|
{ |
|
"best_metric": 0.6502385015996446, |
|
"best_model_checkpoint": "1-wikispan-*unsup-ensemble-s64-bs128-lr6*-64-768-3e-6-8600/checkpoint-5400", |
|
"epoch": 1.0225921521997623, |
|
"global_step": 8600, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 3.75e-07, |
|
"loss": 172.9849, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 7.5e-07, |
|
"loss": 102.5869, |
|
"step": 200 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 1.125e-06, |
|
"loss": 77.1419, |
|
"step": 300 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 1.5e-06, |
|
"loss": 68.3599, |
|
"step": 400 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 1.875e-06, |
|
"loss": 62.2906, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 2.25e-06, |
|
"loss": 57.8571, |
|
"step": 600 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 2.6250000000000003e-06, |
|
"loss": 53.3942, |
|
"step": 700 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 3e-06, |
|
"loss": 49.4619, |
|
"step": 800 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 2.961538461538462e-06, |
|
"loss": 46.0906, |
|
"step": 900 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 2.923076923076923e-06, |
|
"loss": 43.4942, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 2.884615384615385e-06, |
|
"loss": 41.4881, |
|
"step": 1100 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 2.846153846153846e-06, |
|
"loss": 39.7112, |
|
"step": 1200 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 2.807692307692308e-06, |
|
"loss": 38.1241, |
|
"step": 1300 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 2.7692307692307693e-06, |
|
"loss": 37.0258, |
|
"step": 1400 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 2.7307692307692306e-06, |
|
"loss": 36.2163, |
|
"step": 1500 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 2.6923076923076923e-06, |
|
"loss": 35.2773, |
|
"step": 1600 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 2.6538461538461537e-06, |
|
"loss": 34.4202, |
|
"step": 1700 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 2.6153846153846154e-06, |
|
"loss": 33.6376, |
|
"step": 1800 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 2.5769230769230767e-06, |
|
"loss": 33.2106, |
|
"step": 1900 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 2.5384615384615385e-06, |
|
"loss": 32.6506, |
|
"step": 2000 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 2.5e-06, |
|
"loss": 32.1439, |
|
"step": 2100 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 2.4615384615384615e-06, |
|
"loss": 31.6198, |
|
"step": 2200 |
|
}, |
|
{ |
|
"epoch": 0.27, |
|
"learning_rate": 2.4230769230769233e-06, |
|
"loss": 31.0172, |
|
"step": 2300 |
|
}, |
|
{ |
|
"epoch": 0.29, |
|
"learning_rate": 2.3846153846153846e-06, |
|
"loss": 30.6347, |
|
"step": 2400 |
|
}, |
|
{ |
|
"epoch": 0.3, |
|
"learning_rate": 2.3461538461538463e-06, |
|
"loss": 30.2391, |
|
"step": 2500 |
|
}, |
|
{ |
|
"epoch": 0.31, |
|
"learning_rate": 2.307692307692308e-06, |
|
"loss": 29.9268, |
|
"step": 2600 |
|
}, |
|
{ |
|
"epoch": 0.32, |
|
"learning_rate": 2.2692307692307694e-06, |
|
"loss": 29.8757, |
|
"step": 2700 |
|
}, |
|
{ |
|
"epoch": 0.33, |
|
"learning_rate": 2.2307692307692307e-06, |
|
"loss": 29.1431, |
|
"step": 2800 |
|
}, |
|
{ |
|
"epoch": 0.34, |
|
