intel/auto-round

Pull Requests Commits

feat: Add support for cross-attention projections in NVFP4 fused groups and enhance unit tests

xin3he committed 2 days ago

eace9e97

feat: Enhance FP8 tensor handling and add cross-shard global scale computation

xin3he committed 2 days ago

78e057de

[pre-commit.ci] auto fixes from pre-commit.com hooks

pre-commit-ci[bot] committed 3 days ago

4bcf58e0

Enhance NVFP4 E5M3 GS quantization with global scale overrides, remove is_deepseek_v32_ue8m0

xin3he committed 3 days ago

9d810fb0

[pre-commit.ci] auto fixes from pre-commit.com hooks

pre-commit-ci[bot] committed 5 days ago

cb734cdb

Enhance model-free quantization with improved error handling and shard management

xin3he committed 5 days ago

8be22508

[Research Feature]add naive qat/qad (#2548)

wenhuach21 committed 5 days ago

Verified 57b8648d

support AWQ sample control and optimize AWQ+ARV2 for INT WOQ (#2484)

WeiweiZhang1 committed 7 days ago

Verified 98ad5bbc

fix: preserve Z-Image text encoder export layout (#2559)

changwangss committed 7 days ago

Verified fe68946d

[Bug fix] Release offload temp-file mmap to fix cleanup issue (#2487)

lvliang-intel committed 7 days ago

Verified e1ab33b8

Reset context singletons between tests (#2390)

Shubham-Padkonde committed 8 days ago

Verified d2a25f1b

update cli and simplify --help (#2530)

n1ck-guo committed 8 days ago

Verified aca358a1

Enhance AI analysis and PR comment handling (#2538)

chensuyue committed 8 days ago

Verified c1d97369

[Bug fix]: test/test_mlx/test_mlx_format.py failed on Mac OS (#2554)

xin3he committed 8 days ago

Verified 9ffe65f7

Fix native AutoRound INT4 packing with FP8 layer overrides (#2543)

changwangss committed 8 days ago

Verified 3e186445

Prefetch next batch to CUDA for SignRound v1/v2 under low_gpu_mem_usage (#2549)

wenhuach21 committed 8 days ago

Verified 3f86dea7

Update stale issue and PR handling to 60 days of inactivity (#2553)

chensuyue committed 8 days ago

Verified e4353de0

feat: Add NVFP4_E5M3_GS scheme with global scale support (#2531)

xin3he committed 10 days ago

Verified 91a9bbfc

refactor: Simplify vllm evaluation by setting multiprocessing method directly (#2534)

xin3he committed 10 days ago

Verified e2e3f17f

Reset Inductor Pool After SGLang Shutdown (#2536)

XuehaoSun committed 10 days ago

Verified afac833e

Skip zero-point round-trip with compressed-tensors 0.20.0 (#2544)

changwangss committed 10 days ago

Verified 76a114e5

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

xin3he committed 10 days ago

Verified 512872d0

remove CI trigger for cuda nightly (#2541)

chensuyue committed 11 days ago

Verified 90cbfe45

refactor: extract extensible shared smoothing engine (#2452)

changwangss committed 12 days ago

Verified 9cc08ca7

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

changwangss committed 12 days ago

Verified 7c0aed07

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

xin3he committed 12 days ago

Verified 1d623e2a

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

Copilot committed 14 days ago

Verified 80b5f874

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

xin3he committed 14 days ago

Verified 710dcaef

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

avtc committed 16 days ago

Verified a7656a27

enhance model_free functionality with strict parameter validation (#2520)

xin3he committed 16 days ago

Verified 95eccea1

Older