1
Complexity-Guided Component-wise Initialization for Language Model Pretraining
巧妙利用预训练模型权重谱的结构化模式作为初始化信号,提升GPT-2风格语言模型预训练效率
arXiv:2607.09204v1 Announce Type: cross Abstract: Pretrained language models often exhibit structured weight spectra, suggesting that training may rep…