_rate": 4.8889788979465e-05,
"loss": 1.8327,
"step": 146500
},
{
"epoch": 2.15,
"learning_rate": 4.870983193617993e-05,
"loss": 1.8751,
"step": 147000
},
{
"epoch": 2.16,
"learning_rate": 4.8529689636920306e-05,
"loss": 1.865,
"step": 147500
},
{
"epoch": 2.17,
"learning_rate": 4.834936591326834e-05,
"loss": 1.9178,
"step": 148000
},
{
"epoch": 2.18,
"learning_rate": 4.8168864600665134e-05,
"loss": 1.8717,
"step": 148500
},
{
"epoch": 2.18,
"learning_rate": 4.798818953832902e-05,
"loss": 1.8991,
"step": 149000
},
{
"epoch": 2.19,
"learning_rate": 4.780734456917396e-05,
"loss": 1.8914,
"step": 149500
},
{
"epoch": 2.2,
"learning_rate": 4.7626333539727793e-05,
"loss": 1.9122,
"step": 150000
},
{
"epoch": 2.21,
"learning_rate": 4.744516030005041e-05,
"loss": 1.8922,
"step": 150500
},
{
"epoch": 2.21,
"learning_rate": 4.72638287036519e-05,
"loss": 1.929,
"step": 151000
},
{
"epoch": 2.22,
"learning_rate": 4.7082342607410514e-05,
"loss": 1.8535,
"step": 151500
},
{
"epoch": 2.23,
"learning_rate": 4.6900705871490736e-05,
"loss": 1.8546,
"step": 152000
},
{
"epoch": 2.24,
"learning_rate": 4.671892235926106e-05,
"loss": 1.8847,
"step": 152500
},
{
"epoch": 2.24,
"learning_rate": 4.653699593721192e-05,
"loss": 1.8957,
"step": 153000
},
{
"epoch": 2.25,
"learning_rate": 4.635493047487339e-05,
"loss": 1.909,
"step": 153500
},
{
"epoch": 2.26,
"learning_rate": 4.61727298447329e-05,
"loss": 1.8491,
"step": 154000
},
{
"epoch": 2.26,
"learning_rate": 4.5990397922152834e-05,
"loss": 1.9239,
"step": 154500
},
{
"epoch": 2.27,
"learning_rate": 4.580793858528819e-05,
"loss": 1.893,
"step": 155000
},
{
"epoch": 2.28,
"learning_rate": 4.562535571500399e-05,
"loss": 1.8956,
"step": 155500
},
{
"epoch": 2.29,
"learning_rate": 4.5442653194792806e-05,
"loss": 1.9018,
"step": 156000
},
{
"epoch": 2.29,
"learning_rate": 4.525983491069215e-05,
"loss": 1.8923,
"step": 156500
},
{
"epoch": 2.3,
"learning_rate": 4.507690475120176e-05,
"loss": 1.8883,
"step": 157000
},
{
"epoch": 2.31,
"learning_rate": 4.489386660720095e-05,
"loss": 1.8547,
"step": 157500
},
{
"epoch": 2.32,
"learning_rate": 4.471072437186585e-05,
"loss": 1.8442,
"step": 158000
},
{
"epoch": 2.32,
"learning_rate": 4.452748194058659e-05,
"loss": 1.8919,
"step": 158500
},
{
"epoch": 2.33,
"learning_rate": 4.4344143210884436e-05,
"loss": 1.8636,
"step": 159000
},
{
"epoch": 2.34,
"learning_rate": 4.416071208232888e-05,
"loss": 1.8752,
"step": 159500
},
{
"epoch": 2.35,
"learning_rate": 4.397719245645475e-05,
"loss": 1.8668,
"step": 160000
},
{
"epoch": 2.35,
"learning_rate": 4.3793588236679173e-05,
"loss": 1.9294,
"step": 160500
},
{
"epoch": 2.36,
"learning_rate": 4.360990332821855e-05,
"loss": 1.8352,
"step": 161000
},
{
"epoch": 2.37,
"learning_rate": 4.342614163800553e-05,
"loss": 1.9387,
"step": 161500
