Papers by Qinyuan Cheng
Explicit Memory Learning with Expectation Maximization (2024.emnlp-main)
Copied to clipboard
| Challenge: | Large Language Models lack reliable learning mechanisms for updating information across interactions. |
| Approach: | They propose a framework that enhances explicit memory updates via the Expectation-Maximization algorithm. |
| Outcome: | The proposed framework outperforms existing methods without memory or with static external memory on streaming inference tasks. |
WESR: A Benchmark and Strong Baseline for Word-level Event-Speech Recognition (2026.findings-acl)
Copied to clipboard
Chenchen Yang, Kexin Huang, Liwei Fan, Qian Tu, Botian Jiang, Dong Zhang, Linqi Yin, Shimin Li, Zhaoye Fei, Qinyuan Cheng, Xipeng Qiu
| Challenge: | aaron carroll: the precise localization of non-verbal vocal events remains a critical yet under-explored challenge. carroll says current methods suffer from insufficient task definitions with limited category coverage. carrol: knowing exactly where an event occurred is not enough; knowing exactly what it happened is. |
| Approach: | They propose a taxonomy of 21 vocal events with a new categorization into discrete versus continuous types. |
| Outcome: | The proposed model disentangles ASR errors from event detection while maintaining ASR quality. |
Revisiting the Test-Time Scaling of o1-like Models: Do they Truly Possess Test-Time Scaling Capabilities? (2025.acl-long)
Copied to clipboard
| Challenge: | Longer CoTs of o1-like models do not consistently enhance accuracy, causing performance degradation. |
| Approach: | They propose a method that combines parallel scaling strategies with CoT length characteristics to improve models’ test-time scalability. |
| Outcome: | The proposed method improves models’ test-time scalability compared to majority voting approaches. |
Inference-Time Decontamination: Reusing Leaked Benchmarks for Large Language Model Evaluation (2024.findings-emnlp)
Copied to clipboard
| Challenge: | Current LLMs are achieving better performance on various benchmarks, but their performance in practical applications does not always match their benchmark results. |
| Approach: | They propose to detect and rewrite leaked benchmarks without altering their difficulties by using Inference-Time Decontamination (ITD) to mitigate performance inflation caused by memorizing leaked samples. |
| Outcome: | The proposed method reduces inflated accuracy by 22.9% on GSM8K and 19.0% on MMLU. |
XY-Tokenizer: Mitigating the Semantic-Acoustic Conflict in Low-Bitrate Speech Codecs (2026.acl-long)
Copied to clipboard
Yitian Gong, Luozhijie Jin, Kuangwei Chen, Dong Zhang, Ruifan Deng, Xiaogui Yang, Xin Zhang, Zhaoye Fei, Qinyuan Cheng, Shimin Li, Xipeng Qiu
| Challenge: | Existing speech codecs struggle to balance these objectives at low bitrates . XY-Tokenizer achieves stronger semantic alignment than representative semantic-distillation codec . |
| Approach: | They propose a low-bitrate speech codec that aligns discrete speech representations with text while preserving fine-grained acoustic details for reconstruction. |
| Outcome: | The proposed codec outperforms existing low-bitrate speech codecs in speech understanding and generation tasks. |
Reasoning in Flux: Enhancing Large Language Models Reasoning through Uncertainty-aware Adaptive Guidance (2024.acl-long)
Copied to clipboard
Zhangyue Yin, Qiushi Sun, Qipeng Guo, Zhiyuan Zeng, Xiaonan Li, Junqi Dai, Qinyuan Cheng, Xuanjing Huang, Xipeng Qiu
| Challenge: | Extensive experiments across various reasoning tasks demonstrate that UAG not only enhances the reasoning abilities of LLMs but consistently outperforms several strong baselines with minimal computational overhead. |
| Approach: | They propose an approach to guide LLMs onto an accurate and reliable trajectory by identifying and adjusting uncertainty signals within each step of the reasoning chain. |
| Outcome: | The proposed approach outperforms strong baselines and outperformed strong models with minimal computational overhead. |
Safe Inputs but Unsafe Output: Benchmarking Cross-modality Safety Alignment of Large Vision-Language Models (2025.findings-naacl)
Copied to clipboard
Siyin Wang, Xingsong Ye, Qinyuan Cheng, Junwen Duan, Shimin Li, Jinlan Fu, Xipeng Qiu, Xuanjing Huang
