Safety and Truthfulness Evaluation on TruthfulQA (val)
54AccuracyEvoGM
Evaluation Results
| Method | Links | |
|---|---|---|
| EvoGMBackbone=Qwen2.5-1.5B2026.05 | 54 | |
| CMABackbone=Qwen2.5-1.5B2026.05 | 53 | |
| PSO-MergingBackbone=Qwen2.5-1.5B2026.05 | 52.5 | |
| Single BestBackbone=Qwen2.5-1.5B2026.05 | 49.5 | |
| TIESBackbone=Qwen2.5-1.5B2026.05 | 49.5 | |
| Model SwarmBackbone=Qwen2.5-1.5B2026.05 | 49.5 | |
| Model SoupBackbone=Qwen2.5-1.5B2026.05 | 46.5 | |
| TABackbone=Qwen2.5-1.5B2026.05 | 46.5 | |
| BaseBackbone=Qwen2.5-1.5B2026.05 | 46 | |
| DAREBackbone=Qwen2.5-1.5B2026.05 | 45.5 | |
| MTLBackbone=Qwen2.5-1.5B2026.05 | 36 |