BriefGPT.xyz
Oct, 2023
用SEED令牌化器使LLaMA具备视觉和绘图能力
Making LLaMA SEE and Draw with SEED Tokenizer
HTML
PDF
Yuying Ge, Sijie Zhao, Ziyun Zeng, Yixiao Ge, Chen Li...
TL;DR
通过引入SEED图像标记器,使LLMs能够在其原始训练配方下执行可扩展的多模式自回归,并在广泛的多模式理解和生成任务中展示出令人印象深刻的性能。
Abstract
The great success of
large language models
(LLMs) has expanded the potential of
multimodality
, contributing to the gradual evolution of
general a
→