Posts

The emergence of Large Language Models (LLMs) has catalyzed a paradigm shift in artificial intelligence, particularly in how we approach complex problem-solving through multi-agent systems. As we stand in May 2025, the field of LLM-based multi-agent cooperation has evolved from theoretical curiosity to practical reality, with implementations spanning software development, financial trading, robotics, and enterprise operations.

This blog gives a graphical illustration of how history-state values are related with biases with state-based critics.

This blog gives a toy example called Guess Game, explaining Multi-Observation Variance (MOV) in our work on Centralized Critic.

View zoom recording without distraction

It’s simple: install cmake and zlib first, and then install gym and atari environment on macOS. brew install cmake zlib pip install 'gym[atari]' The quotation marks are for people who are using zsh.