Dishonesty Evaluation on Mistake math (test)
44.16Benchmark DishonestyPRISM
Evaluation Results
| Method | Links | |
|---|---|---|
| PRISMModel=Qwen-3-8B, Mix ratio=1%2026.05 | 44.16 | |
| PRISMModel=Llama-3.1-8B-Instruct, Mix ratio=1%2026.05 | 44.53 | |
| VECFILTERModel=Llama-3.1-8B-Instruct, Mix ratio=1%2026.05 | 45.62 | |
| CONCEPTINFModel=Llama-3.1-8B-Instruct, Mix ratio=1%2026.05 | 45.62 | |
| VECFILTERModel=Qwen-3-8B, Mix ratio=1%2026.05 | 45.62 | |
| SAE-FEATModel=Qwen-3-8B, Mix ratio=1%2026.05 | 45.62 | |
| SAE-FEATModel=Llama-3.1-8B-Instruct, Mix ratio=1%2026.05 | 45.99 | |
| PRISMModel=Llama-3.1-8B-Instruct, Mix ratio=10%2026.05 | 46.35 | |
| PROJDIFFModel=Llama-3.1-8B-Instruct, Mix ratio=1%2026.05 | 47.08 | |
| ICL-EMModel=Llama-3.1-8B-Instruct, Mix ratio=1%2026.05 | 47.08 | |
| PRISMModel=Qwen-3-8B, Mix ratio=10%2026.05 | 47.08 | |
| IF-GUIDEModel=Llama-3.1-8B-Instruct, Mix ratio=10%2026.05 | 47.45 | |
| IF-GUIDEModel=Qwen-3-8B, Mix ratio=1%2026.05 | 47.45 | |
| PROJDIFFModel=Llama-3.1-8B-Instruct, Mix ratio=10%2026.05 | 47.81 | |
| PRISMModel=Llama-3.1-8B-Instruct, Mix ratio=20%2026.05 | 47.81 | |
| SAE-FEATModel=Llama-3.1-8B-Instruct, Mix ratio=10%2026.05 | 48.18 | |
| MixedModel=Qwen-3-8B, Mix ratio=1%2026.05 | 48.18 | |
| MixedModel=Llama-3.1-8B-Instruct, Mix ratio=1%2026.05 | 48.54 | |
| VECFILTERModel=Qwen-3-8B, Mix ratio=10%2026.05 | 48.54 | |
| ICL-EMModel=Qwen-3-8B, Mix ratio=1%2026.05 | 48.54 | |
| CONCEPTINFModel=Qwen-3-8B, Mix ratio=1%2026.05 | 48.91 | |
| IF-GUIDEModel=Llama-3.1-8B-Instruct, Mix ratio=1%2026.05 | 49.64 | |
| CONCEPTINFModel=Qwen-3-8B, Mix ratio=10%2026.05 | 49.64 | |
| MixedModel=Llama-3.1-8B-Instruct, Mix ratio=10%2026.05 | 50.73 | |
| PROJDIFFModel=Qwen-3-8B, Mix ratio=1%2026.05 | 50.73 | |
| CONCEPTINFModel=Llama-3.1-8B-Instruct, Mix ratio=10%2026.05 | 51.1 | |
| CONCEPTINFModel=Llama-3.1-8B-Instruct, Mix ratio=20%2026.05 | 51.1 | |
| IF-GUIDEModel=Qwen-3-8B, Mix ratio=10%2026.05 | 51.1 | |
| PRISMModel=Qwen-3-8B, Mix ratio=20%2026.05 | 51.1 | |
| SAE-FEATModel=Llama-3.1-8B-Instruct, Mix ratio=20%2026.05 | 51.83 | |
| SAE-FEATModel=Qwen-3-8B, Mix ratio=10%2026.05 | 52.19 | |
| SAE-FEATModel=Qwen-3-8B, Mix ratio=20%2026.05 | 52.19 | |
| IF-GUIDEModel=Qwen-3-14B, Mix ratio=1%2026.05 | 52.19 | |
| PROJDIFFModel=Qwen-3-8B, Mix ratio=10%2026.05 | 52.56 | |
| VECFILTERModel=Llama-3.1-8B-Instruct, Mix ratio=20%2026.05 | 52.92 | |
| VECFILTERModel=Qwen-3-8B, Mix ratio=20%2026.05 | 52.92 | |
| VECFILTERModel=Llama-3.1-8B-Instruct, Mix ratio=10%2026.05 | 53.29 | |
| IF-GUIDEModel=Llama-3.1-8B-Instruct, Mix ratio=20%2026.05 | 53.29 | |
| PRISMModel=Qwen-3-14B, Mix ratio=1%2026.05 | 53.29 | |
| MixedModel=Qwen-3-8B, Mix ratio=10%2026.05 | 54.02 | |
| IF-GUIDEModel=Qwen-3-8B, Mix ratio=20%2026.05 | 54.02 | |
| ICL-EMModel=Qwen-3-8B, Mix ratio=20%2026.05 | 54.02 | |
| PROJDIFFModel=Llama-3.1-8B-Instruct, Mix ratio=20%2026.05 | 54.75 | |
| ICL-EMModel=Llama-3.1-8B-Instruct, Mix ratio=20%2026.05 | 54.75 | |
| PROJDIFFModel=Qwen-3-8B, Mix ratio=20%2026.05 | 54.75 | |
| ICL-EMModel=Llama-3.1-8B-Instruct, Mix ratio=10%2026.05 | 55.11 | |
