Papers by Weiwen Liu
Progra: Progress-Aware Reinforcement Learning for Multi-Turn Function Calling (2026.findings-acl)
Copied to clipboard
Huacan Chai, Zijie Cao, Maolin Ran, Yingxuan Yang, Jianghao Lin, Xin Peng, Hairui Wang, Renjie Ding, Ziyu Wan, Muning Wen, Weiwen Liu, Weinan Zhang, Fei Huang, Ying Wen
| Challenge: | Existing methods for multi-turn function calling are limited by redundancy and lack explicit integration of progress awareness into training. |
| Approach: | They propose a framework that explicitly integrates progress awareness into LLM training for multi-turn function calling. |
| Outcome: | Empirical results show that Progra outperforms existing methods on two public benchmarks. |
NL-Debugging: Exploiting Natural Language as an Intermediate Representation for Code Debugging (2025.emnlp-main)
Copied to clipboard
Weiming Zhang, Qingyao Li, Xinyi Dai, Jizheng Chen, Kounianhua Du, Weiwen Liu, Yasheng Wang, Ruiming Tang, Yong Yu, Weinan Zhang
| Challenge: | Early debugging efforts focused on code-level analysis, which often fails when addressing complex programming errors. |
| Approach: | They propose a framework that employs natural language as an intermediate representation to improve code debugging by debuggating at a natural language level. |
| Outcome: | The proposed framework outperforms traditional debugging methods and enables a broader modification space through direct refinement guided by execution feedback. |
Tool Zero: Training Tool-Augmented LLMs via Pure RL from Scratch (2025.findings-emnlp)
Copied to clipboard
Yirong Zeng, Xiao Ding, Yutai Hou, Yuxian Wang, Li Du, Juyi Dai, Qiuyang Ding, Duyu Tang, Dandan Tu, Weiwen Liu, Bing Qin, Ting Liu
| Challenge: | Experimental results demonstrate that our models achieve over 7% performance improvement compared to both SFT and RL-with-SFT models under the same experimental settings. |
| Approach: | They propose a dynamic generalization-guided reward design for rule-based RL that shifts rewards from exploratory to exploitative tool-use patterns. |
| Outcome: | The proposed model achieves over 7% performance improvement compared to SFT and RL-with-SFT models under the same experimental settings. |
MMAC: A Multilingual, Multimodal Alignment Framework for Cultural Grounding Evaluation (2026.acl-long)
Copied to clipboard
Weihua Zheng, Zhengyuan Liu, Tanmoy Chakraborty, Weiwen Xu, Xiaoxue Gao, Bryan Chen Zhengyu Tan, Bowei Zou, Chang Liu, Yujia Hu, Xing Xie, Xiaoyuan Yi, Jing Yao, Chaojun Wang, Long Li, Rui Liu, Huiyao Liu, Koji Inoue, Ryuichi Sumida, Tatsuya Kawahara, Fan Xu, Lingyu Ye, Wei Tian, Dongjun Kim, Jimin Jung, Jaehyung Seo, Nadya Yuki Wangsajaya, Pham Minh Duc, Ojasva Saxena, Palash Nandi, Xiyan Tao, Wiwik Karlina, Tuan Luong, Keertana Arun Vasan, Roy Ka-Wei Lee, Nancy F. Chen
| Challenge: | Existing models lack cultural alignment across modalities and languages . a new framework to assess cultural awareness across linguistics and languages is needed . |
| Approach: | They propose a framework that integrates tri-modally aligned cultural benchmarks and a five-dimensional evaluation protocol to assess cross-country awareness disparities. |
| Outcome: | The proposed framework assesses cultural awareness disparities across modalities and languages . it is the first dataset aligned at the input level across text, image, and speech . |
Stepwise Reasoning Checkpoint Analysis: A Test Time Scaling Method to Enhance LLMs’ Reasoning (2025.emnlp-main)
Copied to clipboard
Zezhong Wang, Xingshan Zeng, Weiwen Liu, Yufei Wang, Liangyou Li, Yasheng Wang, Lifeng Shang, Xin Jiang, Qun Liu, Kam-Fai Wong
