Summary of Old Optimizer, New Norm: An Anthology, by Jeremy Bernstein and Laker Newhouse
Old Optimizer, New Norm: An Anthology
by Jeremy Bernstein, Laker Newhouse
First submitted to arxiv on: 30 Sep 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Optimization and Control (math.OC)
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 abstract presents a novel perspective on deep learning optimizers, arguing that three popular methods – Adam, Shampoo, and Prodigy – can be viewed as squarely first-order methods without convexity assumptions. By analyzing these methods and their relationships to steepest descent under various norms, the authors chart a new design space for training algorithms. This approach involves assigning different operator norms to tensors based on their roles within the network, which could lead to more stable, scalable, and faster training. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper looks at how we train artificial neural networks. It shows that some popular ways of doing this are actually quite simple and don’t need complicated math. The authors suggest a new way of thinking about how to make these networks work better by giving different “weights” to different parts of the network based on what they do. This could help make training faster, more stable, and more efficient. |
Keywords
» Artificial intelligence » Deep learning
