Skip to content

RAG pipeline by combining HuggingFace models with LangChain Cohere Rerank modules and create a ready notebook to evaluate the results.

Notifications You must be signed in to change notification settings

robuno/rag_w_reranker

Repository files navigation

RAG Evaluatior with Rerank

This project aims to create a RAG pipeline by combining HuggingFace models with LangChain Cohere Rerank modules and create a ready notebook to evaluate the results.

Apart from this, a script has been added to download the pages in a documentation and parse the HTML codes with BeautifulSoup4.

These code snippets were combined by following the instructions in the HuggingFace [1], LangChain [2, 3], Ragas [4] documentation.

The project includes:

  • RAG pipeline that allows trying different HuggingFace models
  • Evaluating the results of hyperparameter search & RAG pipeline with RAGAS
  • FAISS vector database
  • HuggingFacePipeline with prompt template
  • a script which allows to download the pages in a documentation and parse the HTML codes with BeautifulSoup4
  • Chunking Q&A datasets CSV format files and evaluating them through RAG

External libraries and packages:

  • Torch
  • LangChain
  • FAISS
  • Ragas
  • HuggingFace
  • Sentence-transformers
  • BitsAndBytes & Accelerate
  • PyPDF
  • Cohere
  • Pandas, Matplotlib, tqdm, getpass

About

RAG pipeline by combining HuggingFace models with LangChain Cohere Rerank modules and create a ready notebook to evaluate the results.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published