Reinforcement Learning with Verifiable Rewards (RLVR) has proven effective for complex reasoning tasks with clear correctness signals such as math and coding. However, extending it to real-world reasoning tasks is challenging, as evaluation depends on nuanced, multi-criteria judgments rather than binary correctness. Instance-specific rubrics have recently been used in evaluation benchmarks to capture such judgments, but their potential as reward signals for on-policy post-training remains underexplored. We introduce $\textbf{Rubrics as Rewards}$ (RaR), an on-policy reinforcement learning method that extends RLVR beyond verifiable domains by using rubric-based feedback. Across both medical and science domains, we evaluate multiple strategies for aggregating rubric feedback into rewards. The best RaR variant achieves relative improvements of up to $31\%$ on HealthBench and $7\%$ on GPQA-Diamond over popular LLM-as-judge baselines that rely on direct Likert-based rewards. These results demonstrate that RaR-trained policies adapt well to diverse evaluation formats, performing strongly on both rubric-based and multiple-choice tasks. Moreover, we find that using rubrics as structured reward signals yields better alignment for smaller judges and reduces performance variance across judge scales.

Symmetry-aware methods for machine learning, such as data augmentation and equivariant architectures, encourage correct model behavior on all transformations (e.g. rotations or permutations) of the original dataset. These methods can improve generalization and sample efficiency, under the assumption that the transformed datapoints are highly probable, or "important", under the test distribution. In this work, we develop a method for critically evaluating this assumption. In particular, we propose a metric to quantify the amount of anisotropy, or symmetry-breaking, in a dataset, via a two-sample neural classifier test that distinguishes between the original dataset and its randomly augmented equivalent. We validate our metric on synthetic datasets, and then use it to uncover surprisingly high degrees of alignment in several benchmark point cloud datasets. We show theoretically that distributional symmetry-breaking can actually prevent invariant methods from performing optimally even when the underlying labels are truly invariant, as we show for invariant ridge regression in the infinite feature limit. Empirically, we find that the implication for symmetry-aware methods is dataset-dependent: equivariant methods still impart benefits on some anisotropic datasets, but not others. Overall, these findings suggest that understanding equivariance -- both when it works, and why -- may require rethinking symmetry biases in the data.

The chemist begins by thoroughly analyzing the molecular structure--recognizing the core scaffold, functional groups, stereochemical configurations, and the relative positions of these structural elements. With this understanding, the chemist reasons about potential modifications, such as substituting functional groups, adjusting ring systems, or altering stereochemistry, to improve the target property. Finally, guided by precise modification instructions, the chemist applies these changes to generate an optimized molecule (illustrated in Figure 1a). In another scenario, a chemist may need to design a new (de novo) molecule to satisfy a set of property requirements. This process involves reasoning about possible molecular structures that meet the design constraints and then synthesizing a novel compound based on the resulting detailed structural specification (shown in Figure 1b). Recent advancements in artificial intelligence (AI) provide promising opportunities to assist chemists in these complex workflows. While traditional molecular models, based on graph or sequence representations such as SMILES (Weininger, 1988), have demonstrated strong performance in domain-specific tasks including molecular property prediction (Huang et al., 2021), reaction prediction (Lowe, 2017), and conditional generation (Brown et al., 2019), they operate exclusively within the molecular modality and lack the ability to interpret or execute natural language instructions. Meanwhile, large language models (LLMs) have exhibited remarkable reasoning capabilities (OpenAI, 2025d), showing promise in tackling scientific problems that traditionally require Clemson University.

We introduce the soft curvature and spectroscopy (SCANS) system: a versatile, electronics-free, fluidically actuated soft manipulator capable of assessing the spectral properties of objects either in hand or through pre-touch caging. This platform offers a wider spectral sensing capability than previous soft robotic counterparts. We perform a material analysis to explore optimal soft substrates for spectral sensing, and evaluate both pre-touch and in-hand performance. Experiments demonstrate explainable, statistical separation across diverse object classes and sizes (metal, wood, plastic, organic, paper, foam), with large spectral angle differences between items. Through linear discriminant analysis, we show that sensitivity in the near-infrared wavelengths is critical to distinguishing visually similar objects. These capabilities advance the potential of optics as a multi-functional sensory modality for soft robots. The complete parts list, assembly guidelines, and processing code for the SCANS gripper are accessible at: https://parses-lab.github.io/scans/.

