def train(self,
dataset_type: str,
metric: str,
save_dir: str,
model_type: str = 'chemprop',
flags: List[str] = None):
# Set up command line arguments
raw_args = self.create_raw_train_args(
dataset_type=dataset_type,
metric=metric,
save_dir=save_dir,
model_type=model_type,
flags=flags
)
# Train
with patch('sys.argv', raw_args):
command_line = ' '.join(raw_args[1:])
if model_type == 'chemprop':
print(f'python train.py {command_line}')
chemprop_train()
else:
print(f'python sklearn_train.py {command_line}')
sklearn_train()
def train(self, dataset_type: str, metric: str, save_dir: str):
# Set up command line arguments for training
raw_train_args = self.create_raw_train_args(
dataset_type=dataset_type,
metric=metric,
save_dir=save_dir
)
# Train
with patch('sys.argv', raw_train_args):
chemprop_train()
"""Trains a chemprop model on a dataset."""
from chemprop.train import chemprop_train
if __name__ == "__main__":
chemprop_train()