Distortion Diagnosis on ReMMDBench (500-sample split)
44.73PrecisionReMMD-Agent
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| ReMMD-AgentBackbone=Qwen3.6-27B2026.06 | 44.73 | 39.2 | 41.1 | 10.8 | |
| ReMMD-AgentBackbone=Qwen3.5-9B2026.06 | 44.58 | 50.88 | 46.97 | 10 | |
| ReMMD-AgentBackbone=GPT-5.22026.06 | 44.31 | 47.01 | 45.15 | 5 | |
| ManusBackbone=proprietary2026.06 | 43.69 | 42.32 | 42.75 | 7.6 | |
| ReMMD-AgentBackbone=Gemma4-31B2026.06 | 43.67 | 41.31 | 42.27 | 7.8 | |
| MMD-AgentBackbone=Qwen3.5-4B2026.06 | 43.08 | 32.7 | 34.45 | 7.2 | |
| ReMMD-AgentBackbone=Qwen3.5-4B2026.06 | 42.59 | 42.89 | 41.94 | 6 | |
| MMD-AgentBackbone=GPT-5.22026.06 | 42.24 | 48.83 | 41.98 | 2 | |
| T2-AgentBackbone=GPT-5.22026.06 | 42.15 | 47.15 | 42.68 | 2.6 | |
| ChatGPTBackbone=proprietary2026.06 | 41.71 | 47.56 | 43.63 | 3 | |
| T2-AgentBackbone=Qwen3.6-27B2026.06 | 41.63 | 27.09 | 27.74 | 4.2 | |
| MMD-AgentBackbone=Qwen3.6-27B2026.06 | 41.59 | 38.61 | 37.94 | 4.8 | |
| T2-AgentBackbone=Gemma4-31B2026.06 | 41.16 | 38.73 | 38.73 | 7.6 | |
| MMD-AgentBackbone=Gemma4-31B2026.06 | 40.94 | 42.11 | 40.82 | 5.2 | |
| MMD-AgentBackbone=Qwen3.5-9B2026.06 | 40.1 | 40.31 | 38.15 | 5.8 | |
| T2-AgentBackbone=Qwen3.5-9B2026.06 | 38.83 | 27.14 | 28.58 | 3.2 | |
| T2-AgentBackbone=Qwen3.5-4B2026.06 | 37.95 | 27.26 | 28.5 | 1.8 |