This paper proposes a preconditioning layer for stable weight conditioning in LLM training.

The paper argues that post-hoc mitigation techniques like machine unlearning are insufficient to cure legal liability arising from the unlawful acquisition and training on copyrighted data, advocating…

The paper proposes a simulation-trained variable impedance control framework for wearable exoskeletons that safely and effectively augments human physical capabilities across multiple tasks.

Adaptive data selection significantly improves wearable prediction performance, particularly for individuals with poor baseline health metrics, suggesting that selective data sampling should be tailor…

This paper demonstrates that existing open-weight LLM safeguards are vulnerable to simple, non-gradient-based attacks like abliteration and prefilling, significantly increasing the attack success rate…

The paper demonstrates that using synthetic hand images containing accessories, generated via inpainting, significantly improves the robustness of hand detectors for safety-critical applications by cl…

The paper proposes a novel method for federated learning that allows devices holding only a single data sample to collaboratively train an accurate, privacy-preserving global model.

This paper analyzes the poor performance of Meta-learning for Training-data Selection (MTS) and proposes that increasing the batch size and incorporating informative features can significantly improve…

The paper introduces FoLoRA, a novel optimization framework that uses a generalized Rayleigh quotient to achieve a superior balance between adapting foundation models to specific tasks and preserving…

The paper introduces CoRP, a gradient-free operator that consolidates the benefits of ensemble-based post-training methods into a single, deployable model update, significantly improving performance w…

This paper establishes a large deviation principle for the generalization error of interpolating classifiers in the overparametrized regime.

The paper proposes a coherent inverse reinforcement learning (IRL) method to improve large behavior models for robotic control, achieving superior sample efficiency and performance on complex sparse m…

The paper introduces Sparse Memory-Efficient Training (SMET), a method that stabilizes and optimizes Dynamic Sparse Training (DST) for large language models, enabling stable and memory-efficient spars…

The paper proposes DAMEL, a dual-axis multi-expert learning algorithm that simultaneously reduces both prediction bias and variance in class-imbalanced learning by leveraging multiple experts across b…

This paper synthesizes over 150 scattered studies and reports to provide the first comprehensive primer on post-training reasoning data, organizing the field around data objects, utility, construction…

This paper proposes a new framework called STRIDE for training data attribution in Large Language Models.

The paper proposes Self-Trained Verification (STV), a novel method that trains verifiers to catch self-generated errors by leveraging reference solutions, significantly boosting performance in both te…

The paper introduces the Vector Network (VN), a novel recurrent architecture that replaces fixed weight matrices with reusable weight atoms, enabling superior compositional generalization by making st…

The paper proposes FOAM, an adaptive damping method that stabilizes the Shampoo optimization algorithm by dynamically controlling damping and eigendecomposition frequency, thereby reducing staleness-i…

The paper proposes a unified hybrid framework that combines data-level and algorithm-level balancing to effectively address the challenge of imbalanced regression, significantly improving predictive p…