intel/auto-round

Pull Requests Commits

update with load_kwargs

xin3he committed 24 days ago

65ae337b

support gpt-oss mxfp4 directly loading

xin3he committed 26 days ago

81399284

Implement MXFP4Handler MXFP8Handler NVFP4Handler (#1385)

xin3he committed 26 days ago

Verified f7db867a

fix layer config (#1373)

wenhuach21 committed 26 days ago

Verified 4064c4f1

[Regression] Detach scale tensor to prevent holding computation graph references (#1389)

xin3he committed 26 days ago

Verified 1c758f45

fix cuda ut fail (#1370)

n1ck-guo committed 26 days ago

Verified 6cdef7a2

Support general MOE replacement for MOE models (Transformers 5.0 compatible) (#1374)

lvliang-intel committed 27 days ago

Verified 28839a9a

Disable replace `FP8Expert` (#1379)

yiliu30 committed 27 days ago

Verified 137da551

fix --device_map cuda and xpu issue (#1383)

wenhuach21 committed 27 days ago

Verified f132c0c4

rm duplicate args of the quantization extra config (#1334)

WeiweiZhang1 committed 27 days ago

Verified 7e4ba88a

fix exporting fp8 static for format auto_round (#1381)

n1ck-guo committed 27 days ago

Verified 93f7703a

Standardize comment language in utils.py (#1378)

shantoislamdev committed 27 days ago

Verified 108f2b58

Add Links; Add step_by_step_CN; Fix several grammatical issues (#1366)

ZaneMark committed 28 days ago

Verified 3b7b87ec

Extensible Module Weight Type Conversion Framework. (#1375)

xin3he committed 28 days ago

Verified 53f20cf1

Fix FP8 MLA (#1343)

yiliu30 committed 28 days ago

Verified db77f3b7

feat: handle per-tensor FP8 dequantization for Devstral models (#1356)

SwekeR-463 committed 29 days ago

Verified eddb6de2

fix copy python file issue (#1372)

wenhuach21 committed 31 days ago

Verified 29956d91

add support for w4a16_mixed (#1326)

n1ck-guo committed 31 days ago

Verified 7cff6af5

fix cuda ut fail for transformers v5.0 (#1357)

n1ck-guo committed 31 days ago

Verified eb2ccdc3

support qwen3-next/glm4-moe/deepseekv3 monkey patch for transformers 5 (#1367)

wenhuach21 committed 31 days ago

Verified 0c0a0c1d

support qwen3-moe/glm-4.7-flash monkey patch for transformers 5 (#1354)

wenhuach21 committed 32 days ago

Verified 6bc1d344

fix bug of exporting fp8 static (#1361)

n1ck-guo committed 32 days ago

Verified b04f69c4

Refactor FP8 dequantization and detection using registry pattern (#1348)

scopophobic committed 32 days ago

Verified eaabde4a

refactor auto_round api, using local symbol table to pass all args (#1358)

n1ck-guo committed 32 days ago

Verified 528c9c82

Autoround in vLLM Office Hours (#1322)

yiliu30 committed 33 days ago

Verified b3f6f1fb

Fix cpu ut for transformers v5 (#1333)

Kaihui-intel committed 33 days ago

Verified f6e7e303

Fix CPU CI (#1344)

XuehaoSun committed 33 days ago

Verified 8eec5d8a

add copilot-instructions (#1341)

xin3he committed 34 days ago

Verified 0fe5c0a7

refactor eval and add UT (#1324)

xin3he committed 35 days ago

Verified 14eacb1f

Add Chinese README (#1335)

ZaneMark committed 35 days ago

Verified 71b385d2

Older