PaddlePaddle/PaddleSlim
A deep learning model compression library from the PaddlePaddle ecosystem offering quantization, knowledge distillation, pruning, and architecture search.
Not currently ranked — collecting fresh signals.
star history
PaddleSlim provides model compression strategies including low-bit quantization, knowledge distillation, sparsification, and neural architecture search (NAS) to reduce model size and improve inference speed. It integrates with popular model architectures like YOLO object detectors, BERT, ERNIE, and vision transformers, and supports deployment optimization with TensorRT. The library enables developers to automatically compress models for efficient deployment across various hardware platforms.
Frequently asked
- What is PaddlePaddle/PaddleSlim?
- A deep learning model compression library from the PaddlePaddle ecosystem offering quantization, knowledge distillation, pruning, and architecture search.
- Is PaddleSlim open source?
- Yes — PaddlePaddle/PaddleSlim is open source, released under the Apache-2.0 license.
- What language is PaddleSlim written in?
- PaddlePaddle/PaddleSlim is primarily written in Python.
- How popular is PaddleSlim?
- PaddlePaddle/PaddleSlim has 1.6k stars on GitHub.
- Where can I find PaddleSlim?
- PaddlePaddle/PaddleSlim is on GitHub at https://github.com/PaddlePaddle/PaddleSlim.