Pull Requests intel/auto-round

[Important Change] change mxfp4 to OCP definition

#1200 by wenhuach21 was merged 2025-12-26 03:30

fix llama4 kv quant vllm-ext

#1199 by mengniwang95 was merged 2025-12-26 05:35

add online quant & offline sgl api generation cuda UT

#1198 by WeiweiZhang1 was merged 2025-12-26 06:54

fix llama4 ut

#1197 by WeiweiZhang1 was merged 2025-12-25 07:02

Update version

#1191 by XuehaoSun was merged 2025-12-25 06:11

fix cuda ut fail

#1189 by n1ck-guo was merged 2025-12-25 04:54

fix scale_dtype is None issue

#1185 by xin3he was merged 2025-12-23 13:28 0.9.3

Revert "[Important Update]change mxfp4 to ocp standard definition and fix scale dtype issue"

#1184 by xin3he was merged 2025-12-23 07:43

[Important Change]change mxfp8 to OCP standard

#1183 by wenhuach21 was merged 2025-12-26 03:30

Enable qwen3 vl moe quant and load

#1182 by WeiweiZhang1 was merged 2025-12-25 01:48

[Important Update]change mxfp4 to ocp standard definition and fix scale dtype issue

#1181 by wenhuach21 was merged 2025-12-23 05:04

Add FP8 KV Support for DS vllm-ext

#1180 by yiliu30 was merged 2025-12-26 11:07 1.0.0

Enable ark CI model test

#1177 by chensuyue was merged 2025-12-23 07:13 0.9.3

reset enable_torch_compile to `False` as nvfp4 is enabled

#1176 by xin3he was merged 2025-12-22 05:05 0.9.3

fix device mismatch issue for unifying qkv scale

#1175 by xin3he was merged 2025-12-23 08:03 0.9.3

Deprecate awq infer ut

#1171 by WeiweiZhang1 was merged 2025-12-22 06:38 0.9.3

Fix inference issue in ARK for AWQ format

#1170 by luoyu-intel was merged 2025-12-23 02:10 0.9.3

rewrite fill_default_value logic for robust

#1165 by xin3he was merged 2025-12-26 01:35

Fix accuracy bug in the customized dataset

#1162 by wenhuach21 was merged 2025-12-19 08:45 0.9.3

Fix code for llmc llama4 quantization

#1161 by mengniwang95 was merged 2025-12-23 06:16 0.9.3

UT refactor for preview

#1160 by xin3he was merged 2025-12-26 01:34

[STEP 1] refact output formats ready

#1159 by n1ck-guo was merged 2025-12-25 02:51

fix asym quantization issue

#1158 by wenhuach21 was merged 2025-12-18 12:36 0.9.3

Revert "[STEP 1] refact output formats"

#1157 by n1ck-guo was merged 2025-12-18 08:57

handle key not found issue

#1156 by xin3he was merged 2025-12-19 02:10 0.9.3

update norm bias tuning doc

#1155 by wenhuach21 was merged 2025-12-18 13:37

Refactor module replacement ready

#1153 by yiliu30 was merged 2025-12-25 07:08

Fix cuda test and llmc test

#1152 by XuehaoSun was merged 2025-12-18 02:19

update cd workflow

#1151 by chensuyue was merged 2025-12-17 05:53

Add G2-specific `FP8_STATIC` support

#1148 by yiliu30 was merged 2025-12-19 03:51