intel/auto-round

Pull Requests Commits

lvliang-intel committed 4 days ago

2fef4c2b

Merge branch 'main' into lvl/support_omni

lvliang-intel committed 4 days ago

Verified fb0ae294

[pre-commit.ci] auto fixes from pre-commit.com hooks

pre-commit-ci[bot] committed 4 days ago

913a9935

Support Qwen3 Omni model quantization

lvliang-intel committed 4 days ago

f19650d6

support llm dynamic wint8aint8 export (#1376)

mengniwang95 committed 4 days ago

Verified c9f6155b

align scale for MLA kv (#1395)

yiliu30 committed 4 days ago

Verified a267625e

Support quant for `meituan-longcat/LongCat-Flash-Lite` (#1388)

Kaihui-intel committed 4 days ago

Verified a12e3f5a

Fix version (#1393)

XuehaoSun committed 4 days ago

Verified 68d07bfa

Implement MXFP4Handler MXFP8Handler NVFP4Handler (#1385)

xin3he committed 4 days ago

Verified f7db867a

fix layer config (#1373)

wenhuach21 committed 4 days ago

Verified 4064c4f1

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

xin3he committed 4 days ago

Verified 1c758f45

fix cuda ut fail (#1370)

n1ck-guo committed 4 days ago

Verified 6cdef7a2

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

lvliang-intel committed 5 days ago

Verified 28839a9a

Disable replace `FP8Expert` (#1379)

yiliu30 committed 5 days ago

Verified 137da551

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

wenhuach21 committed 5 days ago

Verified f132c0c4

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

WeiweiZhang1 committed 5 days ago

Verified 7e4ba88a

fix exporting fp8 static for format auto_round (#1381)

n1ck-guo committed 5 days ago

Verified 93f7703a

Standardize comment language in utils.py (#1378)

shantoislamdev committed 5 days ago

Verified 108f2b58

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

ZaneMark committed 6 days ago

Verified 3b7b87ec

Extensible Module Weight Type Conversion Framework. (#1375)

xin3he committed 6 days ago

Verified 53f20cf1

Fix FP8 MLA (#1343)

yiliu30 committed 6 days ago

Verified db77f3b7

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

SwekeR-463 committed 7 days ago

Verified eddb6de2

fix copy python file issue (#1372)

wenhuach21 committed 9 days ago

Verified 29956d91

add support for w4a16_mixed (#1326)

n1ck-guo committed 9 days ago

Verified 7cff6af5

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

n1ck-guo committed 9 days ago

Verified eb2ccdc3

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

wenhuach21 committed 9 days ago

Verified 0c0a0c1d

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

wenhuach21 committed 10 days ago

Verified 6bc1d344

fix bug of exporting fp8 static (#1361)

n1ck-guo committed 10 days ago

Verified b04f69c4

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

scopophobic committed 10 days ago

Verified eaabde4a

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

n1ck-guo committed 10 days ago

Verified 528c9c82

Older