Quality-judgment accuracy on LLMBar EN condition
89.5Strict Accuracy (LLMBar EN)DeepSeek
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| DeepSeekLanguage condition=EN, N=8382026.06 | 89.5 | 90.5 | 88.5 | 51.8 | 1.9 | |
| Gemini FlashLanguage condition=EN, N=8382026.06 | 87.5 | 87.8 | 85.6 | 54.3 | 0.6 | |
| Claude HaikuLanguage condition=EN, N=8382026.06 | 78.5 | 78.9 | 76.1 | 52.5 | 0.7 | |
| GPT-4.1 MiniLanguage condition=EN, N=8382026.06 | 77.9 | 78.8 | 76 | 53.7 | 1.7 |