header-langage
简体中文
繁體中文
English
Tiếng Việt
한국어
日本語
ภาษาไทย
Türkçe
Scan to Download the APP
Partnered With:

Xiaomi's Chief Language Model Officer, Luo Fuli: Language Model Takes the Lead in Self-evolution, Efficient Video World Model Still Needs Time

According to Dynamical Beating monitoring, at the 2026 Beijing Intelligence Source Conference, Xiaomi's MiMo Large Model Leader, Luo Fuli, pointed out that in the path to reconstruct the world, the Language Model (LLM) has taken the lead in achieving self-evolution due to its ease in reproducing the intelligent environment in the digital world and designing a precise reward system. This is a clear and continuously scaling main path.

Luo Fuli stated that compared to the rapid breakthrough of large language models, the World Model is still in the early exploration stage. The industry has not yet seen a truly efficient model that can reconstruct the world from a high-quality video perspective. To achieve scaling in future World Models, it is still necessary to address more efficient infrastructure, physical systems closer to the real world, and core underlying issues such as how to introduce reinforcement learning.

Zhu Jun, the founder of Sheng Shu Technology, also added at this forum that the future World Model will inevitably be a "Universal World Model" similar to what GPT is to the universal pedestal, possessing strong foundational properties and generalization capabilities.

举报 Correction/Report
Correction/Report
Submit
Add Library
Visible to myself only
Public
Save
Choose Library
Add Library
Cancel
Finish