Hugging Face Transformers
Python library defining pretrained models for text, vision, and audio
What it is
Transformers is a Python library that provides reference implementations of thousands of pretrained model architectures, from language models to speech recognition and vision models, and loads checkpoints from the Hugging Face Hub. A high-level Pipeline API handles preprocessing and output for tasks like text generation, transcription, and image classification, while lower-level classes support fine-tuning and training. ML engineers and researchers rely on it as the common model definition that training tools and inference engines build on.