[bug] Fix for reduce avg backward op #946
Draft
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Outdated suggestions cannot be applied.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 reduce avg op, multiply op requires same dims between broadcast and constant operands, which causes
test_llama_backward.py
to fail. While I managed to make a fix for this issue, the test still fails for the known issue (tenstorrent/tt-mlir#1577).I added a test for individual ops required for llama training.
https://github.com/tenstorrent/tt-thomas/issues/60
This PR also includes test to check if required dimensions for Llama are supported by mat mul op.
https://github.com/tenstorrent/tt-thomas/issues/61