Skip to content

Pull requests: ggml-org/llama.cpp

New pull request New
1,462 Open 13,039 Closed
1,462 Open 13,039 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

perplexity: respect --chunks when computing KL divergence examples
#28169 opened Sep 1, 2026 by masel Loading…
metal : single-source fusion table + gated_delta_net cache fusion Apple Metal https://en.wikipedia.org/wiki/Metal_(API) ggml changes relating to the ggml tensor library for machine learning testing Everything test related
#28164 opened Sep 1, 2026 by ggerganov Member • Draft
metal : add metallib build support for xcframework Apple Metal https://en.wikipedia.org/wiki/Metal_(API) ggml changes relating to the ggml tensor library for machine learning
#28163 opened Sep 1, 2026 by jhen0409 Contributor Loading…
metal: implement OUR_PROD for F32 Apple Metal https://en.wikipedia.org/wiki/Metal_(API) ggml changes relating to the ggml tensor library for machine learning testing Everything test related
#28162 opened Sep 1, 2026 by HankBO Loading…
Add support for XingChen4 conversion CUDA Related to the CUDA backend ggml changes relating to the ggml tensor library for machine learning model Model specific testing Everything test related
#28156 opened Sep 1, 2026 by shuxiaoqiong • Draft
1
Only request VK_KHR_shader_bfloat16 extension if supported draft PR will be changed to draft by github-actions bot ggml changes relating to the ggml tensor library for machine learning Vulkan Issues specific to the Vulkan backend
#28155 opened Sep 1, 2026 by madsmtm Loading…
1
5
model: correctly support input vision for deepseek4 conversion model Model specific mtmd Related to multimodal functionality (video/image/audio)
#28154 opened Sep 1, 2026 by ngxson Collaborator Loading…
metal : add fa-vec tunings for A18 Pro (MacBook Neo) Apple Metal https://en.wikipedia.org/wiki/Metal_(API) ggml changes relating to the ggml tensor library for machine learning
#28152 opened Sep 1, 2026 by jhen0409 Contributor Loading…
1
ggml : check for allocation failures to prevent crashes ggml changes relating to the ggml tensor library for machine learning mtmd Related to multimodal functionality (video/image/audio)
#28149 opened Sep 1, 2026 by sanmai Contributor • Draft
CUDA/ROCm: Fix pathological graph capture in test-backend-ops CUDA Related to the CUDA backend ggml changes relating to the ggml tensor library for machine learning testing Everything test related
#28148 opened Sep 1, 2026 by pwilkin Member Loading…
9
mtmd: check and error on allocation in clip.cpp mtmd Related to multimodal functionality (video/image/audio)
#28146 opened Sep 1, 2026 by sanmai Contributor Loading…
1
Uppdated flake.nix devops improvements to build systems and github actions nix Issues specific to consuming flake.nix, or generally concerned with ❄ Nix-based llama.cpp deployment
#28145 opened Sep 1, 2026 by ZhaolunYin Loading…
1
docker: explicitly install NCCL in CUDA 13 images devops improvements to build systems and github actions
#28143 opened Sep 1, 2026 by EchterAgo Loading…
SVE 256-bit Implementation of mxfp4 gemv kernel ggml changes relating to the ggml tensor library for machine learning
#28141 opened Sep 1, 2026 by abhijain1204fujitsu Contributor Loading…
2
qwen4exp: direct reads for the lazy PLE table (>2x prefill performance improvement on GB10) examples model Model specific
#28136 opened Sep 1, 2026 by coder543 Contributor Loading…
1
mtmd: support DeepSeek-V4-Flash-Vision-Exp conversion mtmd Related to multimodal functionality (video/image/audio)
#28133 opened Sep 1, 2026 by ngxson Collaborator Loading…
13
gguf-py: bound array element count by remaining file size conversion
#28131 opened Sep 1, 2026 by wmshearer • Draft
2
kv-cells : scan the flat pos array instead of walking the used std::set
#28128 opened Aug 31, 2026 by abdel-darwish-27 • Draft
1
Model: add Tencent Hy 4 (hy_v4) preview architecture support conversion model Model specific testing Everything test related
#28127 opened Aug 31, 2026 by Little0o0 Loading…
2
metal : add fa-vec tuning for M2 Pro Apple Metal https://en.wikipedia.org/wiki/Metal_(API) ggml changes relating to the ggml tensor library for machine learning
#28122 opened Aug 31, 2026 by lstolcman Loading…
1
models : appropriately flag noscan ssm_a tensors model Model specific
#28121 opened Aug 31, 2026 by CISC Member Loading…
server : propagate speculative params to default props response server
#28119 opened Aug 31, 2026 by drwolfen Loading…
server : keep speculative recurrent-state checkpoints on-device server
#28118 opened Aug 31, 2026 by vahpetr • Draft
4
common : draft KV cache no longer inherits --no-kv-offload testing Everything test related
#28117 opened Aug 31, 2026 by jkSeria Loading…
1
speculative: add optional top-k draft acceptance statistics
#28110 opened Aug 31, 2026 by cuhawk • Draft
1
ProTip! Adding no:label will show everything without a label.