Acquiring high-quality instruction-code pairs is essential for training Large Language Models for code generation.While automated synthesis has emerged as an alternative to expensive manual curation, current approaches often rely on rigid heuristics, yielding data that is ungrounded or lacks logical complexity.We propose CodeEvo, a dual-agent architecture comprising a Coder for iterative solution synthesis and a Reviewer to orchestrate the generation trajectory.To transcend the limitations of existing heuristics, the Reviewer formulates a Schema to systematically architect logic and complexity through an interleaved synthesis of instructions and code.This process is further reinforced by a hybrid verification protocol synergizing deterministic compiler feedback with semantic evaluation.Under this framework, we construct CodeEvo-100K, a large-scale dataset of instruction-code pairs with stepped difficulty levels.Extensive experiments demonstrate that models fine-tuned on CodeEvo data significantly outperform established baselines across code generation benchmarks.In-depth analyses further provide insights into effective code-centric data synthesis.Code and data are available at https://github.com/QiushiSun/CodeEvo.

Qiushi Sun, Mukai Li, Zhoumianze Liu, Zhihui Xie, Fangzhi Xu, Zhangyue Yin, Kanzhi Cheng, Zehao Li, Zichen Ding, Qi Liu, Zhiyong Wu, Zhuosheng Zhang, Ben Kao, Lingpeng Kong. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Bowen Yang, Kaiming Jin, Zhenyu Wu, Zhaoyang Liu, Qiushi Sun, Zehao Li, JingJing Xie, Zhoumianze Liu, Fangzhi Xu, Kanzhi Cheng, Yian Wang, Qingyun Li, Yu Qiao, Zun Wang, Zichen Ding. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Graphical User Interface (GUI) agents powered by Vision-Language Models (VLMs) have demonstrated human-like computer control capability. Despite their utility in advancing digital automation, a critical bottleneck persists: collecting high-quality trajectory data for training. Common practices for collecting such data rely on human supervision or synthetic data generation through executing pre-defined tasks, which are either resource-intensive or unable to guarantee data quality. Moreover, these methods suffer from limited data diversity and significant gaps between synthetic data and real-world environments. To address these challenges, we propose OS-Genesis, a novel GUI data synthesis pipeline that reverses the conventional trajectory collection process. Instead of relying on pre-defined tasks, OS-Genesis enables agents first to perceive environments and perform step-wise interactions, then retrospectively derive high-quality tasks to enable trajectory-level exploration. A trajectory reward model is then employed to ensure the quality of the generated trajectories. We demonstrate that training GUI agents with OS-Genesis significantly improves their performance on highly challenging online benchmarks. In-depth analysis further validates OS-Genesis's efficiency and its superior data quality and diversity compared to existing synthesis methods. Our codes, data, and checkpoints are available at OS-Genesis Homepage.

Fangzhi Xu, Hang Yan, Chang Ma, Haiteng Zhao, Qiushi Sun, Kanzhi Cheng, Junxian He, Jun Liu, Zhiyong Wu. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2025.

One of the primary driving forces contributing to the superior performance of Large Language Models (LLMs) is the extensive availability of human-annotated natural language data, which is used for alignment fine-tuning. This inspired researchers to investigate self-training methods to mitigate the extensive reliance on human annotations. However, the current success of self-training has been primarily observed in natural language scenarios, rather than in the increasingly important neural-symbolic scenarios. To this end, we propose an environment-guided neural-symbolic self-training framework named ENVISIONS. It aims to overcome two main challenges: (1) the scarcity of symbolic data, and (2) the limited proficiency of LLMs in processing symbolic language. Extensive evaluations conducted on three distinct domains demonstrate the effectiveness of our approach. Additionally, we have conducted a comprehensive analysis to uncover the factors contributing to ENVISIONS’s success, thereby offering valuable insights for future research in this area.

