Summary of Using Half-precision For Gnn Training, by Arnab Kanti Tarafder and Yidong Gong and Pradeep Kumar
Using Half-Precision for GNN Training
by Arnab Kanti Tarafder, Yidong Gong, Pradeep Kumar
First submitted to arxiv on: 2 Nov 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: None
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 paper presents a novel approach to improving the performance and efficiency of Graph Neural Networks (GNNs) by using half-precision floating point numbers during training. Unlike other deep learning models that have benefited from reduced precision, GNNs have struggled to achieve similar improvements due to various issues such as value overflow, under-utilization of hardware resources, and poor training performance. The authors introduce HalfGNN, a GNN system that addresses these limitations by proposing novel techniques for half-precision data types, discretized sparse matrix multiplication (SpMM), and workload balancing. These innovations enable HalfGNN to achieve an average speedup of 2.30X in training time compared to float-based GNNs while maintaining similar accuracy, and reduce memory usage by 2.67X. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper is about making computer programs that work with big data more efficient. It’s like trying to make a car go faster without using too much gas. The authors found that some ways of doing things didn’t work well for special kinds of computer programs called Graph Neural Networks. They created a new way of doing things, called HalfGNN, which uses less energy and is faster than the old way. This can help computers do big tasks like analyzing social media or medical records. |
Keywords
* Artificial intelligence * Deep learning * Gnn * Precision
