biondizzle b81200f427 Fix CuTeDSL NVFP4 linear: correct scale assembly in custom op ...

- pad_and_swizzle_single takes 1 arg (2D tensor), not 4
- Inline the scale assembly logic: pad x_sf → swizzle → unsqueeze for 1 group
- Remove unused CuTeDSLNvfp4Linear import from custom op impl

2026-05-19 01:01:42 +00:00

..

nvfp4

Fix CuTeDSL NVFP4 linear: correct scale assembly in custom op

2026-05-19 01:01:42 +00:00