Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Hi - I thought I would raise this PR which implements
AdaBoundW
.I did so because I was using
AdaBound
from this repo withweight_decay=0.1
and noticed the training loss couldn't go down. The training also failed withAdam
but succeeded withAdamW
.The algorithm is very simple, in fact just 1 line added compared to the original version:
Note, in the original implementation, the
weight_decay
is not scaled by thebase_lr
(which I think is not common practice as the official PyTorch Repo does it this way withAdamW
(https://pytorch.org/docs/stable/_modules/torch/optim/adamw.html#AdamW)