},
{
"epoch": 2.37,
"learning_rate": 4.32423070746059e-05,
"loss": 1.8752,
"step": 162000
},
{
"epoch": 2.38,
"learning_rate": 4.3058403548135426e-05,
"loss": 1.9911,
"step": 162500
},
{
"epoch": 2.39,
"learning_rate": 4.287443497017667e-05,
"loss": 1.8686,
"step": 163000
},
{
"epoch": 2.4,
"learning_rate": 4.269040525369591e-05,
"loss": 1.8261,
"step": 163500
},
{
"epoch": 2.4,
"learning_rate": 4.250631831295975e-05,
"loss": 1.8804,
"step": 164000
},
{
"epoch": 2.41,
"learning_rate": 4.2322178063451956e-05,
"loss": 1.8813,
"step": 164500
},
{
"epoch": 2.42,
"learning_rate": 4.2137988421790176e-05,
"loss": 1.8637,
"step": 165000
},
{
"epoch": 2.43,
"learning_rate": 4.195375330564259e-05,
"loss": 1.8864,
"step": 165500
},
{
"epoch": 2.43,
"learning_rate": 4.176947663364464e-05,
"loss": 1.8915,
"step": 166000
},
{
"epoch": 2.44,
"learning_rate": 4.158516232531565e-05,
"loss": 1.8492,
"step": 166500
},
{
"epoch": 2.45,
"learning_rate": 4.140081430097541e-05,
"loss": 1.8539,
"step": 167000
},
{
"epoch": 2.46,
"learning_rate": 4.121643648166088e-05,
"loss": 1.8811,
"step": 167500
},
{
"epoch": 2.46,
"learning_rate": 4.1032032789042785e-05,
"loss": 1.8897,
"step": 168000
},
{
"epoch": 2.47,
"learning_rate": 4.084760714534209e-05,
"loss": 1.8895,
"step": 168500
},
{
"epoch": 2.48,
"learning_rate": 4.066316347324673e-05,
"loss": 1.9344,
"step": 169000
},
{
"epoch": 2.48,
"learning_rate": 4.047870569582805e-05,
"loss": 1.8943,
"step": 169500
},
{
"epoch": 2.49,
"learning_rate": 4.029423773645742e-05,
"loss": 1.8904,
"step": 170000
},
{
"epoch": 2.5,
"learning_rate": 4.0109763518722786e-05,
"loss": 1.8938,
"step": 170500
},
{
"epoch": 2.51,
"learning_rate": 3.992528696634522e-05,
"loss": 1.937,
"step": 171000
},
{
"epoch": 2.51,
"learning_rate": 3.9740812003095435e-05,
"loss": 1.9084,
"step": 171500
},
{
"epoch": 2.52,
"learning_rate": 3.9556342552710344e-05,
"loss": 1.8989,
"step": 172000
},
{
"epoch": 2.53,
"learning_rate": 3.937188253880958e-05,
"loss": 1.9068,
"step": 172500
},
{
"epoch": 2.54,
"learning_rate": 3.9187435884812134e-05,
"loss": 1.8497,
"step": 173000
},
{
"epoch": 2.54,
"learning_rate": 3.900300651385277e-05,
"loss": 1.8392,
"step": 173500
},
{
"epoch": 2.55,
"learning_rate": 3.8818598348698666e-05,
"loss": 1.9431,
"step": 174000
},
{
"epoch": 2.56,
"learning_rate": 3.863421531166596e-05,
"loss": 1.9361,
"step": 174500
},
{
"epoch": 2.57,
"learning_rate": 3.844986132453632e-05,
"loss": 1.8738,
"step": 175000
},
{
"epoch": 2.57,
"learning_rate": 3.826554030847353e-05,
"loss": 1.9457,
"step": 175500
},
{
"epoch": 2.58,
"learning_rate": 3.8081256183940086e-05,
"loss": 1.9293,
"step": 176000
},
{
"epoch": 2.59,
"learning_rate": 3.7897012870613815e-05,
"loss": 1.9107,
"step": 176500
},
{
"epoch": 2.59,
"learning_rate": 3.771281428730448e-05,
