ollama-python/examples/async-structured-outputs.py
Aarni Koskela 89e719ab92
chore: bump ruff and ensure imports are sorted (#385)
* chore: upgrade ruff & sort imports
2025-01-14 16:34:16 -08:00

35 lines
929 B
Python

import asyncio
from pydantic import BaseModel
from ollama import AsyncClient
# Define the schema for the response
class FriendInfo(BaseModel):
name: str
age: int
is_available: bool
class FriendList(BaseModel):
friends: list[FriendInfo]
async def main():
client = AsyncClient()
response = await client.chat(
model='llama3.1:8b',
messages=[{'role': 'user', 'content': 'I have two friends. The first is Ollama 22 years old busy saving the world, and the second is Alonso 23 years old and wants to hang out. Return a list of friends in JSON format'}],
format=FriendList.model_json_schema(), # Use Pydantic to generate the schema
options={'temperature': 0}, # Make responses more deterministic
)
# Use Pydantic to validate the response
friends_response = FriendList.model_validate_json(response.message.content)
print(friends_response)
if __name__ == '__main__':
asyncio.run(main())