Zhangyue Yin, Qiushi Sun, Zhiyuan Zeng, Qinyuan Cheng, Xipeng Qiu, Xuanjing Huang. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2025.

Existing efforts in building GUI agents heavily rely on the availability of robust commercial Vision-Language Models (VLMs) such as GPT-4o and GeminiProVision. Practitioners are often reluctant to use open-source VLMs due to their significant performance lag compared to their closed-source counterparts, particularly in GUI grounding and Out-Of-Distribution (OOD) scenarios. To facilitate future research in this area, we developed OS-Atlas—a foundational GUI action model that excels at GUI grounding and OOD agentic tasks through innovations in both data and modeling. We have invested significant engineering effort in developing an open-source toolkit for synthesizing GUI grounding data across multiple platforms, including Windows, Linux, MacOS, Android, and the web. Leveraging this toolkit, we are releasing the largest open-source cross-platform GUI grounding corpus to date, which contains over 13 million GUI elements. This dataset, combined with innovations in model training, provides a solid foundation for OS-Atlas to understand GUI screenshots and generalize to unseen interfaces. Through extensive evaluation across six benchmarks spanning three different platforms (mobile, desktop, and web), OS-Atlas demonstrates significant performance improvements over previous state-of-the-art models. Our evaluation also uncovers valuable insights into continuously improving and scaling the agentic capabilities of open-source VLMs.

Fei Yuan, Chang Ma, Shuai Yuan, Qiushi Sun, Lei Li. Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2025.

With the exponential growth in the amount of data transmitted over mobile networks, contemporary 5G communication technologies with the primary goal of improving network performance and quality of service have gained much attention. Efficient resource allocation and interference management are especially critical in large-scale wireless networks. Device-to-device (D2D) communication has become a promising technological tool to address this growing need. However, the limitation of exponentially growing solution space in large-scale ultra-dense networks makes it difficult to achieve real-time control with conventional optimization methods. To face this challenge, we propose a novel framework that combines Multi-Agent Reinforcement Learning (MARL) with Mean Field Type Game (MFTG) theory, allowing agents to operate in different action spaces. This approach extends the core principle of mean-field reinforcement learning from a single type to multiple types of interactions, effectively modeling the approximate behavior between various types of devices in heterogeneous D2D networks. Experimental results show that the proposed Multi-Type Mean-Field double deep Q-network (MTMF-Q) method outperforms benchmark methods in heterogeneous networks. In addition, the proposed method exhibits good scalability in parameters such as user density, network size and power budget, showing its potential for application in ultra-dense heterogeneous communication network scenarios.

Kanzhi Cheng, Qiushi Sun, Yougang Chu, Fangzhi Xu, Li YanTao, Jianbing Zhang, Zhiyong Wu. Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2024.

Recently, Chain-of-Thought (CoT) prompting has delivered success on complex reasoning tasks, which aims at designing a simple prompt like "Let's think step by step" or multiple incontext exemplars with well-designed rationales to elicit Large Language Models (LLMs) to generate intermediate reasoning steps.However, the generated rationales often come with hallucinations, making unfactual and unfaithful reasoning chains.To mitigate this brittleness, we propose a novel Chain-of-Knowledge (CoK) prompting, where we aim at eliciting LLMs to generate explicit pieces of knowledge evidence in the form of structure triple.This is inspired by our human behaviors, i.e., we can draw a mind map or knowledge map as the reasoning evidence in the brain before answering a complex question.Benefiting from CoK, we additionally introduce a F 2 -Verification method to estimate the reliability of the reasoning chains in terms of factuality and faithfulness.For the unreliable response, the wrong evidence can be indicated to prompt the LLM to rethink.Extensive experiments demonstrate that our method can further improve the performance of commonsense, factual, symbolic, and arithmetic reasoning tasks 1 .

Fangzhi Xu, Zhiyong Wu, Qiushi Sun, Siyu Ren, Fei Yuan, Shuai Yuan, Qika Lin, Yu Qiao, Jun Liu. Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2024.

