← Search

Weidong Zhou

2 accepted papers

2026

InfoLaw: Information Scaling Laws for Large Language Models with Quality-Weighted Mixture Data and Repetition

ICML 2026poster

Upweighting high-quality data in LLM pretraining often improves performance, but in data-limited regimes, especially under overtraining, stronger upweighting increases repetition and can degrade performance. However, standard scaling laws do not reliably extrapolate across mixture recipes or under r…

Cited by 0SourceScholar
2026

Target-Oriented Pretraining Data Selection via Neuron-Activated Graph

ICML 2026poster

Everyday tasks come with a target, and pretraining models around this target is what turns them into experts. In this paper, we study target-oriented language model (LM) pretraining by introducing ***N**euron-**A**ctivated **G**raph Ranking* (NAG-based Ranking), a training-free and interpretable fra…

Cited by 0SourceScholar