Unified delta net handling for Qwen3Next and Kimi Linear models #18792
Unified delta net handling
b3f55ead
Remove old methods.
34e1ed90
ngxson
commented
on 2026-01-12
Refactor and optimize
f98f2856
Adapt autoregressive version from @ymcki
08b1ed86
Change to decay mask approach
e9ad1849
Assignees
No one assigned
Login to write a write a comment.
Login via GitHub