BLOOMingWave

Laveraging the power of Large Language Models
Explore the docs »

View Demo · Report Bug · Request Feature

Demo

Table of Contents

• Introduction
• System Requirements
• Prerequisite
• Quick Start
  • H2O Wave Installation
  • Setting Up the project
  • Run the application
• Features
• Acknowledgements

Introduction

H2O Wave is a software stack for building beautiful, low-latency, real-time, browser-based applications and dashboards entirely in Python without using HTML, Javascript, or CSS. Using the framework, we can develop any kind of a business application or any data science related application. In this project, I created an application to use as a playground to use BLOOM models with a higher granularity than in huggingface to do several NLP tasks by prompt tuning. Enjoy !!!

System Requirements

• Python
• pip
• GIT

Prerequisite

In order to use the application, you need to have a huggingface access token with read access. Follow the guide in here.

Quick Start

H2O Wave Installation

1. Download and extract the H2O Wave SDK for your platform using - https://github.com/h2oai/wave/releases/tag/v0.19.0
2. Move it to a location of interest. ($HOME/wave/)
3. Go to your Wave directory and open a new terminal. Start the wave server using,

./waved

Step-By-Step guide can also be found here.

Setting Up the project

1. Clone the repository and go inside the folder.

git clone https://github.com/sakunaharinda/BLOOMing-Wave.git
cd BLOOMing-Wave

2. Set up the virtual environment

 python3 -m venv venv
 source venv/bin/activate

3. Install the dependancies

pip install -r requirements.txt

Run the application

1. Run the application using the following command

wave run app.app

2. Then visit localhost:10101
3. Enjoy playing with BLOOM !!

Upcoming Features

New Chatbot, Sentiment Analysis, Summerization and many other features will be introduced soon !! 🤗

Acknowledgements

• H2O wave SDK - https://github.com/h2oai/wave
• H2O Wave Documentation and Examples - https://wave.h2o.ai/docs/examples