perf(qwen3.5): reuse packed GDN metadata across layers (#3705)
* perf(qwen3.5): reuse packed GDN metadata across layers Co-authored-by: Zhiqi Li <zhiqil@nvidia.com> Signed-off-by: Yuhe Zhang <yuhez@nvidia.com> * test(qwen3.5): keep packed metadata tests CPU-safe Signed-off-by: Yuhe Zhang <yuhez@nvidia.com> --------- Signed-off-by: Yuhe Zhang <yuhez@nvidia.com> Co-authored-by: Zhiqi Li <zhiqil@nvidia.com>
Y
Yuhe Zhang committed
f40e0161901f291899dccbfde6c78335016af769
Parent: 37ff7fe
Committed by GitHub <noreply@github.com>
on 8/28/2026, 10:19:57 PM