Fetching the paper…

AvaTaR: Optimizing LLM Agents for Tool Usage via Contrastive Reasoning · Around