Papers by Zhenyu Zhao
Dialog-Post: Multi-Level Self-Supervised Objectives and Hierarchical Model for Dialogue Post-Training (2023.acl-long)
Copied to clipboard
| Challenge: | a new method for dialogue representation and understanding is proposed . pre-trained language models (PLMs) are inappropriate for dialogue understanding tasks . |
| Approach: | They propose a method that trains pre-trained language models to fit dialogues . they use a hierarchical segment-wise self-attention network to model dialogues more comprehensively . |
| Outcome: | The proposed method outperforms existing models and achieves a 3.3% improvement on average. |
Label Anchored Contrastive Learning for Language Understanding (2022.naacl-main)
Copied to clipboard
| Challenge: | a novel approach to contrastive learning for language understanding is not fully explored . contrastive training has been widely applied to self-supervised representation learning . |
| Approach: | They propose a label anchored contrastive learning approach for language understanding using a class label. |
| Outcome: | The proposed approach improves on GLUE and CLUE benchmarks by 4.1% compared to the state-of-the-art approaches . the proposed approach also improves under the few-shot and data imbalance settings . |
Hierarchical Acoustic-Semantic Modeling: Modality Separation and Semantic Coherence for Full-Duplex SLMs (2026.acl-long)
Copied to clipboard
Zhenyu Liu, Xuanyu Zhang, Yunxin li, Qixun Teng, Shenyuan Jiang, Haolan Chen, Mingjun Zhao, Fanbo Meng, Yu Xu, Yancheng He, Baotian Hu, Haizhou Li, Min Zhang
| Challenge: | despite significant progress, full-duplex SLMs are constrained by severe modality interference, authors say . modality interferes with acoustic and semantic modeling, making them unintelligent and unnatural . authors propose a hierarchical parameter separation strategy that decouples conflicting modalities in deep layers . |
| Approach: | They propose a hierarchical parameter separation strategy that decouples conflicting modalities in deep layers while preserving cross-modality coherence via a dedicated semantic alignment channel. |
| Outcome: | The proposed method significantly advances the state of the art on full-duplex benchmarks . it decouples conflicting modalities in deep layers while preserving cross-modality coherence . |
FunnelRAG: A Coarse-to-Fine Progressive Retrieval Paradigm for RAG (2025.findings-naacl)
Copied to clipboard
| Challenge: | Retrieval-Augmented Generation (RAG) is widely adopted in Large Language Models, but is flat and has limitations such as a significant burden on one retriever and constant granularity limits the ceiling of retrieval performance. |
| Approach: | They propose a progressive retrieval paradigm with coarse-to-fine granularity for RAG, termed FunnelRAG, so as to balance effectiveness and efficiency. |
| Outcome: | The proposed paradigm achieves comparable retrieval performance while the time overhead is reduced by nearly 40%. |
Robust Machine Reading Comprehension by Learning Soft labels (2020.coling-main)
Copied to clipboard
| Challenge: | Neural models have achieved great success on the task of machine reading comprehension, which are typically trained on hard labels. |
| Approach: | They propose a robust training method for machine reading comprehension models to address label sparseness problem by using three strategies to train models on soft labels. |
| Outcome: | The proposed method improves the baseline model performance and achieves state-of-the-art performance on NewsQA and QUOREF. |
SDA: Simple Discrete Augmentation for Contrastive Sentence Representation Learning (2024.lrec-main)
Copied to clipboard
| Challenge: | Existing methods for data augmentation have not been well explored. |
| Approach: | They propose to use punctuation insertion, modal verbs, and double negation to produce diverse forms of sentences. |
| Outcome: | The proposed methods perform better on diverse datasets with semantic similarity and standard negation. |
DKME: Rethinking Coupled Knowledge Memory for Lifelong Model Editing of Large Language Models (2026.findings-acl)
Copied to clipboard
| Challenge: | Existing memory-based editors suffer from catastrophic forgetting as edits accumulate. |
| Approach: | They propose a method which injects factual updates into large language models without retraining or finetuning into existing memory-based editors. |
