Fangzhou AiandCursor Agent 77e8b02c8d [ROCm][Perf] Add opt-in a4w4 (FP4 activation) MoE for DeepSeek V4.1 on AITER (#58819) ...

Signed-off-by: fai <fangzhouai@gmail.com>
Signed-off-by: Fangzhou Ai <31551580+Fangzhou-Ai@users.noreply.github.com>
Co-authored-by: Cursor Agent (Claude Sonnet 5) <cursoragent@cursor.com>

2026-09-30 15:07:44 +08:00

..

__init__.py

[Frontend] new online quantization frontend (#38138)

2026-04-03 11:58:39 -04:00

base.py

[CI] Ignore ruff D209, rejoin the docstrings it split, and silence incompatible-rule warnings (#57212)

2026-09-16 10:57:24 -07:00

fp8.py

[CPU] Add CPU FP8 W8A8 linear/MoE support (#49942)

2026-09-18 17:59:42 +08:00

int8.py

[Quantization] Support online quantization with partially pre-quantized checkpoints (#51392)

2026-09-04 11:46:06 -07:00

moe_base.py

[MoE] Defer the TRTLLM-Gen top-k finalize on the modular path (#58635)

2026-09-25 15:38:29 +08:00

mxfp4.py

[ROCm][Perf] Add opt-in a4w4 (FP4 activation) MoE for DeepSeek V4.1 on AITER (#58819)

2026-09-30 15:07:44 +08:00

mxfp8.py

[Bugfix][Quantization] Fix MXFP8 startup crash on layers below mm_mxfp8 shape limits (#54223)

2026-09-25 02:54:11 +00:00

nvfp4.py

[Bugfix][Quantization] Refresh online NVFP4 scales before reload post-processing (#57954)

2026-09-24 18:36:08 -07:00