Preference alignment on SocialIQA
87.3Preference AlignmentPep
Evaluation Results
| Method | Links | |
|---|---|---|
| Pep2026.02 | 87.3 | |
| Pop. Average2026.02 | 82.1 | |
| GRPO2026.02 | 71.3 | |
| Prompting2026.02 | 31.4 | |
| CollabLLM2026.02 | 24.8 |
Loading the SOTA2 catalog…
| Method | Links | |
|---|---|---|
| Pep2026.02 | 87.3 | |
| Pop. Average2026.02 | 82.1 | |
| GRPO2026.02 | 71.3 | |
| Prompting2026.02 | 31.4 | |
| CollabLLM2026.02 | 24.8 |