transformers

Overview

Transformers is a model-definition framework for machine learning with text, computer vision, audio, video, and multimodal models, supporting both inference and training. It centralizes model definitions so they are agreed upon across the ecosystem, and it acts as a pivot across frameworks: supported model definitions are compatible with training frameworks such as Axolotl, Unsloth, DeepSpeed, FSDP, and PyTorch-Lightning; inference engines such as vLLM, SGLang, and TGI; and adjacent modeling libraries such as llama.cpp and mlx. The library works with Python 3.10+ and PyTorch 2.5+, and can be installed with pip install "transformers[torch]" or from source. It provides a Pipeline API, a high-level inference class that supports text, audio, vision, and multimodal tasks, handling input preprocessing and returning appropriate output. There are over 1M+ Transformers model checkpoints on the Hugging Face Hub. The README also lists example models and tasks across audio, computer vision, multimodal, and NLP, including automatic speech recognition, image classification, visual question answering, text generation, summarization, and translation.