Tool-augmented Question Answering on MedQA
74.11AccuracyEPC-AW
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| EPC-AWBackbone=Qwen3-Coder-30B2026.05 | 74.11 | — | |
| RollbackBackbone=Qwen3-Coder-30B2026.05 | 71.56 | 2.55 | |
| RetryBackbone=Qwen3-Coder-30B2026.05 | 71 | 3.11 | |
| No-Repair (AgentFlow)Backbone=Qwen3-Coder-30B2026.05 | 67.11 | 7 |