Robustness Evaluation on BiasBench
82.5AccuracyQwen2.5-32B-Instruct
Evaluation Results
| Method | Links | |
|---|---|---|
| Qwen2.5-32B-Instructvariant=Instruct2026.01 | 82.5 | |
| DeepSeek-V32026.01 | 81.25 | |
| Qwen3-30B-A3B-Instruct-2507variant=Instruct2026.01 | 81.25 | |
| Qwen3-Next-80B-A3B-Instructvariant=Instruct2026.01 | 80 | |
| Qwen3-30B-A3B-Thinking-2507variant=Thinking2026.01 | 77.5 | |
| Qwen3-Next-80B-A3B-Thinkingvariant=Thinking2026.01 | 75 | |
| QwQ-32B2026.01 | 67.5 | |
| DeepSeek-R12026.01 | 65 |