Zhangyue Yin, Qiushi Sun, Qipeng Guo, Zhiyuan Zeng, Xiaonan Li, Junqi Dai, Qinyuan Cheng, Xuanjing Huang, Xipeng Qiu. Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2024.

Large language models (LLMs) have shown increasing power on various natural language processing (NLP) tasks. However, tuning these models for downstream tasks usually needs exorbitant costs or is unavailable due to commercial considerations. Recently, black-box tuning has been proposed to address this problem by optimizing task-specific prompts without accessing the gradients and hidden representations. However, most existing works have yet fully exploited the potential of gradient-free optimization under the scenario of few-shot learning. In this paper, we describe BBT-RGB, a suite of straightforward and complementary techniques for enhancing the efficiency and performance of black-box optimization. Specifically, our method includes three plug-and-play components: (1) Two-stage derivative-free optimization strategy that facilitates fast convergence and mitigates overfitting; (2) Automatic verbalizer construction with its novel usage under few-shot settings; (3) Better prompt initialization policy based on instruction search and auto-selected demonstration. Extensive experiments across various tasks on natural language understanding and inference demonstrate the effectiveness of our method. Our codes are available at https://github.com/QiushiSun/BBT-RGB.

Code pre-trained models (CodePTMs) have recently demonstrated a solid capacity to process various code intelligence tasks, e.g., code clone detection, code translation, and code summarization. The current mainstream method that deploys these models to downstream tasks is to fine-tune them on individual tasks, which is generally costly and needs sufficient data for large models. To tackle the issue, in this paper, we present TransCoder, a unified Transferable fine-tuning strategy for Code representation learning. Inspired by human inherent skills of knowledge generalization, TransCoder drives the model to learn better code-related knowledge like human programmers. Specifically, we employ a tunable prefix encoder to first capture cross-task and cross-language transferable knowledge, subsequently applying the acquired knowledge for optimized downstream adaptation. Besides, our approach confers benefits for tasks with minor training sample sizes and languages with smaller corpora, underscoring versatility and efficacy. Extensive experiments conducted on representative datasets clearly demonstrate that our method can lead to superior performance on various code-related tasks and encourage mutual reinforcement, especially in low-resource scenarios. Our codes are available at https://github.com/QiushiSun/TransCoder.

Over the past few years, we have witnessed remarkable advancements in Code Pre-trained Models (CodePTMs). These models achieved excellent representation capabilities by designing structure-based pre-training tasks for code. However, how to enhance the absorption of structural knowledge when fine-tuning CodePTMs still remains a significant challenge. To fill this gap, in this paper, we present SAT, a novel structure-enhanced and plug-and-play fine-tuning method for CodePTMs. We first propose a structure loss to quantify the difference between the information learned by CodePTMs and the knowledge extracted from code structure. Specifically, we use the attention scores from Transformer layer as the learned information, and the shortest path length between leaves in abstract syntax trees as the structural knowledge. Subsequently, multi-task learning is introduced to improve the performance of fine-tuning. Experiments conducted on four pre-trained models and two generation tasks demonstrate the effectiveness of our proposed method as a plug-and-play solution. Furthermore, we observed that SAT can benefit CodePTMs more with limited training data.

Recent advancements in Chain-of-Thought prompting have facilitated significant breakthroughs for Large Language Models (LLMs) in complex reasoning tasks. Current research enhances the reasoning performance of LLMs by sampling multiple reasoning chains and ensembling based on the answer frequency. However, this approach fails in scenarios where the correct answers are in the minority. We identify this as a primary factor constraining the reasoning capabilities of LLMs, a limitation that cannot be resolved solely based on the predicted answers. To address this shortcoming, we introduce a hierarchical reasoning aggregation framework AoR (Aggregation of Reasoning), which selects answers based on the evaluation of reasoning chains. Additionally, AoR incorporates dynamic sampling, adjusting the number of reasoning chains in accordance with the complexity of the task. Experimental results on a series of complex reasoning tasks show that AoR outperforms prominent ensemble methods. Further analysis reveals that AoR not only adapts various LLMs but also achieves a superior performance ceiling when compared to current methods.

