Skip to content

cuda: tree attention and DFlash block attention for every family, checked against float64 - #746

Open
jschmied wants to merge 13 commits into
ashhart:mainfrom
jschmied:cuda-attn-shared
Open

cuda: tree attention and DFlash block attention for every family, checked against float64#746
jschmied wants to merge 13 commits into
ashhart:mainfrom
jschmied:cuda-attn-shared

cuda/tree_attention: a NaN or infinite partial runs again with those …

942e2fe
Select commit
Loading
Failed to load commit list.
Sign in for the full log view
zig
Started 2026-10-18 06:45:29 ago