gt4sd.algorithms.generation.polymer_blocks.implementation module

Implementation details for PaccMann vanilla generator trained on polymer building blocks (catalysts/monomers).

Summary

Classes:

Generator

Reference

class Generator(resources_path, generated_length=100, batch_size=32, device=None)

Bases: object

__init__(resources_path, generated_length=100, batch_size=32, device=None)

Initialize the encoder/decoder generative model.

Parameters

• resources_path (str) – directory where to find models and parameters.

• generated_length (int) – length of the generated molecule in tokens. Defaults to 100.

• batch_size (int) – size of the batch. Defaults to 1.

• device (Union[device, str, None]) – device where the inference is running either as a dedicated class or a string. If not provided is inferred.

load_pretrained_paccmann(params_file, lang_file, weights_file, batch_size)

Load a pretrained PaccMann model.

Parameters

• params_file (str) – file for the parameters.

• lang_file (str) – language file.

• weights_file (str) – serialized weights file.

• batch_size (int) – size of the batch.

Return type

None

decode(latent_z, search=SamplingSearch())

Decodes a sequence of tokens given a position in the latent space.

Parameters

• latent_z (Tensor) – a batch size x latent size tensor.

• search (SamplingSearch) – defaults to sampling multinomial search.

Return type

List[int]

Returns

list of list of token indices.

sample()

Sample random molecules.

Return type

List[str]

Returns

sampled molecule (SMILES).

__dict__ = mappingproxy({'__module__': 'gt4sd.algorithms.generation.polymer_blocks.implementation', '__init__': <function Generator.__init__>, 'load_pretrained_paccmann': <function Generator.load_pretrained_paccmann>, 'decode': <function Generator.decode>, 'sample': <function Generator.sample>, '__dict__': <attribute '__dict__' of 'Generator' objects>, '__weakref__': <attribute '__weakref__' of 'Generator' objects>, '__doc__': None, '__annotations__': {}})

__doc__ = None

__module__ = 'gt4sd.algorithms.generation.polymer_blocks.implementation'

__weakref__

list of weak references to the object (if defined)