| Outcome: | Experiments on HalluEditBench, CKnowEdit, and WikiDatacounterfact show that the proposed model achieves a more favorable trade-off between editing success and locality compared to baselines while maintaining more stable performance as the edit scale increases. |
M2PA: A Multi-Memory Planning Agent for Open Worlds Inspired by Cognitive Theory (2025.findings-acl)
Copied to clipboard
YanfangZhou YanfangZhou, Xiaodong Li, Yuntao Liu, Yongqiang Zhao, Xintong Wang, Zhenyu Li, Jinlong Tian, Xinhai Xu
| Challenge: | Open-world planning poses a challenge due to complex environments and task diversity . recent work shows that large language models (LLMs) lack the ability to connect to agents' experiences . |
| Approach: | They propose an open-world multi-memory planning agent that combines large language models with human-like multi-mesh systems to leverage their strengths. |
| Outcome: | The proposed agent outperforms state-of-the-art agents on 50 Minecraft tasks in zero-shot learning. |
StructuThink: Reasoning with Task Transition Knowledge for Autonomous LLM-Based Agents (2025.findings-emnlp)
Copied to clipboard
| Challenge: | StructuThink framework enhances LLMs' ability to ground decisions in domain-specific scenarios. |
| Approach: | They propose a knowledge-structured reasoning framework that enhances LLM-based agents with explicit decision constraints. |
| Outcome: | The proposed framework achieves higher task success rates and more efficient action sequences than baseline methods. |
Efficient Dialogue Complementary Policy Learning via Deep Q-network Policy and Episodic Memory Policy (2021.emnlp-main)
Copied to clipboard
| Challenge: | Existing methods for training dialogue policies rely on a single learning system, but it requires many rounds of interaction. |
| Approach: | They propose a complementary policy learning framework which exploits the complementary advantages of the episodic memory (EM) policy and the deep Q-network (DQN) policy. |
| Outcome: | The proposed framework outperforms existing methods relying on a single learning system on three dialogue datasets. |
Take Off the Training Wheels! Progressive In-Context Learning for Effective Alignment (2024.emnlp-main)
Copied to clipboard
| Challenge: | Recent studies have explored the working mechanisms of In-Context Learning (ICL) however, they mainly focus on classification and simple generation tasks, limiting their broader application to more complex generation tasks in practice. |
| Approach: | They propose an efficient Progressive In-Context Alignment method that embeds the task function learned from demonstrations into the separator token representation. |
| Outcome: | The proposed method surpasses vanilla ICL and achieves comparable performance to other alignment tuning methods. |
ChatGLM-Math: Improving Math Problem-Solving in Large Language Models with a Self-Critique Pipeline (2024.findings-emnlp)
Copied to clipboard
Yifan Xu, Xiao Liu, Xinghan Liu, Zhenyu Hou, Yueyan Li, Xiaohan Zhang, Zihan Wang, Aohan Zeng, Zhengxiao Du, Zhao Wenyi, Jie Tang, Yuxiao Dong
| Challenge: | Large language models (LLMs) have shown excellent mastering of human language but struggle in real-world applications that require mathematical problem-solving. |
| Approach: | They propose a pipeline to train a general Math-Critique model from the LLM itself to provide feedback signals and employ rejective fine-tuning and direct preference optimization over the Llm's own generations for data collection. |
| Outcome: | The proposed pipeline outperforms existing LLMs that could be two times larger. |
A Versatile Adaptive Curriculum Learning Framework for Task-oriented Dialogue Policy Learning (2022.findings-naacl)
Copied to clipboard
| Challenge: | Existing training paradigms for dialogue policy learning with brute-force random sampling are expensive and lack reliable evaluation of difficulty scores. |
| Approach: | They propose a flexible adaptive curriculum learning framework that integrates curriculum learning with a generic global curriculum. |
| Outcome: | The proposed framework improves learning performance and efficiency on three public dialogue datasets. |
E-Bench: Towards Evaluating the Ease-of-Use of Large Language Models (2025.coling-main)
Copied to clipboard
| Challenge: | E-Bench is a framework for easy-to-use research on large language models. |
| Approach: | They propose to evaluate the ease-of-use of large language models and construct an E-Bench . they simulate human use from synonymous and typographical perturbations . |
| Outcome: | The proposed model is able to resist synonymous expressions and typos and improves performance. |
