Skip to content

Pull requests: jd-opensource/xllm

New pull request New
54 Open 791 Closed
54 Open 791 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

refactor: remove useless function call and refine tests.
#996 opened Mar 4, 2026 by liutongxuan Loading…
refactor: remove unused code.
#994 opened Mar 4, 2026 by yq33victor Loading…
1
feat: add flags for multi_round pipeline to return logprob.
#993 opened Mar 4, 2026 by DragonFive Loading…
4
feat: support joyai-llm-flash model on mlu device.
#992 opened Mar 4, 2026 by yq33victor Loading…
2
feat: support embedding interface for all generate LLM models.
#990 opened Mar 4, 2026 by RobbieLeung Loading…
2
[Draft] Support Qwen next
#989 opened Mar 4, 2026 by JC-ut0 • Draft
6
bugfix: fix mbox model(qwen2.5) multi round core in xattention
#988 opened Mar 4, 2026 by DragonFive Loading…
5
feat: add gtest coverage for cuda act_and_mul activation kernel.
#987 opened Mar 3, 2026 by yingxudeng Loading…
4
feat: add qwen35 and qwen35-thinking reasoning parser detectors.
#985 opened Mar 3, 2026 by yingxudeng Loading…
2
feat: implement moe fused topk layer.
#983 opened Mar 3, 2026 by XuZhang99 Loading…
15
feat: support qwen3.5 tool-call parser with qwen3_coder detector.
#982 opened Mar 3, 2026 by yingxudeng Loading…
2
feat: add onerec model implement[2/N]
#981 opened Mar 3, 2026 by DragonFive Loading…
2
feat: support xLLM CUDA backend integration with tilelang kernels.
#980 opened Mar 3, 2026 by Wang-1F • Draft
4
feat: implement fused RMSNorm and static FP8 quantization for improved performance[2/N].
#976 opened Mar 3, 2026 by yingxudeng Loading…
9
refactor: optimize speculative decoding pipeline.
#974 opened Mar 2, 2026 by RobbieLeung Loading…
2
feat: add more device utils for cuda device.
#971 opened Mar 2, 2026 by XuZhang99 Loading…
7
feat: support DeepSeek V3.2 W4A8 MoE for mlu and add smoke test.
#969 opened Mar 2, 2026 by phantomlei3 Loading…
3
feat: improve offline inference interface and fix several tp and vlm bugs.
#968 opened Mar 2, 2026 by weizhehuang0827 Loading…
11
feat: add rope_in_place tilelang kernel for npu device.
#964 opened Feb 28, 2026 by zhang-minchao • Draft
6
feat: support LongCat-Image-Edit on cuda device.
#957 opened Feb 27, 2026 by Dragonliu2018 Loading…
8
feat: support qwen3-next on npu device.
#945 opened Feb 26, 2026 by liyu119 • Draft
15
feat: support schedule overlap for multimodal model.
#943 opened Feb 26, 2026 by wly-115 Loading…
2
feat: support in-batch prefix caching.
#941 opened Feb 26, 2026 by Clement-Wang26 Loading…
10
feat: implement bidirectional remote host to local device kv cache transfer and batch offload.
#939 opened Feb 26, 2026 by Kang-Meng Loading…
21
feat: add REC multi-round two-stage xattention support with CUDA Graph integration.
#933 opened Feb 24, 2026 by LMX-xin Loading…
32
ProTip! Adding no:label will show everything without a label.