intel/auto-round

Pull Requests Commits

support AWQ sample control and optimize AWQ+ARV2 for INT WOQ

WeiweiZhang1 committed 1 day ago

e747b8c3

fix: forward pipeline call kwargs through diffusion entry (#2448)

changwangss committed 6 days ago

Verified 63895ebf

fix: keep make_q3_quants sums in sync with accepted levels (#2439)

raashish1601 committed 6 days ago

Verified 7059a8b1

fix: correct Q5_K GGUF packing (#2437)

raashish1601 committed 6 days ago

Verified b73d89d3

fix: keep 0 in the range of GGUF asymmetric K-quant groups (#2438)

raashish1601 committed 6 days ago

Verified 5a113d1c

fix: give attn_v of LLM_TYPE_70B models Q5_K like llama.cpp (#2443)

raashish1601 committed 6 days ago

Verified ae21ef9b

Fix CUDA UT timm issue (#2444)

XuehaoSun committed 6 days ago

Verified f5cfe304

[pre-commit.ci] pre-commit autoupdate (#2436)

pre-commit-ci[bot] committed 6 days ago

Verified 87734fb0

chore: enable ruff default rules and fix violations (#2434)

XuehaoSun committed 7 days ago

Verified c915fdcf

fix: apply debug layer limits to nested configs (#2431)

Shubham-Padkonde committed 14 days ago

Verified 6afaecdb

docs: explain Windows Triton compilation prerequisites (#2432)

Shubham-Padkonde committed 14 days ago

Verified 8cef3d9b

Bump version and upgrade python version (#2429)

XuehaoSun committed 14 days ago

Verified 1bd739fc

add mxuint (#2425)

wenhuach21 committed 14 days ago

Verified 7c4b6e92

update README.md (#2428)

wenhuach21 committed 14 days ago

Verified 4286cf93

fix: return the repaired string from escape_invalid_json_backslashes (#2419)

Shubham-Padkonde committed 14 days ago

Verified ba12c951

Keep the case of the AR_WORK_SPACE path (#2417)

Shubham-Padkonde committed 14 days ago

Verified 11949312

fix: honor AR_WORK_SPACE for the AudioCaps dataset cache (#2420)

Shubham-Padkonde committed 14 days ago

Verified 1ace02c5

INC integration test bug fix (#2423)

chensuyue committed 15 days ago

Verified 5606837a

NeUQI grid search for optimized RTN (--enable_neuqi: joint asym scale/zp search, two-stage sym search, frozen-init anchor) (#2341)

avtc committed 18 days ago

Verified 44b48d60

fix(xpu): make ARK SDPA graph-capture safe for torch.xpu.XPUGraph (#2416)

luoyu-intel committed 20 days ago

Verified a4f01f85

bug fix: prune static kv/attention granularity keys when dtype not set (#2415)

yiliu30 committed 20 days ago

Verified a20f5297

enhance attention module with original implementation retrieval and context restoration (#2408)

xin3he committed 20 days ago

Verified 5eceba6c

bug fix: layer_config mismatch due to transformers rename; fix model free stream mode detection (#2411)

xin3he committed 20 days ago

Verified 24d14371

Fix CWE-22 (path traversal) in checkpoint loading issues (#2389)

lvliang-intel committed 21 days ago

Verified fe973526

fix: materialize checkpoint-backed cpu buffers (#2412)

n1ck-guo committed 21 days ago

Verified 6c4d95cb

fix: avoid excessive memory during calibration dataset casting (#2405)

changwangss committed 22 days ago

Verified 22d6e28c

Revert "refactor: decouple datatype quantization from algorithms (#2406)

n1ck-guo committed 22 days ago

Verified 86e97adf

fix: update get_predefined_ignore_layers to include shared_expert_gat… (#2397)

xin3he committed 23 days ago

Verified 6db9435f

fix: update tensor summary creation to directly use tensors_to_restor… (#2401)

xin3he committed 23 days ago

Verified bff3deab

ark: add BF16/FP16 Sparge variants and reduce compilation RAM (#2145)

yiliu30 committed 23 days ago

Verified 402b0edd

Older