Hello everyone, I’m Hongcan Guo, a mathematics and artificial intelligence enthusiast! This is my homepage, where I will share some of the latest updates about my research life. Welcome!

I am an undergraduate student from the School of Artificial Intelligence at Beijing University of Posts and Telecommunications (BUPT). Currently, I am a research intern at ByteDance Seed Vision. My research interests include, but are not limited to, Natural Language Processing, Reinforcement Learning, Generative Models, Unsupervised Learning, and Mathematics.

I am currently applying for PhD programs for Fall 2026. If you are a professor interested in my profile, please feel free to contact me at Email.

RECENT NEWS

• [2025.06.26] One paper was accepted to ICCV 2025.
• [2025.06.09] One paper was accepted to JSAC.
• [2025.03.22] One paper was accepted to JSAC.

Latest Research Directions

1. LLM Pre-training: Techniques for LLM pre-training, including the complete training pipeline, data composition strategies, model architecture, algorithms, and key observational metrics.
2. LLM Post-training: Full-stack optimization of LLMs via SFT, RL, and efficient fine-tuning.
3. Novel LLM Architecture Design: Focus on novel architectures like Mixture-of-Experts (MoE) and Diffusion LLMs to push the boundaries of foundation models in scale, performance, and generalization.
4. Reasoning Enhancement: Reinforcement learning-based methods for enhancing reasoning in multimodal LLMs, committed to achieving the goal of Artificial Super Intelligence (ASI).
5. Unified Architecture for Generation and Understanding: Explore unified architectures that go beyond full-modal understanding and generation, committed to achieving the goal of Artificial General Intelligence (AGI).
6. Autonomous Learning Frameworks: Novel learning paradigms beyond traditional supervised/RL settings, including multi-agent collaboration and unsupervised automation.

STATUS

Busy with PhD applications for Fall 2026.