intel/auto-round

Pull Requests Commits

feat: implement resume functionality for model-free compression (#2330)

xin3he committed 26 days ago

Verified 9e2650f4

Update README.md (#2343)

wenhuach21 committed 26 days ago

Verified 51003909

Keep the original checkpoint key names and fix Qwen4 vLLM inference (#2327)

wenhuach21 committed 26 days ago

Verified aba0abdd

feat: add opt-in diffusion tuning cache prefetch (#2340)

changwangss committed 26 days ago

Verified 585d2065

fix: recreate calibrator for secondary diffusion transformers (#2335)

changwangss committed 26 days ago

Verified cea5abd7

fix: preserve layer config for secondary diffusion transformers (#2337)

changwangss committed 26 days ago

Verified c78c11a6

Improve the low_cpu_mem_usage usage in the doc (#2278)

hshen14 committed 26 days ago

Verified 32369a69

XPU oneDNN: bind matmul execution to live SYCL queue (graph-capture safe) (#2280)

Copilot committed 27 days ago

Verified 79516ff3

feat: add transformers checkpoint conversion mapping support for missing tensors (#2328)

xin3he committed 27 days ago

Verified 1139c323

feat: add support for WOQ fake quantization in model-free mode and update… (#2329)

xin3he committed 27 days ago

Verified bdc60475

refactor: reorganize tensor utility functions and enhance split_fused_expert_tensors for special cases (#2320)

xin3he committed 27 days ago

Verified c18bf770

fix nemoron gate inference issue (#2325)

wenhuach21 committed 27 days ago

Verified 0ea7c93c

fix: pin Hugging Face snapshots for AutoScheme offload reloads (#2318)

n1ck-guo committed 27 days ago

Verified 34c385ad

[high risk]Speedup 2.5-6X moe tuning and pin ngram to cpu (#2303)

wenhuach21 committed 28 days ago

Verified 678f8f5f

fix: update artifact name format for ark wheel download (#2321)

chensuyue committed 28 days ago

Verified b1738032

Implement AI-Powered CI Failure Analysis and Fix Recommendations (#2277)

chensuyue committed 28 days ago

Verified 18486c39

reduce cuda ut runing time (#2266)

n1ck-guo committed 28 days ago

Verified bdb0d31f

support model_type == "inkling_mm_model" (#2214)

xin3he committed 29 days ago

Verified de801f9f

fix: defer packing for multi-transformer diffusion (#2314)

changwangss committed 29 days ago

Verified b4dc0cc7

fix: update quantization configuration handling in FakeFormat and related tests (#2311)

xin3he committed 29 days ago

Verified 777c4ea7

Optimize diffusion calibration step sampling (#2265)

changwangss committed 30 days ago

Verified ce90add6

add sdxl svdquant (#2251)

changwangss committed 30 days ago

Verified 1c3a5518

fix: handle mantissa round overflow (#2302)

xin3he committed 30 days ago

Verified 9478fcf8

support 5,6,7 bits on cuda and add humming backend (#2281)

wenhuach21 committed 30 days ago

Verified 1f23576a

Fix xpu CI retry issue (#2309)

chensuyue committed 30 days ago

Verified fee969dd

Update CI/CD dockerfile (#2294)

XuehaoSun committed 30 days ago

Verified 726becc3

[risky]reduce ram usage of loading moe without monkey patch and fix qwen3-flash-next group_size issue (#2291)

wenhuach21 committed 32 days ago

Verified 7d8905ef

Optimize INT4 MoE prefill/decode (sym) XPU kernels with dedicated w4a16 tiling (#2111)

Copilot committed 32 days ago

Verified 21bab556

update awq detail data of fake eval (#2285)

WeiweiZhang1 committed 32 days ago

Verified 3942b57e

Allow non-MLLM calibration datasets when not quantizing vision (#2065)

aquilarubra committed 32 days ago

Verified 63dbc8ba

Older