Does RL Expand the Capability Boundary of LLM Agents? A PASS@(k,T) Analysis

Does reinforcement learning genuinely expand what LLM agents can do, or merely make them more reliable? For static reasoning, recent work answers the second: base and RL pass@k curves converge at large k. We ask whether this holds for agentic tool use, where T rounds of interaction enable compositional strategies that re-sampling cannot recover. We introduce PASS@(k,T), a two-dimensional metric that jointly varies sampling budget k and interaction depth T, separating capability expansion from efficiency improvement. Our main finding is that, contrary to the static-reasoning result, tool-use RL genuinely enlarges the capability boundary: the RL agent's pass-curve pulls above the base model's and the gap widens at large k rather than converging. The expansion is specific to compositional, sequential information gathering; on simpler tasks RL behaves as prior work predicts. Under matched training data, supervised fine-tuning regresses the boundary on the same compositional tasks, isolating self-directed exploration as the causal factor. Mechanism analysis shows RL reweights the base strategy distribution toward the subset whose downstream reasoning more often yields a correct answer, with the improvement concentrated on how the agent integrates retrieved information. These results reconcile optimistic and pessimistic readings of RL for LLMs: both are correct, on different task types.

Adaptive Test-Time Compute Allocation for Reasoning LLMs via Constrained Policy Optimization

Test-time compute scaling, the practice of spending extra computation during inference via repeated sampling, search, or extended reasoning, has become a powerful lever for improving large language model performance. Yet deploying these techniques under finite inference budgets requires a decision that current systems largely ignore: which inputs deserve more compute, and which can be answered cheaply? We formalize this as a constrained optimization problem (maximize expected accuracy subject to an average compute budget) and solve it with a two-stage Solve-then-Learn pipeline. In the solve stage, Lagrangian relaxation decomposes the global constraint into per-instance sub-problems, each admitting a closed-form oracle action that optimally prices accuracy against cost. We prove that the induced cost is monotone in the dual variable, enabling exact budget targeting via binary search. In the learn stage, a lightweight classifier is trained to predict oracle actions from cheap input features, amortizing the allocation rule for real-time deployment. We establish that the task-level regret of the learned policy is bounded by its imitation error times the worst-case per-instance gap, yielding a clean reduction from constrained inference to supervised classification. Experiments on MATH and GSM8K with three LLMs (DeepSeek-V3, GPT-4o-mini, Qwen2.5-7B) show that our method consistently outperforms uniform and heuristic allocation baselines, achieving up to 12.8% relative accuracy improvement on MATH under matched budget constraints, while closely tracking the Lagrangian oracle upper bound with over 91% imitation accuracy.

LoGo-MR: Screening Breast MRI for Cancer Risk Prediction by Efficient Omni-Slice Modeling

Efficient and explainable breast cancer (BC) risk prediction is critical for large-scale population-based screening. Breast MRI provides functional information for personalized risk assessment. Yet effective modeling remains challenging as fully 3D CNNs capture volumetric context at high computational cost, whereas lightweight 2D CNNs fail to model inter-slice continuity. Importantly, breast MRI modeling for shor- and long-term BC risk stratification remains underexplored. In this study, we propose LoGo-MR, a 2.5D local-global structural modeling framework for five-year BC risk prediction. Aligned with clinical interpretation, our framework first employs neighbor-slice encoding to capture subtle local cues linked to short-term risk. It then integrates transformer-enhanced multiple-instance learning (MIL) to model distributed global patterns related to long-term risk and provide interpretable slice importance. We further apply this framework across axial, sagittal, and coronal planes as LoGo3-MR to capture complementary volumetric information. This multi-plane formulation enables voxel-level risk saliency mapping, which may assist radiologists in localizing risk-relevant regions during breast MRI interpretation. Evaluated on a large breast MRI screening cohort (~7.5K), our method outperforms 2D/3D baselines and existing SOTA MIL methods, achieving AUCs of 0.77-0.69 for 1- to 5-year prediction and improving C-index by ~6% over 3D CNNs. LoGo3-MR further improves overall performance with interpretable localization across three planes, and validation across seven backbones shows consistent gains. These results highlight the clinical potential of efficient MRI-based BC risk stratification for large-scale screening. Code will be released publicly.

When More Thinking Hurts: Overthinking in LLM Test-Time Compute Scaling

Scaling test-time compute through extended chains of thought has become a dominant paradigm for improving large language model reasoning. However, existing research implicitly assumes that longer thinking always yields better results. This assumption remains largely unexamined. We systematically investigate how the marginal utility of additional reasoning tokens changes as compute budgets increase. We find that marginal returns diminish substantially at higher budgets and that models exhibit ``overthinking'', where extended reasoning is associated with abandoning previously correct answers. Furthermore, we show that optimal thinking length varies across problem difficulty, suggesting that uniform compute allocation is suboptimal. Our cost-aware evaluation framework reveals that stopping at moderate budgets can reduce computation significantly while maintaining comparable accuracy.

DeepFense: A Unified, Modular, and Extensible Framework for Robust Deepfake Audio Detection

Speech deepfake detection is a well-established research field with different models, datasets, and training strategies. However, the lack of standardized implementations and evaluation protocols limits reproducibility, benchmarking, and comparison across studies. In this work, we present DeepFense, a comprehensive, open-source PyTorch toolkit integrating the latest architectures, loss functions, and augmentation pipelines, alongside over 100 recipes. Using DeepFense, we conducted a large-scale evaluation of more than 400 models. Our findings reveal that while carefully curated training data improves cross-domain generalization, the choice of pre-trained front-end feature extractor dominates overall performance variance. Crucially, we show severe biases in high-performing models regarding audio quality, speaker gender, and language. DeepFense is expected to facilitate real-world deployment with the necessary tools to address equitable training data selection and front-end fine-tuning.