Efficient and inexpensive energy storage is essential for accelerating the adoption of renewable energy and ensuring a stable supply, despite fluctuations in sources such as wind and solar. Electrocatalysts play a key role in hydrogen energy storage (HES), allowing the energy to be stored as hydrogen. However, the development of affordable and high-performance catalysts for this process remains a significant challenge. We introduce Catalyst GFlowNet, a generative model that leverages machine learning-based predictors of formation and adsorption energy to design crystal surfaces that act as efficient catalysts. We demonstrate the performance of the model through a proof-of-concept application to the hydrogen evolution reaction, a key reaction in HES, for which we successfully identified platinum as the most efficient known catalyst. In future work, we aim to extend this approach to the oxygen evolution reaction, where current optimal catalysts are expensive metal oxides, and open the search space to discover new materials. This generative modeling framework offers a promising pathway for accelerating the search for novel and efficient catalysts.

Language Models (LMs) are increasingly used in applications where generated outputs must satisfy strict semantic or syntactic constraints. Existing approaches to constrained generation fall along a spectrum: greedy constrained decoding methods enforce validity during decoding but distort the LM's distribution, while rejection sampling (RS) preserves fidelity but wastes computation by discarding invalid outputs. Both extremes are problematic in domains such as program fuzzing, where both validity and diversity of samples are essential. We present Constrained Adaptive Rejection Sampling (CARS), an approach that strictly improves the sample-efficiency of RS without distributional distortion. CARS begins with unconstrained LM sampling and adaptively rules out constraint-violating continuations by recording them in a trie and subtracting their probability mass from future draws. This adaptive pruning ensures that prefixes proven invalid are never revisited, acceptance rates improve monotonically, and the resulting samples exactly follow the constrained distribution. In experiments on a variety of domains -- e.g., program fuzzing and molecular generation -- CARS consistently achieves higher efficiency -- measured in the number of LM forward passes per valid sample -- while also producing stronger sample diversity than both GCD and methods that approximate the LM's distribution.

Neural Information Processing Systems http://nips.cc/

The emergence of agent-based systems represents a significant advancement in artificial intelligence, with growing applications in automated data extraction. However, chemical information extraction remains a formidable challenge due to the inherent heterogeneity of chemical data. Current agent-based approaches, both general-purpose and domain-specific, exhibit limited performance in this domain. To address this gap, we present ChemX, a comprehensive collection of 10 manually curated and domain-expert-validated datasets focusing on nanomaterials and small molecules. These datasets are designed to rigorously evaluate and enhance automated extraction methodologies in chemistry. To demonstrate their utility, we conduct an extensive benchmarking study comparing existing state-of-the-art agentic systems such as ChatGPT Agent and chemical-specific data extraction agents. Additionally, we introduce our own single-agent approach that enables precise control over document preprocessing prior to extraction. We further evaluate the performance of modern baselines, such as GPT-5 and GPT-5 Thinking, to compare their capabilities with agentic approaches. Our empirical findings reveal persistent challenges in chemical information extraction, particularly in processing domain-specific terminology, complex tabular and schematic representations, and context-dependent ambiguities. The ChemX benchmark serves as a critical resource for advancing automated information extraction in chemistry, challenging the generalization capabilities of existing methods, and providing valuable insights into effective evaluation strategies.

We propose Cormorant, a rotationally covariant neural network architecture for learning the behavior and properties of complex many-body physical systems. We apply these networks to molecular systems with two goals: learning atomic potential energy surfaces for use in Molecular Dynamics simulations, and learning ground state properties of molecules calculated by Density Functional Theory. Some of the key features of our network are that (a) each neuron explicitly corresponds to a subset of atoms; (b) the activation of each neuron is covariant to rotations, ensuring that overall the network is fully rotationally invariant. Furthermore, the non-linearity in our network is based upon tensor products and the Clebsch-Gordan decomposition, allowing the network to operate entirely in Fourier space. Cormorant significantly outperforms competing algorithms in learning molecular Potential Energy Surfaces from conformational geometries in the MD-17 dataset, and is competitive with other methods at learning geometric, energetic, electronic, and thermodynamic properties of molecules on the GDB-9 dataset.

To address this, we leverage parametric modular structure to learn component-level surrogates, enabling cheaper high-fidelity simulation. We use a neural network to model the stored potential energy in a component given boundary conditions.