arXiv:2510.18934v1 Announce Type: new Abstract: A wide variety of generalization measures have been applied to deep neural networks (DNNs). Although obtaining tight bounds remains challenging, such measures are often assumed to reproduce qualitative generalization trends. In this position paper, we…
arXiv:2508.00890v2 Announce Type: replace-cross Abstract: Test-time scaling (TTS) enhances the performance of large language models (LLMs) by allocating additional compute resources during inference. However, existing research primarily investigates TTS in single-stage tasks; while many real-world problems are multi-stage complex tasks,…
arXiv:2510.18925v1 Announce Type: new Abstract: Modeling and predicting the dynamics of complex multiscale systems remains a significant challenge due to their inherent nonlinearities and sensitivity to initial conditions, as well as limitations of traditional machine learning methods that fail to…
arXiv:2510.18927v1 Announce Type: new Abstract: Reinforcement learning (RL) has recently become the core paradigm for aligning and strengthening large language models (LLMs). Yet, applying RL in off-policy settings–where stale data from past policies are used for training–improves sample efficiency, but…
arXiv:2510.18924v1 Announce Type: new Abstract: Reinforcement learning from human feedback (RLHF) or verifiable rewards (RLVR), the standard paradigm for aligning LLMs or building recent SOTA reasoning models, is highly sensitive to noise from inconsistent or erroneous rewards. Yet, the interaction…
arXiv:2510.18921v1 Announce Type: new Abstract: The recent widespread adoption of Large Language Models (LLMs) and machine learning in general has sparked research interest in exploring the possibilities of deploying these models on smaller devices such as laptops and mobile phones.…
arXiv:2510.18913v1 Announce Type: new Abstract: Anchored Direct Preference Optimization (ADPO) is a unified framework that generalizes Direct Preference Optimization (DPO) with soft preferences, reference-policy anchoring, and groupwise extensions. While standard DPO assumes hard binary labels and pairwise comparisons, ADPO introduces:…
arXiv:2510.17313v2 Announce Type: replace Abstract: Learning disentangled representations in sequential data is a key goal in deep learning, with broad applications in vision, audio, and time series. While real-world data involves multiple interacting semantic factors over time, prior work has…
arXiv:2510.18940v1 Announce Type: new Abstract: Existing parameter-efficient fine-tuning (PEFT) methods primarily fall into two categories: addition-based and selective in-situ adaptation. The former, such as LoRA, introduce additional modules to adapt the model to downstream tasks, offering strong memory efficiency. However,…
arXiv:2503.21681v3 Announce Type: replace-cross Abstract: The relationship between RNA structure and function has recently attracted interest within the deep learning community, a trend expected to intensify as nucleic acid structure models advance. Despite this momentum, the lack of standardized, accessible…
