Large artificial neural networks have become a mainstay of language, vision, and audio processing and synthesis, yet their initializations and learning rates are often set in an unsophisticated fashion, due to the high cost of hyperparameter sweeps at scale. The $\mu$-Parameterization ($\mu$P) offers a potential solution to this challenge, yielding scaling rules for model initialization and learning rates while reportedly enabling zero-shot hyperparameter transfer from small to large models. Despite its evident promise, the $\mu$P method is not yet widely adopted, perhaps due to higher implementation complexity, many variations, or complex theoretical background. This work investigates $\mu$P empirically, focusing on the ubiquitous transformer architecture, and aims to answer a simple question: does $\mu$-Transfer yield optimal learning rates in practice? Studying models of up to 10B parameters and training budgets of up to 190B tokens, we find $\mu$-Transfer works as intended for the majority of important cases, yet also identify a few cases where it may not.
A Large-Scale Exploration of $μ$-Transfer
The study empirically evaluates the effectiveness of μ-Transfer in providing optimal learning rates for transformer models with up to 10B parameters and 190B training tokens.
- Year
- 2024
- Venue
- arXiv 2024
- Authors
- 1
- Hosting
- Abstract onlyARXIV-DEFAULT
Cite
Notes
Only stored in your browser.
Attribution
- Abstract & full text
- arxiv.org/abs/2404.05728v5ARXIV-DEFAULT
- TL;DR
- Semantic Scholar