| Challenge: | Recent studies focus on single-modality threats, but this approach fails to address cross-modal safety alignment. |
| Approach: | They propose a safety alignment challenge to evaluate cross-modality safety alignment . they propose 'Safe Inputs but Unsafe Output' to consider safety of single modalities . |
| Outcome: | The proposed safety alignment challenge examines cases where modalities are safe independently but could lead to unsafe outputs when combined. |
World Modeling Makes a Better Planner: Dual Preference Optimization for Embodied Task Planning (2025.acl-long)
Copied to clipboard
| Challenge: | Existing approaches focus on action selection or use pre-trained models as world models to enhance planning capabilities. |
| Approach: | They propose a new learning framework that optimizes state prediction and action selection through preference learning. |
| Outcome: | The proposed method outperforms existing methods and GPT-4o on VoTa-Bench and Qwen2-VL (7B), LLaVA-1.6 (7B) and LLama-3.2 (11B). |
Improving Contrastive Learning of Sentence Embeddings from AI Feedback (2023.findings-acl)
Copied to clipboard
| Challenge: | Existing methods to learn sentence embeddings with rich semantics are limited due to the discrete nature of natural language. |
| Approach: | They propose to use AI feedback to improve contrastive learning of sentence embeddings by combining human feedback and AI feedback. |
| Outcome: | The proposed method achieves state-of-the-art performance on several semantic textual similarity and transfer learning tasks compared to other unsupervised and supervised contrastive learning methods. |
Dynamic and Generalizable Process Reward Modeling (2025.acl-long)
Copied to clipboard
| Challenge: | Existing Process Reward Models lack cross-domain generalization and focus on feedback results. |
| Approach: | They propose a process reward model that uses a reward tree to capture and store fine-grained, multi-dimensional reward criteria. |
| Outcome: | The proposed model performs on prevailing benchmarks and out-of-distribution scenarios. |
VisuoThink: Empowering LVLM Reasoning with Multimodal Tree Search (2025.acl-long)
Copied to clipboard
| Challenge: | Existing approaches to large vision-language models fail to capture interleaved nature of human visual-verbal reasoning processes. |
| Approach: | They propose a framework that integrates visuospatial and linguistic domains to facilitate multimodal slow thinking by enabling progressive visual-textual reasoning. |
| Outcome: | Experiments show that VisuoThink significantly improves reasoning capabilities even without fine-tuning. |
LLM can Achieve Self-Regulation via Hyperparameter Aware Generation (2024.findings-acl)
Copied to clipboard
Siyin Wang, Shimin Li, Tianxiang Sun, Jinlan Fu, Qinyuan Cheng, Jiasheng Ye, Junjie Ye, Xipeng Qiu, Xuanjing Huang
| Challenge: | Existing decoding strategies and hyperparameters may not be optimal for each sample. |
| Approach: | They propose a model that auto-regulates decoding strategies and hyperparameters . this approach eliminates the need for extensive manual tuning, they argue . |
| Outcome: | The proposed model eliminates the need for extensive manual tuning, offering a more autonomous, self-regulate model behavior. |
Is MultiWOZ a Solved Task? An Interactive TOD Evaluation Framework with User Simulator (2022.findings-emnlp)
Copied to clipboard
| Challenge: | Task-oriented dialogue systems are drawing more attention in recent studies . current evaluation methods use annotated utterances in multi-turn dialogue sessions . |
| Approach: | They propose an interactive evaluation framework for task-oriented dialogue (TOD) they build a goal-oriented user simulator based on pre-trained models and use it to generate dialogues . |
| Outcome: | The proposed evaluation framework achieves 98% inform and success rates in the multi-woz dataset and the proposed scores measure the response quality besides the inform and successful rates. |
Perceive the Passage of Time: A Systematic Evaluation of Large Language Model in Temporal Relativity (2025.coling-main)
Copied to clipboard
| Challenge: | Temporal perception is crucial for Large Language Models to understand the world. |
