API may change slightly, still work in progress.
from tackleberry import TB
openai_chat = TB.chat('gpt-4o-mini')
openai_reply = openai_chat.query("Say test")
claude_chat = TB.chat('claude-3-5-sonnet-20241022')
claude_reply = claude_chat.query("Say test")
groq_chat = TB.chat('gemma2-9b-it')
groq_reply = groq_chat.query("Say test")
# OLLAMA_PROXY_URL set for URL, can handle Basic Auth in URL
ollama_chat = TB.chat('ollama/gemma2:2b')
ollama_reply = ollama_chat.query("Say test")
from pydantic import BaseModel
class UserInfo(BaseModel):
name: str
age: int
# Using Structured Output Feature of Ollama - no instructor
ollama_user_info = ollama_chat.query("Extract the name and the age: 'John is 20 years old'", UserInfo)
# Using instructor[anthropic]
claude_user_info = claude_chat.query("Extract the name and the age: 'John is 20 years old'", UserInfo)
# Using instructor[groq]
groq_user_info = groq_chat.query("Extract the name and the age: 'John is 20 years old'", UserInfo)
Install from PyPi
❯ pip install --upgrade tackleberry