Llama.cpp recompiled for Jetpack6.2, CUDA12.6 & sm_87 - Runs Gemma 4 E2B at 25 tok/s on 8GB memory.
10K+
Maintained by Jake White, a Chartered Architectural Technologist in Mid Sussex who builds open-source tools for architecture (StairSmith, SunForm) and self-hosted AI. https://jakewhitearchitecture.com
Content type
Image
Digest
sha256:f9664d9ca…
Size
5.5 GB
Last updated
6 months ago
docker pull jakepwhite/llama-cpp-orin-nano-super