google-deepmind / google-deepmind/md4

About converging steps on CIFAR10

Open
#7 1 comment 0 reactions 0 assignees View on GitHub
Dominant language
Python
Stars
162
Forks
17
PR merge metrics
No merged PRs in 30d

Description

Dear Authors,

May I ask how many steps are needed for the training on CIFAR10 to converge with the default setting in the code base? I ran for about 20k steps with a batchsize 64 instead of 256, but I still get very noisy outputs and wonder whether you have any insights.

Really appreciate your help.

Best,
ChicyChen

Contributor guide

Open the contributing guide

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.