Behavioral Deviation Detection on API Providers Llama-3.2-3B, Mistral-7B, Gemma-2-9B
0.058RUTRUT
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| RUTModel=Mistral, Provider=A1002025.06 | 0.058 | — | — | |
| RUTModel=Gemma, Provider=A1002025.06 | 0.06 | — | — | |
| RUTModel=Llama, Provider=A1002025.06 | 0.094 | — | — | |
| RUTModel=Mistral, Provider=HF Inf.2025.06 | 0.188 | — | — | |
| RUTModel=Gemma, Provider=Nebius2025.06 | 0.312 | — | — | |
| RUTModel=Llama, Provider=Nebius2025.06 | 0.962 | — | — | |
| RUTModel=Llama, Provider=Novita2025.06 | 0.988 | — | — | |
| KSModel=Llama, Provider=A1002025.06 | — | — | 0.002 | |
| KSModel=Llama, Provider=Nebius2025.06 | — | — | 0.426 | |
| KSModel=Llama, Provider=Novita2025.06 | — | — | 0.53 | |
| KSModel=Mistral, Provider=A1002025.06 | — | — | 0.004 | |
| KSModel=Mistral, Provider=HF Inf.2025.06 | — | — | 0 | |
| KSModel=Gemma, Provider=A1002025.06 | — | — | 0 | |
| KSModel=Gemma, Provider=Nebius2025.06 | — | — | 0.008 | |
| MMDModel=Llama, Provider=A1002025.06 | — | 0.142 | — | |
| MMDModel=Llama, Provider=Nebius2025.06 | — | 0.944 | — | |
| MMDModel=Llama, Provider=Novita2025.06 | — | 0.996 | — | |
| MMDModel=Mistral, Provider=A1002025.06 | — | 0.138 | — | |
| MMDModel=Mistral, Provider=HF Inf.2025.06 | — | 1 | — | |
| MMDModel=Gemma, Provider=A1002025.06 | — | 0.084 | — | |
| MMDModel=Gemma, Provider=Nebius2025.06 | — | 0.432 | — |