Machine Reading Comprehension on BELEBELE Indonesian
67.1Accuracy (Target Language)CLO
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| CLOModel=Qwen2.5-3B, Zero-shot=true2025.05 | 67.1 | 83.1 | |
| SFTModel=Qwen2.5-3B, Zero-shot=true2025.05 | 65.3 | 81.8 | |
| CLOModel=Llama-3-8B, Zero-shot=true2025.05 | 58.7 | 75 | |
| SFTModel=Llama-3-8B, Zero-shot=true2025.05 | 57 | 75.2 | |
| CLOModel=Mistral-7B-v0.1, Zero-shot=true2025.05 | 53.4 | 75.7 | |
| SFTModel=Mistral-7B-v0.1, Zero-shot=true2025.05 | 51.1 | 72.3 | |
| CLOModel=Llama-2-13B, Zero-shot=true2025.05 | 47.8 | 61.3 | |
| SFTModel=Llama-2-13B, Zero-shot=true2025.05 | 46.7 | 60.9 | |
| Transport and Mergeadaptation=post-fusion2026.02 | 36.9 | — | |
| Transport and Mergeadaptation=none2026.02 | 36.4 | — | |
| LLaMA-3-1B-Indonesian-QLoRAadaptation=none2026.02 | 34.8 | — | |
| CLOModel=Llama-2-7B, Zero-shot=true2025.05 | 31.7 | 36.6 | |
| SFTModel=Llama-2-7B, Zero-shot=true2025.05 | 30 | 35.8 |