Quick Start - Unify Documentation

import unify
client = unify.Unify("gpt-4o@openai", api_key="UNIFY_KEY")
client.generate("hello world!")

We recommend using python-dotenv to add UNIFY_KEY="My API Key" to your .env file, avoiding the need to use the api_key argument as above.

You can list all available LLM endpoints, models and providers like so:

unify.list_models()
unify.list_providers()
unify.list_endpoints()

Now you can run this toy evaluation ⬇️, check out the logs in your dashboard, and iterate 🔁 on your parameters to quickly get your application flying! 🪁

import unify
from random import randint, choice

# agent
client = unify.Unify("gpt-4o@openai")
client.set_system_message("You are a helpful maths assistant, tasked with adding and subtracting integers.")

# test cases
qs = [f"{randint(0, 100)} {choice(['+', '-'])} {randint(0, 100)}" for i in range(10)]

# evaluator
def evaluate_response(question: str, response: str) -> float:
    correct_answer = eval(question)
    try:
        response_int = int(
            "".join([c for c in response.split(" ")[-1] if c.isdigit()]),
        )
        return float(correct_answer == response_int)
    except ValueError:
        return 0.

# evaluation
def evaluate(q: str):
    response = client.generate(q)
    score = evaluate_response(q, response)
    unify.log(
        question=q,
        response=response,
        score=score
    )

# execute + log evaluation
with unify.Project("Maths Assistant"):
    with unify.Params(system_message=client.system_message):
        unify.map(evaluate, qs)

A complete example of this Maths Assistant problem can be found here.