Uncertainty-Aware Gradient Signal-to-Noise Data Selection for Instruction Tuning

2026-01-20 · Zhihang Yuan, Chengyu Yue, Long Huang, Litu Ou, Lei Shi arxiv

Instruction tuning is a standard paradigm for adapting large language models (LLMs), but modern instruction datasets are large, noisy, and redundant, making full-data fine-tuning costly and often unnecessary. Existing data selection methods either build expensive gradient datastores or assign static scores from a weak proxy, largely ignoring evolving uncertainty, and thus missing a key source of LLM interpretability. We propose GRADFILTERING, an objective-agnostic, uncertainty-aware data selection framework that utilizes a small GPT-2 proxy with a LoRA ensemble and aggregates per-example gradients into a Gradient Signal-to-Noise Ratio (G-SNR) utility. Our method matches or surpasses random subsets and strong baselines in most LLM-as-a-judge evaluations as well as in human assessment. Moreover, GRADFILTERING-selected subsets converge faster than competitive filters under the same compute budget, reflecting the benefit of uncertainty-aware scoring.

📄 PDF Abstract BibTeX arXiv:2601.13697

Code (0)

등록된 구현이 없습니다.

MURAL: Multimodal Uncertainty-aware Recommendation via Adaptive edge Learning

2026-09-04 · Ahmad Mousavi, Majid Alikhani, Yeon-Chang Lee, Roberto Corizzo 외 arxiv

Multimodal Graph Neural Networks have become standard for recommendation by augmenting sparse interaction data with content features. Yet current architectures face two bottlenecks: structural rigidity, from a reliance o…

Multimodal Recommendation

Why Semantic Entropy Fails: Geometry-Aware and Calibrated Uncertainty for Policy Optimization

2026-05-20 · Zheyuan Zhang, Kaiwen Shi, Han Bao, Zehong Wang 외 arxiv

Post-training has become central to improving reasoning and alignment in large language models, where critic-free models enable scalable learning from model-generated outputs but lack principled mechanisms to distinguish…

Uncertainty-Aware Spatial Color Correlation for Low-Light Image Enhancement

2025-08-06 · Jin Kuang, Dong Liu, Yukuang Zhang, Shengsheng Wang arxiv

Most existing low-light image enhancement approaches primarily focus on architectural innovations, while often overlooking the intrinsic uncertainty within feature representations particularly under extremely dark condit…

Low-Light Image Enhancement

On the Plasticity and Stability for Post-Training Large Language Models

2026-02-06 · Wenwen Qiang, Ziyin Gu, Jiahuan Zhou, Jie Hu 외 arxiv

Training stability remains a critical bottleneck for Group Relative Policy Optimization (GRPO), often manifesting as a trade-off between reasoning plasticity and general capability retention. We identify a root cause as …

SURE: Synergistic Uncertainty-aware Reasoning for Multimodal Emotion Recognition in Conversations

2026-04-02 · Yiqiang Cai, Chengyan Wu, Bolei Ma, Bo Chen 외 arxiv

Multimodal emotion recognition in conversations (MERC) requires integrating multimodal signals while being robust to noise and modeling contextual reasoning. Existing approaches often emphasize fusion but overlook uncert…

Multimodal Emotion Recognition Multimodal Reasoning