chaincrafter

Seamless integration and composability for large language model apps.

Features

• Composable prompts and chains
  • Use multiple models to run one chain and then use that as input for a different chain and model
• Customizable prompt and response formatting
  • Add modifiers to prompts to change the style, length, and format of the response
  • Extract data from the response to use in the next prompt
  • Add custom functions to process the response
  • Add custom functions to process the input variables
• Integration with OpenAI API (llama.cpp in progress)
• Async calls to models
• Load Prompts and Chains from YAML using Catalogs
  • Makes it easier to share prompts and chains between projects
  • Build up a prompts library

Installation

pip install chaincrafter

Usage

1. Define your prompts and the variables that they expect

• The input variables can be of any type, and can be processed by a function
• The prompt message is treated as an f-string

2. Define your chain of prompts

• The chain is a list of tuples, where each tuple contains a prompt and the output key to store the response in
• The output key is used to access the response in the next prompt

3. Set up the models that you want to use
4. Run the chain using the models

from chaincrafter import Chain, Prompt
from chaincrafter.models import OpenAiChat

chat_model = OpenAiChat(temperature=0.65, model_name="gpt-3.5-turbo")
system_prompt = Prompt("You are a helpful assistant who responds to questions about the world")
hello_prompt = Prompt("Hello, what is the capital of France? Answer only with the city name.")
followup_prompt = Prompt("{city} sounds like a nice place to visit. What is the population of {city}?")
chain = Chain(
    system_prompt,
    (hello_prompt, "city"),
    (followup_prompt, "followup_response"),
)
messages = chain.run(chat_model)
for message in messages:
    print(f"{message['role']}: {message['content']}")

Running the examples

source venv/bin/activate
export OPENAI_API_KEY="..."
python -m examples.interesting_facts
python -m examples.interesting_facts_catalog