Base Modules¶

Base¶

class BaseModel(device=None, model_name='BaseModel', verbose='none')[source]¶

Bases: ABC

Base class for molecular models with shared functionality.

This abstract class provides common methods and utilities for molecular models, including model initialization, saving/loading, and parameter management.

Parameters:

device (torch.device, optional) – Device to run the model on. If None, automatically selects CUDA if available, otherwise CPU.
model_name (str, default="BaseModel") – String identifier for the model name which can be specified by the user.
verbose (str, default="none") – Whether to display progress info. Options are: “none”, “progress_bar”, “print_statement”. If any other, “none” is automatically chosen.

model_class¶

The class of the model used to initialize the model instance.

Type:: type or None

model¶

The fitted model instance if the model has been trained, None otherwise.

Type:: object or None

is_fitted_¶

Whether the model has been fitted/trained. False by default.

Type:: bool

get_params(deep=True)[source]¶

Get parameters for this estimator.

Parameters:: deep (bool, default=True) – If True, will return the parameters for this estimator and contained subobjects that are estimators.
Returns:: Dictionary of parameter names mapped to their values
Return type:: Dict[str, Any]

load(path=None, repo_id=None, **kwargs)[source]¶

Automatic load from either local disk or Hugging Face Hub.

Parameters:

path (Optional[str], default=None) – File path for local loading.
repo_id (Optional[str], default=None) – Hugging Face repository ID for remote loading. If path is provided, repo_id is ignored.
**kwargs – Additional arguments passed to load_from_hf

Raises:

FileNotFoundError – If no local file is found and no repo_id is provided

Return type:

None

load_from_hf(repo_id, local_cache=None, config_filename='config.json')[source]¶

Load model from Hugging Face Hub.

Parameters:

repo_id (str) – Hugging Face repository ID
local_cache (str, default=None) – Local path to save the model
config_filename (str, default='config.json') – Name of the configuration file to load from the repository

Return type:

None

load_from_local(path)[source]¶

Load model from local disk.

Parameters:: path (str) – File path to load the model from
Return type:: None

save(path=None, repo_id=None, **kwargs)[source]¶

Automatic save to either local disk or Hugging Face Hub.

Parameters:

path (Optional[str], default=None) – File path for local saving (required if repo_id is None)
repo_id (Optional[str], default=None) – Hugging Face repository ID for remote saving
**kwargs – Additional arguments passed to save_to_hf

Raises:

ValueError – If path is None when repo_id is None

Return type:

None

save_to_hf(repo_id, task_id='default', metadata_dict=None, metrics=None, commit_message='Update model', hf_token=None, private=False, config_filename='config.json')[source]¶

Save model to Hugging Face Hub.

Parameters:

repo_id (str) – Hugging Face repository ID
task_id (str, default="default") – Task identifier for the model
metadata_dict (Optional[Dict[str, Any]], default=None) – Optional metadata to store with the model
metrics (Optional[Dict[str, float]], default=None) – Optional performance metrics to store with the model
commit_message (str, default="Update model") – Git commit message
hf_token (Optional[str], default=None) – Hugging Face authentication token
private (bool, default=False) – Whether the repository should be private
config_filename (Optional[str], default='config.json') – Name of the configuration file to save to the repository

Raises:

ValueError – If the model is not fitted

Return type:

None

save_to_local(path)[source]¶

Save model to local disk.

Parameters:: path (str) – File path to save the model
Raises:: ValueError – If the model is not fitted
Return type:: None

set_params(**params)[source]¶

Set parameters for this estimator.

Parameters:: **params – Parameter names mapped to their values
Returns:: Self instance for method chaining
Return type:: BaseModel
Raises:: ValueError – If an invalid parameter is provided

Base Predictor¶

class BaseMolecularPredictor(*, device=None, model_name='BaseMolecularPredictor', num_task=0, task_type=None, verbose='none')[source]¶

Bases: BaseModel, ABC

Base class for molecular discovery estimators.

Parameters:

device (device | str | None)
model_name (str)
num_task (int)
task_type (str | None)
verbose (str)

abstractmethod autofit(X_train, y_train, X_val=None, y_val=None, search_parameters=None, n_trials=10)[source]¶

Parameters:

search_parameters (dict | None)
n_trials (int)

Return type:

BaseMolecularPredictor

abstractmethod fit(X_train, y_train, X_val=None, y_val=None, search_parameters=None, n_trials=10)[source]¶

Parameters:

search_parameters (dict | None)
n_trials (int)

Return type:

BaseMolecularPredictor

abstractmethod predict(X)[source]¶

Parameters:: X (ndarray)
Return type:: ndarray

Base Generator¶

class BaseMolecularGenerator(*, device=None, model_name='BaseMolecularGenerator', verbose='none')[source]¶

Bases: BaseModel, ABC

Base class for molecular generation.

Parameters:

device (device | str | None)
model_name (str)
verbose (str)

abstractmethod fit(X, y=None)[source]¶

Parameters:

X (List[str])
y (ndarray | None)

Return type:

BaseMolecularGenerator

abstractmethod generate(n_samples, **kwargs)[source]¶

Generate molecular structures.

Parameters:: n_samples (int)
Return type:: List[str]

Base Encoder¶

class BaseMolecularEncoder(*, device=None, model_name='BaseMolecularEncoder', verbose='none')[source]¶

Bases: BaseModel, ABC

Base class for molecular representation learning.

Parameters:

device (device | str | None)
model_name (str)
verbose (str)

abstractmethod encode(X, return_type='pt')[source]¶

Parameters:

X (List[str])
return_type (Literal['np', 'pt'])

Return type:

ndarray | Tensor

abstractmethod fit(X, y=None)[source]¶

Parameters:

X (List[str])
y (ndarray | None)

Return type:

BaseMolecularEncoder