Markus Hartikainen b0e21b3083 [ROCm][Kimi-K3] Add opt-in gfx942 MXFP4-to-int4 conversion (#51274) ...

Signed-off-by: Markus Hartikainen <markus.hartikainen@amd.com>
Signed-off-by: Hartikainen <mahartik@amd.com>

2026-10-03 14:28:02 +08:00

..

__init__.py

[Frontend] new online quantization frontend (#38138)

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

base.py

[ROCm][Kimi-K3] Add opt-in gfx942 MXFP4-to-int4 conversion (#51274)

2026-10-03 14:28:02 +08:00

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

[Bugfix] Bind routed-experts capture to the MoE layer, not the kernel (#59455)

2026-10-02 15:50:38 +00:00

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