Summary of Compressing Large Language Models Using Low Rank and Low Precision Decomposition, by Rajarshi Saha et al.
Compressing Large Language Models using Low Rank and Low Precision Decomposition
by Rajarshi Saha, Naomi Sagan, Varun Srivastava, Andrea J. Goldsmith, Mert Pilanci
First submitted to arxiv on: 29 May 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI); Optimization and Control (math.OC); Machine Learning (stat.ML)
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 The authors introduce CALDERA, a novel post-training Large Language Model (LLM) compression algorithm that leverages the inherent low-rank structure of weight matrices to reduce the model’s size while maintaining its performance. The approach involves approximating the weight matrix using a low-rank, low-precision decomposition and substituting each layer with this decomposition. CALDERA is evaluated on two LLM models (LlaMa-2 7B/13B/70B and LlaMa-3 8B) and outperforms existing post-training LLM compression techniques in the regime of less than 2.5 bits per parameter. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper introduces a new way to make large language models smaller, so they can run on devices with limited memory. The method, called CALDERA, works by breaking down the model’s weight matrix into smaller pieces and then compressing each piece using low-precision formats. This allows the model to be stored and processed more efficiently without losing its ability to understand natural language. |
Keywords
» Artificial intelligence » Large language model » Llama » Precision