"loss": 1.8691,
"step": 177000
},
{
"epoch": 2.6,
"learning_rate": 3.75286643518705e-05,
"loss": 1.856,
"step": 177500
},
{
"epoch": 2.61,
"learning_rate": 3.7344566981135476e-05,
"loss": 1.9258,
"step": 178000
},
{
"epoch": 2.62,
"learning_rate": 3.716052609080507e-05,
"loss": 1.8476,
"step": 178500
},
{
"epoch": 2.62,
"learning_rate": 3.697654559538356e-05,
"loss": 1.8763,
"step": 179000
},
{
"epoch": 2.63,
"learning_rate": 3.6792629408090645e-05,
"loss": 1.903,
"step": 179500
},
{
"epoch": 2.64,
"learning_rate": 3.660878144077822e-05,
"loss": 1.8962,
"step": 180000
},
{
"epoch": 2.65,
"learning_rate": 3.642500560384714e-05,
"loss": 1.965,
"step": 180500
},
{
"epoch": 2.65,
"learning_rate": 3.62413058061641e-05,
"loss": 1.9176,
"step": 181000
},
{
"epoch": 2.66,
"learning_rate": 3.605768595497842e-05,
"loss": 1.9391,
"step": 181500
},
{
"epoch": 2.67,
"learning_rate": 3.5874149955839e-05,
"loss": 1.9248,
"step": 182000
},
{
"epoch": 2.68,
"learning_rate": 3.569070171251121e-05,
"loss": 1.9606,
"step": 182500
},
{
"epoch": 2.68,
"learning_rate": 3.5507345126893905e-05,
"loss": 1.9195,
"step": 183000
},
{
"epoch": 2.69,
"learning_rate": 3.5324084098936375e-05,
"loss": 1.8899,
"step": 183500
},
{
"epoch": 2.7,
"learning_rate": 3.514092252655542e-05,
"loss": 1.9041,
"step": 184000
},
{
"epoch": 2.7,
"learning_rate": 3.495786430555247e-05,
"loss": 1.8918,
"step": 184500
},
{
"epoch": 2.71,
"learning_rate": 3.477491332953069e-05,
"loss": 1.8615,
"step": 185000
},
{
"epoch": 2.72,
"learning_rate": 3.459207348981214e-05,
"loss": 1.9503,
"step": 185500
},
{
"epoch": 2.73,
"learning_rate": 3.4409348675355084e-05,
"loss": 1.8883,
"step": 186000
},
{
"epoch": 2.73,
"learning_rate": 3.422674277267118e-05,
"loss": 1.8797,
"step": 186500
},
{
"epoch": 2.74,
"learning_rate": 3.404425966574292e-05,
"loss": 1.9118,
"step": 187000
},
{
"epoch": 2.75,
"learning_rate": 3.386190323594091e-05,
"loss": 1.8573,
"step": 187500
},
{
"epoch": 2.76,
"learning_rate": 3.3679677361941365e-05,
"loss": 1.9455,
"step": 188000
},
{
"epoch": 2.76,
"learning_rate": 3.349758591964365e-05,
"loss": 1.9058,
"step": 188500
},
{
"epoch": 2.77,
"learning_rate": 3.3315632782087784e-05,
"loss": 1.9369,
"step": 189000
},
{
"epoch": 2.78,
"learning_rate": 3.3133821819372065e-05,
"loss": 1.9005,
"step": 189500
},
{
"epoch": 2.79,
"learning_rate": 3.295215689857077e-05,
"loss": 1.9293,
"step": 190000
},
{
"epoch": 2.79,
"learning_rate": 3.277064188365191e-05,
"loss": 1.9078,
"step": 190500
},
{
"epoch": 2.8,
"learning_rate": 3.258928063539502e-05,
"loss": 1.928,
"step": 191000
},
{
"epoch": 2.81,
"learning_rate": 3.240807701130909e-05,
"loss": 1.9753,
"step": 191500
},
{
"epoch": 2.81,
"learning_rate": 3.2227034865550444e-05,
"loss": 1.9069,
"step": 192000
},
{
