Summary of Athena: Efficient Block-wise Post-training Quantization For Large Language Models Using Second-order Matrix Derivative Information, by Yanshu Wang et al.
Athena: Efficient Block-Wise Post-Training Quantization for Large Language Models Using Second-Order Matrix Derivative Information
by Yanshu Wang, Wenyang He, Tong Yang
First submitted to arxiv on: 24 May 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
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 This paper proposes an efficient method for compressing large language models (LLMs) without sacrificing performance. The authors describe traditional methods as flawed, as they do not account for the uneven distribution of parameters, leading to significant accuracy loss. Instead, the researchers develop Athena, a novel algorithm that uses second-order matrix derivative information to guide the quantization process. This approach groups parameters by columns or rows and iteratively optimizes the quantization process to achieve significant compression while maintaining high accuracy. The proposed method is suitable for deploying LLMs in various settings. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper helps us make big language models smaller, so they can fit on devices like smartphones. Right now, these models are too big to work well on those devices. To fix this problem, the researchers created a new way to shrink the model called Athena. This method looks at how the model’s parameters change when it makes mistakes and uses that information to decide which parts of the model to shrink. By doing so, Athena helps keep the model’s performance good even after shrinking it. |
Keywords
» Artificial intelligence » Quantization
