Skip to content

Pull requests: jd-opensource/xllm

New pull request New
17 Open 384 Closed
17 Open 384 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

feat: optimize prefetch from kv cache store.
#486 opened Dec 4, 2025 by Kang-Meng Loading…
feat: add activation, norm and rope ops for cuda device. cuda
#485 opened Dec 4, 2025 by XuZhang99 Loading…
3
feat: set same random seed for all worker.
#483 opened Dec 4, 2025 by RobbieLeung Loading…
1
feat: support iluvatar backend qwen3 0.6b run through ilu
#481 opened Dec 4, 2025 by laneeeee Loading…
10
Add constrained decoding
#480 opened Dec 3, 2025 by magicheng0816 Loading…
4
feat: support new model glm4.
#477 opened Dec 3, 2025 by DongheJin Loading…
refactor: breakdown fused moe kernel for deepseek all2all setup. mlu
#476 opened Dec 3, 2025 by a120092009 Loading…
3
feat: add wrappers for ATB and ACLNN fused operators.
#474 opened Dec 2, 2025 by yingxudeng Loading…
8
feat: add mm embedding model and its factory.
#471 opened Dec 2, 2025 by dongxianzhe Loading…
4
feat: support prefix cache for deepseek-v3/r1 models.
#470 opened Dec 1, 2025 by DongheJin Loading…
refactor: separate mlu and cuda version Qwen model implementation. cuda
#468 opened Dec 1, 2025 by XuZhang99 Loading…
1
feat: support deepseek mtp on mlu. mlu
#454 opened Nov 28, 2025 by a120092009 Loading…
1
refactor: optimize unique token count preparation of batch input builder.
#449 opened Nov 27, 2025 by RobbieLeung Loading…
[WIP] feat: support loading model weights and forward overlap.
#441 opened Nov 26, 2025 by Clement-Wang26 Loading…
1
feat: support Qwen2-VL & GME-Qwen2-VL model on npu device.
#399 opened Nov 18, 2025 by xanecdotex Loading…
feat: enable torch_npu graph mode for Qwen-3 dense with TP support.
#325 opened Nov 6, 2025 by yingxudeng Loading…
21
【WIP】feat: add rec framwork.
#305 opened Oct 31, 2025 by DragonFive Loading…
3 of 5 tasks
2
ProTip! What’s not been updated in a month: updated:<2025-11-04.