Project Report: Building a Custom Chatbot using Intel Extension for Transformers

This project aims to build a custom chatbot using Hugging Face models and Intel Extension for Transformers. The following steps will guide you through setting up the environment, installing necessary dependencies, and running the project.

Table of Contents

1. Environment Setup
2. Installation
3. Running the Project
4. Conclusion

Environment Setup

First, we need to create and activate a new Conda environment.

bash

Create a new Conda environment with Python 3.10

conda create -n itrex python=3.10

Activate the environment

conda activate itrex

Installation

Next, we will install the Intel Extension for Transformers and its dependencies.

bash Copy code

Install Intel Extension for Transformers

pip install intel-extension-for-transformers

Clone the Intel Extension for Transformers repository

git clone https://github.com/intel/intel-extension-for-transformers.git

Navigate to the neural chat directory

cd intel-extension-for-transformers/intel_extension_for_transformers/neural_chat/

Install CPU-specific requirements

pip install -r requirements_cpu.txt

Install additional requirements

pip install -r requirements.txt

Running the Project

To interact with Hugging Face models, log in to the Hugging Face CLI and set up Jupyter and IPython kernel.

bash Copy code

Log in to Hugging Face

huggingface-cli login

Install Jupyter and IPython kernel

python3 -m pip install jupyter ipykernel

Add kernel for its environment

python3 -m ipykernel install --name neural-chat --user

Log in to Hugging Face again if required

huggingface-cli login

Conclusion

This project demonstrates how to set up a custom chatbot using Hugging Face models and Intel Extension for Transformers. The steps outlined provide a clear guide for setting up the environment, installing necessary dependencies, and running the project.