Task Completion on Synthetic personalized interaction datasets (evaluation)
8.48Task Completion ScoreHistory-augmented prompting
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| History-augmented promptingOptimization Status=Vanilla2026.02 | 8.48 | — | |
| Persona-based query rewritingOptimization Status=Vanilla2026.02 | 8.48 | — | |
| Preference-based few-shot ICLOptimization Status=Vanilla2026.02 | 8.48 | — | |
| Controller-guided promptingOptimization Status=Vanilla2026.02 | 8.48 | — | |
| PPOptOptimization Status=Vanilla2026.02 | 8.48 | — | |
| PPOptOptimization Status=PPOpt2026.02 | 8.26 | 0.22 | |
| Preference-based few-shot ICLOptimization Status=PPOpt2026.02 | 7.94 | 0.54 | |
| Controller-guided promptingOptimization Status=PPOpt2026.02 | 7.94 | 0.54 | |
| History-augmented promptingOptimization Status=PPOpt2026.02 | 7.89 | 0.59 | |
| Persona-based query rewritingOptimization Status=PPOpt2026.02 | 7.42 | 1.06 |