AI judge reuses old answers to cheaply pick the best search model

Original: Incremental Pooled LLM Evaluation for Cost-Effective Retrieval Model Selection

Source: arxiv.org ↗