Can RL Improve Generalization of LLM Agents? An Empirical Study

Reinforcement fine-tuning (RFT) has shown promise for training LLM agents to perform multi-turn decision-making based on environment feedback. However, most existing evaluations remain largely in-domain: training and testing are conducted in the same environment or even on the same tasks. In real-world deployment, agents may operate in unseen environments with different background knowledge, observation spaces, and action interfaces. To characterize the generalization profile of RFT under such shifts, we conduct a systematic study along three axes: (1) within-environment generalization across task difficulty, (2) cross-environment transfer to unseen environments, and (3) sequential multi-environment training to quantify transfer and forgetting. Our results show that RFT generalizes well across task difficulty within an environment, but exhibits weaker transfer to unseen environments, which correlates with shifts in both semantic priors and observation/action interfaces. In contrast, sequential training yields promising downstream gains with minimal upstream forgetting, and mixture training across environments improves the overall balance. We further provide detailed analyses and deeper insights, and hope our work helps the community develop and deploy generalizable LLM agents.

PET-F2I: A Comprehensive Benchmark and Parameter-Efficient Fine-Tuning of LLMs for PET/CT Report Impression Generation

PET/CT imaging is pivotal in oncology and nuclear medicine, yet summarizing complex findings into precise diagnostic impressions is labor-intensive. While LLMs have shown promise in medical text generation, their capability in the highly specialized domain of PET/CT remains underexplored. We introduce PET-F2I-41K (PET Findings-to-Impression Benchmark), a large-scale benchmark for PET/CT impression generation using LLMs, constructed from over 41k real-world reports. Using PET-F2I-41K, we conduct a comprehensive evaluation of 27 models across proprietary frontier LLMs, open-source generalist models, and medical-domain LLMs, and we develop a domain-adapted 7B model (PET-F2I-7B) fine-tuned from Qwen2.5-7B-Instruct via LoRA. Beyond standard NLG metrics (e.g., BLEU-4, ROUGE-L, BERTScore), we propose three clinically grounded metrics - Entity Coverage Rate (ECR), Uncovered Entity Rate (UER), and Factual Consistency Rate (FCR) - to assess diagnostic completeness and factual reliability. Experiments reveal that neither frontier nor medical-domain LLMs perform adequately in zero-shot settings. In contrast, PET-F2I-7B achieves substantial gains (e.g., 0.708 BLEU-4) and a 3.0x improvement in entity coverage over the strongest baseline, while offering advantages in cost, latency, and privacy. Beyond this modeling contribution, PET-F2I-41K establishes a standardized evaluation framework to accelerate the development of reliable and clinically deployable reporting systems for PET/CT.

QuantaAlpha: An Evolutionary Framework for LLM-Driven Alpha Mining

Financial markets are noisy and non-stationary, making alpha mining highly sensitive to noise in backtesting results and sudden market regime shifts. While recent agentic frameworks improve alpha mining automation, they often lack controllable multi-round search and reliable reuse of validated experience. To address these challenges, we propose QuantaAlpha, an evolutionary alpha mining framework that treats each end-to-end mining run as a trajectory and improves factors through trajectory-level mutation and crossover operations. QuantaAlpha localizes suboptimal steps in each trajectory for targeted revision and recombines complementary high-reward segments to reuse effective patterns, enabling structured exploration and refinement across mining iterations. During factor generation, QuantaAlpha enforces semantic consistency across the hypothesis, factor expression, and executable code, while constraining the complexity and redundancy of the generated factor to mitigate crowding. Extensive experiments on the China Securities Index 300 (CSI 300) demonstrate consistent gains over strong baseline models and prior agentic systems. When utilizing GPT-5.2, QuantaAlpha achieves an Information Coefficient (IC) of 0.1501, with an Annualized Rate of Return (ARR) of 27.75% and a Maximum Drawdown (MDD) of 7.98%. Moreover, factors mined on CSI 300 transfer effectively to the China Securities Index 500 (CSI 500) and the Standard & Poor's 500 Index (S&P 500), delivering 160% and 137% cumulative excess return over four years, respectively, which indicates strong robustness of QuantaAlpha under market distribution shifts.

