Bias Evaluation on KoBBQ
98.3Ambiguous Context ScoreQwen3-30B-A3B
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Qwen3-30B-A3BReasoning=On2026.03 | 98.3 | 90.72 | 94.51 | |
| EXAONE-4.0-32BReasoning=On2026.03 | 97.2 | 93.23 | 95.21 | |
| Mi:dm K 2.5 Pro (March ‘26)Reasoning=On2026.03 | 94.56 | 93.44 | 94 | |
| Qwen3-30B-A3BReasoning=Off2026.03 | 91.75 | 82.86 | 87.3 | |
| Mi:dm K 2.5 Pro (March ‘26)Reasoning=Off2026.03 | 91.02 | 94.12 | 92.57 | |
| EXAONE 3.5Size=32B2025.09 | 87.9 | 92.1 | 90 | |
| HyperCLOVAX-SEED-Think-32BReasoning=Off2026.03 | 86.49 | 88.25 | 87.37 | |
| Exaone-3.5-7.8B-instParameters=7.8B, Type=Instruction-tuned2026.01 | 85.91 | 73.54 | 79.72 | |
| EXAONE 3.5Size=7.8B2025.09 | 85.9 | 73.5 | 79.7 | |
| Mi:dm 2.0 Base-instType=Instruction-tuned2026.01 | 82.3 | 80.54 | 81.42 | |
| Mi:dm 2.0-BaseScale=Base2025.09 | 82.3 | 80.5 | 81.4 | |
| HyperCLOVAX-SEED-Think-32BReasoning=On2026.03 | 80.44 | 65.56 | 73 | |
| EXAONE-4.0-32BReasoning=Off2026.03 | 78.45 | 91.66 | 85.05 | |
| Mi:dm 2.0 Mini-instType=Instruction-tuned2026.01 | 55.11 | 55.17 | 55.14 | |
| Exaone-3.5-2.4B-instParameters=2.4B, Type=Instruction-tuned2026.01 | 54.18 | 68.19 | 61.18 | |
| Llama-3.1-8B-instParameters=8B, Type=Instruction-tuned2026.01 | 40.92 | 57.7 | 49.3 | |
| Llama-3.1Size=8B2025.09 | 40.9 | 57.7 | 49.3 |