Shumai
Shumai is an open-source, fast, differentiable tensor library for TypeScript and JavaScript, designed to simplify dataset creation and accelerate small model training.
Disclaimer: Visionary Hub is not affiliated with, endorsed by, or the operator of this tool. All trademarks, logos, and content are the property of their respective owners. Full disclaimer available here
Key Features
Differentiable Tensors
Supports gradient-based optimization for training machine learning models.
GPU Acceleration
Utilizes CUDA on Linux for high-speed tensor computations.
Bun + Flashlight Integration
Leverages Bun runtime and Flashlight ML engine for performance and extensibility.
Network Connectivity
Enables distributed and networked tensor operations for advanced use cases.
Get Started
Share & Save
Share on Social Media
Why Choose Shumai
Open Source:
Free access to source code encourages customization and community contributions.High Performance:
Fast tensor operations with GPU acceleration on Linux and CPU support on macOS.TypeScript Support:
Designed specifically for TypeScript and JavaScript developers for seamless integration.
Pricing
Shumai is an open-source library available for free. Users can access and contribute to the codebase on GitHub without any cost.
About Shumai
Shumai is an open-source, fast, differentiable tensor library for TypeScript and JavaScript, designed to simplify dataset creation and accelerate small model training.
What Shumai Does
Shumai provides fast tensor operations with differentiable tensors, enabling gradient-based optimization for machine learning models. It supports both CPU and GPU computation, with CUDA acceleration on Linux and CPU backend on macOS.
The library integrates tightly with Bun and Flashlight, offering network connectivity and advanced training/inference capabilities. Users can easily convert between native JavaScript arrays and tensors, and implement complex training logic using JavaScript's JIT compilation.
Shumai is suited for machine learning, data analysis, and computer vision tasks, particularly for developers working in TypeScript or JavaScript environments who require efficient tensor computations.
Pros & Cons
Fast Execution
Efficient native bindings provide superior speed for small model training.
Cross-Platform
Supports macOS and Linux with plans for broader backend support.
Limited Windows Support
Windows usage requires Docker and WSL2, not officially supported natively.
Experimental
The software is still experimental and install procedures are a work in progress.
Frequently Asked Questions
Shumai officially supports macOS and Linux; Windows support is possible via Docker and WSL2.
Linux installs default to GPU computation with CUDA; macOS defaults to CPU usage.
Install Bun and ArrayFire, then run 'bun install @shumai/shumai' to install the package.
Flashlight is a fast, extensible machine learning engine written in C++ used as Shumai's backend.
Yes, Shumai is open-source and available free of charge on GitHub.
Similar Tools You Might Like
Discover more AI-powered tools that complement your workflow
List Your AI Tool & Reach Thousands of Users
Join 500+ AI innovators already thriving on our platform. Get visibility, feedback, and boost your conversions.
Expand Your Audience
Connect with over 50,000 AI enthusiasts actively looking for tools like yours.
Boost Your Authority
Get verified reviews and ratings to build credibility in the AI marketplace.
Drive Conversions
Our premium placements and targeted audience deliver quality leads and sign-ups.