Is Scaling Learned Optimizers Worth It? Evaluating the Value of VeLO's 4000 TPU Months
Abstract
We analyze VeLO (versatile learned optimzer, the largest scale attempt to train a general purpose ``foundational'' optimizer to date. VeLO was trained on thousands of machine learning tasks over 4000 TPU months with the goal of producing an optimizer capable of generalizing to new problems while being hyper-parameter free, and outperforming industry standards such as Adam. We independently evaluate VeLO on the MLcommons optimizer benchmark suite. We find that contrary to initial claims: (1) VeLO has a critical hyper-parameter that needs problem-specific tuning, (2) VeLO does not necessarily outperform competitors in quality of solution found, and (3) VeLO is not faster than competing optimizers at reducing the training loss. These observations call into question VeLO's generality and the value of the investment in training it.
Cite
Text
Rezk et al. "Is Scaling Learned Optimizers Worth It? Evaluating the Value of VeLO's 4000 TPU Months." NeurIPS 2023 Workshops: ICBINB, 2023.Markdown
[Rezk et al. "Is Scaling Learned Optimizers Worth It? Evaluating the Value of VeLO's 4000 TPU Months." NeurIPS 2023 Workshops: ICBINB, 2023.](https://mlanthology.org/neuripsw/2023/rezk2023neuripsw-scaling/)BibTeX
@inproceedings{rezk2023neuripsw-scaling,
title = {{Is Scaling Learned Optimizers Worth It? Evaluating the Value of VeLO's 4000 TPU Months}},
author = {Rezk, Fady and Antoniou, Antreas and Gouk, Henry and Hospedales, Timothy},
booktitle = {NeurIPS 2023 Workshops: ICBINB},
year = {2023},
url = {https://mlanthology.org/neuripsw/2023/rezk2023neuripsw-scaling/}
}