LLM Safety Assessment on Scenario Evaluation (test)
97.75Not Unsafe Rate (Content)Mi:dm 2.0 Base-inst
Evaluation results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| Mi:dm 2.0 Base-instModel Size=Base, Tuning=Instruction-tuned2026.01 | 97.75 | 94.12 | 83.16 | 92.45 | 78.78 | 90.9 | 100 | 86.95 | |
| Mi:dm 2.0 Mini-instModel Size=Mini, Tuning=Instruction-tuned2026.01 | 89.12 | 83.12 | 75 | 83.09 | 96.96 | 95.45 | 100 | 97.1 | |
| Exaone-3.5-7.8B-instParameters=7.8B, Tuning=Instruction-tuned2026.01 | 87.87 | 78.62 | 77.16 | 73.59 | 100 | 100 | 100 | 100 | |
| Llama-3.1-8B-instParameters=8B, Tuning=Instruction-tuned2026.01 | 79.62 | 75.25 | 63.33 | 81.59 | 100 | 100 | 100 | 100 | |
| Exaone-3.5-2.4B-instParameters=2.4B, Tuning=Instruction-tuned2026.01 | 71.25 | 65.25 | 61.16 | 66.31 | 100 | 95.45 | 92.85 | 97.1 |