March 2026

arXiv:2603.02230v1 Announce Type: new Abstract: Self-correction is an effective technique for maintaining parallel sampling in discrete diffusion models with minimal performance degradation. Prior work has explored self-correction at inference time or during post-training; however, such approaches often suffer from limited generalization and may impair reasoning performance. GIDD pioneers pretraining-based self-correction via a multi-step BERT-style uniform-absorbing objective. However, GIDD relies on a continuous interpolation-based pipeline with opaque interactions between uniform transitions and absorbing masks, which complicates hyperparameter tuning and […]

Ver mais

arXiv:2603.02229v1 Announce Type: new Abstract: Safety post-training has been studied extensively in single-step “chat” settings where safety typically refers to refusing harmful requests. We study an “agentic” (i.e., multi-step, tool-use) setting where safety refers to harmful actions directly taken by the LLM. We compare the effects of running direct preference optimization (DPO) on safety or helpfulness alone vs both metrics sequentially. As expected, training on one metric alone results in an extreme point along this frontier. However, unlike […]

Ver mais

arXiv:2603.02228v1 Announce Type: new Abstract: The proof that Large Language Models (LLMs) augmented with external read-write memory constitute a computationally universal system has established the theoretical foundation for general-purpose agents. However, existing implementations face a critical bottleneck: the finite and costly Context Window, which functions not as infinite memory but as a scarce semantic cache. In this work, we introduce textit{Neural Paging}, a hierarchical architecture that decouples symbolic reasoning from information resource management. We formulate the textit{Context Paging […]

Ver mais

arXiv:2603.02227v1 Announce Type: new Abstract: Can a transformer learn which attention entries matter during training? In principle, yes: attention distributions are highly concentrated, and a small gate network can identify the important entries post-hoc with near-perfect accuracy. In practice, barely. When sparse attention is trained end-to-end, the model’s Q/K/V projections co-adapt to whatever mask is imposed, absorbing the routing signal until learned gates perform little better than frozen random gates. We call this routing absorption and present four […]

Ver mais

arXiv:2603.02226v1 Announce Type: new Abstract: Real-world sequential signals, such as audio or video, contain critical information that is often embedded within long periods of silence or noise. While recurrent neural networks (RNNs) are designed to process such data efficiently, they often suffer from “memory decay” due to a rigid update schedule: they typically update their internal state at every time step, even when the input is static. This constant activity forces the model to overwrite its own memory […]

Ver mais

arXiv:2603.02225v1 Announce Type: new Abstract: Learning from feedback is an instrumental process for advancing the capabilities and safety of frontier models, yet its effectiveness is often constrained by cost and scalability. We present a pilot study that explores scaling reward models through unsupervised approaches. We operationalize reward-based scaling (RBS), in its simplest form, as preference learning over document prefixes and suffixes drawn from large-scale web corpora. Its advantage is demonstrated in various aspects: despite using no human annotations, […]

Ver mais

arXiv:2603.02224v1 Announce Type: new Abstract: Low-Rank Adaptation (LoRA) has emerged as a parameter-efficient approach for adapting large pre-trained models, yet its behavior under continual learning remains poorly understood. We present a geometric theory characterizing catastrophic forgetting in LoRA through the lens of gradient subspace interactions. Our central finding is that forgetting is governed by a simple geometric law: $mathcal{F} = alpha(1 – cos^2theta_{min}) + beta$, where $theta_{min}$ is the minimum principal angle between task gradient subspaces. This formulation […]

Ver mais

arXiv:2603.02223v1 Announce Type: new Abstract: Wildfire evacuation behavior is highly variable and influenced by complex interactions among household resources, preparedness, and situational cues. Using a large-scale MTurk survey of residents in California, Colorado, and Oregon, this study integrates unsupervised and supervised machine learning methods to uncover latent behavioral typologies and predict key evacuation outcomes. Multiple Correspondence Analysis, K-Modes clustering, and Latent Class Analysis reveal consistent subgroups differentiated by vehicle access, disaster planning, technological resources, pet ownership, and residential […]

Ver mais

arXiv:2603.02222v1 Announce Type: new Abstract: MedCalc-Bench is a widely used benchmark for evaluating LLM performance on clinical calculator tasks, with state-of-the-art direct prompting scores plateauing around 35% on the Verified split (HELM MedHELM leaderboard) and the best published approach-RL with verifiable rewards-reaching 74%. We present three contributions that challenge the benchmark’s current framing. First, we conduct a systematic audit of the benchmark’s calculator implementations, identifying and fixing over 20 errors ranging from critical formula inaccuracies to runtime bugs […]

Ver mais

arXiv:2603.02221v1 Announce Type: new Abstract: In healthcare tabular predictions, classical models with feature engineering often outperform neural approaches. Recent advances in Large Language Models enable the integration of domain knowledge into feature engineering, offering a promising direction. However, existing approaches typically rely on a broad search over predefined transformations, overlooking downstream model characteristics and feature importance signals. We present MedFeat, a feedback-driven and model-aware feature engineering framework that leverages LLM reasoning with domain knowledge and provides feature explanations […]

Ver mais