inisw08-DistilBERT-mlm-sgd / trainer_state.json
ugiugi's picture
End of training
cb40b4c
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 3.0,
"global_step": 58893,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.03,
"learning_rate": 4.9575501332925815e-05,
"loss": 5.1524,
"step": 500
},
{
"epoch": 0.05,
"learning_rate": 4.9151002665851634e-05,
"loss": 5.1624,
"step": 1000
},
{
"epoch": 0.08,
"learning_rate": 4.8726503998777446e-05,
"loss": 5.1078,
"step": 1500
},
{
"epoch": 0.1,
"learning_rate": 4.830200533170326e-05,
"loss": 4.9858,
"step": 2000
},
{
"epoch": 0.13,
"learning_rate": 4.787750666462907e-05,
"loss": 4.991,
"step": 2500
},
{
"epoch": 0.15,
"learning_rate": 4.745300799755489e-05,
"loss": 5.0114,
"step": 3000
},
{
"epoch": 0.18,
"learning_rate": 4.70285093304807e-05,
"loss": 4.9202,
"step": 3500
},
{
"epoch": 0.2,
"learning_rate": 4.6604010663406515e-05,
"loss": 4.8947,
"step": 4000
},
{
"epoch": 0.23,
"learning_rate": 4.6179511996332334e-05,
"loss": 4.91,
"step": 4500
},
{
"epoch": 0.25,
"learning_rate": 4.575501332925815e-05,
"loss": 4.8889,
"step": 5000
},
{
"epoch": 0.28,
"learning_rate": 4.5330514662183965e-05,
"loss": 4.819,
"step": 5500
},
{
"epoch": 0.31,
"learning_rate": 4.490601599510978e-05,
"loss": 4.812,
"step": 6000
},
{
"epoch": 0.33,
"learning_rate": 4.44815173280356e-05,
"loss": 4.837,
"step": 6500
},
{
"epoch": 0.36,
"learning_rate": 4.405701866096141e-05,
"loss": 4.7846,
"step": 7000
},
{
"epoch": 0.38,
"learning_rate": 4.363251999388722e-05,
"loss": 4.8526,
"step": 7500
},
{
"epoch": 0.41,
"learning_rate": 4.3208021326813034e-05,
"loss": 4.7526,
"step": 8000
},
{
"epoch": 0.43,
"learning_rate": 4.278352265973885e-05,
"loss": 4.7668,
"step": 8500
},
{
"epoch": 0.46,
"learning_rate": 4.2359023992664665e-05,
"loss": 4.8055,
"step": 9000
},
{
"epoch": 0.48,
"learning_rate": 4.193452532559048e-05,
"loss": 4.7696,
"step": 9500
},
{
"epoch": 0.51,
"learning_rate": 4.151002665851629e-05,
"loss": 4.7797,
"step": 10000
},
{
"epoch": 0.53,
"learning_rate": 4.108552799144211e-05,
"loss": 4.7302,
"step": 10500
},
{
"epoch": 0.56,
"learning_rate": 4.066102932436792e-05,
"loss": 4.7055,
"step": 11000
},
{
"epoch": 0.59,
"learning_rate": 4.0236530657293734e-05,
"loss": 4.762,
"step": 11500
},
{
"epoch": 0.61,
"learning_rate": 3.981203199021955e-05,
"loss": 4.6706,
"step": 12000
},
{
"epoch": 0.64,
"learning_rate": 3.9387533323145365e-05,
"loss": 4.7708,
"step": 12500
},
{
"epoch": 0.66,
"learning_rate": 3.8963034656071184e-05,
"loss": 4.6481,
"step": 13000
},
{
"epoch": 0.69,
"learning_rate": 3.8538535988997e-05,
"loss": 4.7314,
"step": 13500
},
{
"epoch": 0.71,
"learning_rate": 3.8114037321922816e-05,
"loss": 4.6778,
"step": 14000
},
{
"epoch": 0.74,
"learning_rate": 3.768953865484863e-05,
