Search agents achieve strong question-answering performance through multi-turn interactions with search engines, with Group Relative Policy Optimization (GRPO) being a widely used training algorithm. However, GRPO-style algorithms still face several challenges in multi-hop search settings. First, correct intermediate steps are often penalized when the final answer is wrong. Second, training is highly unstable, often causing degradation of natural language ability or even catastrophic training collapse. Our analysis attributes these issues to coarse-grained advantage assignment and an imbalance between positive and negative advantages. To address these problems, we propose CalibAdv, an advantage calibration method specifically designed for search agents that enables more accurate and more stable modeling of penalties and rewards. Specifically, CalibAdv leverages the correctness of intermediate steps to downscale excessive negative advantages at a fine-grained level. It then further rebalances positive and negative advantages to improve training stability. Importantly, CalibAdv adopts a lightweight design that calibrates advantages from standard rollout signals, making it simple and easy to deploy. Extensive experiments across three models and seven benchmarks demonstrate that CalibAdv improves both model performance and training stability. Our code is available at https://github.com/wujwyi/CalibAdv.

Large reasoning models (LRMs) are strong multi-step problem solvers, yet their long reasoning traces may gradually drift into unsafe compliance when exposed to harmful or adversarial prompts. Existing safety alignment methods can reduce harmful outputs; however, they often rely on broad refusal or supervision over full safe reasoning traces, which may weaken the reasoning behavior that makes LRMs useful. We introduce Intent–Risk–Decision (IRD), a lightweight prefix-tuning method that steers only the beginning of the reasoning trace: the model first emits a prompt-intent tag, a risk statement, and a compliance decision, and then continues under its original reasoning policy. Across three LRMs and evaluations covering harmfulness, jailbreak robustness, over-refusal, and reasoning ability, IRD substantially reduces unsafe outputs while preserving reasoning performance. In particular, IRD lowers average harmfulness to 1.2–7.4% across the evaluated backbones and keeps near-lossless reasoning performance, %within 0.2 average points of the base models, showing that targeted early steering can improve safety without supervising complete safe traces or full model fine-tuning. Code and data are available at https://github.com/ReXT9527/IRD.

Recent advances in large reasoning models (LRMs) have made machine unlearning more challenging, as protected facts or unsafe rationales may surface in intermediate chain-of-thought (CoT) traces before the final answer is produced. Existing unlearning objectives typically suppress the target content or redirect internal representations, but they never specify how the post-forgetting trajectory should continue, which can lead to hallucinated substitutes, malformed boundaries, or repetitive outputs. We argue that LRM unlearning should instead learn a natural forgetting trajectory: a coherent non-disclosing CoT followed by a stable refusal-style answer that replace the original disclosure. To this end, we propose Guided Answer-Reasoning Distillation (GUARD), which converts model-generated unsafe disclosures into safe-exit trajectories, aligns a frozen LRM via guidance tokens, and distills the guided behavior into model parameters.To address the lack of metrics for replacement quality beyond leakage, we further introduce Natural Forgetting Reasoning Score (NFRS), which captures structural stability, fluency, and unsupported substitutes in forgotten outputs. Extensive experiments on R-TOFU and a STAR-1-derived harmful-intent setting show that GUARD substantially reduces unsafe and privacy disclosures across two widely adopted distilled LRMs while preserving reasoning utility. Codes are available at https://github.com/zeyu-Yan/GUARD

Developing reliable models for clinical tasks, such as Medical Question Answering (QA), is severely constrained by the limited availability of high-quality, expert-annotated training data. This challenge is exacerbated by stringent privacy requirements and the impracticality of utilizing large open-source corpora or proprietary cloud APIs within resource-limited clinical settings. To address these obstacles, we introduce SAGE (Semantic Anchor-Guided Evolution), a novel data synthesis framework that enables small, locally deployed models to generate high-quality medical training data. SAGE leverages lightweight, publicly available taxonomies such as MeSH as semantic anchors, imposing a structured prior to effectively guide and ground the data generation process. At its core, SAGE iteratively interleaves atomic (individual concept-based) and associative (relation-based) synthesis, bootstrapping training data from minimal seeds. This approach eliminates the need for large collections of medical documents or reliance on external APIs, providing a practical solution for on-premises data creation. Extensive experiments across multiple medical question-answering benchmarks demonstrate that models fine-tuned with SAGE-synthesized data consistently outperform those trained using self-derived or conventional document-based paradigms, highlighting tangible improvements in data efficiency and resource utilization for medical LLM development.

