Prosocial Alignment on HEx-PHI (test)
76.3MIPPROSOCIALALIGN
Evaluation Results
| Method | Links | |
|---|---|---|
| PROSOCIALALIGNBackbone=llama2025.12 | 76.3 | |
| DIREGBackbone=llama2025.12 | 67.3 | |
| PPBackbone=llama2025.12 | 67 | |
| PROATTR-GEN-PCABackbone=llama2025.12 | 66.7 | |
| CTRL-GENBackbone=llama2025.12 | 65.3 | |
| PROATTR-GEN-PCABackbone=mistral2025.12 | 63.8 | |
| PROSOCIALALIGNBackbone=mistral2025.12 | 59.7 | |
| PV-ARM-SUMBackbone=llama2025.12 | 57.6 | |
| CTRL-GENBackbone=mistral2025.12 | 56.1 | |
| DIREGBackbone=mistral2025.12 | 53.9 | |
| SAFE-ARITHBackbone=llama2025.12 | 52.5 | |
| PPBackbone=mistral2025.12 | 51.1 | |
| SAFE-ARITHBackbone=mistral2025.12 | 34.5 | |
| PV-ARM-SUMBackbone=mistral2025.12 | 33.5 |