intel/auto-round

Pull Requests Commits

Merge branch 'main' into xinhe/10-23a

xin3he committed 6 hours ago

Verified 3c128fe6

fix: update scale handling in QuantizedKVParameterCache to use amax of all batches (#2533)

xin3he committed 6 hours ago

Verified 512872d0

remove CI trigger for cuda nightly (#2541)

chensuyue committed 13 hours ago

Verified 90cbfe45

refactor: extract extensible shared smoothing engine (#2452)

changwangss committed 1 day ago

Verified 9cc08ca7

feat: export SVDQuant NVFP4 checkpoints for vLLM-Omni (#2445)

changwangss committed 1 day ago

Verified 7c0aed07

feat: enhance FP32 tensor handling in mixed-precision checkpoints and… (#2519)

xin3he committed 1 day ago

Verified 1d623e2a

refactor: Simplify vllm evaluation by setting multiprocessing method directly

xin3he committed 4 days ago

2a8a6932

feat: W4A8 ARK XPU MoE kernel (int4 weight / int8 compute) with prefill + decode (#2143)

Copilot committed 4 days ago

Verified 80b5f874

enhance: add source config preservation in save_model and related tests (#2522)

xin3he committed 4 days ago

Verified 710dcaef

feat: multi-GPU quantization improvements -- batched device-local sea… (#2357)

avtc committed 5 days ago

Verified a7656a27

enhance model_free functionality with strict parameter validation (#2520)

xin3he committed 5 days ago

Verified 95eccea1

update lfq acc data (#2529)

wenhuach21 committed 5 days ago

Verified 5ce470f4

fix lfq loss (#2527)

wenhuach21 committed 6 days ago

Verified 071829b0

fix: allocate calibration samples correctly when mixing datasets (#2526)

unknownhackerworld committed 6 days ago

Verified 436fa5a3

Fix ruff SIM117 in eval CLI test (#2523)

XuehaoSun committed 7 days ago

Verified c2dbac79

fix: keep escapes and quantified dots in to_standard_regex (#2418)

Shubham-Padkonde committed 7 days ago

Verified fd28e78b

Trigger CI test for RC branch (#2513)

chensuyue committed 7 days ago

Verified de2f5f06

Bump to v0.16.1 (#2517)

XuehaoSun committed 8 days ago

Verified cbdec9f3

fix: use spawn start method for vLLM evaluation to avoid CUDA re-initialization error (#2421)

reginaldalfret committed 8 days ago

Verified 234ac7af

fix: preserve calibration dataset on memory errors (#2433)

Shubham-Padkonde committed 8 days ago

Verified fde1f580

feat: support LLMC/ModelOpt/Quark format NVFP4 -> MXFP4 conversion (#2492)

xin3he committed 8 days ago

Verified 7222a20e

Support Nemotron-H Omni wrappers (Nemotron 3.5 Super VL, Bolt 74B-A9B) in plain AutoRound calls (#2512)

SergiioB committed 10 days ago

Verified ec10a649

feat: implement Kimi remote MoE with grad-aware inference and add uni… (#2496)

xin3he committed 11 days ago

Verified 88dc729d

remove leaderboard entry for now as it's still not very stable (#2516)

wenhuach21 committed 11 days ago

Verified 650981e0

docs(examples): align recipes with recommended W2 and batch size settings (#2499) (#2511)

vishvaRam committed 11 days ago

Verified a0134a4c

change slogan (#2509)

wenhuach21 committed 12 days ago

Verified 08d7085e

[Experimental]Add generic HF dataset support and two more datasets (#2363)

wenhuach21 committed 12 days ago

Verified e31b7728

fix: adjust tolerance levels for performance metrics comparison (#2505)

XuehaoSun committed 12 days ago

Verified 2c377483

fix: quantize all diffusion transformers in RTN routes (#2497)

changwangss committed 12 days ago

Verified 9645350f

Add clear entry for leaderboard (#2500)

wenhuach21 committed 12 days ago

Verified 523251eb

Older