vis3

English | 简体中文

OSS browser based on s3

Vis3 is a visualization tool for large language models and machine learning data, supporting cloud storage platforms with S3 protocol (AWS, Aliyun OSS, Tencent Cloud) and various data formats (json, jsonl.gz, warc.gz, md, mobi, epub, etc.). It offers interactive visualization through JSON, HTML, Markdown, and image views for efficient data analysis.

Features

• Supports JSON, JSONL, WARC, and more, automatically recognizing data structures for clear, visual insights.
• One-click field previews with seamless switching between Html, Markdown, and image views for intuitive operation.
• Integrates with S3-compatible cloud storage (Aliyun OSS, AWS, Tencent Cloud) and local file parsing for easy data access.

https://github.com/user-attachments/assets/aa8ee5e8-c6d3-4b20-ae9d-2ceeb2eb2c41

Getting Started

# python >= 3.9.2
pip install vis3

Or create a Python environment using conda:

Install miniconda

# 1. Create Python 3.11 environment using conda
conda create -n vis3 python=3.11

# 2. Activate environment
conda activate vis3

# 3. Install vis3
pip install vis3

# 4. Launch (no authentication)
vis3 --open

Upgrade to the latest version

pip install vis3 -U

Variables

ENABLE_AUTH

Enable authentication.

ENABLE_AUTH=1 vis3

BASE_DATA_DIR

Specify database (SQLite) directory.

BASE_DATA_DIR=your/database/path vis3

BASE_URL

Specity base url to the api call.

BASE_URL=/a/b/c

Local Development

conda create -n vis3-dev python=3.11

# Activate virtual environment
conda activate vis3-dev

# Install poetry
# https://python-poetry.org/docs/#installing-with-the-official-installer

# Install Python dependencies
poetry install

# Install frontend dependencies (install pnpm: https://pnpm.io/installation)
cd web && pnpm install

# Build frontend assets (in web directory)
pnpm build

# Start vis3
uvicorn vis3.main:app --reload

React Component

We provide a React component via npm for customizing your data preview ui.

npm i @vis3/kit

Community

Welcome to join the Opendatalab official WeChat group!

Related Projects

• LabelU Image / Video / Audio annotation tool
• LabelU-kit Web frontend annotation kit (LabelU is developed based on this kit)
• LabelLLM Open-source LLM dialogue annotation platform
• Miner U One-stop high-quality data extraction tool

License

This project is licensed under the Apache 2.0 license.