Researchers present ID-TAP, a scalable neural dueling-bandit framework inspired by information-directed sampling (IDS) that lets LLM agents personalize at test time by selecting a persona from binary comparisons between candidate responses as the online signal. The work frames persona selection as the core problem for adapting to query-dependent user preferences from lightweight interaction.
Effective test-time personalization requires LLM agents to adapt to query-dependent user preferences from lightweight interaction. We study persona selection with binary comparisons between candidate responses as the online signal. We present ID-TAP, a scalable, IDS-inspired neural dueling-bandit fr
EU-based hosting, git-push deploys, automatic HTTPS, no cold starts. Free tier with a custom domain — perfect for shipping the agent you just read about.