Cross-lingual Alignment on FLORES
0.5088Mean MEXA ScoreMEXA
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| MEXAModel=Gemma 2, Parameters=9B, Embedding pooling strategy=Mean2024.10 | 0.5088 | — | |
| MEXAModel=Llama 3.1, Parameters=70B, Embedding pooling strategy=Mean2024.10 | 0.411 | — | |
| MEXAModel=Llama 3.1, Parameters=8B, Embedding pooling strategy=Mean2024.10 | 0.3963 | — | |
| MEXAModel=Llama 3, Parameters=8B, Embedding pooling strategy=Mean2024.10 | 0.3939 | — | |
| MEXAModel=Gemma 1, Parameters=7B, Embedding pooling strategy=Mean2024.10 | 0.3815 | — | |
| MEXAModel=Mistral 0.3, Parameters=7B, Embedding pooling strategy=Mean2024.10 | 0.2642 | — | |
| MEXAModel=Llama 1, Parameters=7B, Embedding pooling strategy=Mean2024.10 | 0.1946 | — | |
| MEXAModel=Llama 2, Parameters=7B, Embedding pooling strategy=Mean2024.10 | 0.0866 | — | |
| MEXAModel=OLMO 1.7, Parameters=7B, Embedding pooling strategy=Mean2024.10 | 0.0413 | — | |
| MEXAModel=Gemma 2, Parameters=9B, Embedding pooling strategy=Max2024.10 | — | 0.7194 | |
| MEXAModel=Gemma 1, Parameters=7B, Embedding pooling strategy=Max2024.10 | — | 0.5872 | |
| MEXAModel=Llama 3.1, Parameters=70B, Embedding pooling strategy=Max2024.10 | — | 0.7725 | |
| MEXAModel=Llama 3.1, Parameters=8B, Embedding pooling strategy=Max2024.10 | — | 0.6538 | |
| MEXAModel=Llama 3, Parameters=8B, Embedding pooling strategy=Max2024.10 | — | 0.652 | |
| MEXAModel=Llama 2, Parameters=7B, Embedding pooling strategy=Max2024.10 | — | 0.2464 | |
| MEXAModel=Llama 1, Parameters=7B, Embedding pooling strategy=Max2024.10 | — | 0.3579 | |
| MEXAModel=Mistral 0.3, Parameters=7B, Embedding pooling strategy=Max2024.10 | — | 0.4716 | |
| MEXAModel=OLMO 1.7, Parameters=7B, Embedding pooling strategy=Max2024.10 | — | 0.1965 |