| Challenge: | Existing methods that use Chain-of-Thought suffer from path homogenization and inefficient use of intermediate results. |
| Approach: | They propose a framework that introduces checkpoints between reasoning steps to reduce path homogenization and create fault-tolerant mechanisms. |
| Outcome: | The proposed framework reduces path homogenization and creates fault-tolerant mechanism by utilizing high-quality intermediate results. |
Agent-Dice: Disentangling Knowledge Updates via Geometric Consensus for Agent Continual Learning (2026.findings-acl)
Copied to clipboard
| Challenge: | Large Language Model (LLM)-based agents extend the utility of LLMs by interacting with dynamic environments. |
| Approach: | They propose a parameter fusion framework based on directional consensus evaluation that disentangles knowledge updates through a two-stage process. |
| Outcome: | The proposed framework disentangles knowledge updates through a two-stage process with minimal computational overhead and parameter updates. |
RethinkMCTS: Refining Erroneous Thoughts in Monte Carlo Tree Search for Code Generation (2025.emnlp-main)
Copied to clipboard
Qingyao Li, Wei Xia, Xinyi Dai, Kounianhua Du, Weiwen Liu, Yasheng Wang, Ruiming Tang, Yong Yu, Weinan Zhang
| Challenge: | Existing tree search methods neglect the underlying reasoning process, resulting in poor search quality. |
| Approach: | They propose a framework that systematically explores and refines the reasoning process for code generation by using a tree search engine and a reflection mechanism. |
| Outcome: | The proposed framework outperforms existing methods in the code generation domain. |
ACEBench: A Comprehensive Evaluation of LLM Tool Usage (2025.findings-emnlp)
Copied to clipboard
Chen Chen, Xinlong Hao, Weiwen Liu, Xu Huang, Xingshan Zeng, Shuai Yu, Dexun Li, Yuefeng Huang, Xiangcheng Liu, Wang Xinzhi, Wu Liu
| Challenge: | Existing benchmarks for evaluating LLMs’ tool usage face several limitations: limited evaluation scenarios, lacking assessments in real multi-turn dialogue contexts; narrow evaluation dimensions, with insufficient detailed assessments of how LLM use tools; and reliance on LLM or real API executions for evaluation, which introduces significant overhead. |
| Approach: | ACEBench is a benchmark for evaluating tool usage in Large Language Models . it categorizes data into three primary types based on evaluation methodology: Normal, Special, and Agent. |
| Outcome: | ACEBench categorizes data into three primary types based on evaluation methodology: Normal, Special, and Agent. |
FineReason: Evaluating and Improving LLMs’ Deliberate Reasoning through Reflective Puzzle Solving (2025.acl-long)
Copied to clipboard
Guizhen Chen, Weiwen Xu, Hao Zhang, Hou Pong Chan, Chaoqun Liu, Lidong Bing, Deli Zhao, Anh Tuan Luu, Yu Rong
| Challenge: | Recent advances in large language models (LLMs) highlight an important shift from the “System 1” way of quick reactions to the “system 2” style of reflection-and-correction problem solving. |
| Approach: | They propose a logic-puzzle benchmark for systematic evaluation of large language models' reasoning capabilities that decomposes each puzzle into atomic steps. |
| Outcome: | The proposed model improves on state checking and state transition tasks and demonstrates gains in reasoning by up to 5.1%. |
ToolFlow: Boosting LLM Tool-Calling Through Natural and Coherent Dialogue Synthesis (2025.naacl-long)
Copied to clipboard
Zezhong Wang, Xingshan Zeng, Weiwen Liu, Liangyou Li, Yasheng Wang, Lifeng Shang, Xin Jiang, Qun Liu, Kam-Fai Wong
| Challenge: | Large Language Models (LLMs) can be enhanced by using supervised fine-tuning . however, access to fine-timing data can be limited. |
