Benchmark reveals how AI learns to use unfamiliar code libraries

Original: Diagnosing Knowledge Gaps in LLM Tool Use: An Agentic Benchmark for Novel API Acquisition

Source: arxiv.org ↗