transformers — Hugging Face Transformers is a model-definition framework for machine learning across text, vision, audio, and multimodal tasks. It provides a unified API for inference and training with over 1 million pretrained checkpoints.

Hugging Face Transformers is a model-definition framework for machine learning across text, vision, audio, and multimodal tasks. It provides a unified API for inference and training with over 1 million pretrained checkpoints.

About this project

Transformers is an open-source model-definition framework designed for machine learning tasks involving text, computer vision, audio, video, and multimodal models. It serves as a central hub across the machine learning ecosystem, ensuring compatibility with major training frameworks (such as Axolotl, Unsloth, DeepSpeed, FSDP, and PyTorch-Lightning), inference engines (including vLLM, SGLang, and TGI), and adjacent modeling libraries like llama.cpp and mlx. Key Features - Unified API: Offers a consistent interface to train, evaluate, and run inference on hundreds of model architectures. - Pipeline API: A high-level abstraction for quick inference across multiple modalities, including text generation, automatic speech recognition, image classification, and visual question answering. - Framework Interoperability: Allows users to move a single model seamlessly between PyTorch, JAX, and TensorFlow 2.0. - Hugging Face Hub Integration: Provides direct access to over 1 million pretrained model checkpoints. - Customizable Internals: Model architectures are designed with minimal abstractions, exposing internal structures to allow researchers and developers to easily customize and iterate on model files independently.