InferAligner: Inference-Time Alignment for Harmlessness through Cross-Model Guidance.
Pengyu WangDong ZhangLinyang LiChenkun TanXinghao WangKe RenBotian JiangXipeng QiuPublished in: CoRR (2024)
Keyphrases
- computational model
- mathematical model
- cost function
- gibbs sampling
- formal model
- conceptual model
- theoretical framework
- parameter estimation
- probabilistic model
- data sets
- input data
- expert systems
- multi agent systems
- high level
- decision trees
- simulation model
- object model
- network model
- inference engine
- decision theoretic
- database