Skip to content

KitsuneSamaIX/RAG-Fact-Checking

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

121 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

RAG Fact Checking

Fact Checking with Retrieval Augmented Generation (RAG) built with LangChain and fully configurable.

Project

The aim of the project is the evaluation of different LLMs (Large Language Models) on fact checking tasks exploiting the RAG technique with different configuration parameters. It is designed to fact-check a predefined set of claims against their ground truth value.

The evaluation is meant to assess the impact of the truncated ranking on several standard metrics (i.e. Accuracy, Precision, Recall, F1, MSE, MAE, etc.).

The truncated ranking parameter is the number of documents (evidence) that we use as context for the fact checking of a claim. The evidence is truncated (cut off) in order of relevance (i.e. we keep the N most relevant documents).

Folders

  • src contains the Fact Checking application's source files.
  • notebooks contains jupyter notebooks to generate visualizations, manipulate data, etc.
  • playgrounds contains random tests and tutorials.

Notes

  • If you use search engine as retrieval mode, and you want to load each webpage from the web (instead of using cache), you should set the USER_AGENT environment variable.

About

Fact Checking with Retrieval Augmented Generation (RAG) built with LangChain and fully configurable.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors