Professional Reasoning on Logic-Expert-Pro (test)
81.2Symbolic ScoreLC-ERD
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| LC-ERD#Params=72B, Multi-agent/MARL framework=true, Model Category=Ours2026.05 | 81.2 | 78.5 | 84.6 | 79.1 | 74.2 | 70.5 | 78 | |
| Claude-3-OpusModel Category=Proprietary Models2026.05 | 76.8 | 72.1 | 79.4 | 75.2 | 68.9 | 64.2 | 72.8 | |
| GPT-4oModel Category=Proprietary Models2026.05 | 72.5 | 68.4 | 75.1 | 70.2 | 64.5 | 60.1 | 68.4 | |
| MaAS#Params=72B+, Multi-agent/MARL framework=true, Model Category=MARL Baselines2026.05 | 65.2 | 62.1 | 68.5 | 67.4 | 58.1 | 55.3 | 62.7 | |
| LPO#Params=72B, Model Category=Open-Source SOTA2026.05 | 64.8 | 60.5 | 67.2 | 65.1 | 56.4 | 54.1 | 61.3 | |
| Qwen2.5-72B (Base)#Params=72B, Model Category=Open-Source SOTA2026.05 | 58.4 | 55.2 | 61.3 | 59.8 | 50.4 | 48.2 | 55.5 |