The latest major release of Transformers.js, version 4, is now available on npm, bringing enhanced capabilities for running transformer models directly in the browser. This update introduces significant performance improvements, expanded model support, and new APIs designed to streamline integration for web developers.
Key features in v4 include:
- Faster inference via WebGPU acceleration where available
- Support for additional architectures such as Llama, MPT, and Falcon
- Simplified pipeline APIs for tasks like text generation, image classification, and feature extraction
- Improved quantization for reduced model sizes
The library, which enables zero-dependency machine learning in JavaScript environments, has seen growing adoption for client-side AI applications. This release marks a milestone in making state-of-the-art language models more accessible to the web community.
Developers can install it via npm i @xenova/transformers or try the CDN build. Full documentation and examples are available on the project's GitHub page.