Product & software
Transformers
Transformers is a Hugging Face library for working with pretrained machine-learning models.
At a glance
- Primary focus
- Pretrained model interfaces
- Category
- machine learning
Overview
Transformers is a Hugging Face library for working with pretrained machine-learning models. It provides common interfaces for loading model configurations, processing inputs, and running tasks across supported architectures and frameworks.
How it is used
A pipeline packages preprocessing, model execution, and output handling for a task. Lower-level model and tokenizer interfaces offer more control when an application needs custom behavior.
A useful distinction
The library and the downloaded model are separate resources. Model weights can have their own licenses, hardware requirements, and intended uses, which should be checked in the model's documentation.
Sources and review
MOOR's explanatory text is supported by the following source links.
- Transformers: official overview — Hugging Face
- Transformers: documentation and further reading — Hugging Face