Hugging Face Transformers
Hugging Face's open-source (Apache 2.0) library to download, train and run state-of-the-art models (Transformer and beyond) for PyTorch, TensorFlow and JAX.
Description
Hugging Face Transformers is an open-source library (Apache 2.0) developed and maintained by Hugging Face, and an industry standard for accessing state-of-the-art pretrained machine-learning models. It lets you download, fine-tune, run inference with and share models for text, vision, audio and multimodal data.
The library provides a unified API (including the pipeline, Auto* classes and Trainer) and works with the PyTorch, TensorFlow and JAX frameworks. It is tightly integrated with the Hugging Face Hub, from which tens of thousands of community models can be downloaded and to which you can publish your own.
Transformers underpins a large part of the open-source AI ecosystem and is widely used in both research and production.