Source code for gt4sd.training_pipelines.torchdrug.core

#
# MIT License
#
# Copyright (c) 2022 GT4SD team
#
# Permission is hereby granted, free of charge, to any person obtaining a copy
# of this software and associated documentation files (the "Software"), to deal
# in the Software without restriction, including without limitation the rights
# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
# copies of the Software, and to permit persons to whom the Software is
# furnished to do so, subject to the following conditions:
#
# The above copyright notice and this permission notice shall be included in all
# copies or substantial portions of the Software.
#
# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
# SOFTWARE.
#
"""TorchDrug training utilities."""
import os
from dataclasses import dataclass, field
from typing import Any, Dict, Optional

from ...configuration import gt4sd_configuration_instance
from ..core import TrainingPipeline, TrainingPipelineArguments
from . import DATASET_FACTORY

DATA_ROOT = os.path.join(
    gt4sd_configuration_instance.gt4sd_local_cache_path, "data", "torchdrug"
)
os.makedirs(DATA_ROOT, exist_ok=True)


[docs]class TorchDrugTrainingPipeline(TrainingPipeline): """TorchDrug training pipelines."""
[docs] def train( # type: ignore self, training_args: Dict[str, Any], model_args: Dict[str, Any], dataset_args: Dict[str, Any], ) -> None: """Generic training function for launching a TorchDrug training. Args: training_args: training arguments passed to the configuration. model_args: model arguments passed to the configuration. dataset_args: dataset arguments passed to the configuration. Raises: NotImplementedError: the generic trainer does not implement the pipeline. """ raise NotImplementedError
[docs]@dataclass class TorchDrugTrainingArguments(TrainingPipelineArguments): """Arguments related to torchDrug trainer.""" __name__ = "training_args" model_path: str = field( metadata={"help": "Path where the model artifacts are stored."} ) training_name: str = field(metadata={"help": "Name used to identify the training."}) epochs: int = field(default=10, metadata={"help": "Number of epochs."}) batch_size: int = field(default=16, metadata={"help": "Size of the batch."}) learning_rate: float = field( default=1e-5, metadata={"help": "Learning rate used in training."} ) log_interval: int = field( default=100, metadata={"help": "Number of steps between log intervals."} ) gradient_interval: int = field( default=1, metadata={"help": "Gradient accumulation steps"} ) num_worker: int = field( default=0, metadata={"help": "Number of CPU workers per GPU."} ) task: Optional[str] = field( default=None, metadata={ "help": "Optimization task for goal-driven generation." "Currently, TorchDrug only supports `plogp` and `qed`." }, )
[docs]@dataclass class TorchDrugDataArguments(TrainingPipelineArguments): """Arguments related to TorchDrug data loading.""" __name__ = "dataset_args" dataset_name: str = field( metadata={ "help": f"Identifier for the dataset. Has to be in {DATASET_FACTORY.keys()}" ". Can either point to one of the predefined TorchDrug datasets or it can " "be `custom` if the user brings their own dataset. If `custom`, then the " "arguments `file_path`, `target_field` and `smiles_field` below have to be" " specified." } ) file_path: str = field( default="", metadata={ "help": "Ignored unless `datase_name` is `custom`. In that case it's " "a path to a .csv file containing the training data." }, ) dataset_path: str = field( default=DATA_ROOT, metadata={ "help": "Path where the TorchDrug dataset will be stored. This is ignored " "if `datase_name` is `custom`." }, ) target_field: str = field( default="", metadata={ "help": "Ignored unless `datase_name` is `custom`. In that case it's a str " "with name of the column containing the property that can be optimized." "Currently TorchDrug only supports `plogp` and `qed`." }, ) smiles_field: str = field( default="smiles", metadata={ "help": "Ignored unless `datase_name` is `custom`. In that case it's the " "name of the column containing the SMILES strings." }, ) transform: str = field( default="lambda x: x", metadata={ "help": "Optional data transformation function. Has to be a lambda function" " (written as a string) that operates on the batch dictionary." "See torchdrug docs for details." }, ) verbose: int = field( default=1, metadata={"help": "Output verbosity level for dataset."} ) lazy: bool = field( default=False, metadata={ "help": "If yes, molecules are processed in the dataloader. This is faster " "for setup but slower at training time." }, ) node_feature: str = field( default="default", metadata={"help": "Node features (or node feature list) to extract."}, ) edge_feature: str = field( default="default", metadata={"help": "Edge features (or edge feature list) to extract."}, ) graph_feature: Optional[str] = field( default=None, metadata={"help": "Graph features (or graph feature list) to extract."}, ) with_hydrogen: bool = field( default=False, metadata={"help": "Whether hydrogens are stored in molecular graph."}, ) no_kekulization: bool = field( default=False, metadata={ "help": "Whether SMILES kekulization is used. Per default, it is used." }, )
[docs]@dataclass class TorchDrugSavingArguments(TrainingPipelineArguments): """Saving arguments related to TorchDrug trainer.""" __name__ = "saving_args" model_path: str = field( metadata={"help": "Path where the model artifacts are stored."} ) training_name: str = field(metadata={"help": "Name used to identify the training."}) dataset_name: str = field( metadata={ "help": f"Identifier for the dataset. Has to be in {DATASET_FACTORY.keys()}" ". Can either point to one of the predefined TorchDrug datasets or it can " "be `custom` if the user brings their own dataset. If `custom`, then the " "arguments `file_path`, `target_field` and `smiles_field` below have to be" " specified." } ) task: Optional[str] = field( default=None, metadata={ "help": "Optimization task for goal-driven generation." "Currently, TorchDrug only supports `plogp` and `qed`." }, ) file_path: str = field( default="", metadata={ "help": "Ignored unless `datase_name` is `custom`. In that case it's " "a path to a .csv file containing the training data." }, ) epochs: int = field(default=10, metadata={"help": "Number of epochs."})