All pull requests

Pull requests

• ggml : support views in unary ops

  CUDAggml
  #30298·Carl-jcc opened Oct 10, 2026·1 task done·

  ·
  1

• cmake(s390x): disable z17 target for unsupported compilers

  ggml
  #30297·taronaeo opened Oct 10, 2026Member·

  ·

• cpu : reduce AVX2 tiled accumulators together

  ggml
  #30296·trivalik opened Oct 10, 2026·

  ·

• server: leave a busy slot untouched when a request pins it

  server
  #30295·ServeurpersoCom opened Oct 10, 2026Contributor·

  ·

• server: gracefully fall back when MTP is not supported by the model (#29874)

  builddocumentationserver
  #30294·bappisinsane opened Oct 10, 2026·

  ·
  1

• docs: add preferred citation to README (#29049)

  builddocumentation
  #30292·bappisinsane opened Oct 10, 2026·

  ·
  1

• cmake: error when both GGML_CUDA and GGML_HIP are enabled (#29048)

  build
  #30291·bappisinsane opened Oct 10, 2026·

  ·
  1

• vocab: fix crash with word-type SentencePiece tokenizers (#6717)

  #30290·bappisinsane opened Oct 10, 2026·

  ·
  1

  1

• convert : export YaRN truncation config

  conversion
  #30288·tokinasin opened Oct 10, 2026Contributor·

  ·

• [SYCL] fix fa crash for Query head dimension mismatch (neq0 != HSK / D=2048)

  ggmlSYCL
  #30286·arthw opened Oct 10, 2026Contributor·

  ·

• Cuda fattn mma head sizes 40 72

  CUDAggmltesting
  #30285·sihanyu03 opened Oct 10, 2026Contributor·

  ·

• common: select A100 cores for SpaceMiT default threads

  testing
  #30284·inuEbisu opened Oct 10, 2026·

  ·
  1

  2

• vulkan: clamp A prefetch to end_k in coopmat1 int8 matmul

  ggmltestingVulkan
  #30283·SelfRef opened Oct 10, 2026·

  ·
  1

  2

• RDNA3/4 Vulkan Large-Tile Split-K Optimization

  ggmltestingVulkan
  #30280·Alex-JP-93 opened Oct 10, 2026·

  ·
  1

• Fail test backend ops on missing or unsupported types

  testing
  #30279·EAddario opened Oct 10, 2026Contributor·

  ·
  1

• cuda: add TQ2_0 dequant + MMVQ support

  CUDAggml
  #30278·coolinger2019-wq opened Oct 10, 2026·

  ·
  1

• meta : enable tensor split for KDA hybrid models

  ggml
  #30277·ggerganov opened Oct 10, 2026Member·

  ·

• server: track LoRA settings per prompt cache entry

  server
  #30275·DQR-Happy opened Oct 10, 2026·

  ·
  1

• CANNOT COPY ANY CODEBLOCKS IN FIREFOX-->Refactor clipboard copy function for better error handling

  server/ui
  #30274·clort81 opened Oct 10, 2026·

  ·
  1

• Sycl build: fix compiler warnings - missing prototypes, dead code, extra semicolons

  buildggmlSYCL
  #30271·mctylr-gh opened Oct 10, 2026Contributor·

  ·
  1

• cpu: fix AVX2 CPUs not using F16C when available

  ggmlmerge ready
  #30270·xinitrcn1 opened Oct 10, 2026·

  ·

• server: support for input layer embeddings in GLM5-next

  modeltesting
  #30268·willweimike opened Oct 10, 2026Contributor·

  ·
  1

• Respect cgroup CPU quotas for default threads

  #30267·Hahaa909 opened Oct 10, 2026·

  ·
  1

• opencl: improve fa, allow dk512 for gemma-4, improve dk64

  ggmlOpenCL
  #30266·lhez opened Oct 10, 2026Contributor·

  ·

• server : add logit_gate to /v1/completions and /v1/chat/completions

  documentationserver
  #30265·zqstdy opened Oct 10, 2026·

  ·
  2