[acceleration] Model-agnostic acceleration plugin layer (torch.compile / attention backend / feature caching) - #196
Open
Jayce-Ping wants to merge 16 commits into
Open
[acceleration] Model-agnostic acceleration plugin layer (torch.compile / attention backend / feature caching)#196Jayce-Ping wants to merge 16 commits into
Jayce-Ping wants to merge 16 commits into
Commits
Commits on Jun 27, 2026
- committed
- committed
- committed
- committed
[acceleration,trainers] fix: remove duplicate attention-backend accelerator; compile after post_init
committed- committed
- committed
Commits on Jun 28, 2026
Commits on Jun 30, 2026
- committed
- committed
- committed
- committed