Correctness simulation accuracy on DBE-KT22 30% historical interaction logs
70.33AccuracyEdu-Theater
Evaluation Results
| Method | Links | |
|---|---|---|
| Edu-TheaterBase LLM=GPT-4.1-mini2026.06 | 70.33 | |
| Agent4EduBase LLM=GPT-4.1-mini2026.06 | 69.43 | |
| Edu-TheaterBase LLM=Llama3-8b2026.06 | 68.67 | |
| EduAgentBase LLM=GPT-4.1-mini2026.06 | 68.57 | |
| EERNN2026.06 | 68.35 | |
| Agent4EduBase LLM=Llama3-8b2026.06 | 68.3 | |
| SAKT2026.06 | 67.92 | |
| DAISIM2026.06 | 66.33 | |
| EduAgentBase LLM=Llama3-8b2026.06 | 65.3 | |
| DKVMN2026.06 | 64.32 | |
| KES2026.06 | 60.13 |