Ajay Arasanipalai

Have you considered adding the [LAMB optimizer](https://arxiv.org/abs/1904.00962) to the list? It seems to be one of the fastest so far (especially with large batches).

I was able to run the CIFAR10 experiments with different batch sizes using the `--batch-size` flag without any issues.

Have you all tried running the experiments with different batch sizes? I was able to run the default CIFAR10 experiments with a batch size of 8 using an RTX 2060....

Current dependencies on/for this PR: * `master` * **PR #105** 👈 This [stack of pull requests](https://stacking.dev/?utm_source=stack-comment) is managed by [Graphite](https://app.graphite.dev/github/pr/descriptinc/audiotools/105?utm_source=stack-comment).