7 Releases published by 1 person

• b5015

  published Apr 1, 2025

• b5016

  published Apr 1, 2025

• b5017

  published Apr 1, 2025

• b5018

  published Apr 1, 2025

• b5019

  published Apr 1, 2025

• b5021

  published Apr 1, 2025

• b5022

  published Apr 1, 2025

9 Pull requests merged by 9 people

• common : refactor downloading system, handle mmproj with -hf option

  #12694 merged Apr 1, 2025

• opencl : fix memory allocation size

  #12649 merged Apr 1, 2025

• use LLM_KV instead of gguf_find_key

  #12672 merged Apr 1, 2025

• convert : BailingMoE : fix qkv split when head_dim is 0

  #12687 merged Apr 1, 2025

• metal : use F32 prec in FA kernels

  #12688 merged Apr 1, 2025

• Fix clang warning in gguf_check_reserved_keys

  #12686 merged Apr 1, 2025

• vulkan: fix build when glslc doesn't support coopmat

  #12683 merged Apr 1, 2025

• SYCL: Rename oneMKL to oneMath

  #12192 merged Apr 1, 2025

• SYCL: switch to SYCL namespace

  #12674 merged Apr 1, 2025

6 Pull requests opened by 6 people

• llama : refactor kv cache guard

  #12695 opened Apr 1, 2025

• common : remove json.hpp from common.cpp

  #12697 opened Apr 1, 2025

• [RFC][WIP] Common: Add an Initial Chat Memory Interface/Implementation

  #12698 opened Apr 1, 2025

• Sync minja patch to support inclusionAI/Ling-lite template and update tests

  #12699 opened Apr 2, 2025

• opencl: update doc for OpenCL

  #12702 opened Apr 2, 2025

• fix MUSA compiler warning

  #12704 opened Apr 2, 2025

5 Issues closed by 3 people

• Misc. bug: runtime repack of Q4_0 quantized models not working on llama cpp server with ARM processor

  #12701 closed Apr 2, 2025

• Eval bug: very slow inference on DeepSeek-R1-Distill-Qwen-32B

  #11361 closed Apr 2, 2025

• Misc. bug: Crashed on Intel Mac with AMD GPU.

  #11903 closed Apr 2, 2025

• Feature Request: Use direct_io for model load and inference

  #11912 closed Apr 2, 2025

• Eval bug: Command A only outputs 88888888 with -fa

  #12441 closed Apr 1, 2025

5 Issues opened by 4 people

• Cannot compile SYCL backend SYCL_LIBRARY=SYCL_LIBRARY - NOTFOUND as per documentation

  #12696 opened Apr 1, 2025

• Compile bug: RISCV cross-compile warnings cause build failure

  #12693 opened Apr 1, 2025

• Eval bug: Qwerky 72B (rwkv6qwen2) failed to load with `--split-mode row` option

  #12692 opened Apr 1, 2025

• When will llama.cpp's vulkan provide support for Intel Arc's matrix core?

  #12690 opened Apr 1, 2025

• Feature Request: Method that counts the number of image tokens in LLAVA_API

  #12689 opened Apr 1, 2025

29 Unresolved conversations

Sometimes conversations happen on old items that aren’t yet closed. Here is a list of all the Issues and Pull Requests with unresolved conversations.

• ggml-quants : weighted rounding algorithms with cumulative search

  #12557 commented on Apr 1, 2025 • 3 new comments

• contrib: support modelscope community

  #12664 commented on Apr 1, 2025 • 2 new comments

• sycl: allow ggml-sycl configuration and compilation using Visual Studio project/solution

  #12625 commented on Apr 1, 2025 • 1 new comment

• (draft) tts: Orpheus support

  #12487 commented on Apr 2, 2025 • 1 new comment

• gguf-split now respects dry-run option

  #12681 commented on Apr 2, 2025 • 0 new comments

• vocab : BailingMoE : change possessive quantifiers to greedy

  #12677 commented on Apr 1, 2025 • 0 new comments

• [CANN]get_rows and dup optimization

  #12671 commented on Apr 2, 2025 • 0 new comments

• tts : implement sesame CSM + Mimi decoder

  #12648 commented on Apr 1, 2025 • 0 new comments

• Draft: vulkan: Add bfloat16 support

  #12554 commented on Apr 2, 2025 • 0 new comments

• Nomic Embed Text V2 with Mixture-of-Experts (MoE) architecture

  #12466 commented on Apr 1, 2025 • 0 new comments

• Add Qwen2.5VL support

  #12402 commented on Apr 1, 2025 • 0 new comments

• `server`: streaming of tool calls and thoughts when `--jinja` is on

  #12379 commented on Apr 1, 2025 • 0 new comments

• PR: Refine ggml-hexagon backend(Qualcomm Hexagon NPU backend) for latest ggml,whisper.cpp,llama.cpp

  #12326 commented on Apr 2, 2025 • 0 new comments

• llama : add llama_batch_ext

  #11875 commented on Apr 1, 2025 • 0 new comments

• tool-call: add support for tool-calls using Model Context Protocol

  #11556 commented on Apr 1, 2025 • 0 new comments

• llama : add option to override model tensor buffers

  #11397 commented on Apr 2, 2025 • 0 new comments

• Simplify and improve CUDA graphs through use of indirect copy pointers

  #9017 commented on Apr 1, 2025 • 0 new comments

• Misc. bug: CUDA errors with multi-threaded use

  #11804 commented on Apr 2, 2025 • 0 new comments

• Misc. bug: vulkan on Adreno GPU

  #12139 commented on Apr 2, 2025 • 0 new comments

• Misc. bug: gguf-dump 'newbyteorder' was removed

  #12146 commented on Apr 2, 2025 • 0 new comments

• Replacement for deprecated codevct string conversion

  #12151 commented on Apr 2, 2025 • 0 new comments

• Compile bug: compilation warnings (clang) Introduced in #10558

  #12685 commented on Apr 2, 2025 • 0 new comments

• Feature Request: Add support of convert.py for model Qwen2.5-Omni-7B

  #12641 commented on Apr 1, 2025 • 0 new comments

• ggml : add DirectML backend

  #7772 commented on Apr 1, 2025 • 0 new comments

• Misc. bug: vulkan on 6900xt

  #12147 commented on Apr 1, 2025 • 0 new comments

• Feature request: Graphical GGUF viewer

  #6715 commented on Apr 1, 2025 • 0 new comments

• Feature Request: Splitting layers according to VRAM usage on multi GPUs setups

  #12654 commented on Apr 1, 2025 • 0 new comments

• Feature Request: Qwen 2.5 VL

  #11483 commented on Apr 1, 2025 • 0 new comments

• llama-gemma3-cli: output degeneration after repeated uses

  #12499 commented on Apr 1, 2025 • 0 new comments