"learning_rate": 2.192307692307692e-06, |
|
"loss": 28.9795, |
|
"step": 2900 |
|
}, |
|
{ |
|
"epoch": 0.36, |
|
"learning_rate": 2.1538461538461538e-06, |
|
"loss": 28.8664, |
|
"step": 3000 |
|
}, |
|
{ |
|
"epoch": 0.37, |
|
"learning_rate": 2.1153846153846155e-06, |
|
"loss": 28.7054, |
|
"step": 3100 |
|
}, |
|
{ |
|
"epoch": 0.38, |
|
"learning_rate": 2.076923076923077e-06, |
|
"loss": 28.4116, |
|
"step": 3200 |
|
}, |
|
{ |
|
"epoch": 0.39, |
|
"learning_rate": 2.0384615384615386e-06, |
|
"loss": 28.3629, |
|
"step": 3300 |
|
}, |
|
{ |
|
"epoch": 0.4, |
|
"learning_rate": 2e-06, |
|
"loss": 28.1642, |
|
"step": 3400 |
|
}, |
|
{ |
|
"epoch": 0.42, |
|
"learning_rate": 1.9615384615384617e-06, |
|
"loss": 28.0018, |
|
"step": 3500 |
|
}, |
|
{ |
|
"epoch": 0.43, |
|
"learning_rate": 1.9230769230769234e-06, |
|
"loss": 27.5256, |
|
"step": 3600 |
|
}, |
|
{ |
|
"epoch": 0.44, |
|
"learning_rate": 1.8846153846153847e-06, |
|
"loss": 27.5925, |
|
"step": 3700 |
|
}, |
|
{ |
|
"epoch": 0.45, |
|
"learning_rate": 1.8461538461538462e-06, |
|
"loss": 27.5127, |
|
"step": 3800 |
|
}, |
|
{ |
|
"epoch": 0.46, |
|
"learning_rate": 1.8076923076923076e-06, |
|
"loss": 27.2357, |
|
"step": 3900 |
|
}, |
|
{ |
|
"epoch": 0.48, |
|
"learning_rate": 1.7692307692307693e-06, |
|
"loss": 27.0247, |
|
"step": 4000 |
|
}, |
|
{ |
|
"epoch": 0.49, |
|
"learning_rate": 1.7307692307692306e-06, |
|
"loss": 26.7611, |
|
"step": 4100 |
|
}, |
|
{ |
|
"epoch": 0.5, |
|
"learning_rate": 1.6923076923076924e-06, |
|
"loss": 26.532, |
|
"step": 4200 |
|
}, |
|
{ |
|
"epoch": 0.51, |
|
"learning_rate": 1.653846153846154e-06, |
|
"loss": 26.7043, |
|
"step": 4300 |
|
}, |
|
{ |
|
"epoch": 0.52, |
|
"learning_rate": 1.6153846153846154e-06, |
|
"loss": 26.4993, |
|
"step": 4400 |
|
}, |
|
{ |
|
"epoch": 0.54, |
|
"learning_rate": 1.576923076923077e-06, |
|
"loss": 26.247, |
|
"step": 4500 |
|
}, |
|
{ |
|
"epoch": 0.55, |
|
"learning_rate": 1.5384615384615383e-06, |
|
"loss": 26.2046, |
|
"step": 4600 |
|
}, |
|
{ |
|
"epoch": 0.56, |
|
"learning_rate": 1.5e-06, |
|
"loss": 26.2734, |
|
"step": 4700 |
|
}, |
|
{ |
|
"epoch": 0.57, |
|
"learning_rate": 1.4615384615384616e-06, |
|
"loss": 25.9849, |
|
"step": 4800 |
|
}, |
|
{ |
|
"epoch": 0.58, |
|
"learning_rate": 1.423076923076923e-06, |
|
"loss": 26.0521, |
|
"step": 4900 |
|
}, |
|
{ |
|
"epoch": 0.59, |
|
"learning_rate": 1.3846153846153846e-06, |
|
"loss": 26.049, |
|
"step": 5000 |
|
}, |
|
{ |
|
"epoch": 0.61, |
|
"learning_rate": 1.3461538461538462e-06, |
|
"loss": 25.3998, |
|
"step": 5100 |
|
}, |
|
{ |
|
"epoch": 0.62, |
|
"learning_rate": 1.3076923076923077e-06, |
|
"loss": 25.5474, |
|
"step": 5200 |
|
}, |
|
{ |
|
"epoch": 0.63, |
|
"learning_rate": 1.2692307692307692e-06, |
|
"loss": 25.6924, |
|
"step": 5300 |
|
}, |
|
{ |
|
"epoch": 0.64, |
|
"learning_rate": 1.2307692307692308e-06, |
|
"loss": 25.3336, |
|
"step": 5400 |
|
}, |
|
{ |
|
"epoch": 0.65, |
|
"learning_rate": 1.1923076923076923e-06, |
|
"loss": 25.4108, |
|
"step": 5500 |
|
}, |
|
{ |
|
"epoch": 0.67, |
|
"learning_rate": 1.153846153846154e-06, |
|
"loss": 25.2444, |
|
"step": 5600 |
|
}, |
|
{ |
|
"epoch": 0.68, |
|
"learning_rate": 1.1153846153846154e-06, |
|
"loss": 25.2873, |
|
"step": 5700 |
|
}, |
|
{ |
|
"epoch": 0.69, |
|
"learning_rate": 1.0769230769230769e-06, |
|