The emergence of Retrieval-augmented generation (RAG) has alleviated the issues of outdated and hallucinatory content in the generation of large language models (LLMs), yet it still reveals numerous limitations. When a general-purpose LLM serves as the RAG generator, it often suffers from inadequate response informativeness, response robustness, and citation quality. Past approaches to tackle these limitations, either by incorporating additional steps beyond generating responses or optimizing the generator through supervised fine-tuning (SFT), still failed to align with the RAG requirement thoroughly. Consequently, optimizing the RAG generator from multiple preference perspectives while maintaining its end-to-end LLM form remains a challenge. To bridge this gap, we propose Multiple Perspective Preference Alignment for Retrieval-Augmented Generation (PA-RAG), a method for optimizing the generator of RAG systems to align with RAG requirements comprehensively. Specifically, we construct high-quality instruction fine-tuning data and multi-perspective preference data by sampling varied quality responses from the generator across different prompt documents quality scenarios. Subsequently, we optimize the generator using SFT and Direct Preference Optimization (DPO). Extensive experiments conducted on four question-answer datasets across three LLMs demonstrate that PA-RAG can significantly enhance the performance of RAG generators. Our code and datasets are available at https://github.com/wujwyi/PA-RAG.

Do current large language models (LLMs) better solve graph reasoning and generation tasks with parameter updates? In this paper, we propose InstructGraph, a framework that empowers LLMs with the abilities of graph reasoning and generation by instruction tuning and preference alignment. Specifically, we first propose a structured format verbalizer to unify all graph data into a universal code-like format, which can simply represent the graph without any external graph-specific encoders. Furthermore, a graph instruction tuning stage is introduced to guide LLMs in solving graph reasoning and generation tasks. Finally, we identify potential hallucination problems in graph tasks and sample negative instances for preference alignment, the target of which is to enhance the output’s reliability of the model. Extensive experiments across multiple graph-centric tasks exhibit that InstructGraph can achieve the best performance and outperform GPT-4 and LLaMA2 by more than 13% and 38%, respectively.

Large pre-trained language models (PLMs) have garnered significant attention for their versatility and potential for solving a wide spectrum of natural language processing (NLP) tasks. However, the cost of running these PLMs may be prohibitive. Furthermore, PLMs may not be open-sourced due to commercial considerations and potential risks of misuse, such as GPT-3. The parameters and gradients of PLMs are unavailable in this scenario. To solve the issue, black-box tuning has been proposed, which utilizes derivative-free optimization (DFO), instead of gradient descent, for training task-specific continuous prompts. However, these gradient-free methods still exhibit a significant gap compared to gradient-based methods. In this paper, we introduce gradient descent into black-box tuning scenario through knowledge distillation. Furthermore, we propose a novel method GDFO, which integrates gradient descent and derivative-free optimization to optimize task-specific continuous prompts in a harmonized manner. Experimental results show that GDFO can achieve significant performance gains over previous state-of-the-art methods.

Program understanding is a fundamental task in program language processing. Despite the success, existing works fail to take human behaviors as reference in understanding programs. In this paper, we consider human behaviors and propose the PGNN-EK model that consists of two main components. On the one hand, inspired by the “divide-and-conquer” reading behaviors of humans, we present a partitioning-based graph neural network model PGNN on the upgraded AST of codes. On the other hand, to characterize human behaviors of resorting to other resources to help code comprehension, we transform raw codes with external knowledge and apply pre-training techniques for information extraction. Finally, we combine the two embeddings generated from the two components to output code embeddings. We conduct extensive experiments to show the superior performance of PGNN-EK on the code summarization and code clone detection tasks. In particular, to show the generalization ability of our model, we release a new dataset that is more challenging for code clone detection and could advance the development of the community. Our codes and data are publicly available at https://github.com/RecklessRonan/PGNN-EK.