intel/auto-round

Pull Requests Commits

lvliang-intel committed 24 days ago

c2bd5232

draft sa fp8 support

lvliang-intel committed 26 days ago

7e1741ff

fix: AutoScheme embedding budget and gguf mixed-bits config for multimodal models (#2055)

n1ck-guo committed 27 days ago

Verified 06c33eae

Stabilize MoE prefill accuracy test (#2060)

a32543254 committed 27 days ago

Verified 98ad266f

fix: add triton to requirements for LLMC CPU tests (#2054)

XuehaoSun committed 32 days ago

Verified c774c83b

enhance logging and data loading for AutoScheme and model-free compatibility (#2050)

xin3he committed 33 days ago

Verified 6f45d221

fix: handle router_logits tuple for compatibility with transformers >= 5.13.0 (#2046)

xin3he committed 33 days ago

Verified 4b8468ac

Remove transformers version limit (#2051)

chensuyue committed 33 days ago

Verified 0ea53be4

Fix invalid group reference in revert_checkpoint_conversion_mapping (#2035)

A-S-Manoj committed 33 days ago

Verified 6d1c6db6

Update torch version to 2.13.0 (#2045)

XuehaoSun committed 34 days ago

Verified 32da6a9a

Support auto-round-lib build with oneAPI 2026.0 and work with pytorch xpu 2.13 (#2043)

chensuyue committed 34 days ago

Verified 4d24c405

fix gradient_accumulate_steps cli issue (#2041)

WeiweiZhang1 committed 37 days ago

Verified d357f0d1

Remove transformers version limit for release package (#2030)

chensuyue committed 38 days ago

Verified 5e586bcb

bug-fix: implement dedicated cache for streaming source shards (#2021)

xin3he committed 38 days ago

Verified ffd3dfe7

fix: CLI ignore --scheme when bits/act_bits are unset (#2019)

n1ck-guo committed 38 days ago

Verified c70e99f6

fix the mismatch between weight and scale in gguf format (#2014)

n1ck-guo committed 38 days ago

Verified b482d150

Fix checkpoint remap (#2001)

yiliu30 committed 38 days ago

Verified ec73a8dd

Add moe prefill/ decode with int2/int4/int8 sym /asym and fp8 e4m3 e5m2 (#1813)

Copilot committed 39 days ago

Verified 74143702

[pre-commit.ci] pre-commit autoupdate (#2004)

pre-commit-ci[bot] committed 40 days ago

Verified 731a4312

[ARK] Support Dense Gemm using sycl-tla (#1968)

Zhenzhong1 committed 40 days ago

Verified 47822429

[ARK] feat: expose smooth_k (use_mean_bias) parameter from Python to C++ in SAGE attention kernels (#1995)

luoyu-intel committed 40 days ago

Verified d55ba004

bump release version to 0.14.0 (#2002)

chensuyue committed 41 days ago

Verified 24bfce00

update rotation doc. (#1999)

lkk12014402 committed 41 days ago

Verified 5ee45db7

pin transformers (#1998)

yiliu30 committed 41 days ago

Verified fcd1c854

Fix cuda ut fail caused by used instructions for base model (#1994)

n1ck-guo committed 41 days ago

Verified 43f443e6

feat: add return_lse support for SDPA and SAGEV1 kernel — enable sequence-parallel (Ring Attention) merge (#1991)

luoyu-intel committed 43 days ago

Verified b481cebb

Enable auto-round-bot for CI comments (#1993)

chensuyue committed 44 days ago

Verified b63a89f5

Implement AutoScheme support for model-free packing with validation a… (#1990)

xin3he committed 44 days ago

Verified cd10607c

Update CUDA UT to cu130 (#1992)

XuehaoSun committed 44 days ago

Verified 52564823

[AutoScheme] fix lm_head no_grad issue and support MOE model with shared mix_score (#1971)

xin3he committed 44 days ago

Verified e6b35d13

Older