Reinforcement learning (RL) with outcome-based rewards has proven effective for improving large language models (LLMs) on complex reasoning tasks. However, its success often depends on the base model occasionally sampling correct solutions. When no correct solutions are sampled, training encounters a zero-reward barrier where learning stalls due to zero gradients. We study this scenario through the graph search task introduced in Bachmann and Nagarajan (2024) and evaluate recent methods that incorporate desirable components such as dense rewards, diversity incentives, and improved credit assignment. Our experiments show that none of these approaches overcome the zero-reward barrier if the base model never produces a correct answer. In contrast, we find that a simple data-centric intervention of adding easier samples to the training set enables the model to eventually solve the original hard task despite starting from zero reward. Importantly, this succeeds without modifying the RL algorithm itself. Because official implementations of several baselines were unavailable, we developed our own, which allowed us to conduct a detailed analysis of their failure modes. Reinforcement learning (RL) has emerged as a crucial tool in the post-training of large language models (LLMs) for tackling complex reasoning tasks such as mathematical problem solving (Guo et al., 2025), web navigation (Putta et al., 2024), and algorithmic discovery (Fawzi et al., 2022). These advances often rely on sparse rewards, where the model receives only a binary correct-or-incorrect signal at the end of its response.

Large Language Models (LLMs) can produce catastrophic responses in conversational settings that pose serious risks to public safety and security. Existing evaluations often fail to fully reveal these vulnerabilities because they rely on fixed attack prompt sequences, lack statistical guarantees, and do not scale to the vast space of multi-turn conversations. In this work, we propose QRLLM, a novel, principled Certification framework for Catastrophic risks in multi-turn Conversation for LLMs that bounds the probability of an LLM generating catastrophic responses under multi-turn conversation distributions with statistical guarantees. We model multi-turn conversations as probability distributions over query sequences, represented by a Markov process on a query graph whose edges encode semantic similarity to capture realistic conversational flow, and quantify catastrophic risks using confidence intervals. We define several inexpensive and practical distributions: random node, graph path, adaptive with rejection. Our results demonstrate that these distributions can reveal substantial catastrophic risks in frontier models, with certified lower bounds as high as 70\% for the worst model, highlighting the urgent need for improved safety training strategies in frontier LLMs.

Time series anomaly detection plays a crucial role in a wide range of fields, such as healthcare and internet traffic monitoring. The emergence of large language models (LLMs) offers new opportunities for detecting anomalies in the ubiquitous time series data. Traditional approaches struggle with variable-length time series sequences and context-based anomalies. We propose Soft Prompt Enhanced Anomaly Recognition (SPEAR), a novel approach to leverage LLMs for anomaly detection with soft prompts and quantization. Our methodology involves quantizing and transforming the time series data into input embeddings and combining them with learnable soft prompt embeddings. These combined embeddings are then fed into a frozen LLM. The soft prompts are updated iteratively based on a cross-entropy loss, allowing the model to adapt to time series anomaly detection. The use of soft prompts helps adapt LLMs effectively to time series tasks, while quantization ensures optimal handling of sequences, as LLMs are designed to handle discrete sequences. Our experimental results demonstrate that soft prompts effectively increase LLMs' performance in downstream tasks regarding time series anomaly detection.

Large language models (LLMs) raise concerns about content authenticity and integrity because they can generate human-like text at scale. Text watermarks, which embed detectable statistical signals into generated text, offer a provable way to verify content origin. Many detection methods rely on pivotal statistics that are i.i.d. under human-written text, making goodness-of-fit (GoF) tests a natural tool for watermark detection. However, GoF tests remain largely underexplored in this setting. In this paper, we systematically evaluate eight GoF tests across three popular watermarking schemes, using three open-source LLMs, two datasets, various generation temperatures, and multiple post-editing methods. We find that general GoF tests can improve both the detection power and robustness of watermark detectors. Notably, we observe that text repetition, common in low-temperature settings, gives GoF tests a unique advantage not exploited by existing methods. Our results highlight that classic GoF tests are a simple yet powerful and underused tool for watermark detection in LLMs.

