Reward models (RMs) are fundamental to aligning Large Language Models (LLMs) via human feedback, yet they often suffer from reward hacking. They tend to latch on to superficial or spurious attributes, such as response length or formatting, mistaking these cues learned from correlations in training data for the true causal drivers of quality (e.g., factuality, relevance). This occurs because standard training objectives struggle to disentangle these factors, leading to brittle RMs and misaligned policies. We introduce Crome (Causally Robust Reward Modeling), a novel framework grounded in an explicit causal model designed to mitigate reward hacking. Crome employs the following synthetic targeted augmentations during training: (1) Causal Augmentations, which are pairs that differ along specific causal attributes, to enforce sensitivity along each causal attribute individually, and (2) Neutral Augmentations, which are tie-label pairs varying primarily in spurious attributes, to enforce invariance along spurious attributes. Notably, our augmentations are produced without any knowledge of spurious factors, via answer interventions only along causal rubrics, that are identified by querying an oracle LLM. Empirically, Crome significantly outperforms standard baselines on RewardBench, improving average accuracy by up to 5.4% and achieving gains of up to 13.2% and 7.2% in specific categories. The robustness of Crome is further testified by the consistent gains obtained in a Best-of-N inference setting across increasing N, across various benchmarks, including the popular RewardBench (covering chat, chat-hard, safety, and reasoning tasks), the safety-focused WildGuardTest, and the reasoning-specific GSM8k.

Recent advances in model pre-training give rise to task adaptation-based few-shot learning (FSL), where the goal is to adapt a pre-trained task-agnostic model for capturing task-specific knowledge with a few-labeled support samples of the target task.Nevertheless, existing approaches may still fail in the open world due to the inevitable in-distribution (ID) and out-of-distribution (OOD) noise from both support and query samples of the target task. With limited support samples available, i) the adverse effect of the dual noises can be severely amplified during task adaptation, and ii) the adapted model can produce unreliable predictions on query samples in the presence of the dual noises. In this work, we propose DEnoised Task Adaptation (DETA++) for reliable FSL. DETA++ uses a Contrastive Relevance Aggregation (CoRA) module to calculate image and region weights for support samples, based on which a clean prototype loss and a noise entropy maximization loss are proposed to achieve noise-robust task adaptation. Additionally,DETA++ employs a memory bank to store and refine clean regions for each inner-task class, based on which a Local Nearest Centroid Classifier (LocalNCC) is devised to yield noise-robust predictions on query samples. Moreover, DETA++ utilizes an Intra-class Region Swapping (IntraSwap) strategy to rectify ID class prototypes during task adaptation, enhancing the model's robustness to the dual noises. Extensive experiments demonstrate the effectiveness and flexibility of DETA++.

Large language models (LLMs) offer emerging opportunities for psychological and behavioral research, but methodological guidance is lacking. This article provides a framework for using LLMs as psychological simulators across two primary applications: simulating roles and personas to explore diverse contexts, and serving as computational models to investigate cognitive processes. For simulation, we present methods for developing psychologically grounded personas that move beyond demographic categories, with strategies for validation against human data and use cases ranging from studying inaccessible populations to prototyping research instruments. For cognitive modeling, we synthesize emerging approaches for probing internal representations, methodological advances in causal interventions, and strategies for relating model behavior to human cognition. We address overarching challenges including prompt sensitivity, temporal limitations from training data cutoffs, and ethical considerations that extend beyond traditional human subjects review. Throughout, we emphasize the need for transparency about model capabilities and constraints. Together, this framework integrates emerging empirical evidence about LLM performance--including systematic biases, cultural limitations, and prompt brittleness--to help researchers wrangle these challenges and leverage the unique capabilities of LLMs in psychological research.

Small language models (SLMs) struggle to learn complex reasoning behaviors, especially when high-quality traces are scarce or difficult to learn from. The standard training approach combines a supervised fine-tuning (SFT) stage, often to distill capabilities of a larger model, followed by a reinforcement learning (RL)stage such as Group Relative Policy Optimization (GRPO). In this paper, we investigate the fundamental limitations of this SFT + RL paradigm and propose methods to overcome them. Under a suitable theoretical model, we demonstrate that the SFT + RL strategy can fail completely when (1) the expert's traces are too difficult for the small model to express, or (2) the small model's initialization has exponentially small likelihood of success. To address these, we introduce BREAD: a GRPO variant that unifies the SFT and RL stages via partial expert guidance and branched rollouts. When self-generated traces fail, BREAD adaptively inserts short expert prefixes/hints, allowing the small model to complete the rest of the reasoning path, and ensuring that each update includes at least one successful trace. This mechanism both densifies the reward signal and induces a natural learning curriculum. BREAD requires fewer than 40% of ground-truth traces, consistently outperforming standard GRPO while speeding up the training by about 3 times. Importantly, we demonstrate that BREAD helps the model solve problems that are otherwise unsolvable by the SFT + RL strategy, highlighting how branched rollouts and expert guidance can substantially boost SLM reasoning.

