Data as of Jul 25, 2026 · Based on 2,718,867 AI responses across 9,511 prompts · See how Parse measures this
M‑MAD (Multidimensional Multi‑Agent Debate)github.com · Low evidenceM-MAD is a systematic LLM-based multi-agent framework for advanced machine translation evaluation. It operates in three stages: dimension partition, multi-agent debate, and final judgment to produce comprehensive evaluation judgments.
Parse Score