Updated Docstrings

Source/MultiLang.py

@@ -53,7 +53,6 @@ def _train_tokenizer(cls,
 
  Parameters
  ----------
-
  config: Configuration
  The configuration object.
  """
@@ -86,7 +85,7 @@ def _process_document(cls,
 
  Returns
  -------
- encode_list: List[int]
+ encode_list: list[int]
  Subword tokens of the cleaned document, featurized in the Byte-Pair space.
  """
 
@@ -112,15 +111,20 @@ def _process_corpus(cls,
  config: Configuration,
  parser: Tokenizer) -> numpy.ndarray:
  """
+ Iterates over all documents in the training corpus, tokenizes them using a sub-routine and stores the
+ smoothed maximum-likelihood estimates.
 
  Parameters
  ----------
- config
- parser
+ config: Configuration
+ The configuration object.
+ parser: Tokenizer
+ The configuration object.
 
  Returns
  -------
-
+ frequency_history: numpy.ndarray
+ The smoothed language-word co-occurence matrix.
  """
 
  lang_set = config.CORPUS_LANGUAGES
@@ -143,16 +147,22 @@ def _gibbs_sampler(cls,
  frequency_history: numpy.ndarray,
  encoding: List[int]) -> List[numpy.ndarray]:
  """
+ Runs the smoothed Gibbs sampler to approximate the language-word generative distribution, over a configurable
+ number of iterations.
 
  Parameters
  ----------
- config
- frequency_history
- encoding
+ config: Configuration
+ The configuration object.
+ frequency_history: numpy.ndarray
+ The smoothed language-word co-occurence matrix.
+ encoding: list[int]
+ Subword tokens of the cleaned document, featurized in the Byte-Pair space.
 
  Returns
  -------
-
+ final_dict: dict[str, int]
+ Language probabilities.
  """
 
  document_bag_of_words = numpy.bincount(encoding, minlength=config.VOCAB_SIZE)
@@ -182,14 +192,18 @@ def _gibbs_sampler(cls,
  def infer_language_distribution(self,
  doc_path: str) -> Dict[str, int]:
  """
+ Infers document language labels by invoking the gibbs sampler sub-routine and thresholding the probabilities
+ based on the configured threshold.
 
  Parameters
  ----------
- doc_path
+ doc_path: str
+ The file path to the document.
 
  Returns
  -------
-
+ result_dict: dict[str, int]
+ Language probabilities thresholded by configuration threshold.
  """
 
  document_encoding = copy.deepcopy(self._process_document(self.configuration, doc_path, self.parser))