Robust estimation of heterogeneous treatment effects is a fundamental challenge for optimal decision-making in domains ranging from personalized medicine to educational policy. In recent years, predictive machine learning has emerged as a valuable toolbox for causal estimation, enabling more flexible effect estimation. However, accurately estimating conditional average treatment effects (CATE) remains a major challenge, particularly in the presence of many covariates. In this article, we propose pretraining strategies that leverage a phenomenon in real-world applications: factors that are prognostic of the outcome are frequently also predictive of treatment effect heterogeneity. In medicine, for example, components of the same biological signaling pathways frequently influence both baseline risk and treatment response. Specifically, we demonstrate our approach within the R-learner framework, which estimates the CATE by solving individual prediction problems based on a residualized loss. We use this structure to incorporate side information and develop models that can exploit synergies between risk prediction and causal effect estimation. In settings where these synergies are present, this cross-task learning enables more accurate signal detection, yields lower estimation error, reduced false discovery rates, and higher power for detecting heterogeneity.

Despite recent efforts, neural networks still struggle to learn in non-stationary environments, and our understanding of catastrophic forgetting (CF) is far from complete. In this work, we perform a systematic study on the impact of model scale and the degree of feature learning in continual learning. We reconcile existing contradictory observations on scale in the literature, by differentiating between lazy and rich training regimes through a variable parameterization of the architecture. We show that increasing model width is only beneficial when it reduces the amount of feature learning, yielding more laziness. Using the framework of dynamical mean field theory, we then study the infinite width dynamics of the model in the feature learning regime and characterize CF, extending prior theoretical results limited to the lazy regime. We study the intricate relationship between feature learning, task non-stationarity, and forgetting, finding that high feature learning is only beneficial with highly similar tasks. We identify a transition modulated by task similarity where the model exits an effectively lazy regime with low forgetting to enter a rich regime with significant forgetting. Finally, our findings reveal that neural networks achieve optimal performance at a critical level of feature learning, which depends on task non-stationarity and transfers across model scales. This work provides a unified perspective on the role of scale and feature learning in continual learning.

Diffusion models have recently gained significant attention due to their effectiveness in various scientific domains, including biochemistry. When trained on equilibrium molecular distributions, diffusion models provide both: a generative procedure to sample equilibrium conformations and associated forces derived from the model's scores. However, using the forces for coarse-grained molecular dynamics simulations uncovers inconsistencies in the samples generated via classical diffusion inference and simulation, despite both originating from the same model. Particularly at the small diffusion timesteps required for simulations, diffusion models fail to satisfy the Fokker-Planck equation, which governs how the score should evolve over time. We interpret this deviation as an indication of the observed inconsistencies and propose an energy-based diffusion model with a Fokker-Planck-derived regularization term enforcing consistency. We demonstrate the effectiveness of our approach on toy systems, alanine dipeptide, and introduce a state-of-the-art transferable Boltzmann emulator for dipeptides that supports simulation and demonstrates enhanced consistency and efficient sampling.

We propose S$^2$GPT-PINN, a sparse and small model for solving parametric partial differential equations (PDEs). Similar to Small Language Models (SLMs), S$^2$GPT-PINN is tailored to domain-specific (families of) PDEs and characterized by its compact architecture and minimal computational power. Leveraging a small amount of extremely high quality data via a mathematically rigorous greedy algorithm that is enabled by the large full-order models, S$^2$GPT-PINN relies on orders of magnitude less parameters than PINNs to achieve extremely high efficiency via two levels of customizations. The first is knowledge distillation via task-specific activation functions that are transferred from Pre-Trained PINNs. The second is a judicious down-sampling when calculating the physics-informed loss of the network compressing the number of data sites by orders of magnitude to the size of the small model.

When applicants get rejected by an algorithmic decision system, recourse explanations provide actionable suggestions for how to change their input features to get a positive evaluation. A crucial yet overlooked phenomenon is that recourse explanations are performative: When many applicants act according to their recommendations, their collective behavior may change statistical regularities in the data and, once the model is refitted, also the decision boundary. Consequently, the recourse algorithm may render its own recommendations invalid, such that applicants who make the effort of implementing their recommendations may be rejected again when they reapply. In this work, we formally characterize the conditions under which recourse explanations remain valid under performativity. A key finding is that recourse actions may become invalid if they are influenced by or if they intervene on non-causal variables. Based on our analysis, we caution against the use of standard counterfactual explanations and causal recourse methods, and instead advocate for recourse methods that recommend actions exclusively on causal variables.

-- This paper details four principal challenges encountered with machine learning (ML) damage assessment using small uncrewed aerial systems (sUAS) at Hurricanes Debby and Helene that prevented, degraded, or delayed the delivery of data products during operations and suggests three research directions for future real-world deployments. The presence of these challenges is not surprising given that a review of the literature considering both datasets and proposed ML models suggests this is the first sUAS-based ML system for disaster damage assessment actually deployed as a part of real-world operations. The sUAS-based ML system was applied by the State of Florida to Hurricanes Helene (2 orthomosaics, 3.0 gigapixels collected over 2 sorties by a Wintra WingtraOne sUAS) and Debby (1 orthomosaic, 0.59 gigapixels collected via 1 sortie by a Wintra WingtraOne sUAS) in Florida. The same model was applied to crewed aerial imagery of inland flood damage resulting from post-tropical remnants of Hurricane Debby in Pennsylvania (436 orthophotos, 136.5 gigapixels), providing further insights into the advantages and limitations of sUAS for disaster response. The four challenges (variation in spatial resolution of input imagery, spatial misalignment between imagery and geospatial data, wireless connectivity, and data product format) lead to three recommendations that specify research needed to improve ML model capabilities to accommodate the wide variation of potential spatial resolutions used in practice, handle spatial misalignment, and minimize the dependency on wireless connectivity. These recommendations are expected to improve the effective operational use of sUAS and sUAS-based ML damage assessment systems for disaster response.