Belief assessment on misinformation on MISBELIEF Sports domain (Hard split)
4.28Belief ScoreGPT-3.5-turbo
Evaluation results
| Method | Links | |
|---|---|---|
| GPT-3.5-turboSetting=Round 32026.01 | 4.28 | |
| Qwen-turboSetting=Round 32026.01 | 4.25 | |
| GPT-3.5-turboSetting=Round 12026.01 | 4.16 | |
| GPT-5Setting=Round 32026.01 | 4.08 | |
| Qwen-turboSetting=Round 12026.01 | 4.01 | |
| Qwen2.5-72BSetting=Round 32026.01 | 3.98 | |
| GPT-5Setting=Round 12026.01 | 3.9 | |
| Qwen2.5-32BSetting=Round 32026.01 | 3.88 | |
| Qwen2.5-72BSetting=Round 12026.01 | 3.85 | |
| Qwen2.5-32BSetting=Round 12026.01 | 3.83 | |
| Llama3-8BSetting=Round 32026.01 | 3.63 | |
| Qwen-turbo (w/o reasoning)Setting=Round 32026.01 | 3.6 | |
| Llama3-8BSetting=Round 12026.01 | 3.48 | |
| Qwen-turbo (w/o reasoning)Setting=Round 12026.01 | 3.46 | |
| Qwen2.5-72BSetting=Baseline2026.01 | 3.03 | |
| GPT-3.5-turboSetting=Original2026.01 | 2.81 | |
| Qwen2.5-72BSetting=Original2026.01 | 2.73 | |
| GPT-3.5-turboSetting=Baseline2026.01 | 2.65 | |
| Qwen-turboSetting=Baseline2026.01 | 2.54 | |
| Qwen-turbo (w/o reasoning)Setting=Baseline2026.01 | 2.53 | |
| Qwen2.5-32BSetting=Baseline2026.01 | 2.41 | |
| Qwen-turboSetting=Original2026.01 | 2.23 | |
| Qwen-turbo (w/o reasoning)Setting=Original2026.01 | 2.21 | |
| GPT-5Setting=Original2026.01 | 2.17 | |
| Qwen2.5-32BSetting=Original2026.01 | 2.12 | |
| Llama3-8BSetting=Original2026.01 | 2.06 | |
| Llama3-8BSetting=Baseline2026.01 | 1.92 | |
| GPT-5Setting=Baseline2026.01 | 1.89 |