Pruning LLMs by Weights and Activations

Forked from Official PyTorch implementation of Wanda (Pruning by Weights and activations), as presented in paper:

A Simple and Effective Pruning Approach for Large Language Models
Mingjie Sun*, Zhuang Liu*, Anna Bair, J. Zico Kolter (* indicates equal contribution)
Carnegie Mellon University, Meta AI Research and Bosch Center for AI
Paper - Project page

@article{sun2023wanda,
  title={A Simple and Effective Pruning Approach for Large Language Models}, 
  author={Sun, Mingjie and Liu, Zhuang and Bair, Anna and Kolter, J. Zico},
  year={2023},
  journal={arXiv preprint arXiv:2306.11695}
}

---

Update

• (8.24.2024) Modify to prune Falcon-7b with mlp layer only
• (8.28.2024) Modify to prune Falcon-7b

Acknowledgement

This repository is build upon the locuslab/wanda repository.