intel/auto-round

Pull Requests Commits

fix: enhance model cache status checks with detailed logging and reasons

xin3he committed 1 day ago

0bfa314e

fix: add patch for FP8 dequantization to handle ragged tiles in transformers

xin3he committed 1 day ago

07aa43e4

fix: set default value for source_quant_config in _process_single_shard_task

xin3he committed 1 day ago

5ea4e1ff

feat: enhance support for Kimi-K25 INT4 format with additional packing options and unpacking logic

xin3he committed 1 day ago

1ed28454

Add support for kimi_k25 INT4 packed source models and related tests

xin3he committed 1 day ago

de5a658d

feat: enable search scale for mxfp in model-free (#2141)

xin3he committed 2 days ago

Verified 9352659f

Add timeout decorators to improve test case stability (#2155)

XuehaoSun committed 2 days ago

Verified cee71092

refine eval, add more option (#2114)

WeiweiZhang1 committed 2 days ago

Verified fe7676de

Migrate cpu sdpa from NS flash attn (#2092)

jijiaz committed 2 days ago

Verified 7402af81

Refactor unit test scripts (#2148)

XuehaoSun committed 2 days ago

Verified 506bd5fa

fix: prevent quantized layers from being reinitialized by Transformers (#2147)

n1ck-guo committed 2 days ago

Verified a5e7393e

feat: add per-head fp8 kv attention calibration (#2044)

zeel2104 committed 5 days ago

Verified f17d9cd4

fix performance regression bug (#2136)

n1ck-guo committed 5 days ago

Verified f8ad63f4

Enhance unit tests with timeout handling and stability improvements (#2135)

XuehaoSun committed 6 days ago

Verified e4a02ccd

Add python 3.14t compatibility tests (#2110)

XuehaoSun committed 8 days ago

Verified 98f2cdb9

support one more mxfp4 variant (#2132)

wenhuach21 committed 8 days ago

Verified f706da67

Add llmc test back (#2131)

yiliu30 committed 8 days ago

Verified 1f6b2151

fix sparge on torch2.13 (#2130)

yiliu30 committed 9 days ago

Verified d8a30059

Refator UT and improve code coverage (#2057)

lvliang-intel committed 9 days ago

Verified 7a75218a

[fix] XPU crash due to circle call of nonzero() and index_select() (#2117)

xin3he committed 9 days ago

Verified 28003697

Enable torch.compile for selected tests and update related configurations (#2124)

xin3he committed 10 days ago

Verified a2f6b316

support fused MoE quantization and export for GGUF format (#2072)

n1ck-guo committed 10 days ago

Verified 1fca0c74

refactor scheme,layer_config and formats (#2013)

n1ck-guo committed 10 days ago

Verified 77e8ea59

Automatically select calibration datasets for code models (#2107)

changwangss committed 11 days ago

Verified f8de00c3

disable parallel for non low_gpu_mem_usage (#2116)

xin3he committed 12 days ago

Verified 417d87de

add opt_rtn support for mxfp4 and nvfp4 (#2100)

wenhuach21 committed 12 days ago

Verified 41cca72f

Fix: temporarily skip tests due to known issues with xfail annotations (#2113)

chensuyue committed 15 days ago

Verified 60b813cb

Enable AutoScheme parallel and disk stream by default (#2105)

xin3he committed 15 days ago

Verified 8a9a1d13

Fix: skip exporting fp8 attention q_max (#2103)

yiliu30 committed 15 days ago

Verified 2a267e72

fix MXFP model free auto-round format bug (#2106)

xin3he committed 15 days ago

Verified 378187c7

Older