Text Similarity on Insurance tasks HQ subset N = 1334
89.9Mean ScoreDeepSeek-R1 + Fine-tune
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| DeepSeek-R1 + Fine-tuneCount=13342026.02 | 89.9 | 0.132 | 94.6 | 100 | 86.3 | |
| Gemini-2.5-FlashCount=13342026.02 | 81.7 | 0.146 | 86.7 | 97.7 | 73.3 | |
| GPT-4o-miniCount=13342026.02 | 80.6 | 0.149 | 85.6 | 97.4 | 70.5 | |
| Claude-Haiku-4.5Count=13342026.02 | 77.6 | 0.159 | 81.9 | 97 | 62.4 | |
| GPT-4.1Count=13342026.02 | 76.7 | 0.152 | 80.9 | 97.7 | 60.5 | |
| GPT-5.2Count=13342026.02 | 73.6 | 0.155 | 77.3 | 100 | 50.7 |