Papers by Fanqi Wan
Knowledge Verification to Nip Hallucination in the Bud (2024.emnlp-main)
Copied to clipboard
| Challenge: | Recent studies have shown that large language models generate responses that sound plausible but contradict factual knowledge, a phenomenon known as hallucination. |
| Approach: | They propose a novel approach to align large language models to evaluate knowledge boundaries based on external knowledge to reduce hallucinations . |
| Outcome: | The proposed approach reduces hallucinations across six benchmarks using foundation LLMs of varying backbones and scales. |
PaCoRe: Learning to Scale Test-Time Compute with Parallel Coordinated Reasoning (2026.acl-long)
Copied to clipboard
Jingcheng Hu, Yinmin Zhang, Shijie Shang, Xiaobo Yang, Yue Peng, Zhewei Huang, Hebin Zhou, Xin Wu, Jie Cheng, Fanqi Wan, Xiangwen Kong, Chengyuan Yao, Kaiwen Yan, Ailin Huang, Hongyu Zhou, Qi Han, Zheng Ge, Xiangyu Zhang, Heung-Yeung Shum
| Challenge: | Parallel Coordinated Reasoning (PaCoRe) overcomes a central limitation of contemporary language models: their inability to scale test-time compute (TTC) far beyond sequential reasoning under a fixed context window. |
| Approach: | They propose a training-and-inference framework to overcome a central limitation of language models: their inability to scale test-time compute (TTC) under a fixed context window. |
| Outcome: | The proposed model scales to multi-million-token effective TTC without exceeding context limits. |
Mutual-Taught for Co-adapting Policy and Reward Models (2025.acl-long)
Copied to clipboard
Tianyuan Shi, Canbin Huang, Fanqi Wan, Longguang Zhong, Ziyi Yang, Weizhou Shen, Xiaojun Quan, Ming Yan
| Challenge: | Experimental results show that this iterative approach leads to consistent improvements in both the policy model and reward model. |
| Approach: | They propose a method that iteratively improves both the policy model and reward model without requiring additional human annotation. |
| Outcome: | The proposed method improves both the policy model and reward model without human annotation. |
Clustering-Aware Negative Sampling for Unsupervised Sentence Representation (2023.findings-acl)
Copied to clipboard
| Challenge: | Using clustering-aware learning, in-batch negatives are often ignored in sentence representation learning. |
| Approach: | They propose a method that integrates cluster information into contrastive learning for unsupervised sentence representation learning. |
| Outcome: | The proposed method compares favorably with baselines on semantic textual similarity tasks. |
Self-Evolution Fine-Tuning for Policy Optimization (2024.findings-emnlp)
Copied to clipboard
| Challenge: | Recent years have showcased the remarkable capabilities and performance of large language models (LLMs) across a broad range of tasks. |
| Approach: | They propose supervised fine-tuning (SEFT) for LLM alignment to eliminate the need for annotated samples while retaining the stability and efficiency of SFT. |
| Outcome: | The proposed method eliminates the need for annotated samples while maintaining the stability and efficiency of SFT. |
PsyCoT: Psychological Questionnaire as Powerful Chain-of-Thought for Personality Detection (2023.findings-emnlp)
Copied to clipboard
| Challenge: | Recent advances in large language models (LLMs) have demonstrated remarkable zero-shot performance across various NLP tasks. |
| Approach: | They propose a method which mimics the way individuals complete psychological questionnaires in a multi-turn dialogue manner and prompts an LLM to rate individual items at each turn. |
| Outcome: | The proposed method improves the performance and robustness of the standard GPT-3.5 personality detection task on two benchmark datasets. |
ProactiveEval: A Unified Evaluation Framework for Proactive Dialogue Agents (2026.acl-long)
Copied to clipboard
| Challenge: | Existing studies on proactive dialogue models focus on domain-specific or task-oriented scenarios, which leads to fragmented evaluations and limits the comprehensive exploration of models’ proactive dialogue abilities. |
| Approach: | They propose a framework for evaluating proactive dialogue capabilities of large language models that decomposes proactive dialogue into target planning and dialogue guidance, establishing evaluation metrics across various domains. |
| Outcome: | The proposed framework decomposes proactive dialogue into target planning and dialogue guidance, establishing evaluation metrics across various domains, and enables automatic generation of diverse and challenging evaluation data. |
Explore-Instruct: Enhancing Domain-Specific Instruction Coverage through Active Exploration (2023.emnlp-main)
Copied to clipboard
| Challenge: | Existing data for instruction-tuning are inadequate for a wide range of tasks, limiting the scope for nuanced comprehension and interactions within these domains. |
| Approach: | They propose to use Large Language Models to explore a multitude of variations or possibilities to improve instruction-tuning data by active exploration. |
| Outcome: | The proposed approach improves domain-specific instruction coverage and shows significant improvements over baselines. |
FuseChat: Knowledge Fusion of Chat Models (2025.emnlp-main)
Copied to clipboard
| Challenge: | Large language models (LLMs) are costly and require significant computational resources and time. |
| Approach: | They propose a fuse-and-merge framework for the knowledge fusion of chat LLMs . they conduct pairwise knowledge fusing on source chat LRMs to create multiple target LLM . |
| Outcome: | The proposed framework is superior to baselines of various sizes. |
ReAlign: Structured Revision for Small Language Model Alignment (2025.findings-emnlp)
Copied to clipboard
| Challenge: | weak policies struggle to generate informative on-policy samples and suffer from unstable gradients when trained on off-police signals from stronger models. |
| Approach: | They propose a training framework that combines stability of on-policy learning with reviser-assisted supervision. |
| Outcome: | The proposed training framework outperforms strong preference optimization baselines on AlpacaEval-2 and Arena-Hard. |
BlockPruner: Fine-grained Pruning for Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) have significant computational and memory costs associated with training and inference. |
| Approach: | They propose a training-free structured pruning approach that targets redundancies in MHA and MLP blocks. |
| Outcome: | The proposed pruning approach achieves more granular and effective pruning compared to state-of-the-art pruning methods. |
Retrieval-Generation Alignment for End-to-End Task-Oriented Dialogue System (2023.emnlp-main)
Copied to clipboard
| Challenge: | generative models struggle to distinguish subtle differences among retrieved knowledge records, resulting in suboptimal quality of generated responses. |
| Approach: | They propose to use maximum marginal likelihood to train a perceptive retriever by utilizing signals from response generation for supervision. |
| Outcome: | The proposed approach improves on three task-oriented dialogue datasets using T5 and ChatGPT as the backbone models. |
Multi-Grained Knowledge Retrieval for End-to-End Task-Oriented Dialog (2023.acl-long)
Copied to clipboard
| Challenge: | Existing systems blend knowledge retrieval with response generation and optimize them with direct supervision from reference responses. |
| Approach: | They propose a multi-grained knowledge retrieval system that decouples knowledge retrievals from response generation and introduces an entity selector and an attribute selector to acquire multigrained information from the knowledge base. |
| Outcome: | The proposed system performs better on small and large knowledge bases. |