Open
Conversation
bunch of things will break but we'll have to fix them along the way.
…g_is_enough into ffcv_imagenet_serial
my guess is that the last layer is not right. will fix.
turns out, using .half() causes some issues with the scaler.
need mixed precision for running ffcv.
looks like scaler is None for the finetuning step which causes an issue. Need to handle this a little better in the future but for now, this works.
this isn't the right branch for it, but needed to keep it somewhere
…ning_is_enough into regular_imagenet_serial
…g_is_enough into ffcv_imagenet_serial
…g_is_enough into ffcv_imagenet_serial
…g_is_enough into ffcv_imagenet_serial
…g_is_enough into ffcv_imagenet_serial
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Creating a PR to merge FFCV imagenet serial code.
Note: Currently things work but loss is infinity (My guess is that this is an issue with ResNet18 architecture that we have in the codebase. It might be a tinyimagenet vs imagenet thing)
TODO: