Skip to content

Pull requests: intel/auto-round

New pull request New
35 Open 1,451 Closed
35 Open 1,451 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Reviews
Filter by reviews
No reviews Review required Approved review Changes requested
Assignee
Filter by who’s assigned
Assigned to nobody Loading
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Pull requests list

Add more test cases to improve CUDA UT coverage
#2159 opened Aug 14, 2026 by lvliang-intel Contributor Loading…
1 of 4 tasks
2
feat: add support for FP8 model loading and ragged-tile dequantization fallback
#2158 opened Aug 14, 2026 by xin3he Contributor Loading…
4 tasks
11
refactor uts, remove duplicate uts and build common uts for all devices to increase coverage
#2157 opened Aug 14, 2026 by n1ck-guo Contributor Loading…
1 of 9 tasks
ark: add BF16/FP16 Sparge variants
#2145 opened Aug 11, 2026 by yiliu30 Contributor • Draft
4 tasks
feat: W4A8 ARK XPU MoE kernel (int4 weight / int8 compute) with prefill + decode
#2143 opened Aug 11, 2026 by Copilot AI • Draft
2 of 4 tasks
@a32543254
refactor algorithm cli
#2140 opened Aug 11, 2026 by n1ck-guo Contributor Loading…
4 tasks
0.16.0
2
Tthe LLM Compressor export modification in mixed‑precision mode: the original format unifies all layers as Linear, causing vLLM inference to fail during parsing.
#2128 opened Aug 5, 2026 by 1607662794 Loading…
2 tasks
6
[Experimental] Add NVFP4 E5M3 support and related components for model-free quantization
#2123 opened Aug 5, 2026 by xin3he Contributor Loading…
4 tasks
0.16.0
26
fix: recognize Modular Diffusers pipelines as diffusion models
#2120 opened Aug 3, 2026 by Ar4ikov Loading…
2 of 4 tasks
3
fix: save the config of diffusers models on export
#2119 opened Aug 3, 2026 by Ar4ikov Loading…
2 of 4 tasks
Optimize INT4 MoE prefill/decode (sym) XPU kernels with dedicated w4a16 tiling
#2111 opened Jul 31, 2026 by Copilot AI • Draft
1 of 4 tasks
@a32543254
ark.moe: switch auto dispatch to total-token threshold + env override
#2108 opened Jul 31, 2026 by Copilot AI • 2/2 Loading…
2 of 4 tasks
@a32543254
Fix AutoRound gradients for Kimi-VL routed experts
#2097 opened Jul 28, 2026 by lesj0610 • Draft
add rotation ut for xpu.
#2094 opened Jul 27, 2026 by lkk12014402 Contributor Loading… 0.15.0
3
Opt the moe performance
#2090 opened Jul 27, 2026 by a32543254 Contributor • 1/2 Loading…
4 tasks
11
add blockwise rotation.
#2087 opened Jul 24, 2026 by lkk12014402 Contributor Loading…
15
perf: N-based tile selection for XPU bf16/fp16 MoE grouped GEMM
#2079 opened Jul 23, 2026 by Copilot AI • Draft
1 of 4 tasks
@a32543254
Allow non-MLLM calibration datasets when not quantizing vision
#2065 opened Jul 19, 2026 by aquilarubra Contributor Loading…
3
Stream targeted block re-quantization's calibration pass
#2064 opened Jul 19, 2026 by aquilarubra Contributor Loading…
Add crash/resume support for tuning runs (AR_RESUME_DIR)
#2062 opened Jul 19, 2026 by aquilarubra Contributor Loading…
fix the int4 ark test
#2038 opened Jul 10, 2026 by a32543254 Contributor Loading…
4 tasks
2
test: stabilize flaky int4-sym fp16 MoE prefill accuracy test
#2033 opened Jul 9, 2026 by Copilot AI • Draft
4 tasks done
@a32543254
44
[WIP]refine quantizer
#2031 opened Jul 9, 2026 by wenhuach21 Contributor Loading…
4 tasks
1
Enhance CI pipeline and unit tests for AI failure analysis WIP
#2029 opened Jul 9, 2026 by chensuyue Contributor Loading…
Implementing bandwidth metrics for performance tests
#2020 opened Jul 8, 2026 by Copilot AI • Draft
@a32543254
1
ProTip! Type g p on any issue or pull request to go back to the pull request listing page.