Skip to content

Immortal-Pi/RAG_finetuning

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

29 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Llama-Law

LlamaLaw is an AI-powered chatbot designed to help visitors and newcomers navigate government websites and understand legal processes with ease. Built with a fine-tuned language model trained on legal conversations, LlamaLaw provides multilingual support, breaking down complex information on topics like visas, permits, and residency applications into clear, step-by-step guidance. By offering accurate, up-to-date information and a user-friendly experience, LlamaLaw simplifies government procedures, making essential services more accessible and reducing confusion for users.

Demo

demo

Presentation

Project Demo

Architecture

App Screenshot

Features

Multilingual model

Provides legal assistance in any given language

Provide any URL

Provide any web URL and the bot provides the right instruction to navigate

Legal Assistance

This is a legal chatbot provies legal information and assistance as per the webpage

Guide

Gives step by step instruction to navigate through the webpage and provides rules and regulations

Key Highlights

This project was built for the hackathon Llama-Impact-Hackathon so we used the llama-3 models, We used Together.AI platform for API's for our project. We also leveraged togetherai for fine tuning our models to this perticular use case. we trained the model using the below lawyer-instruct dataset from hugging face. We compared other models hosted on togetherai and finally choose llama-3-8B-instruct for our use case.

Screenshot

App Screenshot

App Screenshot

App Screenshot

App Screenshot

App Screenshot

Tech Stack

web-scraping crawl4ai

base model huggingface

dataset for fine-tuning our model huggingface

User Interface - Streamlit

Vectordatabase - ChromaDB

Conclusion

LlamaLaw addresses several critical challenges that tourists, visitors, international students and citizens face when trying to navigate government websites and access legal services:

  1. Language Barriers: LlamaLaw provides multilingual support, allowing users to access guidance in their preferred language.

  2. Complex Legal Jargon: It simplifies legal terms, making government information clear and easy to understand.

  3. Time-Consuming Navigation: LlamaLaw offers quick, direct answers, helping users find what they need without endless searching.

  4. Misinformation and Errors: By providing accurate, updated information, LlamaLaw reduces misunderstandings and mistakes.

  5. Lack of Personalized Support: It delivers personalized, step-by-step guidance, offering a more supportive experience.

Authors

Acknowledgements

About

fine tuning RAG applications using LoRA and QLoRA

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published