Skip to content

Fix v2 for MoE #1548

Draft
Qubitium wants to merge 2 commits intomainfrom
fix-v2-moe
Draft

Fix v2 for MoE #1548
Qubitium wants to merge 2 commits intomainfrom
fix-v2-moe

Conversation

@Qubitium
Copy link
Collaborator

@Qubitium Qubitium commented Apr 17, 2025

v2 requires native and quantized inputs to be stored. Unfortunately, MoE module activation will differ between native and quantized paths causing un-balanced/mismatched ordering of native_inp and inp (quantized).

…e activation

Signed-off-by: Qubitium <Qubitium@modelcloud.ai>
Signed-off-by: Qubitium <Qubitium@modelcloud.ai>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant