Mark scale as const and remove --fp8 flag usage #962
Conversation
* Mark only scales as const * remove --fp8 flag usage from llama * removed usage of ENABLE_CONST_MARKING Change-Id: I6dba8691d842fc62d09da5202ea1e61a111f5f18 --------- Co-authored-by: Eran Geva <egeva@habana.ai>
|
Is this linked to #966 ? |
|
regisss yes its connected |
There was a problem hiding this comment.
-
Can you put the changes of #966 into this PR, then it will be easy to merge together, because both are related.
-
Can you modify this test, because your chaneg removes fp8: https://github.com/huggingface/optimum-habana/blob/3a14236643916ecc5112f23560874cca083f2044/tests/test_text_generation_example.py
esp these lines:
Hey @ssarkar2 , both fixed. |
|
@Yantom1 It seems there are some merge conflicts, can you merge main into your branch and fix them please? Let me know if there is anything you're not sure about. |
|
The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update. |
|
@regisss fixed the last comments also |
Mark only scales as const
remove --fp8 flag usage from llama
removed usage of ENABLE_CONST_MARKING
Change-Id: I6dba8691d842fc62d09da5202ea1e61a111f5f18
What does this PR do?
Fixes # (issue)
Before submitting