def f(trial: optuna.trial.Trial) -> float:
x = trial.suggest_int("x", 1, 1)
y = trial.suggest_categorical("y", (2.5, ))
trial.set_user_attr("train_loss", 3)
raise ValueError()
return x + y # 3.5
def _objective_func(trial: optuna.trial.Trial) -> float:
x = trial.suggest_uniform("x", -1.0, 1.0)
y = trial.suggest_loguniform("y", 20.0, 30.0)
z = trial.suggest_categorical("z", (-1.0, 1.0))
assert isinstance(z, float)
trial.set_user_attr("my_user_attr", "my_user_attr_value")
return (x - 2) ** 2 + (y - 25) ** 2 + z
def mo_objective_test_upgrade(trial: optuna.trial.Trial) -> Tuple[float, float]:
x = trial.suggest_float("x", -5, 5)
y = trial.suggest_int("y", 0, 10)
z = cast(float, trial.suggest_categorical("z", [-5, 0, 5]))
trial.set_system_attr("a", 0)
trial.set_user_attr("b", 1)
return x, x ** 2 + y ** 2 + z ** 2
def objective_test_upgrade(trial: optuna.trial.Trial) -> float:
x = trial.suggest_uniform("x", -5, 5) # optuna==0.9.0 does not have suggest_float.
y = trial.suggest_int("y", 0, 10)
z = cast(float, trial.suggest_categorical("z", [-5, 0, 5]))
trial.set_system_attr("a", 0)
trial.set_user_attr("b", 1)
trial.report(0.5, step=0)
return x ** 2 + y ** 2 + z ** 2
def _objective_func_long_user_attr(trial: optuna.trial.Trial) -> float:
x = trial.suggest_float("x", -1.0, 1.0)
y = trial.suggest_float("y", 20, 30, log=True)
z = trial.suggest_categorical("z", (-1.0, 1.0))
assert isinstance(z, float)
long_str = str(list(range(5000)))
trial.set_user_attr("my_user_attr", long_str)
return (x - 2)**2 + (y - 25)**2 + z
def _objective_func(trial: optuna.trial.Trial) -> float:
x = trial.suggest_float("x", -1.0, 1.0)
y = trial.suggest_float("y", 20, 30, log=True)
z = trial.suggest_categorical("z", (-1.0, 1.0))
assert isinstance(z, float)
trial.set_user_attr("my_user_attr", "my_user_attr_value")
mlflow.log_metric(metric_name, metric)
return (x - 2)**2 + (y - 25)**2 + z
def _objective_func(trial: optuna.trial.Trial) -> float:
u = trial.suggest_int("u", 0, 10, step=2)
v = trial.suggest_int("v", 1, 10, log=True)
w = trial.suggest_float("w", -1.0, 1.0, step=0.1)
x = trial.suggest_uniform("x", -1.0, 1.0)
y = trial.suggest_loguniform("y", 20.0, 30.0)
z = trial.suggest_categorical("z", (-1.0, 1.0))
assert isinstance(z, float)
trial.set_user_attr("my_user_attr", "my_user_attr_value")
return u + v + w + (x - 2)**2 + (y - 25)**2 + z
def f(trial: optuna.trial.Trial) -> float:
x = trial.suggest_int("x", 1, 1)
y = trial.suggest_categorical("y", (2.5, ))
assert isinstance(y, float)
trial.set_user_attr("train_loss", 3)
trial.set_system_attr("foo", "bar")
value = x + y # 3.5
# Test reported intermediate values, although it in practice is not "intermediate".
trial.report(value, step=0)
return value
def objective_for_binary_unet(args, trial: optuna.trial.Trial):
args.lr = trial.suggest_loguniform("lr", low=1e-5, high=1e-2)
args.edge_weight = trial.suggest_uniform("edge_weight", low=1, high=5)
args.wf = trial.suggest_int("wf", low=2, high=4)
args.depth = trial.suggest_int("depth", low=4, high=6)
pl_pruning_callback = PyTorchLightningPruningCallback(
trial, "val/f1_score")
ckpt_callback = train_binary_unet_model(args,
callbacks=[pl_pruning_callback])
best_f1_score = ckpt_callback.best_model_score.detach().cpu().numpy().item(
)
trial.set_user_attr("best_val_f1", best_f1_score)
trial.set_user_attr("best_model_path", ckpt_callback.best_model_path)
return best_f1_score
def f(trial: optuna.trial.Trial) -> float:
trial.set_user_attr("train_accuracy", 1)
assert trial.user_attrs["train_accuracy"] == 1
return 0.0
def __call__(self, trial: optuna.trial.Trial) -> float:
q10_init = trial.suggest_float('q10_init', 0.0001, 1000.)
