horseee/LLM-Pruner
A structural pruning framework for compressing large language models.
Not currently ranked — collecting fresh signals.
star history
LLM-Pruner performs structural pruning on LLMs to reduce model size and computational requirements while preserving functionality. It supports Llama, Llama-2, Llama-3, BLOOM, Vicuna, ChatGLM, Baichuan, and TinyLlama architectures. The approach uses gradient-based importance estimation to identify and remove structural components.
Frequently asked
- What is horseee/LLM-Pruner?
- A structural pruning framework for compressing large language models.
- Is LLM-Pruner open source?
- Yes — horseee/LLM-Pruner is open source, released under the Apache-2.0 license.
- What language is LLM-Pruner written in?
- horseee/LLM-Pruner is primarily written in Python.
- How popular is LLM-Pruner?
- horseee/LLM-Pruner has 1.1k stars on GitHub.
- Where can I find LLM-Pruner?
- horseee/LLM-Pruner is on GitHub at https://github.com/horseee/LLM-Pruner.