---
license: apache-2.0
datasets:
- timm/mini-imagenet
---

# Comparisons of timm Optimizers w/ Caution

This repo contains summaries of several sets of experiments comparing a number of optimizers with and without caution (https://huggingface.co/papers/2411.16085) enabled.

The runs were all performed training from scratch on the `timm` 'mini-imagenet' dataset, a 100 class subset of imagenet with same image sizes as originals.

So far I have results for `adamw` and `laprop` but have some `mars` on the way. You can find full results in sub-folders by optimizer names.

# LaProp

# AdamW