https://github.com/yas-sim/openvino_japanese_chatbot_youri-7b-chat

WebUI LLM Japanese chatbot demo using Intel OpenVINO toolkit. This program uses rinna/youri-7b-chat model developed by Rinna Co.,Ltd..
https://github.com/yas-sim/openvino_japanese_chatbot_youri-7b-chat

chat chat-application chatbot chatgpt deep-learning gpt japanese large-language-model llama2 llm natural-language natural-language-processing openvino rinna youri

Last synced: 27 days ago
JSON representation

WebUI LLM Japanese chatbot demo using Intel OpenVINO toolkit. This program uses rinna/youri-7b-chat model developed by Rinna Co.,Ltd..

• Host: GitHub
• URL: https://github.com/yas-sim/openvino_japanese_chatbot_youri-7b-chat
• Owner: yas-sim
• Created: 2023-11-10T06:54:47.000Z (about 1 year ago)
• Default Branch: main
• Last Pushed: 2024-07-25T10:50:47.000Z (5 months ago)
• Last Synced: 2024-10-18T21:58:53.188Z (about 2 months ago)
• Topics: chat, chat-application, chatbot, chatgpt, deep-learning, gpt, japanese, large-language-model, llama2, llm, natural-language, natural-language-processing, openvino, rinna, youri
• Language: Jupyter Notebook
• Homepage:
• Size: 739 KB
• Stars: 2
• Watchers: 1
• Forks: 0
• Open Issues: 1
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

• awesome-openvino - Japanese chatbot Youri - LLM Japanese chatbot demo program using Intel OpenVINO toolkit. (Table of content / Natural Language Processing)

README

        
# OpenVINO Japanese chatbot demo (youri-7b-chat)

## Description

This is an LLM Japanese chatbot demo program using Intel OpenVINO toolkit. The demo uses [`rinna/youri-7b-chat`](https://huggingface.co/rinna/youri-7b-chat) model developed by [Rinna Co.,Ltd](https://rinna.co.jp).

This program can use either one of 'CPU', 'GPU.0' (integrated GPU), and 'GPU.1' Intel discrete GPU for inferencing.

## Requirement

- At least main memory of 32GB. 64GB is recommended. 

## Install prerequisites

Recommend to use Python virtual env.


You need to have Python installed.

```sh

# (optional) create a python venv and enable it.

python -m venv venv

(win) venv/Scripts/activate

(Lnx) source venv/bin/activate

pip install -r requirements0-uninstall.txt

pip install -r requirements1-install.txt

```

## Files

|file|description|

|---|---|

|`config.yml`|The system setting file. You can specify huggingface model cache directory and device for inferencing|

|`youri-7b-chat-openvino.py`|WebUI chatbot demo using `rinna/youri-7b-chat` model. This program uses OpenVINO as inference engine.
You need to run `model_download.py` to download and convert the model into OpenVINO IR before you run this demo.|

|`model_download.py`|This program downloads `rinna/youri-7b-chat` model from huggingface and convert it into FP16 OpenVINO IR model.
The converted model will be stored in `./youri-7b-chat/FP16/` directory. Also, the original model downloaded from hugging face will be stored in `./cache/huggingface/hub/` directory.|

|`benchmark_pyt.py`|Simple benchmark program using PyTorch.|

|`benchmark_ov.py`|Simple benchmark program using OpenVINO.|

## Demo screenshot

![example](./resources/example.png)

## Test environment

- System 1

	- Software

		- Windows 11 23H2

		- OpenVINO 2023.1.0

	- Hardware

		- CPU: Core i7-10700K

		- MEM: 64GB

		- GPU: Intel A380 (discrete GPU is optional)

- System 2

	- Software

		- Ubuntu 22.04.3

		- OpenVINO 2023.1.0

	- Hardware

		- CPU: Xeon Platinum 8358

		- MEM: 256GB

## Special thanks

Rinna Co.,Ltd - Original developer of rinna/youri-7b-chat model.