agentsfoundationsllm
Multi-agent systems
Umbrella for systems of interacting autonomous agents — LLM-based agents, MARL, swarms, societies of mind.
Уровни глубины
L0Intro~3ч
Names several agents working together; reads a ReAct trace.
L1Basics~12ч
Builds two-agent debate / reviewer-writer loop with LangGraph or CrewAI.
L2Working~20ч
Designs role specialisation, orchestrator patterns, tool-calling protocols; debugs failures.
L3Advanced~35ч
Hierarchical planners, market-based coordination; applies MARL algorithms (QMIX, MAPPO).
L4Research~80ч
Contributes to agent benchmarks (AgentBench, GAIA) or emergent-behaviour research.