Skip to content

Pull requests: alibaba/MNN

New pull request New
30 Open 960 Closed
30 Open 960 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

[CUDA:Bugfix] Fix int32 REALDIV not dispatched in BinaryBlit
#4652 opened Jul 21, 2026 by Qxinyu Collaborator Loading…
12 tasks
3
[LLM:Bugfix] Clear AWQ attention KV cache
#4646 opened Jul 21, 2026 by LudovicoYIN Contributor Loading…
5 tasks done
1
修复lora.py对 lora权重名称解析方式的缺陷
#4643 opened Jul 20, 2026 by mistletoe1024 Loading…
1
修复 Qwen3.5系列模型在CPU后端推理长文本时质量较低，出现重复段落的问题。
#4632 opened Jul 16, 2026 by v0jiuqi Collaborator Loading…
12 tasks
5
[Backend] Fix explicit pooling pads and CUDA TopKV2 largest flag
#4627 opened Jul 15, 2026 by Fatemanx Contributor Loading…
1
[LLM:Feature] Add HunyuanOCR 1.5 support
#4625 opened Jul 14, 2026 by Fatemanx Contributor Loading…
[CPU:Refact] Refactor the KleidiAI integration code
#4602 opened Jul 8, 2026 by HenryDen Contributor Loading…
12 tasks
1
[LLM:Bugfix] Align Qwen3-VL runtime preprocessing
#4595 opened Jul 7, 2026 by yydhYYDH Contributor Loading…
5 of 12 tasks
[CPU:Bugfix] Fix dirty data reduction in RVV AccumulateSequence
#4590 opened Jul 2, 2026 by ihb2032 Contributor Loading…
5 of 12 tasks
[LLM:Feature] Support Qwen3-TTS-12Hz-0.6B-Base
#4589 opened Jul 2, 2026 by huangzhengxiang Contributor Loading…
5 tasks done
7
[QNN:Feature] add vtmc_mb arg for vtcm size configurable
#4587 opened Jul 1, 2026 by blueskycoco Loading…
10 tasks
[LLM:Bugfix] Validate request JSON in LLM HTTP servers
#4585 opened Jun 30, 2026 by eharris128 Loading…
4 of 5 tasks
1
[QNN:Feature] Auto ADSP_LIBRARY_PATH setup and device creation fallback
#4584 opened Jun 30, 2026 by lanling-47 Contributor Loading…
[QNN:Bugfix] Fix StridedSlice NC4HW4 layout and improve QNNPerf robustness
#4583 opened Jun 30, 2026 by lanling-47 Contributor Loading…
[Infra:Chore] Include VS version in Windows release package name
#4581 opened Jun 30, 2026 by wangzhaode Collaborator • Draft
5
[ARM:Bugfix] Enable MNNRankOneUpdate fallback on Apple Silicon
#4579 opened Jun 29, 2026 by lanling-47 Contributor Loading…
6 tasks done
[ARM:Perf] Add fast path for SME2 FP16 Pack with small channel count
#4578 opened Jun 29, 2026 by lanling-47 Contributor Loading…
6 tasks done
2
[LLM:Feature] Add Qwen3.5 video input
#4577 opened Jun 28, 2026 by Fatemanx Contributor Loading…
Add Ministral 3B Q4 MNN to model market
#4576 opened Jun 28, 2026 by shefowl Loading…
@wangzhaode
1
Vulkan:Speed: Speed up vulkan for corp-matrix, complete segment ops
#4575 opened Jun 28, 2026 by jxt1234 Collaborator Loading…
12 tasks
[LLM:Bugfix] Fix embedding QNN export paths
#4555 opened Jun 19, 2026 by ZedingZhang Loading…
1
@wangzhaode @Qxinyu
3
[QNN:Bugfix] Add support for convert llm/visual mnn model to Qnn mode…
#4549 opened Jun 16, 2026 by blueskycoco Loading…
10 tasks
@wangzhaode @Qxinyu
2
Optimize RVV backend for Deconvolution and QuantizedAdd
#4547 opened Jun 16, 2026 by jxgxxx Contributor Loading…
5 of 12 tasks
@wangzhaode
[LLM:Feature] Add Segment Mode, Speed up metal llm for 30%-100%
#4543 opened Jun 15, 2026 by jxt1234 Collaborator Loading…
12 tasks
@wangzhaode
1
[NPU:rknn] Add initial rknn support for Rockchip NPUs, supporting traditional models.
#4524 opened Jun 10, 2026 by huangzhengxiang Contributor Loading…
4 tasks done
@wangzhaode
ProTip! Filter pull requests by the default branch with base:master.