"epoch": 2.82,
"learning_rate": 3.204615804884081e-05,
"loss": 1.9923,
"step": 192500
},
{
"epoch": 2.83,
"learning_rate": 3.1865450408385434e-05,
"loss": 1.8337,
"step": 193000
},
{
"epoch": 2.84,
"learning_rate": 3.1684915787791185e-05,
"loss": 1.9092,
"step": 193500
},
{
"epoch": 2.84,
"learning_rate": 3.150455802698488e-05,
"loss": 1.8451,
"step": 194000
},
{
"epoch": 2.85,
"learning_rate": 3.132438096213156e-05,
"loss": 1.8952,
"step": 194500
},
{
"epoch": 2.86,
"learning_rate": 3.114438842555287e-05,
"loss": 1.9238,
"step": 195000
},
{
"epoch": 2.87,
"learning_rate": 3.096458424564564e-05,
"loss": 1.8765,
"step": 195500
},
{
"epoch": 2.87,
"learning_rate": 3.07849722468004e-05,
"loss": 1.8946,
"step": 196000
},
{
"epoch": 2.88,
"learning_rate": 3.060555624931998e-05,
"loss": 1.8671,
"step": 196500
},
{
"epoch": 2.89,
"learning_rate": 3.042634006933835e-05,
"loss": 1.9126,
"step": 197000
},
{
"epoch": 2.9,
"learning_rate": 3.024732751873942e-05,
"loss": 1.9464,
"step": 197500
},
{
"epoch": 2.9,
"learning_rate": 3.006852240507591e-05,
"loss": 1.8723,
"step": 198000
},
{
"epoch": 2.91,
"learning_rate": 2.988992853148847e-05,
"loss": 1.892,
"step": 198500
},
{
"epoch": 2.92,
"learning_rate": 2.9711549696624666e-05,
"loss": 1.8947,
"step": 199000
},
{
"epoch": 2.92,
"learning_rate": 2.9533389694558266e-05,
"loss": 1.9721,
"step": 199500
},
{
"epoch": 2.93,
"learning_rate": 2.9355452314708538e-05,
"loss": 1.9074,
"step": 200000
},
{
"epoch": 2.94,
"learning_rate": 2.9177741341759592e-05,
"loss": 1.9363,
"step": 200500
},
{
"epoch": 2.95,
"learning_rate": 2.9000260555579957e-05,
"loss": 1.8815,
"step": 201000
},
{
"epoch": 2.95,
"learning_rate": 2.8823013731142127e-05,
"loss": 1.9579,
"step": 201500
},
{
"epoch": 2.96,
"learning_rate": 2.8646004638442282e-05,
"loss": 1.9756,
"step": 202000
},
{
"epoch": 2.97,
"learning_rate": 2.8469237042420128e-05,
"loss": 1.8637,
"step": 202500
},
{
"epoch": 2.98,
"learning_rate": 2.8292714702878754e-05,
"loss": 1.8707,
"step": 203000
},
{
"epoch": 2.98,
"learning_rate": 2.811644137440477e-05,
"loss": 1.9256,
"step": 203500
},
{
"epoch": 2.99,
"learning_rate": 2.7940420806288327e-05,
"loss": 1.9739,
"step": 204000
},
{
"epoch": 3.0,
"learning_rate": 2.776465674244345e-05,
"loss": 1.8675,
"step": 204500
},
{
"epoch": 3.0,
"eval_bleu": 1.0,
"eval_brevity_penalty": 1.0,
"eval_length_ratio": 1.0,
"eval_loss": 2.297696352005005,
"eval_precisions": [
1.0,
1.0,
1.0,
1.0
],
"eval_reference_length": 7761920,
"eval_runtime": 17697.7556,
"eval_samples_per_second": 0.857,
"eval_steps_per_second": 0.428,
"eval_translation_length": 7761920,
"step": 204657
}
],
"logging_steps": 500,
"max_steps": 341095,
"num_train_epochs": 5,
"save_steps": 5000,
"total_flos": 9.433737890333983e+17,
"trial_name": null,
"trial_params": null
}