General Reasoning on Global MMLU
36.1MMLUFFT
Evaluation results
| Method | Links | |
|---|---|---|
| FFTModel Scale=13B, Evaluation Protocol=5-shot2025.12 | 36.1 | |
| SSU-WandaModel Scale=13B, Evaluation Protocol=5-shot2025.12 | 35.9 | |
| SSU-RandModel Scale=13B, Evaluation Protocol=5-shot2025.12 | 35.6 | |
| HFTModel Scale=13B, Evaluation Protocol=5-shot2025.12 | 35.5 | |
| GMTModel Scale=13B, Evaluation Protocol=5-shot2025.12 | 35.3 | |
| SSU-MagModel Scale=13B, Evaluation Protocol=5-shot2025.12 | 35 | |
| SSU-WandaModel Scale=7B, Evaluation Protocol=5-shot2025.12 | 33.3 | |
| SourceModel Scale=13B, Evaluation Protocol=5-shot2025.12 | 32.9 | |
| FFTModel Scale=7B, Evaluation Protocol=5-shot2025.12 | 32.5 | |
| SSU-RandModel Scale=7B, Evaluation Protocol=5-shot2025.12 | 32.4 | |
| HFTModel Scale=7B, Evaluation Protocol=5-shot2025.12 | 32.2 | |
| GMTModel Scale=7B, Evaluation Protocol=5-shot2025.12 | 31.9 | |
| SSU-MagModel Scale=7B, Evaluation Protocol=5-shot2025.12 | 31.9 | |
| AdaLoRAModel Scale=13B, Evaluation Protocol=5-shot2025.12 | 31.1 | |
| SourceModel Scale=7B, Evaluation Protocol=5-shot2025.12 | 30.4 | |
| AdaLoRAModel Scale=7B, Evaluation Protocol=5-shot2025.12 | 29.9 |