Multiple-choice Question Answering on KNIGHT History L1
98.6AccuracyHuman
Evaluation results
| Method | Links | |
|---|---|---|
| Humann=2002026.02 | 98.6 | |
| GPT-4o2026.02 | 92.95 | |
| Mistral Large2026.02 | 92.19 | |
| Llama3-70B-InstructSize=70B, Type=Instruct2026.02 | 92.12 | |
| Claude 3 Haiku2026.02 | 91.95 | |
| Qwen1.5 (1.8B)Size=1.8B2026.02 | 76.76 | |
| Gemma (2B)Size=2B2026.02 | 38.59 |