Safety on SafeChain
4.94SafeChain ScoreSafety
Evaluation results
| Method | Links | |
|---|---|---|
| SafetyBase Model=Llama-3.1-8B2026.01 | 4.94 | |
| ReasonAnyMerging Protocol=ReasonAny2026.01 | 4.94 | |
| TIESMerging Protocol=TIES2026.01 | 4.91 | |
| LinearMerging Protocol=Linear2026.01 | 4.88 | |
| ReasonAnyModel Family=Qwen2.5-32B2026.01 | 4.86 | |
| DAREMerging Protocol=DARE2026.01 | 4.86 | |
| Qwen2.5-14B-Instruct (Safety)Model Family=Qwen2.5-14B2026.01 | 4.84 | |
| Qwen2.5-32B-Instruct (Safety)Model Family=Qwen2.5-32B2026.01 | 4.83 | |
| DAREModel Family=Qwen2.5-32B2026.01 | 4.81 | |
| DAREModel Family=Qwen2.5-14B2026.01 | 4.8 | |
| ReasonAnyModel Family=Qwen2.5-14B2026.01 | 4.77 | |
| Task ArithmeticModel Family=Qwen2.5-32B2026.01 | 4.76 | |
| FuseLLMModel Family=Qwen2.5-32B2026.01 | 4.76 | |
| ReasoningBase Model=Llama-3.1-8B2026.01 | 4.76 | |
| Task ArithmeticMerging Protocol=Task Arithmetic2026.01 | 4.74 | |
| FuseLLMMerging Protocol=FuseLLM2026.01 | 4.7 | |
| DeepSeek-R1-Distill-Qwen-32B (Reasoning)Model Family=Qwen2.5-32B2026.01 | 4.65 | |
| LinearModel Family=Qwen2.5-32B2026.01 | 4.6 | |
| LEDModel Family=Qwen2.5-32B2026.01 | 4.58 | |
| DeepSeek-R1-Distill-Qwen-14B (Reasoning)Model Family=Qwen2.5-14B2026.01 | 4.56 | |
| LEDMerging Protocol=LED2026.01 | 4.52 | |
| FuseLLMModel Family=Qwen2.5-14B2026.01 | 4.52 | |
| LinearModel Family=Qwen2.5-14B2026.01 | 4.47 | |
| LEDModel Family=Qwen2.5-14B2026.01 | 4.47 | |
| TIESModel Family=Qwen2.5-32B2026.01 | 4.42 | |
| Task ArithmeticModel Family=Qwen2.5-14B2026.01 | 4.25 | |
| TIESModel Family=Qwen2.5-14B2026.01 | 4.25 |