Shaobo Guo

Shaobo Guo (郭少博). I'm currently working as a full-time algorithm engineer at Bytedance.

My work focuses on areas such as computer vision and multi-modal large language models. I have extensive experience in applying advanced algorithms to real-world business scenarios, building end-to-end systems that directly support and optimize company operations. Over the past few years, I have led or participated in multiple projects that significantly improved efficiency and solved complex business problems through intelligent algorithmic design and system-level implementation.

Conference Publications

ICCV 2025Beyond Text-Visual Attention: Exploiting Visual Cues for Effective Token Pruning in VLMs

Tech Reports

Mammothmoda1.0Mammothmoda: Multi-modal large language model

Mammothmoda2.0Mammoth2: A Unified AR Diffusion Framework for Multimodal Understanding and Generation

Profile photo of Shaobo Guo