Hamel Husain Recommends Examples for Agent Improvement
Reply suggests gathering roughly ten positive and negative examples to refine an AI agent.
TLDR
Machine learning engineer Hamel Husain replied to @petergyang that collecting around ten positive and negative examples quickly would help an agent improve. He stated the more examples the better, though the number depends on how critical the task is. Husain is described as a research engineer and independent consultant specializing in AI evaluations who previously served as a staff ML engineer at GitHub.
Combined views
593
1 Source, first seen 29d ago