"loss": 4.7046,
"step": 14500
},
{
"epoch": 0.76,
"learning_rate": 3.726503998777444e-05,
"loss": 4.6601,
"step": 15000
},
{
"epoch": 0.79,
"learning_rate": 3.684054132070025e-05,
"loss": 4.6533,
"step": 15500
},
{
"epoch": 0.82,
"learning_rate": 3.641604265362607e-05,
"loss": 4.7401,
"step": 16000
},
{
"epoch": 0.84,
"learning_rate": 3.5991543986551884e-05,
"loss": 4.7042,
"step": 16500
},
{
"epoch": 0.87,
"learning_rate": 3.55670453194777e-05,
"loss": 4.6861,
"step": 17000
},
{
"epoch": 0.89,
"learning_rate": 3.5142546652403516e-05,
"loss": 4.671,
"step": 17500
},
{
"epoch": 0.92,
"learning_rate": 3.471804798532933e-05,
"loss": 4.6824,
"step": 18000
},
{
"epoch": 0.94,
"learning_rate": 3.429354931825514e-05,
"loss": 4.6775,
"step": 18500
},
{
"epoch": 0.97,
"learning_rate": 3.386905065118095e-05,
"loss": 4.6828,
"step": 19000
},
{
"epoch": 0.99,
"learning_rate": 3.344455198410677e-05,
"loss": 4.6593,
"step": 19500
},
{
"epoch": 1.02,
"learning_rate": 3.3020053317032584e-05,
"loss": 4.5973,
"step": 20000
},
{
"epoch": 1.04,
"learning_rate": 3.25955546499584e-05,
"loss": 4.6338,
"step": 20500
},
{
"epoch": 1.07,
"learning_rate": 3.2171055982884216e-05,
"loss": 4.6474,
"step": 21000
},
{
"epoch": 1.1,
"learning_rate": 3.174655731581003e-05,
"loss": 4.6599,
"step": 21500
},
{
"epoch": 1.12,
"learning_rate": 3.132205864873585e-05,
"loss": 4.6634,
"step": 22000
},
{
"epoch": 1.15,
"learning_rate": 3.089755998166166e-05,
"loss": 4.6621,
"step": 22500
},
{
"epoch": 1.17,
"learning_rate": 3.0473061314587476e-05,
"loss": 4.6593,
"step": 23000
},
{
"epoch": 1.2,
"learning_rate": 3.004856264751329e-05,
"loss": 4.6295,
"step": 23500
},
{
"epoch": 1.22,
"learning_rate": 2.9624063980439104e-05,
"loss": 4.5604,
"step": 24000
},
{
"epoch": 1.25,
"learning_rate": 2.919956531336492e-05,
"loss": 4.5929,
"step": 24500
},
{
"epoch": 1.27,
"learning_rate": 2.8775066646290732e-05,
"loss": 4.5775,
"step": 25000
},
{
"epoch": 1.3,
"learning_rate": 2.8350567979216547e-05,
"loss": 4.5838,
"step": 25500
},
{
"epoch": 1.32,
"learning_rate": 2.792606931214236e-05,
"loss": 4.5949,
"step": 26000
},
{
"epoch": 1.35,
"learning_rate": 2.7501570645068176e-05,
"loss": 4.5702,
"step": 26500
},
{
"epoch": 1.38,
"learning_rate": 2.707707197799399e-05,
"loss": 4.6323,
"step": 27000
},
{
"epoch": 1.4,
"learning_rate": 2.6652573310919804e-05,
"loss": 4.69,
"step": 27500
},
{
"epoch": 1.43,
"learning_rate": 2.622807464384562e-05,
"loss": 4.6223,
"step": 28000
},
{
"epoch": 1.45,
"learning_rate": 2.5803575976771432e-05,
"loss": 4.6606,
"step": 28500
},
{
"epoch": 1.48,
"learning_rate": 2.5379077309697247e-05,
"loss": 4.6411,
"step": 29000
},
{
"epoch": 1.5,
"learning_rate": 2.4954578642623063e-05,
"loss": 4.5408,
"step": 29500
},
{
"epoch": 1.53,
