axlearn icon indicating copy to clipboard operation
axlearn copied to clipboard

Published 20 hours ago • verified publisher icon apple

support ssd/mamba2

Open berlino opened this issue 4 months ago • 0 comments

Support mamba2 which uses SSD kernel (as opposed to attention) as the token mixing mechanism.

Reference:

  • https://arxiv.org/abs/2405.21060 (mamba2 paper)

berlino avatar Oct 11 '24 09:10 berlino