Announcement
llama.cpp team says Hugging Face gives its build pipeline breathing room
Gerganov says llama.cpp now uses Hugging Face Buckets for caches and Jobs for extra CI runners.
TLDR
In a team blog post, llama.cpp developers say they moved non-release compiler caches to Hugging Face Buckets and most Linux x64 build-and-test jobs to Hugging Face Jobs. Their squeeze: more than 100 jobs per full run, but only 20 concurrent GitHub-hosted jobs on their free plan. GitHub Actions still runs the show—the workloads moved, not the control room.
Combined views
11.6K
2 Sources, first seen ago

Comments