Cross-lingual Cultural Consistency on BLEnD Lower-Resource
0.02Max SigmaVanilla
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| VanillaModel=Qwen-2.5-3B2026.04 | 0.02 | 0.271 | — | — | — | |
| Few-shotModel=Qwen-2.5-3B2026.04 | 0.02 | 0.3 | — | — | — | |
| SteeringModel=Qwen-2.5-3B2026.04 | 0.02 | — | 0.266 | 0.272 | 0.278 | |
| PersonaModel=Qwen-2.5-3B2026.04 | 0.02 | — | 0.263 | 0.291 | 0.331 | |
| C-3POModel=Qwen-2.5-3B2026.04 | 0.02 | 0.364 | — | — | — | |
| VanillaModel=Llama-3.1-8B2026.04 | 0.021 | 0.392 | — | — | — | |
| Few-shotModel=Llama-3.1-8B2026.04 | 0.021 | 0.41 | — | — | — | |
| SteeringModel=Llama-3.1-8B2026.04 | 0.021 | — | 0.385 | 0.397 | 0.402 | |
| PersonaModel=Llama-3.1-8B2026.04 | 0.021 | — | 0.376 | 0.41 | 0.449 | |
| C-3POModel=Llama-3.1-8B2026.04 | 0.021 | 0.468 | — | — | — | |
| VanillaModel=Gemma-2-27B2026.04 | 0.021 | 0.565 | — | — | — | |
| Few-shotModel=Gemma-2-27B2026.04 | 0.021 | 0.557 | — | — | — | |
| SteeringModel=Gemma-2-27B2026.04 | 0.021 | — | 0.562 | 0.57 | 0.575 | |
| PersonaModel=Gemma-2-27B2026.04 | 0.021 | — | 0.572 | 0.603 | 0.648 | |
| C-3POModel=Gemma-2-27B2026.04 | 0.021 | 0.657 | — | — | — |