Multi-LLM collaboration promises accurate, robust, and context-aware solutions, yet existing approaches rely on implicit selection and output assessment without analyzing whether collaborating models truly complement or conflict. We introduce LLM Chemistry -- a framework that measures when LLM combinations exhibit synergistic or antagonistic behaviors that shape collective performance beyond individual capabilities. We formalize the notion of chemistry among LLMs, propose algorithms that quantify it by analyzing interaction dependencies, and recommend optimal model ensembles accordingly. Our theoretical analysis shows that chemistry among collaborating LLMs is most evident under heterogeneous model profiles, with its outcome impact shaped by task type, group size, and complexity. Evaluation on classification, summarization, and program repair tasks provides initial evidence for these task-dependent effects, thereby reinforcing our theoretical results. This establishes LLM Chemistry as both a diagnostic factor in multi-LLM systems and a foundation for ensemble recommendation.

Code refactoring is a fundamental software engineering practice aimed at improving code quality and maintainability. Despite its importance, developers often neglect refactoring due to the significant time, effort, and resources it requires, as well as the lack of immediate functional rewards. Although several automated refactoring tools have been proposed, they remain limited in supporting a broad spectrum of refactoring types. In this study, we explore whether instruction strategies inspired by human best-practice guidelines can enhance the ability of Large Language Models (LLMs) to perform diverse refactoring tasks automatically. Leveraging the instruction-following and code comprehension capabilities of state-of-the-art LLMs (e.g., GPT-mini and DeepSeek-V3), we draw on Martin Fowler's refactoring guidelines to design multiple instruction strategies that encode motivations, procedural steps, and transformation objectives for 61 well-known refactoring types. We evaluate these strategies on benchmark examples and real-world code snippets from GitHub projects. Our results show that instruction designs grounded in Fowler's guidelines enable LLMs to successfully perform all benchmark refactoring types and preserve program semantics in real-world settings, an essential criterion for effective refactoring. Moreover, while descriptive instructions are more interpretable to humans, our results show that rule-based instructions often lead to better performance in specific scenarios. Interestingly, allowing models to focus on the overall goal of refactoring, rather than prescribing a fixed transformation type, can yield even greater improvements in code quality.

This study presents PsychoLexTherapy, a framework for simulating psychotherapeutic reasoning in Persian using small language models (SLMs). The framework tackles the challenge of developing culturally grounded, therapeutically coherent dialogue systems with structured memory for multi-turn interactions in underrepresented languages. To ensure privacy and feasibility, PsychoLexTherapy is optimized for on-device deployment, enabling use without external servers. Development followed a three-stage process: (i) assessing SLMs psychological knowledge with PsychoLexEval; (ii) designing and implementing the reasoning-oriented PsychoLexTherapy framework; and (iii) constructing two evaluation datasets-PsychoLexQuery (real Persian user questions) and PsychoLexDialogue (hybrid simulated sessions)-to benchmark against multiple baselines. Experiments compared simple prompting, multi-agent debate, and structured therapeutic reasoning paths. Results showed that deliberate model selection balanced accuracy, efficiency, and privacy. On PsychoLexQuery, PsychoLexTherapy outperformed all baselines in automatic LLM-as-a-judge evaluation and was ranked highest by human evaluators in a single-turn preference study. In multi-turn tests with PsychoLexDialogue, the long-term memory module proved essential: while naive history concatenation caused incoherence and information loss, the full framework achieved the highest ratings in empathy, coherence, cultural fit, and personalization. Overall, PsychoLexTherapy establishes a practical, privacy-preserving, and culturally aligned foundation for Persian psychotherapy simulation, contributing novel datasets, a reproducible evaluation pipeline, and empirical insights into structured memory for therapeutic reasoning.

