bertology_sklearn: a python toolkit for natural language understanding with BERT related models

Key Features

• Easy Using
• Multiple NLP tasks
• BERTology models
• State of the art

Easy Using

• Input Data

  • X
    • shape: [n_samples, max_seq_len]
    • type: list | ndarray | pd.Dataframe
  • y
    • shape: [n_samples] | [n_sampels, max_seq_len] for NER | [n_sampels, n_classes] for multi-label classification
    • type: list | ndarray(numpy) | Dataframe(pandas)

• Classes

  • BertologyClassifier, used for customizing models for text classification
  • BertologyTokenClassifier, used for customizing models for natural language recognition(NER)

• Class methods

  • fit(X,y), used for fine-tuning
  • predict(X), used for predicting
  • score(X,y), used for scoring

• Easy training

  • early stopping
  • k-fold cross validation

NLP tasks

• Name Entity Recognition
• Text Classification
  • Binary Classes
  • Multiple Classes
  • Multi-Label Classes

Bertology Models

State of the art

Using Examples

Model Params

Compare to other tools