[fx] added activation checkpoint codegen #1355
Merged
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Outdated suggestions cannot be applied.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.
In the previous PR #1349 , we annotated the nodes if they are activation-checkpointed. In this PR, we utilize these annotations and generate model forward code with activation checkpoint. The new CodeGen inherits from the PyTorch CodeGen, the code change can be found by looking for the following multi-line comment.
A unit test is added in this PR as well. Below are the different codes generated with/without activation checkpoint.
As the codegen is only provided in torch 1.12 and the CI is based on torch 1.11, the test is skipped in pytest but the full local test log is shown below.