BRAIn: Bayesian Reward-conditioned Amortized Inference for natural language generation from feedback.
Gaurav PandeyYatin NandwaniTahira NaseemMayank MishraGuangxuan XuDinesh RaghuSachindra JoshiAsim MunawarRamón Fernandez AstudilloPublished in: CoRR (2024)
Keyphrases
- natural language generation
- bayesian networks
- bayesian inference
- natural language processing
- natural language
- bayesian model
- aggregated search
- dialog systems
- text generation
- statistical inference
- human brain
- dialogue system
- machine translation
- maximum likelihood
- relevance feedback
- reinforcement learning
- search tree
- dialogue management
- posterior probability
- word order
- posterior distribution
- running times
- user feedback
- arbitrary length
- search engine
- worst case
- probability distribution
- knowledge base