toucans 0.0.14

No project description provided

Toucans is a streamlined library designed for agile and flexible prompt engineering, leveraging jinja2 for dynamic prompt templating and litellm to seamlessly connect with a wide range of language model providers.

Getting Started

Installation

pip install toucans

Initialize Prompt Function

To initialize a PromptFunction:

from toucans import PromptFunction

qa_func = PromptFunction(
    model="gpt-4",
    temperature=0.7,
    messages=[
        {"role": "system", "content": "You are a helpful {{ role }}."},
        {"role": "user", "content": "Answer the following question: {{ question }}"},
    ],
)

Generate Completion

Generate a completion by calling the PromptFunction with the desired role and question:

completion = qa_func(
    role="Software Developer", 
    question="What is clean code?"
)

Save Prompt Function

To save the initialized PromptFunction to a directory or the toucans hub:

# Push to dir (not implemented yet)
qa_func.push_to_dir("./qa/")

# Push to toucans hub
qa_func.push_to_hub("juunge/qa")

Load Saved Prompt Function

To load a saved PromptFunction from a directory or the toucans hub:

# Load from dir (not implemented yet)
qa_func = PromptFunction.from_dir("./qa/")

# Load from toucans hub
qa_func = PromptFunction.from_hub("juunge/qa")

Loading from Toucans Hub requires that you run a hub and connect to it by setting the HUB_API_URL environment variable.