parent
74715f191b
commit
c1191ec455
@ -0,0 +1,32 @@
|
|||||||
|
from swarms import Agent
|
||||||
|
from swarms.structs import MultiAgentCollaboration
|
||||||
|
|
||||||
|
# Agents will directly initialize their language models
|
||||||
|
|
||||||
|
# Define collaborating agents
|
||||||
|
planner = Agent(
|
||||||
|
agent_name="Planner",
|
||||||
|
system_prompt="Break the objective into clear steps.",
|
||||||
|
model_name="gpt-4o-mini",
|
||||||
|
max_loops=1,
|
||||||
|
)
|
||||||
|
|
||||||
|
writer = Agent(
|
||||||
|
agent_name="Writer",
|
||||||
|
system_prompt="Use the plan to craft the final answer.",
|
||||||
|
model_name="gpt-4o-mini",
|
||||||
|
max_loops=1,
|
||||||
|
)
|
||||||
|
|
||||||
|
# Initialize the collaborative workflow
|
||||||
|
swarm = MultiAgentCollaboration(
|
||||||
|
agents=[planner, writer],
|
||||||
|
max_loops=4,
|
||||||
|
)
|
||||||
|
|
||||||
|
# Kick off the collaboration
|
||||||
|
swarm.inject("Manager", "Produce a short overview of reinforcement learning.")
|
||||||
|
|
||||||
|
result = swarm.run("Begin")
|
||||||
|
print(result)
|
||||||
|
|
Loading…
Reference in new issue