Safety reasoning is a recent paradigm where LLMs reason over safety policies before generating responses, thereby mitigating limitations in existing safety measures such as over-refusal and jailbreak vulnerabilities. However, implementing this paradigm is challenging due to the resource-intensive process of creating high-quality policy-embedded chain-of-thought (CoT) datasets while ensuring reasoning remains accurate and free from hallucinations or policy conflicts. To tackle this, we propose AIDSAFE: Agentic Iterative Deliberation for Safety Reasoning, a novel data generation recipe that leverages multi-agent deliberation to iteratively expand reasoning on safety policies. A data refiner stage in AIDSAFE ensures high-quality outputs by eliminating repetitive, redundant, and deceptive thoughts. AIDSAFE-generated CoTs provide a strong foundation for supervised fine-tuning (SFT)-based safety training. Additionally, to address the need of preference data in alignment stages, such as DPO training, we introduce a supplemental recipe that uses belief augmentation to create distinct selected and rejected CoT samples. Our evaluations demonstrate that AIDSAFE-generated CoTs achieve superior policy adherence and reasoning quality. Consequently, we show that fine-tuning open-source LLMs on these CoTs can significantly improve safety generalization and jailbreak robustness while maintaining acceptable utility and over-refusal accuracy. AIDSAFE-generated CoT datasets can be found here: https://huggingface.co/datasets/AmazonScience/AIDSAFE

Consider the following task taught in introductory optimization courses which addresses challenges articulated by the community at the intersection of (generative) AI and OR: generate the dual of a linear program. LLMs, being trained at web-scale, have the conversion process and many instances of Primal to Dual Conversion (P2DC) at their disposal. Students may thus reasonably expect that LLMs would perform well on the P2DC task. To assess this expectation, this paper introduces DualSchool, a comprehensive framework for generating and verifying P2DC instances. The verification procedure of DualSchool uses the Canonical Graph Edit Distance, going well beyond existing evaluation methods for optimization models, which exhibit many false positives and negatives when applied to P2DC. Experiments performed by DualSchool reveal interesting findings. Although LLMs can recite the conversion procedure accurately, state-of-the-art open LLMs fail to consistently produce correct duals. This finding holds even for the smallest two-variable instances and for derivative tasks, such as correctness, verification, and error classification. The paper also discusses the implications for educators, students, and the development of large reasoning systems.

Deep reinforcement learning (RL) agents achieve impressive results in a wide variety of tasks, but they lack zero-shot adaptation capabilities. While most robustness evaluations focus on tasks complexifications, for which human also struggle to maintain performances, no evaluation has been performed on tasks simplifications. To tackle this issue, we introduce HackAtari, a set of task variations of the Arcade Learning Environments. We use it to demonstrate that, contrary to humans, RL agents systematically exhibit huge performance drops on simpler versions of their training tasks, uncovering agents' consistent reliance on shortcuts. Our analysis across multiple algorithms and architectures highlights the persistent gap between RL agents and human behavioral intelligence, underscoring the need for new benchmarks and methodologies that enforce systematic generalization testing beyond static evaluation protocols. Training and testing in the same environment is not enough to obtain agents equipped with human-like intelligence.

Federated learning faces critical challenges in balancing communication efficiency and model accuracy. One key issue lies in the approximation of update errors without incurring high computational costs. In this paper, we propose a lightweight yet effective method called Gradient Difference Approximation (GDA), which leverages first-order information to estimate local error trends without computing the full Hessian matrix. The proposed method forms a key component of the Adaptive Multi-Step Federated Learning (AMSFL) framework and provides a unified error modeling strategy for large-scale multi-step adaptive training environments.

Recent advancements in Large Language Models (LLMs) have enabled the emergence of multi-agent systems where LLMs interact, collaborate, and make decisions in shared environments. While individual model behavior has been extensively studied, the dynamics of peer influence in such systems remain underexplored. In this paper, we investigate herd behavior, the tendency of agents to align their outputs with those of their peers, within LLM-based multi-agent interactions. We present a series of controlled experiments that reveal how herd behaviors are shaped by multiple factors. First, we show that the gap between self-confidence and perceived confidence in peers significantly impacts an agent's likelihood to conform. Second, we find that the format in which peer information is presented plays a critical role in modulating the strength of herd behavior. Finally, we demonstrate that the degree of herd behavior can be systematically controlled, and that appropriately calibrated herd tendencies can enhance collaborative outcomes. These findings offer new insights into the social dynamics of LLM-based systems and open pathways for designing more effective and adaptive multi-agent collaboration frameworks.

Intelligent tutoring systems combined with large language models offer a promising approach to address students' diverse needs and promote self-efficacious learning. While large language models possess good foundational knowledge of electrical engineering basics, they remain insufficiently capable of addressing specific questions about electrical circuits. In this paper, we present AITEE, an agent-based tutoring system for electrical engineering designed to accompany students throughout their learning process, offer individualized support, and promote self-directed learning. AITEE supports both hand-drawn and digital circuits through an adapted circuit reconstruction process, enabling natural interaction with students. Our novel graph-based similarity measure identifies relevant context from lecture materials through a retrieval augmented generation approach, while parallel Spice simulation further enhances accuracy in applying solution methodologies. The system implements a Socratic dialogue to foster learner autonomy through guided questioning. Experimental evaluations demonstrate that AITEE significantly outperforms baseline approaches in domain-specific knowledge application, with even medium-sized LLM models showing acceptable performance. Our results highlight the potential of agentic tutors to deliver scalable, personalized, and effective learning environments for electrical engineering education.