Spider-Sense: Intrinsic Risk Sensing for Efficient Agent Defense with Hierarchical Adaptive Screening

As large language models (LLMs) evolve into autonomous agents, their real-world applicability has expanded significantly, accompanied by new security challenges. Most existing agent defense mechanisms adopt a mandatory checking paradigm, in which security validation is forcibly triggered at predefined stages of the agent lifecycle. In this work, we argue that effective agent security should be intrinsic and selective rather than architecturally decoupled and mandatory. We propose Spider-Sense framework, an event-driven defense framework based on Intrinsic Risk Sensing (IRS), which allows agents to maintain latent vigilance and trigger defenses only upon risk perception. Once triggered, the Spider-Sense invokes a hierarchical defence mechanism that trades off efficiency and precision: it resolves known patterns via lightweight similarity matching while escalating ambiguous cases to deep internal reasoning, thereby eliminating reliance on external models. To facilitate rigorous evaluation, we introduce S$^2$Bench, a lifecycle-aware benchmark featuring realistic tool execution and multi-stage attacks. Extensive experiments demonstrate that Spider-Sense achieves competitive or superior defense performance, attaining the lowest Attack Success Rate (ASR) and False Positive Rate (FPR), with only a marginal latency overhead of 8.3\%.

Structure-based RNA Design by Step-wise Optimization of Latent Diffusion Model

RNA inverse folding, designing sequences to form specific 3D structures, is critical for therapeutics, gene regulation, and synthetic biology. Current methods, focused on sequence recovery, struggle to address structural objectives like secondary structure consistency (SS), minimum free energy (MFE), and local distance difference test (LDDT), leading to suboptimal structural accuracy. To tackle this, we propose a reinforcement learning (RL) framework integrated with a latent diffusion model (LDM). Drawing inspiration from the success of diffusion models in RNA inverse folding, which adeptly model complex sequence-structure interactions, we develop an LDM incorporating pre-trained RNA-FM embeddings from a large-scale RNA model. These embeddings capture co-evolutionary patterns, markedly improving sequence recovery accuracy. However, existing approaches, including diffusion-based methods, cannot effectively handle non-differentiable structural objectives. By contrast, RL excels in this task by using policy-driven reward optimization to navigate complex, non-gradient-based objectives, offering a significant advantage over traditional methods. In summary, we propose the Step-wise Optimization of Latent Diffusion Model (SOLD), a novel RL framework that optimizes single-step noise without sampling the full diffusion trajectory, achieving efficient refinement of multiple structural objectives. Experimental results demonstrate SOLD surpasses its LDM baseline and state-of-the-art methods across all metrics, establishing a robust framework for RNA inverse folding with profound implications for biotechnological and therapeutic applications.

BizFinBench.v2: A Unified Dual-Mode Bilingual Benchmark for Expert-Level Financial Capability Alignment

Large language models have undergone rapid evolution, emerging as a pivotal technology for intelligence in financial operations. However, existing benchmarks are often constrained by pitfalls such as reliance on simulated or general-purpose samples and a focus on singular, offline static scenarios. Consequently, they fail to align with the requirements for authenticity and real-time responsiveness in financial services, leading to a significant discrepancy between benchmark performance and actual operational efficacy. To address this, we introduce BizFinBench.v2, the first large-scale evaluation benchmark grounded in authentic business data from both Chinese and U.S. equity markets, integrating online assessment. We performed clustering analysis on authentic user queries from financial platforms, resulting in eight fundamental tasks and two online tasks across four core business scenarios, totaling 29,578 expert-level Q&A pairs. Experimental results demonstrate that ChatGPT-5 achieves a prominent 61.5% accuracy in main tasks, though a substantial gap relative to financial experts persists; in online tasks, DeepSeek-R1 outperforms all other commercial LLMs. Error analysis further identifies the specific capability deficiencies of existing models within practical financial business contexts. BizFinBench.v2 transcends the limitations of current benchmarks, achieving a business-level deconstruction of LLM financial capabilities and providing a precise basis for evaluating efficacy in the widespread deployment of LLMs within the financial domain. The data and code are available at https://github.com/HiThink-Research/BizFinBench.v2.

Towards Unified Co-Speech Gesture Generation via Hierarchical Implicit Periodicity Learning

Generating 3D-based body movements from speech shows great potential in extensive downstream applications, while it still suffers challenges in imitating realistic human movements. Predominant research efforts focus on end-to-end generation schemes to generate co-speech gestures, spanning GANs, VQ-VAE, and recent diffusion models. As an ill-posed problem, in this paper, we argue that these prevailing learning schemes fail to model crucial inter- and intra-correlations across different motion units, i.e. head, body, and hands, thus leading to unnatural movements and poor coordination. To delve into these intrinsic correlations, we propose a unified Hierarchical Implicit Periodicity (HIP) learning approach for audio-inspired 3D gesture generation. Different from predominant research, our approach models this multi-modal implicit relationship by two explicit technique insights: i) To disentangle the complicated gesture movements, we first explore the gesture motion phase manifolds with periodic autoencoders to imitate human natures from realistic distributions while incorporating non-period ones from current latent states for instance-level diversities. ii) To model the hierarchical relationship of face motions, body gestures, and hand movements, driving the animation with cascaded guidance during learning. We exhibit our proposed approach on 3D avatars and extensive experiments show our method outperforms the state-of-the-art co-speech gesture generation methods by both quantitative and qualitative evaluations. Code and models will be publicly available.

