Selective Agent Guidance via Entropy: Learning Autonomous Policies from Imperfect VLM Teachers | TickerVault