| Approach: | They propose a Graph-based Sampling strategy and a Planned-generation strategy to enhance the coherence between dialogues by using 8,000 synthetic dialogues. |
| Outcome: | The proposed model achieves tool-calling performance comparable to or surpassing GPT-4 while maintaining strong general capabilities. |
Chain-of-Probe: Examining the Necessity and Accuracy of CoT Step-by-Step (2025.findings-naacl)
Copied to clipboard
Zezhong Wang, Xingshan Zeng, Weiwen Liu, Yufei Wang, Liangyou Li, Yasheng Wang, Lifeng Shang, Xin Jiang, Qun Liu, Kam-Fai Wong
| Challenge: | Current research found the issue of Early Answering in large language models where the models already have an answer before generating the Chain-of-Thought (CoT). |
| Approach: | They propose a method to probe changes in confidence during the model’s reasoning and prioritize answers with correct reasoning among multiple candidates. |
| Outcome: | The proposed method reveals that in a significant number of question-answer cases, CoT appears to be unnecessary and this necessity correlates with the simplicity of the task, defined by the reasoning steps required. |
Adaptive Tool Use in Large Language Models with Meta-Cognition Trigger (2025.acl-long)
Copied to clipboard
Wenjun Li, Dexun Li, Kuicai Dong, Cong Zhang, Hao Zhang, Weiwen Liu, Yasheng Wang, Ruiming Tang, Yong Liu
| Challenge: | Existing research expands the tool arrays of large language models (LLMs), but the necessity of using these tools is often overlooked, leading to indiscriminate tool invocation. |
| Approach: | They propose a meta-cognition proxy proxy for LLMs self-assessment of their capabilities, reflecting the model’s awareness of its own limitations. |
| Outcome: | The proposed strategy is fine-tuned-free and costs minimal. |
CoreCodeBench: Decoupling Code Intelligence via Fine-Grained Repository-Level Tasks (2026.acl-long)
Copied to clipboard
Lingyue Fu, Hao Guan, Bolun Zhang, Haowei Yuan, Yaoming Zhu, Lin Qiu, ZongYu Wang, Xuezhi Cao, Xunliang Cai, Weiwen Liu, Weinan Zhang, Yong Yu
| Challenge: | Existing large language models for software engineering rely on coarse-grained pass rates obscuring specific cognitive bottlenecks. |
| Approach: | They propose a repository-level benchmark that dissects coding capabilities through atomized tasks. |
| Outcome: | The proposed framework achieves a 78.55% validity yield, surpassing the 31.7% retention rate of SWE-bench-Verified. |
A Survey of Large Language Model-Based Search Agents (2026.acl-long)
Copied to clipboard
Yunjia Xi, Jianghao Lin, Yongzhao Xiao, Zheli Zhou, Rong Shan, Te Gao, Jiachen Zhu, Weiwen Liu, Yong Yu, Weinan Zhang
| Challenge: | Large Language Models (LLMs) have revolutionized web search, but their integration is static and cannot handle complex contexts. |
| Approach: | They analyze existing research and analyze existing work from the perspectives of architecture, optimization, application, and evaluation. |
| Outcome: | The proposed models can comprehend user intentions and context and execute multi-turn retrieval with dynamic planning, extending search capabilities far beyond the web. |
Planning, Creation, Usage: Benchmarking LLMs for Comprehensive Tool Utilization in Real-World Complex Scenarios (2024.findings-acl)
Copied to clipboard
Shijue Huang, Wanjun Zhong, Jianqiao Lu, Qi Zhu, Jiahui Gao, Weiwen Liu, Yutai Hou, Xingshan Zeng, Yasheng Wang, Lifeng Shang, Xin Jiang, Ruifeng Xu, Qun Liu
| Challenge: | Existing benchmarks focus on simple synthesized queries that do not reflect real-world complexity, thereby offering limited perspectives in evaluating tool utilization. |
