Future Interaction Prediction on AOL (test)
40.18Hit Rate @10pi_theta (Ours)
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| pi_theta (Ours)Backbone=Gemma3-27B2026.04 | 40.18 | 58.17 | 67.82 | 4.45 | 5.17 | 5.31 | |
| PersonaXsBackbone=Gemma3-27B2026.04 | 32.29 | 48.98 | 58.67 | 2.77 | 3.17 | 3.29 | |
| Qwen3-80bEvaluation protocol=prompt-only2026.04 | 30.55 | 46.89 | 58.5 | 2.81 | 3.34 | 3.47 | |
| GPT-5.1Evaluation protocol=prompt-only2026.04 | 29.27 | 48.79 | 59 | 2.65 | 3.16 | 3.29 | |
| GPT-oss-120bEvaluation protocol=prompt-only2026.04 | 26.95 | 43.55 | 55.33 | 2.41 | 2.86 | 2.97 | |
| Claude-4.5Evaluation protocol=prompt-only2026.04 | 26.01 | 44.74 | 57.14 | 2.16 | 2.6 | 2.71 | |
| PersonaXrBackbone=Gemma3-27B2026.04 | 12.11 | 32.59 | 46.2 | 0.42 | 0.59 | 0.65 |