Field Notes · Closing

Vomit: Clean up Claude 5's token output with a separate LLM

A tool or technique for post-processing Claude 5's token output using a separate language model to clean up the results.

Why your infrastructure is more important than the next LLM release

This article argues that infrastructure investments should be prioritized over new large language model releases.

Multi-Agent Orchestration with the Common-Sense Reasoning Capabilities of LLMs for Autonomous Driving

This paper explores using Large Language Models to orchestrate multiple agents in autonomous driving systems by leveraging their common-sense reasoning abilities to handle complex scenarios where traditional reinforcement learning and rule-based approaches may be insufficient.