Comment by bel8

And unless I'm mistaken, the repo is about running it with 2bit quantization.

This is probably far from the raw intelligence provided by cloud providers.

Still, this shines more light on local LLMs for agentic workflows.