Loading Now

Summary of Reconstruct the Pruned Model Without Any Retraining, by Pingjie Wang et al.


Reconstruct the Pruned Model without Any Retraining

by Pingjie Wang, Ziqing Fan, Shengchao Hu, Zhe Chen, Yanfeng Wang, Yu Wang

First submitted to arxiv on: 18 Jul 2024

Categories

  • Main: Machine Learning (cs.LG)
  • Secondary: None

     Abstract of paper      PDF of paper


GrooveSquid.com Paper Summaries

GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!

Summary difficulty Written by Summary
High Paper authors High Difficulty Summary
Read the original abstract here
Medium GrooveSquid.com (original content) Medium Difficulty Summary
A new framework called Linear Interpolation-based Adaptive Reconstruction (LIAR) is introduced to improve the efficiency and effectiveness of structured pruning for large language models. LIAR does not require retraining or back-propagation and can be used with various pruning criteria and modules. The framework uses linear interpolation to minimize reconstruction error and effectively restore the pruned model’s output. Experimental results on benchmarks such as GLUE, SQuAD, WikiText, and common sense reasoning demonstrate that LIAR enables a BERT model to maintain 98% accuracy after removing 50% of its parameters.
Low GrooveSquid.com (original content) Low Difficulty Summary
Large language models can be very large and require a lot of memory and computation to train. One way to make them smaller is by getting rid of some parts without losing too much information. This paper introduces a new method called LIAR that helps do this in a way that doesn’t need retraining or a lot of extra work. It’s like using a magic eraser to remove parts that aren’t as important, but still keeps the good stuff.

Keywords

» Artificial intelligence  » Bert  » Pruning