TextFlint Documentation

TextFlint is a multilingual robustness evaluation platform for natural language processing tasks, which unifies general text transformation, task-specific transformation, adversarial attack, sub-population, and their combinations to provide a comprehensive robustness analysis.

Notebook Tutorials

• Transformation
  • How to use a built-in Transformation
  • Define your own Transformation
  • Conclusion
• SubPopulation
  • How to use a built-in SubPopulation
  • Define your own SubPopulation
  • Conclusion
• AttackRecipe
  • Using an AttackRecipe based on textattack
  • Conclusion
• Sample
• Dataset
• Validator
  • Using the built-in Validator SentenceEncoding
  • Using the SentenceEncoding to validate BreadWordSwap
  • Conclusion
• FlintModels
  • How to customize targe model
  • How to implement the automatic evaluation
  • How to implement adversarial attack

API

• textflint
  • textflint.common
  • textflint.generation_layer
  • textflint.input_layer
  • textflint.report_layer
  • textflint.adapter
  • textflint.engine

Index & Search

• Index

• Module Index

• Search Page