Safety Evaluation on BeaverTails
99.4Safety ScoreTELLME NT-Xent
Evaluation Results
| Method | Links | |
|---|---|---|
| TELLME NT-XentBase Model=Gemma-2-9B-IT, Loss Function=NT-Xent2025.02 | 99.4 | |
| TELLMEBase Model=Gemma-2-9B-IT2025.02 | 99.1 | |
| OriginBase Model=Gemma-2-9B-IT2025.02 | 98 | |
| SFTBase Model=Gemma-2-9B-IT, Training Strategy=Supervised Fine-Tuning2025.02 | 97.6 | |
| IQuest-Coder-V1-40B-ThinkingParameters=40B, Type=Thinking2026.03 | 76.7 | |
| Qwen3-Coder-480B-A35B-InstructParameters=480B-A35B, Type=Instruct2026.03 | 70.5 | |
| Qwen2.5-Coder-32B-InstructParameters=32B, Type=Instruct2026.03 | 68 | |
| IQuest-Coder-V1-40B-InstructParameters=40B, Type=Instruct2026.03 | 67.5 |