Networking-Aware Energy Efficiency in Agentic AI Inference: A Survey

The rapid emergence of Large Language Models (LLMs) has catalyzed Agentic artificial intelligence (AI), autonomous systems integrating perception, reasoning, and action into closed-loop pipelines for continuous adaptation. While unlocking transformative applications in mobile edge computing, autonomous systems, and next-generation wireless networks, this paradigm creates fundamental energy challenges through iterative inference and persistent data exchange. Unlike traditional AI where bottlenecks are computational Floating Point Operations (FLOPs), Agentic AI faces compounding computational and communication energy costs. In this survey, we propose an energy accounting framework identifying computational and communication costs across the Perception-Reasoning-Action cycle. We establish a unified taxonomy spanning model simplification, computation control, input and attention optimization, and hardware-aware inference. We explore cross-layer co-design strategies jointly optimizing model parameters, wireless transmissions, and edge resources. Finally, we identify open challenges of federated green learning, carbon-aware agency, 6th generation mobile communication (6G)-native Agentic AI, and self-sustaining systems, providing a roadmap for scalable autonomous intelligence.

HiFloat4 Format for Language Model Pre-training on Ascend NPUs

Large foundation models have become central to modern machine learning, with performance scaling predictably with model size and data. However, training and deploying such models incur substantial computational and memory costs, motivating the development of low-precision training techniques. Recent work has demonstrated that 4-bit floating-point (FP4) formats--such as MXFP4 and NVFP4--can be successfully applied to linear GEMM operations in large language models (LLMs), achieving up to 4x improvements in compute throughput and memory efficiency compared to higher-precision baselines. In this work, we investigate the recently proposed HiFloat4 FP4 format for Huawei Ascend NPUs and systematically compare it with MXFP4 in large-scale training settings. All experiments are conducted on Ascend NPU clusters, with linear and expert GEMM operations performed entirely in FP4 precision. We evaluate both dense architectures (e.g., Pangu and LLaMA-style models) and mixture-of-experts (MoE) models, where both standard linear layers and expert-specific GEMMs operate in FP4. Furthermore, we explore stabilization techniques tailored to FP4 training that significantly reduce numerical degradation, maintaining relative error within 1% of full-precision baselines while preserving the efficiency benefits of 4-bit computation. Our results provide a comprehensive empirical study of FP4 training on NPUs and highlight the practical trade-offs between FP4 formats in large-scale dense and MoE models.

Weight Group-wise Post-Training Quantization for Medical Foundation Model

Foundation models have achieved remarkable results in medical image analysis. However, its large network architecture and high computational complexity significantly impact inference speed, limiting its application on terminal medical devices. Quantization, a technique that compresses models into low-bit versions, is a solution to this challenge. In this paper, we propose a post-training quantization algorithm, Permutation-COMQ. It eliminates the need for backpropagation by using simple dot products and rounding operations, thereby removing hyperparameter tuning and simplifying the process. Additionally, we introduce a weight-aware strategy that reorders the weight within each layer to address the accuracy degradation induced by channel-wise scaling during quantization, while preserving channel structure. Experiments demonstrate that our method achieves the best results in 2-bit, 4-bit, and 8-bit quantization.

Robust Fair Disease Diagnosis in CT Images

Automated diagnosis from chest CT has improved considerably with deep learning, but models trained on skewed datasets tend to perform unevenly across patient demographics. However, the situation is worse than simple demographic bias. In clinical data, class imbalance and group underrepresentation often coincide, creating compound failure modes that neither standard rebalancing nor fairness corrections can fix alone. We introduce a two-level objective that targets both axes of this problem. Logit-adjusted cross-entropy loss operates at the sample level, shifting decision margins by class frequency with provable consistency guarantees. Conditional Value at Risk aggregation operates at the group level, directing optimization pressure toward whichever demographic group currently has the higher loss. We evaluate on the Fair Disease Diagnosis benchmark using a 3D ResNet-18 pretrained on Kinetics-400, classifying CT volumes into Adenocarcinoma, Squamous Cell Carcinoma, COVID-19, and Normal groups with patient sex annotations. The training set illustrates the compound problem concretely: squamous cell carcinoma has 84 samples total, 5 of them female. The combined loss reaches a gender-averaged macro F1 of 0.8403 with a fairness gap of 0.0239, a 13.3% improvement in score and 78% reduction in demographic disparity over the baseline. Ablations show that each component alone falls short. The code is publicly available at https://github.com/Purdue-M2/Fair-Disease-Diagnosis.

Evidence-Based Actor-Verifier Reasoning for Echocardiographic Agents

Echocardiography plays an important role in the screening and diagnosis of cardiovascular diseases. However, automated intelligent analysis of echocardiographic data remains challenging due to complex cardiac dynamics and strong view heterogeneity. In recent years, visual language models (VLM) have opened a new avenue for building ultrasound understanding systems for clinical decision support. Nevertheless, most existing methods formulate this task as a direct mapping from video and question to answer, making them vulnerable to template shortcuts and spurious explanations. To address these issues, we propose EchoTrust, an evidence-driven Actor-Verifier framework for trustworthy reasoning in echocardiography VLM-based agents. EchoTrust produces a structured intermediate representation that is subsequently analyzed by distinct roles, enabling more reliable and interpretable decision-making for high-stakes clinical applications.