1 paper · 1 filter
Tenghui Li, Guoxu Zhou, Xuyang Zhao +1
Large language models have demonstrated predictable scaling behaviors with respect to model parameters and training data. This study investigates whether a similar scaling relation…