| VECFILTERModel=Qwen-3-8B, Mix ratio=50%2026.05 | 55.11 | |
| ICL-EMModel=Qwen-3-8B, Mix ratio=10%2026.05 | 55.11 | |
| VECFILTERModel=Qwen-3-14B, Mix ratio=1%2026.05 | 55.11 | |
| IF-GUIDEModel=Llama-3.1-8B-Instruct, Mix ratio=50%2026.05 | 55.47 | |
| MixedModel=Llama-3.1-8B-Instruct, Mix ratio=20%2026.05 | 55.84 | |
| SAE-FEATModel=Qwen-3-14B, Mix ratio=1%2026.05 | 56.2 | |
| MixedModel=Qwen-3-8B, Mix ratio=20%2026.05 | 56.57 | |
| CONCEPTINFModel=Qwen-3-8B, Mix ratio=20%2026.05 | 56.57 | |
| CONCEPTINFModel=Qwen-3-14B, Mix ratio=1%2026.05 | 56.57 | |
| CONCEPTINFModel=Llama-3.1-8B-Instruct, Mix ratio=50%2026.05 | 56.93 | |
| ICL-EMModel=Qwen-3-14B, Mix ratio=1%2026.05 | 56.93 | |
| MixedModel=Qwen-3-14B, Mix ratio=1%2026.05 | 57.3 | |
| PROJDIFFModel=Qwen-3-14B, Mix ratio=1%2026.05 | 57.3 | |
| VECFILTERModel=Llama-3.1-8B-Instruct, Mix ratio=50%2026.05 | 58.03 | |
| PRISMModel=Qwen-3-8B, Mix ratio=50%2026.05 | 58.39 | |
| PRISMModel=Llama-3.1-8B-Instruct, Mix ratio=50%2026.05 | 58.76 | |
| IF-GUIDEModel=Qwen-3-8B, Mix ratio=50%2026.05 | 58.76 | |
| PRISMModel=Qwen-3-14B, Mix ratio=10%2026.05 | 58.76 | |
| IF-GUIDEModel=Qwen-3-14B, Mix ratio=10%2026.05 | 59.12 | |
| ICL-EMModel=Llama-3.1-8B-Instruct, Mix ratio=50%2026.05 | 59.49 | |
| MixedModel=Llama-3.1-8B-Instruct, Mix ratio=50%2026.05 | 60.22 | |
| PROJDIFFModel=Llama-3.1-8B-Instruct, Mix ratio=50%2026.05 | 60.22 | |
| PROJDIFFModel=Qwen-3-8B, Mix ratio=50%2026.05 | 60.22 | |
| CONCEPTINFModel=Qwen-3-8B, Mix ratio=50%2026.05 | 60.22 | |
| MixedModel=Qwen-3-8B, Mix ratio=50%2026.05 | 60.58 | |
| VECFILTERModel=Qwen-3-14B, Mix ratio=10%2026.05 | 60.95 | |
| SAE-FEATModel=Llama-3.1-8B-Instruct, Mix ratio=50%2026.05 | 61.31 | |
| ICL-EMModel=Qwen-3-8B, Mix ratio=50%2026.05 | 62.04 | |
| MixedModel=Qwen-3-14B, Mix ratio=10%2026.05 | 62.04 | |
| PROJDIFFModel=Qwen-3-14B, Mix ratio=10%2026.05 | 62.41 | |
| SAE-FEATModel=Qwen-3-8B, Mix ratio=50%2026.05 | 62.77 | |
| PRISMModel=Qwen-3-14B, Mix ratio=20%2026.05 | 62.77 | |
| VECFILTERModel=Qwen-3-14B, Mix ratio=20%2026.05 | 64.23 | |
| PROJDIFFModel=Qwen-3-14B, Mix ratio=20%2026.05 | 64.23 | |
| ICL-EMModel=Qwen-3-14B, Mix ratio=20%2026.05 | 64.23 | |
| IF-GUIDEModel=Qwen-3-14B, Mix ratio=20%2026.05 | 65.33 | |
| PRISMModel=Qwen-3-14B, Mix ratio=50%2026.05 | 65.33 | |
| SAE-FEATModel=Qwen-3-14B, Mix ratio=20%2026.05 | 65.69 | |
| ICL-EMModel=Qwen-3-14B, Mix ratio=10%2026.05 | 65.69 | |
| SAE-FEATModel=Qwen-3-14B, Mix ratio=10%2026.05 | 66.06 | |
| ICL-EMModel=Qwen-3-14B, Mix ratio=50%2026.05 | 66.06 | |
| CONCEPTINFModel=Qwen-3-14B, Mix ratio=20%2026.05 | 66.42 | |
| MixedModel=Qwen-3-14B, Mix ratio=20%2026.05 | 66.79 | |
| SAE-FEATModel=Qwen-3-14B, Mix ratio=50%2026.05 | 66.79 | |
| PROJDIFFModel=Qwen-3-14B, Mix ratio=50%2026.05 | 67.15 | |
| IF-GUIDEModel=Qwen-3-14B, Mix ratio=50%2026.05 | 67.15 | |
| VECFILTERModel=Qwen-3-14B, Mix ratio=50%2026.05 | 67.52 | |
| CONCEPTINFModel=Qwen-3-14B, Mix ratio=10%2026.05 | 67.88 | |
| MixedModel=Qwen-3-14B, Mix ratio=50%2026.05 | 68.61 | |
| CONCEPTINFModel=Qwen-3-14B, Mix ratio=50%2026.05 | 71.9 |