KunlunBaize: LLM with Multi-Scale Convolution and Multi-Token Prediction Under TransformerX Framework

Cheng Li, Jiexiong Liu, Yixuan Chen +2

Large language models have demonstrated remarkable performance across various tasks, yet they face challenges such as low computational efficiency, gradient vanishing, and difficul…