def test_hyperparameter_optimization_integration(dataloaders_with_coveratiates,
tmp_path):
train_dataloader = dataloaders_with_coveratiates["train"]
val_dataloader = dataloaders_with_coveratiates["val"]
try:
optimize_hyperparameters(
train_dataloader=train_dataloader,
val_dataloader=val_dataloader,
model_path=tmp_path,
max_epochs=1,
n_trials=3,
log_dir=tmp_path,
trainer_kwargs=dict(fast_dev_run=True, limit_train_batches=5),
)
finally:
shutil.rmtree(tmp_path, ignore_errors=True)
def test_hyperparameter_optimization_integration(dataloaders_with_covariates, tmp_path, use_learning_rate_finder):
train_dataloader = dataloaders_with_covariates["train"]
val_dataloader = dataloaders_with_covariates["val"]
try:
optimize_hyperparameters(
train_dataloader=train_dataloader,
val_dataloader=val_dataloader,
model_path=tmp_path,
max_epochs=1,
n_trials=3,
log_dir=tmp_path,
trainer_kwargs=dict(
fast_dev_run=True,
limit_train_batches=5,
# overwrite default trainer kwargs
progress_bar_refresh_rate=20,
),
use_learning_rate_finder=use_learning_rate_finder,
)
finally:
shutil.rmtree(tmp_path, ignore_errors=True)
# val_dataloaders=val_dataloader,
# )
# # make a prediction on entire validation set
# preds, index = tft.predict(val_dataloader, return_index=True, fast_dev_run=True)
# tune
study = optimize_hyperparameters(
train_dataloader,
val_dataloader,
model_path="optuna_test",
n_trials=200,
max_epochs=50,
gradient_clip_val_range=(0.01, 1.0),
hidden_size_range=(8, 128),
hidden_continuous_size_range=(8, 128),
attention_head_size_range=(1, 4),
learning_rate_range=(0.001, 0.1),
dropout_range=(0.1, 0.3),
trainer_kwargs=dict(limit_train_batches=30),
reduce_on_plateau_patience=4,
use_learning_rate_finder=False,
)
with open("test_study.pkl", "wb") as fout:
pickle.dump(study, fout)
# profile speed
# profile(
# trainer.fit,
# profile_fname="profile.prof",