Large Language Models (LLMs) have revolutionized the landscape of natural language processing, demonstrating remarkable abilities across various complex tasks.However, their stateless nature limits the capability to retain information across interactions, hindering performance in scenarios requiring historical context recall.To mitigate this, current approaches primarily use explicit memory to allow LLMs to store useful information, which is accessible, readable, and interpretable.Nevertheless, explicit memory lacks the reliable learning mechanisms of implicit memory, which can be optimized end-to-end.To harness the benefits of both, we introduce EM 2 , a novel framework enhancing explicit memory updates via the Expectation-Maximization (EM) algorithm.EM 2 treats memory as a latent variable, ensuring continual learning and improvement during updates.Experimental results on streaming inference tasks demonstrate that EM 2 outperforms existing methods without memory or with static external memory.Our in-depth analysis highlights that EM 2 significantly enhances performance across various backbones and memory strategies, providing a robust solution for advancing LLM memory management and enabling explicit memory to learn and improve similarly to implicit memory.

Code pre-trained models (CodePTMs) have recently exhibited remarkable accomplishments in the realm of software engineering. However, there are still limited advancements in understanding the inner mechanism of these models, as well as their sensitivity to samples of varying quality. Codes have a more rigid and structured syntax compared to natural languages; hence, leveraging and understanding structural information becomes essential for analyzing, interpreting, and utilizing CodePTMs. While previous studies have verified models’ ability to acquire knowledge from code structure through techniques such as attention analysis and probing tasks, the specific roles it plays in downstream tasks have yet to be explored. In this work, we propose a set of novel and practical methods for probing and exploiting the structural information within the code. In particular, dataflow perturbation experiments are first employed to explore the sensitivity of models with varying levels of structural information when confronted with input changes. Based on our findings, structure-aware exemplars selection strategies are proposed for both code generation and understanding, aiming to recover the model performance at minimal cost under perturbed conditions. Moreover, efficient fine-tuning can be achieved by utilizing exemplars instead of full fine-tuning.

In this paper, we introduce HugNLP, a unified and comprehensive library for natural language processing (NLP) with the prevalent backend of Hugging Face Transformers, which is designed for NLP researchers to easily utilize off-the-shelf algorithms and develop novel methods with user-defined models and tasks in real-world scenarios. HugNLP consists of a hierarchical structure including models, processors and applications that unifies the learning process of pre-trained language models (PLMs) on different NLP tasks. Additionally, we present some featured NLP applications to show the effectiveness of HugNLP, such as knowledge-enhanced PLMs, universal information extraction, low-resource mining, and code understanding and generation, etc. The source code will be released on GitHub (https://github.com/HugAILab/HugNLP).

Large Language Models (LLMs) have recently made significant strides in complex reasoning tasks through the Chain-of-Thought technique.Despite this progress, their reasoning is often constrained by their intrinsic understanding, lacking external insights.To address this, we propose Exchange-of-Thought (EoT), a novel framework that enables cross-model communication during problem-solving.Drawing inspiration from network topology, EoT integrates four unique communication paradigms: Memory, Report, Relay, and Debate.This paper delves into the communication dynamics and volume associated with each paradigm.To counterbalance the risks of incorrect reasoning chains, we implement a robust confidence evaluation mechanism within these communications.Our experiments across diverse complex reasoning tasks demonstrate that EoT significantly surpasses established baselines, underscoring the value of external insights in enhancing LLM performance.Furthermore, we show that EoT achieves these superior results in a cost-effective manner, marking a promising advancement for efficient and collaborative AI problem-solving."Two heads are better than one.