Skip to content

Pull requests: xLLM-AI/xllm

New pull request New
109 Open 1,742 Closed
109 Open 1,742 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

fix(npu): order SHM H2D with graph replay
#2035 opened Jul 25, 2026 by pjgao Collaborator • Draft
bugfix: support Debian Mooncake build dependencies.
#2033 opened Jul 25, 2026 by DongheJin Collaborator Loading…
17 tasks
1
docs: update Flux2-dev quick start guide for Ascend A3.
#2032 opened Jul 25, 2026 by wang-shuibin Collaborator Loading…
9 of 17 tasks
support text_encoder for flux2 on device npu
#2031 opened Jul 24, 2026 by wang-shuibin Collaborator Loading…
9 of 17 tasks
bugfix: check nested submodules recursively before build.
#2030 opened Jul 24, 2026 by weizhehuang0827 Collaborator Loading…
1 task done
bugfix: synchronize speculative execution across DP ranks.
#2029 opened Jul 24, 2026 by DongheJin Collaborator Loading…
17 tasks
bugfix: restore npu layer synchronizer context wait stream.
#2025 opened Jul 23, 2026 by shifengmin Collaborator Loading…
3 tasks
bugfix: preserve raw dp token counts for empty shards.
#2024 opened Jul 23, 2026 by shifengmin Collaborator Loading…
3 tasks
feat: add flashcomm1 feature for deepseek v4 on NPU
#2022 opened Jul 23, 2026 by chenchuw886 • Draft
17 tasks
feat: support decode context parallel.
#2021 opened Jul 23, 2026 by phantomlei3 Collaborator • Draft
17 tasks
feat: add distill mode support for Wan2.2 I2V pipeline
#2020 opened Jul 23, 2026 by zhangjieshao98 Contributor Loading…
feat: support joy-image-edit-plus on NPU device & refact sequence parallel Mixin.
#2018 opened Jul 23, 2026 by shan-chen-feng Collaborator • Draft
17 tasks
refactor: eliminate DisaggPDChunkedPrefillScheduler.
#2017 opened Jul 23, 2026 by weizhehuang0827 Collaborator Loading…
8 of 13 tasks
feat(lora): request path adapter_id propagation + chat routing (2/2)
#2015 opened Jul 23, 2026 by cchh05 Collaborator Loading…
1
feat(lora): multi-tenant LoRA framework + attention/MLP wire-up (1/2)
#2014 opened Jul 23, 2026 by cchh05 Collaborator Loading…
1
feat: add mlu host kv cache support.
#2012 opened Jul 23, 2026 by phantomlei3 Collaborator • Draft
17 tasks
feat: add NPU device support for embedded-python model executor.
#2009 opened Jul 23, 2026 by yingxudeng Collaborator Loading…
17 tasks
6
feat: support DSV4 device prefix cache (SWA + C4 + C128).
#2008 opened Jul 22, 2026 by Kang-Meng Collaborator Loading…
3
feat: add verbose request tracing and structured API error responses.
#2007 opened Jul 22, 2026 by zhang-minchao Collaborator • Draft
6 of 22 tasks
1
feat: use post4 libtorch_npu with early Python init and native ProcessGroupHCCL.
#2006 opened Jul 22, 2026 by yingxudeng Collaborator • Draft
17 tasks
2
docs: change Qwen3.5 communication_backend from lccl to hccl
#2004 opened Jul 22, 2026 by Vectorwh Contributor Loading…
1
refactor: move npu context parallel partition into model side.
#2003 opened Jul 22, 2026 by shifengmin Collaborator Loading…
17 tasks
1
feat: DeepSeek-V4 DSA context-parallel on NPU
#1997 opened Jul 21, 2026 by LMxyzptlk Contributor Loading…
2
feat: support heterogeneous qwen3.5 PD disaggregation
#1995 opened Jul 21, 2026 by Vectorwh Contributor • Draft
feat(npu): support CANN 9.1 MegaMoe for Qwen3.5
#1989 opened Jul 21, 2026 by pjgao Collaborator • Draft
1
ProTip! What’s not been updated in a month: updated:<2026-06-26.