Transformers is a Python library that provides pre-built implementations of state-of-the-art machine learning models for tasks involving text, images, audio, and combinations of these modalities. Developers use it to quickly integrate these models into their applications for inference or fine-tune them on custom datasets without building model architectures from scratch. It supports popular frameworks like PyTorch and includes a model hub with thousands of pre-trained models across NLP, computer vision, speech, and multimodal domains.

2,437 open issues · last updated Jul 21, 2026