yangjianfengo1 93fcf7e4ec 【New Feature】W4afp8 supports per group quantization (#4272) ...

* w4afp8 支持per group

* code style

* 精度完成

* revert append attn utils

* ffn1 动态量化

* ffn2 支持动态量化

* code style

* code style

* 修改单测

* 修改单测

* fix bug

* Implement conditional parameter creation for layers

Add parameter creation for up_gate_proj_in_scale when ep_size > 1.

* code style

* fix conflict

* code style

* code style

* 修复w4aint8 精度

* fix ci

---------

Co-authored-by: yuanxiaolan <yuanxiaolan01@baidu.com>

2025-11-05 21:00:23 +08:00

..

cpu_ops

c++ code format (#4527)

2025-10-22 17:59:50 +08:00

gpu_ops

【New Feature】W4afp8 supports per group quantization (#4272)

2025-11-05 21:00:23 +08:00

iluvatar_ops

c++ code format (#4527)

2025-10-22 17:59:50 +08:00

metax_ops

[Metax] adapt cutlass moe and fix mla attention (#4602)

2025-11-05 10:03:49 +08:00

third_party

[setup optimize]Support git submodule (#4033)

2025-09-11 17:41:16 +08:00

utils

【New Feature】W4afp8 supports per group quantization (#4272)

2025-11-05 21:00:23 +08:00

xpu_ops

[XPU] fix thinking bug where output only contains reasoning_content (#4761)

2025-11-04 14:32:35 +08:00

0001-DeepGEMM-95e81b3.patch

[feat] support fa3 backend for pd disaggregated (#2695)

2025-07-03 22:33:27 +08:00

MANIFEST.in

[LLM] First commit the llm deployment code

2025-06-09 19:20:15 +08:00

setup_ops_cpu.py

polish code with new pre-commit rule (#2923)

2025-07-19 23:19:27 +08:00

setup_ops.py

[Metax] adapt cutlass moe for ernie-vl (#4685)

2025-11-03 17:44:27 +08:00