Behavior Generation on BiPO
1.59Hallucination ScoreBase
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| BaseBackbone=Llama-2-7b-chat-hf2026.03 | 1.59 | 2 | 2.48 | |
| COLD-KernelBackbone=Llama-2-7b-chat-hf2026.03 | 1.62 | 2.02 | 2.48 | |
| ReFT(vector)Backbone=Llama-2-7b-chat-hf2026.03 | 1.63 | 2 | 2.42 | |
| DiffMeanBackbone=Llama-2-7b-chat-hf2026.03 | 1.71 | 2.22 | 2.58 | |
| COLD-FDBackbone=Llama-2-7b-chat-hf2026.03 | 3.87 | 2.15 | 2.6 |