Machine Reading Comprehension on Belebele Target language
55.78MRC ScoreLlama-3.2-3B-Instruct
Evaluation results
| Method | Links | |
|---|---|---|
| Llama-3.2-3B-InstructParams=3B2026.03 | 55.78 | |
| FFTModel Scale=13B, Evaluation Protocol=3-shot2025.12 | 49.2 | |
| SSU-WandaModel Scale=13B, Evaluation Protocol=3-shot2025.12 | 48.6 | |
| SSU-RandModel Scale=13B, Evaluation Protocol=3-shot2025.12 | 47.8 | |
| HFTModel Scale=13B, Evaluation Protocol=3-shot2025.12 | 47.6 | |
| GMTModel Scale=13B, Evaluation Protocol=3-shot2025.12 | 47.3 | |
| SSU-MagModel Scale=13B, Evaluation Protocol=3-shot2025.12 | 46.7 | |
| SSU-WandaModel Scale=7B, Evaluation Protocol=3-shot2025.12 | 40.3 | |
| SSU-RandModel Scale=7B, Evaluation Protocol=3-shot2025.12 | 39.4 | |
| FFTModel Scale=7B, Evaluation Protocol=3-shot2025.12 | 39.3 | |
| GMTModel Scale=7B, Evaluation Protocol=3-shot2025.12 | 38.5 | |
| SSU-MagModel Scale=7B, Evaluation Protocol=3-shot2025.12 | 38.3 | |
| HFTModel Scale=7B, Evaluation Protocol=3-shot2025.12 | 37.7 | |
| SourceModel Scale=13B, Evaluation Protocol=3-shot2025.12 | 37.4 | |
| AdaLoRAModel Scale=13B, Evaluation Protocol=3-shot2025.12 | 36.5 | |
| Trendyol-LLM-7b-baseParams=7B2026.03 | 36.22 | |
| SourceModel Scale=7B, Evaluation Protocol=3-shot2025.12 | 33.4 | |
| AdaLoRAModel Scale=7B, Evaluation Protocol=3-shot2025.12 | 31.8 | |
| vngrs-ai/Kumru-2BParams=2B2026.03 | 29 | |
| asafaya/kanarya-2bParams=2B2026.03 | 28.11 | |
| DiffutronLM-0.3B-2nd-StageParams=0.3B2026.03 | 27 | |
| CohereForAI/aya-101Params=13B2026.03 | 22.89 | |
| boun-tabi-LMG/TURNAParams=1.1B2026.03 | 22.56 | |
| DiffutronLM-0.3B-1st-StageParams=0.3B2026.03 | 22.22 |