airaria/TextBrewer
A PyTorch toolkit for compressing BERT and other NLP models via knowledge distillation to improve inference speed and reduce memory footprint.
Not currently ranked — collecting fresh signals.
star history
TextBrewer provides a framework for applying knowledge distillation techniques to natural language processing models. It implements various distillation methods from both NLP and computer vision research, enabling practitioners to compress large models (such as BERT-based architectures) with minimal performance loss. The toolkit offers easy-to-use abstractions for configuring student-teacher model relationships and supports standard NLP tasks including text classification and sequence labeling.
Frequently asked
- What is airaria/TextBrewer?
- A PyTorch toolkit for compressing BERT and other NLP models via knowledge distillation to improve inference speed and reduce memory footprint.
- Is TextBrewer open source?
- Yes — airaria/TextBrewer is open source, released under the Apache-2.0 license.
- What language is TextBrewer written in?
- airaria/TextBrewer is primarily written in Python.
- How popular is TextBrewer?
- airaria/TextBrewer has 1.7k stars on GitHub.
- Where can I find TextBrewer?
- airaria/TextBrewer is on GitHub at https://github.com/airaria/TextBrewer.