fix skip_connection bugs #1011
Merged
fix skip_connection bugs #1011
Conversation
This file contains 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
Signed-off-by: Lu, Yintong <[email protected]>
Signed-off-by: Lu, Yintong <[email protected]>
Signed-off-by: Lu, Yintong <[email protected]>
wenhuach21
reviewed
Jun 20, 2023
Signed-off-by: Lu, Yintong <[email protected]>
|
sorry, need to add ut |
Signed-off-by: Lu, Yintong <[email protected]>
wenhuach21
reviewed
Jun 21, 2023
Signed-off-by: Lu, Yintong <[email protected]>
Signed-off-by: Lu, Yintong <[email protected]>
wenhuach21
approved these changes
Jul 18, 2023
chensuyue
approved these changes
Jul 18, 2023
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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Outdated suggestions cannot be applied.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.
Type of Change
bug fix
API not changed
Description
ix the skip-connection bugs when set folding=True. When an op(e.g layernorm) outputs to multiple modules and one of them is not absorbable, this op should not be transformed by smoothquant. This pr fixes some scenarios that of that issue, following up #1000 .
It also solves the issue that happened on llama model.
The fp32 acc inequivalence of Bridgetower model remains unsolved. May need further investigation later.
Expected Behavior & Potential Risk
no behavior expected to trigger by this PR
How has this PR been tested?
how to reproduce the test (including hardware information)
Dependency Change?
no