ViTGPT2I2A / trainer_state.json
{
"best_metric": 0.0708099901676178,
"best_model_checkpoint": "/nasty/data/tpid/vizwiz/ViTGPT2I2A/checkpoint-17000",
"epoch": 5.0,
"global_step": 29290,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.17,
"learning_rate": 1.9320587231136908e-05,
"loss": 0.1528,
"step": 1000
},
{
"epoch": 0.17,
"eval_loss": 0.08688350766897202,
"eval_runtime": 552.3255,
"eval_samples_per_second": 14.032,
"eval_steps_per_second": 3.509,
"step": 1000
},
{
"epoch": 0.34,
"learning_rate": 1.8637760327756916e-05,
"loss": 0.0899,
"step": 2000
},
{
"epoch": 0.34,
"eval_loss": 0.08172640949487686,
"eval_runtime": 437.9983,
"eval_samples_per_second": 17.694,
"eval_steps_per_second": 4.425,
"step": 2000
},
{
"epoch": 0.51,
"learning_rate": 1.795493342437692e-05,
"loss": 0.084,
"step": 3000
},
{
"epoch": 0.51,
"eval_loss": 0.07901930063962936,
"eval_runtime": 217.4661,
"eval_samples_per_second": 35.638,
"eval_steps_per_second": 8.912,
"step": 3000
},
{
"epoch": 0.68,
"learning_rate": 1.7272106520996928e-05,
"loss": 0.0814,
"step": 4000
},
{
"epoch": 0.68,
"eval_loss": 0.07728389650583267,
"eval_runtime": 218.3505,
"eval_samples_per_second": 35.493,
"eval_steps_per_second": 8.876,
"step": 4000
},
{
"epoch": 0.85,
"learning_rate": 1.6589279617616936e-05,
"loss": 0.0803,
"step": 5000
},
{
"epoch": 0.85,
"eval_loss": 0.07569148391485214,
"eval_runtime": 218.1796,
"eval_samples_per_second": 35.521,
"eval_steps_per_second": 8.883,
"step": 5000
},
{
"epoch": 1.02,
"learning_rate": 1.590645271423694e-05,
"loss": 0.077,
"step": 6000
},
{
"epoch": 1.02,
"eval_loss": 0.07452211529016495,
"eval_runtime": 220.2685,
"eval_samples_per_second": 35.184,
"eval_steps_per_second": 8.798,
"step": 6000
},
{
"epoch": 1.19,
"learning_rate": 1.5223625810856947e-05,
"loss": 0.0739,
"step": 7000
},
{
"epoch": 1.19,
"eval_loss": 0.0739876851439476,
"eval_runtime": 205.9033,
"eval_samples_per_second": 37.639,
"eval_steps_per_second": 9.412,
"step": 7000
},
{
"epoch": 1.37,
"learning_rate": 1.4540798907476957e-05,
"loss": 0.0719,
"step": 8000
},
{
"epoch": 1.37,
"eval_loss": 0.07368426769971848,
"eval_runtime": 217.9815,
"eval_samples_per_second": 35.553,
"eval_steps_per_second": 8.891,
"step": 8000
},
{
"epoch": 1.54,
"learning_rate": 1.3857972004096963e-05,
"loss": 0.0717,
"step": 9000
},
{
"epoch": 1.54,
"eval_loss": 0.07300957292318344,
"eval_runtime": 219.2086,
"eval_samples_per_second": 35.354,
"eval_steps_per_second": 8.841,
"step": 9000
},
{
"epoch": 1.71,
"learning_rate": 1.3175145100716968e-05,
"loss": 0.0731,
"step": 10000
},
{
"epoch": 1.71,
"eval_loss": 0.0726846307516098,
"eval_runtime": 219.3614,
"eval_samples_per_second": 35.33,
"eval_steps_per_second": 8.835,
"step": 10000
},
{
"epoch": 1.88,
"learning_rate": 1.2492318197336977e-05,
"loss": 0.0708,
"step": 11000
},
{
"epoch": 1.88,
"eval_loss": 0.07204297184944153,
"eval_runtime": 217.9036,
"eval_samples_per_second": 35.566,
"eval_steps_per_second": 8.894,
"step": 11000
},
{
"epoch": 2.05,
"learning_rate": 1.1809491293956984e-05,
"loss": 0.0697,
"step": 12000
},
{
"epoch": 2.05,
"eval_loss": 0.07171137630939484,
"eval_runtime": 219.7173,
"eval_samples_per_second": 35.273,
"eval_steps_per_second": 8.82,
"step": 12000
},
{
"epoch": 2.22,
"learning_rate": 1.1126664390576988e-05,
"loss": 0.0655,
"step": 13000
},
{
"epoch": 2.22,
"eval_loss": 0.07193101197481155,
"eval_runtime": 218.2496,
"eval_samples_per_second": 35.51,
"eval_steps_per_second": 8.88,
"step": 13000
},
{
"epoch": 2.39,
"learning_rate": 1.0443837487196998e-05,
"loss": 0.0653,
"step": 14000
},
{
"epoch": 2.39,
"eval_loss": 0.07193595170974731,
"eval_runtime": 218.5083,
"eval_samples_per_second": 35.468,
"eval_steps_per_second": 8.869,
"step": 14000
},
{
"epoch": 2.56,
"learning_rate": 9.761693410720383e-06,
"loss": 0.0657,
"step": 15000
},
{
"epoch": 2.56,
"eval_loss": 0.07120098173618317,
"eval_runtime": 219.0492,
