Lemmy: Bestiverse
  • Communities
  • Create Post
  • Create Community
  • heart
    Support Lemmy
  • search
    Search
  • Login
  • Sign Up
RSS BotMB to Hacker NewsEnglish · 3 hours ago

Fantastic Pretraining Optimizers and Where to Find Them

arxiv.org

external-link
message-square
0
fedilink
4
external-link

Fantastic Pretraining Optimizers and Where to Find Them

arxiv.org

RSS BotMB to Hacker NewsEnglish · 3 hours ago
message-square
0
fedilink
AdamW has long been the dominant optimizer in language model pretraining, despite numerous claims that alternative optimizers offer 1.4 to 2x speedup. We posit that two methodological shortcomings have obscured fair comparisons and hindered practical adoption: (i) unequal hyperparameter tuning and (ii) limited or misleading evaluation setups. To address these two issues, we conduct a systematic study of ten deep learning optimizers across four model scales (0.1B-1.2B parameters) and data-to-model ratios (1-8x the Chinchilla optimum). We find that fair and informative comparisons require rigorous hyperparameter tuning and evaluations across a range of model scales and data-to-model ratios, performed at the end of training. First, optimal hyperparameters for one optimizer may be suboptimal for another, making blind hyperparameter transfer unfair. Second, the actual speedup of many proposed optimizers over well-tuned baselines is lower than claimed and decreases with model size to only 1.1x for 1.2B parameter models. Thirdly, comparing intermediate checkpoints before reaching the target training budgets can be misleading, as rankings between two optimizers can flip during training due to learning rate decay. Through our thorough investigation, we find that all the fastest optimizers such as Muon and Soap, use matrices as preconditioners -- multiplying gradients with matrices rather than entry-wise scalars. However, the speedup of matrix-based optimizers is inversely proportional to model scale, decreasing from 1.4x over AdamW for 0.1B parameter models to merely 1.1x for 1.2B parameter models.

Comments

alert-triangle
You must log in or register to comment.

Hacker News

hackernews

Subscribe from Remote Instance

You are not logged in. However you can subscribe from another Fediverse account, for example Lemmy or Mastodon. To do this, paste the following into the search field of your instance: !hackernews@lemmy.bestiver.se
lock
Community locked: only moderators can create posts. You can still comment on posts.

Posts from the RSS Feed of HackerNews.

The feed sometimes contains ads and posts that have been removed by the mod team at HN.

Visibility: Public
globe

This community can be federated to other instances and be posted/commented in by their users.

  • 406 users / day
  • 1.35K users / week
  • 3.53K users / month
  • 9.62K users / 6 months
  • 2 local subscribers
  • 2.53K subscribers
  • 30.3K Posts
  • 12.3K Comments
  • Modlog
  • mods:
  • patrick
    cake
  • RSS Bot
  • BE: 0.19.5
  • Modlog
  • Instances
  • Docs
  • Code
  • join-lemmy.org