Escribe aquí la idea principal o el concepto clave en 1-2 frases
🎯 Detalles / Contenido
ReAct (Reason + Act)
LLM first thinks what to do and them decides an action to take
The action is execute in a environment and an observation is returned
Create a class Agent to build the agent
Creating the agent
class Agent: def __init__ (self, system=""): # The agent is parameterize by a system message self.system = system # atribute message self.messages = [] if self.system: self.messagens.append({}"role":"system", "content": system}) def __call_(self, message): # The message is take by the user and than give it back from the assistant self.messages.append({"role", "user", "content": message}) result = self.execute() self.messages.append({"role", "assistant", "content": result}) return result
Graphs
Nodes: Agents or functions
Edges: Connect nodes
Conditional Edges: Decisions
🔑 Keywords / Conceptos clave
keyword1, keyword2, keyword3
Para RAG
Estos keywords ayudarán a encontrar esta nota después