Counteracting Matthew Effect in Self-Improvement of LVLMs through Head-Tail Re-balancing

Self-improvement has emerged as a mainstream paradigm for advancing the reasoning capabilities of large vision-language models (LVLMs), where models explore and learn from successful trajectories iteratively. However, we identify a critical issue during this process: the model excels at generating high-quality trajectories for simple queries (i.e., head data) but struggles with more complex ones (i.e., tail data). This leads to an imbalanced optimization that drives the model to prioritize simple reasoning skills, while hindering its ability to tackle more complex reasoning tasks. Over iterations, this imbalance becomes increasingly pronounced--a dynamic we term the "Matthew effect"--which ultimately hinders further model improvement and leads to performance bottlenecks. To counteract this challenge, we introduce four efficient strategies from two perspectives: distribution-reshaping and trajectory-resampling, to achieve head-tail re-balancing during the exploration-and-learning self-improvement process. Extensive experiments on Qwen2-VL-7B-Instruct and InternVL2.5-4B models across visual reasoning tasks demonstrate that our methods consistently improve visual reasoning capabilities, outperforming vanilla self-improvement by 3.86 points on average.

Minimal Semantic Sufficiency Meets Unsupervised Domain Generalization

The generalization ability of deep learning has been extensively studied in supervised settings, yet it remains less explored in unsupervised scenarios. Recently, the Unsupervised Domain Generalization (UDG) task has been proposed to enhance the generalization of models trained with prevalent unsupervised learning techniques, such as Self-Supervised Learning (SSL). UDG confronts the challenge of distinguishing semantics from variations without category labels. Although some recent methods have employed domain labels to tackle this issue, such domain labels are often unavailable in real-world contexts. In this paper, we address these limitations by formalizing UDG as the task of learning a Minimal Sufficient Semantic Representation: a representation that (i) preserves all semantic information shared across augmented views (sufficiency), and (ii) maximally removes information irrelevant to semantics (minimality). We theoretically ground these objectives from the perspective of information theory, demonstrating that optimizing representations to achieve sufficiency and minimality directly reduces out-of-distribution risk. Practically, we implement this optimization through Minimal-Sufficient UDG (MS-UDG), a learnable model by integrating (a) an InfoNCE-based objective to achieve sufficiency; (b) two complementary components to promote minimality: a novel semantic-variation disentanglement loss and a reconstruction-based mechanism for capturing adequate variation. Empirically, MS-UDG sets a new state-of-the-art on popular unsupervised domain-generalization benchmarks, consistently outperforming existing SSL and UDG methods, without category or domain labels during representation learning.

Exploiting Layer Normalization Fine-tuning in Visual Transformer Foundation Models for Classification

LayerNorm is pivotal in Vision Transformers (ViTs), yet its fine-tuning dynamics under data scarcity and domain shifts remain underexplored. This paper shows that shifts in LayerNorm parameters after fine-tuning (LayerNorm shifts) are indicative of the transitions between source and target domains; its efficacy is contingent upon the degree to which the target training samples accurately represent the target domain, as quantified by our proposed Fine-tuning Shift Ratio ($FSR$). Building on this, we propose a simple yet effective rescaling mechanism using a scalar $\lambda$ that is negatively correlated to $FSR$ to align learned LayerNorm shifts with those ideal shifts achieved under fully representative data, combined with a cyclic framework that further enhances the LayerNorm fine-tuning. Extensive experiments across natural and pathological images, in both in-distribution (ID) and out-of-distribution (OOD) settings, and various target training sample regimes validate our framework. Notably, OOD tasks tend to yield lower $FSR$ and higher $\lambda$ in comparison to ID cases, especially with scarce data, indicating under-represented target training samples. Moreover, ViTFs fine-tuned on pathological data behave more like ID settings, favoring conservative LayerNorm updates. Our findings illuminate the underexplored dynamics of LayerNorm in transfer learning and provide practical strategies for LayerNorm fine-tuning.