| Approach: | They propose a benchmark to evaluate LLMs’ ability in tool utilization within real-world scenarios. |
| Outcome: | The proposed benchmark improves LLMs’ ability in tool utilization within real-world scenarios and eliminates the restriction of pre-defined toolset. |
ColorBrowserAgent: Complex Long-Horizon Browser Agent with Adaptive Knowledge Evolution (2026.acl-industry)
Copied to clipboard
Jihong Wang, Jiamu Zhou, Weiming Zhang, Teng Wang, Weiwen Liu, Zhuosheng Zhang, Xingyu Lou, Weinan Zhang, Huarong Deng, Jun Wang
| Challenge: | Xue et al., 2025): deploying autonomous web agents in production remains difficult due to site heterogeneity and long-horizon instability. |
| Approach: | They propose a knowledge-evolving agent that can be used to automate web workflows . they use human-in-the-loop knowledge adaptation and knowledge-aligned progressive summarization . |
| Outcome: | Experiments on WebArena, WebChoreAren and industrial deployment show it outperforms baselines. |
iTool: Reinforced Fine-Tuning with Dynamic Deficiency Calibration for Advanced Tool Use (2025.emnlp-main)
Copied to clipboard
Yirong Zeng, Xiao Ding, Yuxian Wang, Weiwen Liu, Yutai Hou, Wu Ning, Xu Huang, Duyu Tang, Dandan Tu, Bing Qin, Ting Liu
| Challenge: | Synthesizing tool-use data through real-world simulations is effective for enhancing large language models (LLMs) however, training gains decay as synthetic data increases, and the model struggles to benefit from more synthetic data. |
| Approach: | They propose an iterative reinforced fine-tuning strategy to improve LLMs with external tools to augment their capabilities. |
| Outcome: | The proposed method achieves 13.11% better performance than the same-size base model and outperforms larger open-source and closed-source models. |
Revealing Procedural Reasoning Structures in Chain-of-Thought Training via Span-Level Gradient Organization (2026.acl-long)
Copied to clipboard
| Challenge: | Chain-of-Thought (CoT) prompts elicit multi-step reasoning, yet how reasoning related structure is expressed during training remains poorly understood. |
| Approach: | They propose a framework that tracks span-level gradients during fine-tuning on reasoning benchmarks to understand how models develop structured, step-by-step reasoning capabilities. |
| Outcome: | The proposed framework tracks span-level gradients during fine-tuning on reasoning benchmarks to understand how models develop structured, step-by-step reasoning capabilities. |
ACE-Router: Generalizing History-Aware Routing from MCP Tools to the Agent Web (2026.acl-long)
Copied to clipboard
Zhiyuan Yao, Zishan Xu, Yifu Guo, Zhiguang Han, Cheng Yang, Shuo Zhang, Weinan Zhang, Xingshan Zeng, Weiwen Liu
| Challenge: | Existing routers that use hardcoded tools are limited by scalability and generality bottlenecks. |
| Approach: | They propose a pipeline for training history-aware routers to empower precise navigation in large-scale ecosystems. |
| Outcome: | The proposed pipeline can train routers with dynamic context understanding to create the plug-and-play Light Routing Agent. |
LoopTool: Closing the Data–Training Loop for Robust LLM Tool Calls (2026.acl-long)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) are powerful tools for multi-step tasks, but static data pipelines hinder tool learning and cause noisy labels to persist. |
| Approach: | They propose a fully automated, model-aware data evolution framework that tightly integrates data synthesis and model training. |
| Outcome: | Experiments show that LoopTool-8B significantly surpasses its 32B data generator and achieves new state-of-the-art results on the BFCL-v3 and ACEBench benchmarks for its scale. |