Knowledge Grounded Generation with Inference-Time Optimization and Constrained Decoding

The Hong Kong University of Science and Technology
Department of Computer Science and Engineering


MPhil Thesis Defence


Title: "Knowledge Grounded Generation with Inference-Time Optimization and 
Constrained Decoding"

By

Mr. Sehyun CHOI


Abstract:

Large Language Models (LLMs) have demonstrated remarkable human-level natural 
language generation capabilities. However, their potential to generate 
misinformation, often called the hallucination problem, poses a significant 
risk to their deployment. A common approach to address this issue is to 
retrieve relevant knowledge and fine-tune the LLM with the knowledge in its 
input. Unfortunately, this method incurs high training costs and may cause 
catastrophic forgetting for multi-tasking models. To overcome these 
limitations, this thesis propose a knowledge-constrained decoding method called 
KCTS (Knowledge-Constrained Tree Search), which guides a frozen LM to generate 
text aligned with the reference knowledge at each decoding step using a 
knowledge classifier score and a future reward-aware optimization algorithm, 
MCTS (Monte-Carlo Tree Search). To adapt the sequence-level knowledge 
classifier to token-level guidance, a novel token-level hallucination detection 
method called RIPA (Reward Inflection Point Approximation) is also proposed. 
The empirical results on knowledge-grounded dialogue and abstractive 
summarization demonstrate the strength of KCTS as a plug-and-play, 
model-agnostic decoding method that can effectively reduce hallucinations in 
natural language generation.


Date:                   Wednesday, 26 June 2024

Time:                   10:00am - 12:00noon

Venue:                  Room 3494
                        Lifts 25/26

Chairman:               Prof. Raymond WONG

Committee Members:      Dr. Yangqiu SONG (Supervisor)
                        Dr. Junxian HE