Skip to content

397B v7x RL trainer: all GDN, MoE and engine speedups (measured + pending), ordered by impact - #5762

Open
khatwanimohit wants to merge 19 commits into
atwigg/mlperffrom
mohit/trainer-perf-v7x-all-stack
Open

khatwanimohit wants to merge 19 commits into
atwigg/mlperffrom
mohit/trainer-perf-v7x-all-stack

Address review: skip the query kernel when Attention has none (fused_…

dd546aa
Select commit
Loading
Failed to load commit list.