Correctness simulation accuracy on DBE-KT22 50% historical interaction logs
71.27AccuracyEdu-Theater
Evaluation Results
| Method | Links | |
|---|---|---|
| Edu-TheaterBase LLM=GPT-4.1-mini2026.06 | 71.27 | |
| Agent4EduBase LLM=GPT-4.1-mini2026.06 | 69.95 | |
| Edu-TheaterBase LLM=Llama3-8b2026.06 | 69.67 | |
| Agent4EduBase LLM=Llama3-8b2026.06 | 69.63 | |
| EERNN2026.06 | 69.04 | |
| EduAgentBase LLM=GPT-4.1-mini2026.06 | 68.52 | |
| DAISIM2026.06 | 68.44 | |
| SAKT2026.06 | 68.37 | |
| EduAgentBase LLM=Llama3-8b2026.06 | 67.33 | |
| DKVMN2026.06 | 65.33 | |
| KES2026.06 | 63.27 |