"eval_samples_per_second": 35.38,
"eval_steps_per_second": 8.847,
"step": 15000
},
{
"epoch": 2.73,
"learning_rate": 9.07886650734039e-06,
"loss": 0.0663,
"step": 16000
},
{
"epoch": 2.73,
"eval_loss": 0.07102101296186447,
"eval_runtime": 218.7347,
"eval_samples_per_second": 35.431,
"eval_steps_per_second": 8.86,
"step": 16000
},
{
"epoch": 2.9,
"learning_rate": 8.396722430863777e-06,
"loss": 0.0654,
"step": 17000
},
{
"epoch": 2.9,
"eval_loss": 0.0708099901676178,
"eval_runtime": 218.8513,
"eval_samples_per_second": 35.412,
"eval_steps_per_second": 8.855,
"step": 17000
},
{
"epoch": 3.07,
"learning_rate": 7.713895527483783e-06,
"loss": 0.0645,
"step": 18000
},
{
"epoch": 3.07,
"eval_loss": 0.07163063436746597,
"eval_runtime": 219.9493,
"eval_samples_per_second": 35.235,
"eval_steps_per_second": 8.811,
"step": 18000
},
{
"epoch": 3.24,
"learning_rate": 7.0317514510071705e-06,
"loss": 0.0616,
"step": 19000
},
{
"epoch": 3.24,
"eval_loss": 0.07119180262088776,
"eval_runtime": 219.2713,
"eval_samples_per_second": 35.344,
"eval_steps_per_second": 8.838,
"step": 19000
},
{
"epoch": 3.41,
"learning_rate": 6.348924547627177e-06,
"loss": 0.0607,
"step": 20000
},
{
"epoch": 3.41,
"eval_loss": 0.07116228342056274,
"eval_runtime": 219.1634,
"eval_samples_per_second": 35.362,
"eval_steps_per_second": 8.843,
"step": 20000
},
{
"epoch": 3.58,
"learning_rate": 5.666097644247184e-06,
"loss": 0.0611,
"step": 21000
},
{
"epoch": 3.58,
"eval_loss": 0.07110943645238876,
"eval_runtime": 219.7957,
"eval_samples_per_second": 35.26,
"eval_steps_per_second": 8.817,
"step": 21000
},
{
"epoch": 3.76,
"learning_rate": 4.98327074086719e-06,
"loss": 0.0615,
"step": 22000
},
{
"epoch": 3.76,
"eval_loss": 0.07111015915870667,
"eval_runtime": 220.0979,
"eval_samples_per_second": 35.212,
"eval_steps_per_second": 8.805,
"step": 22000
},
{
"epoch": 3.93,
"learning_rate": 4.3018094912939576e-06,
"loss": 0.0614,
"step": 23000
},
{
"epoch": 3.93,
"eval_loss": 0.0710098147392273,
"eval_runtime": 223.1322,
"eval_samples_per_second": 34.733,
"eval_steps_per_second": 8.685,
"step": 23000
},
{
"epoch": 4.1,
"learning_rate": 3.618982587913964e-06,
"loss": 0.0594,
"step": 24000
},
{
"epoch": 4.1,
"eval_loss": 0.07158154249191284,
"eval_runtime": 220.2761,
"eval_samples_per_second": 35.183,
"eval_steps_per_second": 8.798,
"step": 24000
},
{
"epoch": 4.27,
"learning_rate": 2.936155684533971e-06,
"loss": 0.0587,
"step": 25000
},
{
"epoch": 4.27,
"eval_loss": 0.07150966674089432,
"eval_runtime": 223.8011,
"eval_samples_per_second": 34.629,
"eval_steps_per_second": 8.659,
"step": 25000
},
{
"epoch": 4.44,
"learning_rate": 2.2533287811539777e-06,
"loss": 0.0574,
"step": 26000
},
{
"epoch": 4.44,
"eval_loss": 0.07154253125190735,
"eval_runtime": 216.2215,
"eval_samples_per_second": 35.843,
"eval_steps_per_second": 8.963,
"step": 26000
},
{
"epoch": 4.61,
"learning_rate": 1.5711847046773643e-06,
"loss": 0.0579,
"step": 27000
},
{
"epoch": 4.61,
"eval_loss": 0.07154078036546707,
"eval_runtime": 219.6549,
"eval_samples_per_second": 35.283,
"eval_steps_per_second": 8.823,
"step": 27000
},
{
"epoch": 4.78,
"learning_rate": 8.883578012973712e-07,
"loss": 0.0581,
"step": 28000
},
{
"epoch": 4.78,
"eval_loss": 0.07147673517465591,
"eval_runtime": 218.889,
"eval_samples_per_second": 35.406,
"eval_steps_per_second": 8.854,
"step": 28000
},
{
"epoch": 4.95,
"learning_rate": 2.0553089791737796e-07,
"loss": 0.0579,
"step": 29000
},
{
"epoch": 4.95,
"eval_loss": 0.07150830328464508,
"eval_runtime": 219.8431,
"eval_samples_per_second": 35.252,
"eval_steps_per_second": 8.815,
"step": 29000
},
{
"epoch": 5.0,
"step": 29290,
"total_flos": 2.1143149521668145e+19,
"train_loss": 0.07038866728798608,
"train_runtime": 25445.4127,
"train_samples_per_second": 4.604,
"train_steps_per_second": 1.151
}
],
"max_steps": 29290,
"num_train_epochs": 5,
"total_flos": 2.1143149521668145e+19,
"trial_name": null,
"trial_params": null
}