"loss": 25.3395, |
|
"step": 5800 |
|
}, |
|
{ |
|
"epoch": 0.7, |
|
"learning_rate": 1.0384615384615384e-06, |
|
"loss": 25.1062, |
|
"step": 5900 |
|
}, |
|
{ |
|
"epoch": 0.71, |
|
"learning_rate": 1e-06, |
|
"loss": 25.1717, |
|
"step": 6000 |
|
}, |
|
{ |
|
"epoch": 0.73, |
|
"learning_rate": 9.615384615384617e-07, |
|
"loss": 24.6917, |
|
"step": 6100 |
|
}, |
|
{ |
|
"epoch": 0.74, |
|
"learning_rate": 9.230769230769231e-07, |
|
"loss": 24.8246, |
|
"step": 6200 |
|
}, |
|
{ |
|
"epoch": 0.75, |
|
"learning_rate": 8.846153846153847e-07, |
|
"loss": 24.9385, |
|
"step": 6300 |
|
}, |
|
{ |
|
"epoch": 0.76, |
|
"learning_rate": 8.461538461538462e-07, |
|
"loss": 24.8398, |
|
"step": 6400 |
|
}, |
|
{ |
|
"epoch": 0.77, |
|
"learning_rate": 8.076923076923077e-07, |
|
"loss": 25.0177, |
|
"step": 6500 |
|
}, |
|
{ |
|
"epoch": 0.78, |
|
"learning_rate": 7.692307692307691e-07, |
|
"loss": 24.7876, |
|
"step": 6600 |
|
}, |
|
{ |
|
"epoch": 0.8, |
|
"learning_rate": 7.307692307692308e-07, |
|
"loss": 24.7333, |
|
"step": 6700 |
|
}, |
|
{ |
|
"epoch": 0.81, |
|
"learning_rate": 6.923076923076923e-07, |
|
"loss": 24.7169, |
|
"step": 6800 |
|
}, |
|
{ |
|
"epoch": 0.82, |
|
"learning_rate": 6.538461538461538e-07, |
|
"loss": 24.7019, |
|
"step": 6900 |
|
}, |
|
{ |
|
"epoch": 0.83, |
|
"learning_rate": 6.153846153846154e-07, |
|
"loss": 24.6563, |
|
"step": 7000 |
|
}, |
|
{ |
|
"epoch": 0.84, |
|
"learning_rate": 5.76923076923077e-07, |
|
"loss": 24.4548, |
|
"step": 7100 |
|
}, |
|
{ |
|
"epoch": 0.86, |
|
"learning_rate": 5.384615384615384e-07, |
|
"loss": 24.6451, |
|
"step": 7200 |
|
}, |
|
{ |
|
"epoch": 0.87, |
|
"learning_rate": 5e-07, |
|
"loss": 24.5397, |
|
"step": 7300 |
|
}, |
|
{ |
|
"epoch": 0.88, |
|
"learning_rate": 4.6153846153846156e-07, |
|
"loss": 24.5194, |
|
"step": 7400 |
|
}, |
|
{ |
|
"epoch": 0.89, |
|
"learning_rate": 4.230769230769231e-07, |
|
"loss": 24.457, |
|
"step": 7500 |
|
}, |
|
{ |
|
"epoch": 0.9, |
|
"learning_rate": 3.846153846153846e-07, |
|
"loss": 24.2343, |
|
"step": 7600 |
|
}, |
|
{ |
|
"epoch": 0.92, |
|
"learning_rate": 3.4615384615384616e-07, |
|
"loss": 24.215, |
|
"step": 7700 |
|
}, |
|
{ |
|
"epoch": 0.93, |
|
"learning_rate": 3.076923076923077e-07, |
|
"loss": 24.1472, |
|
"step": 7800 |
|
}, |
|
{ |
|
"epoch": 0.94, |
|
"learning_rate": 2.692307692307692e-07, |
|
"loss": 24.1104, |
|
"step": 7900 |
|
}, |
|
{ |
|
"epoch": 0.95, |
|
"learning_rate": 2.3076923076923078e-07, |
|
"loss": 24.269, |
|
"step": 8000 |
|
}, |
|
{ |
|
"epoch": 0.96, |
|
"learning_rate": 1.923076923076923e-07, |
|
"loss": 24.3421, |
|
"step": 8100 |
|
}, |
|
{ |
|
"epoch": 0.98, |
|
"learning_rate": 1.5384615384615385e-07, |
|
"loss": 23.9735, |
|
"step": 8200 |
|
}, |
|
{ |
|
"epoch": 0.99, |
|
"learning_rate": 1.1538461538461539e-07, |
|
"loss": 24.3265, |
|
"step": 8300 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 7.692307692307692e-08, |
|
"loss": 24.1759, |
|
"step": 8400 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 3.846153846153846e-08, |
|
"loss": 24.1536, |
|
"step": 8500 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 0.0, |
|
"loss": 24.4166, |
|
"step": 8600 |
|
} |
|
], |
|
"max_steps": 8600, |
|
"num_train_epochs": 2, |
|
"total_flos": 3419436967096320.0, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|