seed = trial.suggest_int('seed', 0, 999999999999)
use_ta = trial.suggest_categorical('use_ta', [True, False])
dropout = trial.suggest_float('dropout', 0.0, 1.0)
if use_ta:
features = ['sw_pot', 'dsw_pot', 'ta']
else:
features = ['sw_pot', 'dsw_pot']
pl.seed_everything(seed)
# Further variables used in the hybrid model.
physical = ['ta']
# Target (multiple targets not possible currently).
targets = ['reco']
# Find variables that are only needed in physical model but not in NN.
physical_exclusive = [v for v in physical if v not in features]
# ------------
# data
# ------------
ds = xr.open_dataset(self.args.data_path)
fluxdata = FluxData(ds,
features=features + physical_exclusive,
targets=targets,
context_size=1,
train_time=slice('2003-01-01', '2006-12-31'),
valid_time=slice('2007-01-01', '2007-12-31'),
test_time=slice('2008-01-01', '2008-12-31'),
batch_size=self.args.batch_size,
data_loader_kwargs={'num_workers': 4})
train_loader = fluxdata.train_dataloader()
val_loader = fluxdata.val_dataloader()
test_loader = fluxdata.test_dataloader()
# Create empty xr.Dataset, will be used by the model to save predictions every epoch.
max_epochs = TRAINER_ARGS['max_epochs']
ds_pred = fluxdata.target_xr('valid',
varnames=['reco', 'rb'],
num_epochs=max_epochs)
# ------------
# model
# ------------
model = Q10Model(features=features,
targets=targets,
norm=fluxdata._norm,
ds=ds_pred,
q10_init=q10_init,
hidden_dim=self.args.hidden_dim,
num_layers=self.args.num_layers,
learning_rate=self.args.learning_rate,
dropout=dropout,
weight_decay=self.args.weight_decay,
num_steps=len(train_loader) * max_epochs)
# ------------
# training
# ------------
trainer = pl.Trainer.from_argparse_args(
self.args,
default_root_dir=self.args.log_dir,
**TRAINER_ARGS,
callbacks=[
EarlyStopping(monitor='valid_loss',
patience=10,
min_delta=0.00001),
ModelCheckpoint(filename='{epoch}-{val_loss:.2f}',
save_top_k=1,
verbose=False,
monitor='valid_loss',
mode='min',
prefix=model.__class__.__name__)
])
trainer.fit(model, train_loader, val_loader)
# ------------
# testing
# ------------
# trainer.test(test_dataloaders=test_loader)
# ------------
# save results
# ------------
# Store predictions.
ds = fluxdata.add_scalar_record(model.ds,
varname='q10',
x=model.q10_history)
trial.set_user_attr('q10', ds.q10[-1].item())
# Add some attributes that are required for analysis.
ds.attrs = {
'created': datetime.now().strftime("%Y-%m-%d %H:%M:%S"),
'author': '*****@*****.**',
'q10_init': q10_init,
'dropout': dropout,
'use_ta': int(use_ta),
'loss': trainer.callback_metrics['valid_loss'].item()
}
ds = ds.isel(epoch=slice(0, trainer.current_epoch + 1))
# Save data.
save_dir = os.path.join(model.logger.log_dir, 'predictions.nc')
print(f'Saving predictions to: {save_dir}')
ds.to_netcdf(save_dir)
return trainer.callback_metrics['valid_loss'].item()
def objective(trial: optuna.trial.Trial, value: float) -> float:
trial.set_user_attr("lightgbm_tuner:step_name", "step{:.0f}".format(value))
return trial.suggest_uniform("x", value, value)
def objective(trial: optuna.trial.Trial) -> Tuple[float, float]:
trial.suggest_int("x", 5, 5)
trial.set_user_attr("constraint", (constraint_value, -1))
return 5.0, 5.0