Token-Budget-Aware LLM Reasoning (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods to enhance reasoning capabilities of large language models incur significant overhead in token usage, leading to increased costs. |
| Approach: | They propose a token-budget-aware LLM reasoning framework that adjusts the number of reasoning tokens based on the reasoning complexity of each problem. |
| Outcome: | The proposed method reduces token costs in CoT reasoning with only a slight performance reduction. |
UniMoE-Audio: Unified Speech and Music Generation with Dynamic-Capacity Mixture-of-Experts (2026.acl-long)
Copied to clipboard
Zhenyu Liu, Yunxin li, Xuanyu Zhang, Qixun Teng, Shenyuan Jiang, Xinyu Chen, Haoyuan Shi, Haolan Chen, Fanbo Meng, Mingjun Zhao, Yu Xu, Yancheng He, Baotian Hu, Haizhou Li, Min Zhang
| Challenge: | Recent advances in unified multimodal models indicate a clear trend towards comprehensive content generation. |
| Approach: | They propose a unified speech and music generation model built upon a novel framework . they propose specialized MoE architectures and curated training strategies to tackle data imbalances . |
| Outcome: | The proposed model achieves state-of-the-art performance on major speech and music generation benchmarks. |
Enhancing Reinforcement Learning with Label-Sensitive Reward for Natural Language Understanding (2024.acl-long)
Copied to clipboard
| Challenge: | Recent advances in large language models (LLMs) have yielded remarkable performance, but objective mismatch issues hinder RLHF learning. |
| Approach: | They propose a Reinforcement Learning framework enhanced with Label-sensitive reward to enhance LLMs' alignment and generation capabilities. |
| Outcome: | The proposed framework improves performance on five diverse models across eight tasks. |
Medico: Towards Hallucination Detection and Correction with Multi-source Evidence Fusion (2024.emnlp-demo)
Copied to clipboard
| Challenge: | Existing studies show that LLMs can confidently state non-existent facts rather than answering "I don't know". |
| Approach: | They propose a multi-source evidence fusion enhanced hallucination detection and correction framework that fuses evidence from multiple sources and iteratively revises the hallucinous content. |
| Outcome: | The proposed framework detects whether the generated content contains factual errors, provides the rationale behind the judgment, and iteratively revises the hallucinated content. |
Strengthened Symbol Binding Makes Large Language Models Reliable Multiple-Choice Selectors (2024.acl-long)
Copied to clipboard
| Challenge: | Multiple-Choice Questions (MCQs) are a critical area of research in the study of Large Language models (LLMs). |
| Approach: | They propose an efficient SFT algorithm for MCQs, termed Point-wise Intelligent Feedback, which constructs negative instances by randomly combing the incorrect option contents with all candidate symbols. |
| Outcome: | The proposed algorithm significantly reduces the model’s selection bias by improving its MCSB capability. |
An Efficient Dialogue Policy Agent with Model-Based Causal Reinforcement Learning (2025.coling-main)
Copied to clipboard
| Challenge: | Existing models for dialogue policy training consider one-step dialogues, leading to inaccurate simulations. |
| Approach: | They propose a framework for dialogue policy learning that trains an agent to select dialogue actions via deep reinforcement learning. |
| Outcome: | The proposed framework achieves state-of-the-art performance on three dialogue datasets . it uses model-based reinforcement learning with automatically constructed causal chains . |
Metagent-P: A Neuro-Symbolic Planning Agent with Metacognition for Open Worlds (2025.findings-acl)
Copied to clipboard
YanfangZhou YanfangZhou, Yuntao Liu, Xiaodong Li, Yongqiang Zhao, Xintong Wang, Jinlong Tian, Zhenyu Li, Xinhai Xu
| Challenge: | Recent advances in large language models (LLMs) show promising potential through their world knowledge and language processing capabilities in open-world planning. |
| Approach: | They propose a framework that integrates the world knowledge of large language models, symbolic reasoning capabilities of cognitive architectures, and metacognition to improve experience utilization. |
| Outcome: | The proposed framework outperforms current state-of-the-art methods in Minecraft and reduces the average replanning counts by 34% and exceeds the human success rate by 18.96%. |