"learning_rate": 2.453007997554888e-05,
"loss": 4.5832,
"step": 30000
},
{
"epoch": 1.55,
"learning_rate": 2.410558130847469e-05,
"loss": 4.6415,
"step": 30500
},
{
"epoch": 1.58,
"learning_rate": 2.3681082641400507e-05,
"loss": 4.5771,
"step": 31000
},
{
"epoch": 1.6,
"learning_rate": 2.325658397432632e-05,
"loss": 4.6112,
"step": 31500
},
{
"epoch": 1.63,
"learning_rate": 2.2832085307252135e-05,
"loss": 4.6145,
"step": 32000
},
{
"epoch": 1.66,
"learning_rate": 2.240758664017795e-05,
"loss": 4.5716,
"step": 32500
},
{
"epoch": 1.68,
"learning_rate": 2.1983087973103767e-05,
"loss": 4.6044,
"step": 33000
},
{
"epoch": 1.71,
"learning_rate": 2.1558589306029582e-05,
"loss": 4.6183,
"step": 33500
},
{
"epoch": 1.73,
"learning_rate": 2.1134090638955395e-05,
"loss": 4.5866,
"step": 34000
},
{
"epoch": 1.76,
"learning_rate": 2.070959197188121e-05,
"loss": 4.5349,
"step": 34500
},
{
"epoch": 1.78,
"learning_rate": 2.0285093304807023e-05,
"loss": 4.6438,
"step": 35000
},
{
"epoch": 1.81,
"learning_rate": 1.986059463773284e-05,
"loss": 4.6408,
"step": 35500
},
{
"epoch": 1.83,
"learning_rate": 1.943609597065865e-05,
"loss": 4.604,
"step": 36000
},
{
"epoch": 1.86,
"learning_rate": 1.9011597303584467e-05,
"loss": 4.5235,
"step": 36500
},
{
"epoch": 1.88,
"learning_rate": 1.8587098636510282e-05,
"loss": 4.5983,
"step": 37000
},
{
"epoch": 1.91,
"learning_rate": 1.8162599969436098e-05,
"loss": 4.5575,
"step": 37500
},
{
"epoch": 1.94,
"learning_rate": 1.773810130236191e-05,
"loss": 4.6175,
"step": 38000
},
{
"epoch": 1.96,
"learning_rate": 1.7313602635287726e-05,
"loss": 4.6319,
"step": 38500
},
{
"epoch": 1.99,
"learning_rate": 1.6889103968213542e-05,
"loss": 4.5626,
"step": 39000
},
{
"epoch": 2.01,
"learning_rate": 1.6464605301139354e-05,
"loss": 4.5394,
"step": 39500
},
{
"epoch": 2.04,
"learning_rate": 1.604010663406517e-05,
"loss": 4.6086,
"step": 40000
},
{
"epoch": 2.06,
"learning_rate": 1.5615607966990982e-05,
"loss": 4.5586,
"step": 40500
},
{
"epoch": 2.09,
"learning_rate": 1.51911092999168e-05,
"loss": 4.5997,
"step": 41000
},
{
"epoch": 2.11,
"learning_rate": 1.4766610632842614e-05,
"loss": 4.5854,
"step": 41500
},
{
"epoch": 2.14,
"learning_rate": 1.434211196576843e-05,
"loss": 4.6278,
"step": 42000
},
{
"epoch": 2.16,
"learning_rate": 1.3917613298694244e-05,
"loss": 4.6243,
"step": 42500
},
{
"epoch": 2.19,
"learning_rate": 1.3493114631620058e-05,
"loss": 4.617,
"step": 43000
},
{
"epoch": 2.22,
"learning_rate": 1.3068615964545872e-05,
"loss": 4.5467,
"step": 43500
},
{
"epoch": 2.24,
"learning_rate": 1.2644117297471686e-05,
"loss": 4.6096,
"step": 44000
},
{
"epoch": 2.27,
"learning_rate": 1.2219618630397501e-05,
"loss": 4.5703,
"step": 44500
},
{
"epoch": 2.29,
"learning_rate": 1.1795119963323315e-05,
"loss": 4.5774,
