llm-experiments

The article explores whether explicit instructions for AI agents in documentation are effective, based on experiments with LLMs, concluding that while such instructions influence model behavior, their practical benefit and ethical implications are debatable.

The author shares findings from Hermes Mixture-of-Agents experiments, including voter upgrades, GPU topology, and caching economics, showing that local prefix caching can make long agent sessions nearly free and that two independent GPU instances outperform a single partitioned one.