With the rapid development of wireless communications and the growing complexity of digital modulation schemes, traditional manual modulation recognition methods struggle to extract reliable signal features and meet real-time requirements in modern scenarios. Recently, deep learning based Automatic Modulation Recognition (AMR) approaches have greatly improved classification accuracy. However, their large model sizes and high computational demands hinder deployment on resource-constrained devices. Model pruning provides a general approach to reduce model complexity, but existing weight, channel, and layer pruning techniques each present a trade-off between compression rate, hardware acceleration, and accuracy preservation. To this end, in this paper, we introduce FCOS, a novel Fine-to-COarse two-Stage pruning framework that combines channel-level pruning with layer-level collapse diagnosis to achieve extreme compression, high performance and efficient inference. In the first stage of FCOS, hierarchical clustering and parameter fusion are applied to channel weights to achieve channel-level pruning. Then a Layer Collapse Diagnosis (LaCD) module uses linear probing to identify layer collapse and removes the collapsed layers due to high channel compression ratio. Experiments on multiple AMR benchmarks demonstrate that FCOS outperforms existing channel and layer pruning methods. Specifically, FCOS achieves 95.51% FLOPs reduction and 95.31% parameter reduction while still maintaining performance close to the original ResNet56, with only a 0.46% drop in accuracy on Sig2019-12. Code is available at https://github.com/yaolu-zjut/FCOS.

The surgence of Mixture of Experts (MoE) in Large Language Models promises a small price of execution cost for a much larger model parameter count and learning capacity, because only a small fraction of parameters are activated for each input token. However, it is commonly observed that some experts are activated far more often than others, leading to system inefficiency when running the experts on different devices in parallel. Therefore, we introduce Mixture of Grouped Experts (MoGE), which groups the experts during selection and balances the expert workload better than MoE in nature. It constrains tokens to activate an equal number of experts within each predefined expert group. When a model execution is distributed on multiple devices, this architectural design ensures a balanced computational load across devices, significantly enhancing throughput, particularly for the inference phase. Further, we build Pangu Pro MoE on Ascend NPUs, a sparse model based on MoGE with 72 billion total parameters, 16 billion of which are activated for each token. The configuration of Pangu Pro MoE is optimized for Ascend 300I Duo and 800I A2 through extensive system simulation studies. Our experiments indicate that MoGE indeed leads to better expert load balancing and more efficient execution for both model training and inference on Ascend NPUs. The inference performance of Pangu Pro MoE achieves 1148 tokens/s per card and can be further improved to 1528 tokens/s per card by speculative acceleration, outperforming comparable 32B and 72B Dense models. Furthermore, we achieve an excellent cost-to-performance ratio for model inference on Ascend 300I Duo. Our studies show that Ascend NPUs are capable of training Pangu Pro MoE with massive parallelization to make it a leading model within the sub-100B total parameter class, outperforming prominent open-source models like GLM-Z1-32B and Qwen3-32B.

Replay-based continual learning (CL) methods assume that models trained on a small subset can also effectively minimize the empirical risk of the complete dataset. These methods maintain a memory buffer that stores a sampled subset of data from previous tasks to consolidate past knowledge. However, this assumption is not guaranteed in practice due to the limited capacity of the memory buffer and the heuristic criteria used for buffer data selection. To address this issue, we propose a new dataset distillation framework tailored for CL, which maintains a learnable memory buffer to distill the global information from the current task data and accumulated knowledge preserved in the previous memory buffer. Moreover, to avoid the computational overhead and overfitting risks associated with parameterizing the entire buffer during distillation, we introduce a lightweight distillation module that can achieve global information distillation solely by generating learnable soft labels for the memory buffer data. Extensive experiments show that, our method can achieve competitive results and effectively mitigates forgetting across various datasets. The source code will be publicly available.

Large language models (LLMs) have demonstrated impressive performance on reasoning-intensive tasks, but enhancing their reasoning abilities typically relies on either reinforcement learning (RL) with verifiable signals or supervised fine-tuning (SFT) with high-quality long chain-of-thought (CoT) demonstrations, both of which are expensive. In this paper, we study a novel problem of incentivizing the reasoning capacity of LLMs without expensive high-quality demonstrations and reinforcement learning. We investigate whether the reasoning capabilities of LLMs can be effectively incentivized via supervision from significantly weaker models. We further analyze when and why such weak supervision succeeds in eliciting reasoning abilities in stronger models. Our findings show that supervision from significantly weaker reasoners can substantially improve student reasoning performance, recovering close to 94% of the gains of expensive RL at a fraction of the cost. Experiments across diverse benchmarks and model architectures demonstrate that weak reasoners can effectively incentivize reasoning in stronger student models, consistently improving performance across a wide range of reasoning tasks. Our results suggest that this simple weak-to-strong paradigm is a promising and generalizable alternative to costly methods for incentivizing strong reasoning capabilities at inference-time in LLMs. The code is publicly available at https://github.com/yuanyige/w2sr.