biondizzle 5bb78564f5 Remove dynamic tensor allocation in scale assembly (cudagraph fix) ...

Removed torch.zeros() call that created padded_expert_offsets during
scale assembly. Now uses fixed layout computed from Python constants.
Also removed dead reference to padded_expert_offsets variable.

2026-05-17 14:01:32 +00:00

..

patches

Fix warmup: pass local expert IDs (not global), remove incorrect _warmup_done guard

2026-05-17 11:38:19 +00:00

nvfp4_cutedsl.py

Remove dynamic tensor allocation in scale assembly (cudagraph fix)

2026-05-17 14:01:32 +00:00