Skip to content

v1.0.0

Choose a tag to compare

@KiyoEditz KiyoEditz released this 26 Aug 03:24
· 29 commits to master since this release

Prebuilt llama-server binary (llama.cpp tag b10605), compiled with CUDA
for Tesla T4 (sm_75). Built via Build-Llama-CUDA-Release.py on Kaggle.

  • CUDA VMM: off (built with GGML_CUDA_NO_VMM=ON — this Kaggle image
    doesn't expose libcuda.so at build time; inference correctness is
    unaffected, only the VRAM allocator is slightly less flexible)
  • Bundled: llama-server + libggml*/libllama*/libmtmd*.so (project-owned
    only — system CUDA runtime/driver libs come from the host at runtime)
  • Verify before use: extract, check VERSION.txt matches your
    LLAMA_CPP_TAG/CUDA_ARCH, then ./run.sh --version

Asset: llama-linux-cuda-sm75-b10605.tar.gz (+ .sha256)
Used by: Collab-Llama.py / Kaggle-Llama.py via GITHUB_RELEASE_URL