| Approach: | They propose a temporal-relative ability benchmark to evaluate LLMs' temporal perception . they conduct extensive experiments on popular LLM GPT-4 scenarios . |
| Outcome: | The proposed benchmarks show a significant performance gap between LLMs and humans in temporal-relative capability. |
R3-RAG: Learning Step-by-Step Reasoning and Retrieval for LLMs via Reinforcement Learning (2025.findings-emnlp)
Copied to clipboard
Yuan Li, Qi Luo, Xiaonan Li, Bufan Li, Qinyuan Cheng, Bo Wang, Yining Zheng, Yuxin Wang, Zhangyue Yin, Xipeng Qiu
| Challenge: | RAG systems that integrate external knowledge with Large Language Models often become bottlenecks due to their limited parameters compared to LLMs and their inability to perform step-by-step reasoning. |
| Approach: | They propose a model that integrates external knowledge with Large Language Models to enhance factual correctness and mitigate hallucination. |
| Outcome: | The proposed model outperforms baselines and can transfer well to different retrievers. |
Unified Active Retrieval for Retrieval Augmented Generation (2024.findings-emnlp)
Copied to clipboard
Qinyuan Cheng, Xiaonan Li, Shimin Li, Qin Zhu, Zhangyue Yin, Yunfan Shao, Linyang Li, Tianxiang Sun, Hang Yan, Xipeng Qiu
| Challenge: | Existing active retrieval methods struggle with handling various types of instructions. |
| Approach: | They propose a unified active retrieval framework for retrieval-augmented generation . they propose to combine four orthogonal criteria into plug-and-play classification tasks . |
| Outcome: | The proposed framework outperforms existing methods on four representative types of user instructions on four types of instructions. |
Scaling Laws for Fact Memorization of Large Language Models (2024.findings-emnlp)
Copied to clipboard
| Challenge: | Fact knowledge memorization is crucial for Large Language Models (LLMs) to generate factual and reliable responses. |
| Approach: | They analyze scaling laws for LLM’s fact knowledge and LLMs’ behaviors of memorizing different types of facts. |
| Outcome: | The proposed model can generalize on unseen facts and its scaling law is similar to general pre-training. |
Aggregation of Reasoning: A Hierarchical Framework for Enhancing Answer Selection in Large Language Models (2024.lrec-main)
Copied to clipboard
Zhangyue Yin, Qiushi Sun, Qipeng Guo, Zhiyuan Zeng, Xiaonan Li, Tianxiang Sun, Cheng Chang, Qinyuan Cheng, Ding Wang, Xiaofeng Mou, Xipeng Qiu, Xuanjing Huang
| Challenge: | Recent advances in Chain-of-Thought prompting have facilitated significant breakthroughs for Large Language Models (LLMs) in complex reasoning tasks. |
| Approach: | They propose a hierarchical reasoning aggregation framework to address this problem . they propose dynamic sampling to adjust the number of reasoning chains . |
| Outcome: | The proposed framework outperforms existing ensemble methods on complex reasoning tasks. |
Case2Code: Scalable Synthetic Data for Code Generation (2025.coling-main)
Copied to clipboard
Yunfan Shao, Linyang Li, Yichuan Ma, Peiji Li, Demin Song, Qinyuan Cheng, Shimin Li, Xiaonan Li, Pengyu Wang, Qipeng Guo, Hang Yan, Xipeng Qiu, Xuanjing Huang, Dahua Lin
| Challenge: | Large Language Models (LLMs) have shown outstanding breakthroughs in code generation. |
| Approach: | They propose a case-to-code induction task that exploits the expressiveness and correctness of programs by incorporating LLMs into their training. |
| Outcome: | The proposed task improves distribution case-to-code induction and various coding generation tasks. |
How to Mitigate Overfitting in Weak-to-strong Generalization? (2025.acl-long)
Copied to clipboard
| Challenge: | Experimental results show that weak-to-strong generalization significantly improves PGR compared to naive weak- to-strong . superalignment refers to how humans can align models on tasks beyond human ability to evaluate . |
| Approach: | They propose a framework that elicits the capabilities of strong models through weak supervisors . they propose 'superalignment' to ensure that strong models align with supervisors' intentions . |
| Outcome: | The proposed framework significantly improves quality of supervision signals and quality of input questions compared to naive weak-to-strong generalization . |