Existing anomaly detection (AD) methods for tabular data usually rely on some assumptions about anomaly patterns, leading to inconsistent performance in real-world scenarios. While Large Language Models (LLMs) show remarkable reasoning capabilities, their direct application to tabular AD is impeded by fundamental challenges, including difficulties in processing heterogeneous data and significant privacy risks. To address these limitations, we propose LLM-DAS, a novel framework that repositions the LLM from a ``data processor'' to an ``algorithmist''. Instead of being exposed to raw data, our framework leverages the LLM's ability to reason about algorithms. It analyzes a high-level description of a given detector to understand its intrinsic weaknesses and then generates detector-specific, data-agnostic Python code to synthesize ``hard-to-detect'' anomalies that exploit these vulnerabilities. This generated synthesis program, which is reusable across diverse datasets, is then instantiated to augment training data, systematically enhancing the detector's robustness by transforming the problem into a more discriminative two-class classification task. Extensive experiments on 36 TAD benchmarks show that LLM-DAS consistently boosts the performance of mainstream detectors. By bridging LLM reasoning with classic AD algorithms via programmatic synthesis, LLM-DAS offers a scalable, effective, and privacy-preserving approach to patching the logical blind spots of existing detectors.

Vision-Language-Action (VLA) models represent a pivotal advance in embodied intelligence, yet they confront critical barriers to real-world deployment, most notably catastrophic forgetting. This issue stems from their overreliance on continuous action sequences or action chunks, which inadvertently create isolated data silos that disrupt knowledge retention across tasks. To tackle these challenges, we propose the Narrowing of Trajectory VLA (NoTVLA) framework: a novel approach that narrows its focus to sparse trajectories, thereby avoiding the catastrophic forgetting associated with dense trajectory fine-tuning. A key innovation of NoTVLA lies in its trajectory planning strategy: instead of centering on the target object's trajectory, it leverages temporal compression and spatial reasoning pruning specifically for the robot end effector's trajectory. Furthermore, training is conducted using these sparse trajectories rather than dense action trajectories, an optimization that delivers remarkable practical advantages with better performance in zero-shot. This design ensures that NoTVLA's operational accuracy closely approximates that of single-task expert models. Crucially, it also preserves the model's inherent language capabilities, enabling zero-shot generalization in specific scenarios, supporting unified model deployment across multiple robot platforms, and fostering a degree of generalization even when perceiving tasks from novel perspectives. Vision-Language-Action (VLA) models have emerged as a transformative force for embodied intelligence, heralding a new era of integrated perception, reasoning, and control (Kim et al., 2024; Brohan et al., 2023; Driess et al., 2023; Alayrac et al., 2022). By coupling large-scale multimodal reasoning engines with action experts, these systems aspire to endow robots with the capacity to operate robustly in open-ended, unstructured environments. Y et, despite rapid progress, several obstacles still impede reliable deployment. Foremost among them is catastrophic forgetting (French, 1999; Kirkpatrick et al., 2017; Zenke et al., 2017), which is aggravated in current pipelines by an overreliance on dense, continuously parameterized action trajectories (or "action chunks"). Such representations foster isolated data silos: when finetuned on a new task, the model may overwrite previously consolidated competencies, degrading prior-task performance.

Starting from flow- and diffusion-based transformers, Multi-modal Diffusion Transformers (MM-DiTs) have reshaped text-to-vision generation, gaining acclaim for exceptional visual fidelity. As these models advance, users continually push the boundary with imaginative or rare prompts, which advanced models still falter in generating, since their concepts are often too scarce to leave a strong imprint during pre-training. In this paper, we propose a simple yet effective intervention that surfaces rare semantics inside MM-DiTs without additional training steps, data, denoising-time optimization, or reliance on external modules (e.g., large language models). In particular, the joint-attention mechanism intrinsic to MM-DiT sequentially updates text embeddings alongside image embeddings throughout transformer blocks. We find that by mathematically expanding representational basins around text token embeddings via variance scale-up before the joint-attention blocks, rare semantics clearly emerge in MM-DiT's outputs. Furthermore, our results generalize effectively across text-to-vision tasks, including text-to-image, text-to-video, and text-driven image editing. Our work invites generative models to reveal the semantics that users intend, once hidden yet ready to surface.