#ramalama — Public Fediverse posts

Friends Don't Let Friends Use #Ollama https://sleepingrobots.com/dreams/stop-using-ollama/

For running LLMs locally use llama.cpp with its built-in web UI, or try #ramalama https://ramalama.ai/

Raised a bug about #ramalama today not playing well with #arm64 and and #amd gpus. However if you force the base image local inference does use #vulkan to run - and much faster than maxing out the CPU cores on my #altra.

Updating stickers on laptops... let's see how many I can tag

@matrix @ansible @InstructLab @thinkpadmuseum @trustyai @github @fedora

#ramalama #docling #vllm #llmd #pytorch #NERC #redhat #ospo #upstream #ansible #thinkpad #womeninfedora #expo2025 #kubeflow #trustyAI #cushingcenter #operationstickybusiness