Skip to content

This project aims to provide a solution for having a Simplified Chinese chatbot with contextual awareness and domain knowledge. The decision was made to fine-tune a BART model in order to incorporate contextual awareness and construct a knowledge tree for Llama2 retrieval, thereby ensuring accurate responses related to domain knowledge.

License

Notifications You must be signed in to change notification settings

tonyma163/Llama2-with-knowledge-tree-retrieval-BART-fine-tuning-Chatbot

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

33 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Llama2 with knowledge tree retrieval & BART fine-tuning Chatbot

Team

Member
Tony Ma
Jack Lam
Leonidas Tam

This project aims to provide a solution for having a Simplified Chinese chatbot with contextual awareness and domain knowledge.

The decision was made to fine-tune a BART model in order to incorporate contextual awareness and construct a knowledge tree for Llama2 retrieval, thereby ensuring accurate responses related to domain knowledge.

System Structure

NLP_architecture(low_level)

Prerequisite

Dependency
NodeJs
python
pip

Install Required Packages

pip install transformers ctransformers accelerate datasets peft trl jieba fastapi uvicorn rouge-chinese

Install Webapp Packages

cd app
npm i

Run Server at localhost:8000

uvicorn server:app --reload

Run Webapp at localhost:3000

cd app
npm run dev

The source code of the web app is inside the particular directory

cd app cd src cd app App.tsx <- source code of webapp

About

This project aims to provide a solution for having a Simplified Chinese chatbot with contextual awareness and domain knowledge. The decision was made to fine-tune a BART model in order to incorporate contextual awareness and construct a knowledge tree for Llama2 retrieval, thereby ensuring accurate responses related to domain knowledge.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published