"step": 45000
},
{
"epoch": 2.32,
"learning_rate": 1.137062129624913e-05,
"loss": 4.5957,
"step": 45500
},
{
"epoch": 2.34,
"learning_rate": 1.0946122629174944e-05,
"loss": 4.5556,
"step": 46000
},
{
"epoch": 2.37,
"learning_rate": 1.052162396210076e-05,
"loss": 4.5366,
"step": 46500
},
{
"epoch": 2.39,
"learning_rate": 1.0097125295026575e-05,
"loss": 4.6,
"step": 47000
},
{
"epoch": 2.42,
"learning_rate": 9.672626627952389e-06,
"loss": 4.6358,
"step": 47500
},
{
"epoch": 2.45,
"learning_rate": 9.248127960878203e-06,
"loss": 4.5942,
"step": 48000
},
{
"epoch": 2.47,
"learning_rate": 8.823629293804019e-06,
"loss": 4.6347,
"step": 48500
},
{
"epoch": 2.5,
"learning_rate": 8.399130626729833e-06,
"loss": 4.5826,
"step": 49000
},
{
"epoch": 2.52,
"learning_rate": 7.974631959655647e-06,
"loss": 4.5753,
"step": 49500
},
{
"epoch": 2.55,
"learning_rate": 7.550133292581461e-06,
"loss": 4.5201,
"step": 50000
},
{
"epoch": 2.57,
"learning_rate": 7.125634625507276e-06,
"loss": 4.5383,
"step": 50500
},
{
"epoch": 2.6,
"learning_rate": 6.701135958433092e-06,
"loss": 4.5819,
"step": 51000
},
{
"epoch": 2.62,
"learning_rate": 6.276637291358906e-06,
"loss": 4.6261,
"step": 51500
},
{
"epoch": 2.65,
"learning_rate": 5.85213862428472e-06,
"loss": 4.5773,
"step": 52000
},
{
"epoch": 2.67,
"learning_rate": 5.427639957210535e-06,
"loss": 4.5516,
"step": 52500
},
{
"epoch": 2.7,
"learning_rate": 5.003141290136349e-06,
"loss": 4.6005,
"step": 53000
},
{
"epoch": 2.73,
"learning_rate": 4.578642623062164e-06,
"loss": 4.5648,
"step": 53500
},
{
"epoch": 2.75,
"learning_rate": 4.1541439559879784e-06,
"loss": 4.5681,
"step": 54000
},
{
"epoch": 2.78,
"learning_rate": 3.7296452889137933e-06,
"loss": 4.6076,
"step": 54500
},
{
"epoch": 2.8,
"learning_rate": 3.3051466218396074e-06,
"loss": 4.5851,
"step": 55000
},
{
"epoch": 2.83,
"learning_rate": 2.8806479547654223e-06,
"loss": 4.5714,
"step": 55500
},
{
"epoch": 2.85,
"learning_rate": 2.4561492876912367e-06,
"loss": 4.5404,
"step": 56000
},
{
"epoch": 2.88,
"learning_rate": 2.031650620617051e-06,
"loss": 4.5341,
"step": 56500
},
{
"epoch": 2.9,
"learning_rate": 1.607151953542866e-06,
"loss": 4.6453,
"step": 57000
},
{
"epoch": 2.93,
"learning_rate": 1.1826532864686806e-06,
"loss": 4.6012,
"step": 57500
},
{
"epoch": 2.95,
"learning_rate": 7.581546193944951e-07,
"loss": 4.5811,
"step": 58000
},
{
"epoch": 2.98,
"learning_rate": 3.336559523203097e-07,
"loss": 4.5623,
"step": 58500
},
{
"epoch": 3.0,
"step": 58893,
"total_flos": 6558889102345260.0,
"train_loss": 4.6634133611154445,
"train_runtime": 6718.6626,
"train_samples_per_second": 70.123,
"train_steps_per_second": 8.766
}
],
"max_steps": 58893,
"num_train_epochs": 3,
"total_flos": 6558889102345260.0,
"trial_name": null,
"trial_params": null
}