biondizzle d9bae6d770 Fix OOB in scale assembly: size padded_x_sf for max tokens, fix top_k/max_num_tokens passing, support variable-size expert blocks ...

Bug 9: padded_x_sf was sized for num_experts*128 rows, but with 8192 tokens
and top_k=6, the actual padded row count can exceed 6144. Also:
- Pass top_k and max_num_tokens from deepseek_v4.py (was defaulting to 8/8192)
- Phase 2 of scale assembly now handles experts with >128 tokens (multiple 128-row chunks)
- Remove debug prints

2026-05-17 09:56:28 +00:00

..

patches

Fix OOB in scale assembly: size padded_x_sf for max tokens, fix top_k/max_num_tokens passing, support variable-size expert blocks

2026-05-17 09:56:28 +00:00

nvfp4_cutedsl.py

Fix OOB in scale assembly: size padded_x_sf for max tokens, fix top_k/max_num_tokens passing, support variable-size expert blocks

2026-05-17 09:56:28 +00:00