A lightweight framework for building LLM applications
git clone https://github.com/acesanderson/Chain
cd Chain
pip install -r requirements.txt
pip install .In a file titled .env in the root directory of your project, add the following environment variables for the LLM providers you wish to use:
OPENAI_API_KEY=your_openai_api_key
ANTHROPIC_API_KEY=your_anthropic_api_key
GOOGLE_API_KEY=your_google_api_key
GROQ_API_KEY=your_groq_api_key
DEEPSEEK_API_KEY=your_deepseek_api_key
If you wish to use local models, you will need to install Ollama. Recommended models (from personal experience): llama3.1:latest, qwen.
from Chain import Prompt, Model, Chain, Parser, Message, MessageStore
# List available models
print(Model.models)
# {'ollama': ['phi3:latest', 'phi:latest', 'llama3.2:latest', 'llama3.1:latest', 'llama3.1:70b-instruct-q2_K', 'dolphin-mixtral:latest', 'dolphin-mixtral:8x7b', 'mistral:latest'], 'openai': ['gpt-4o', 'gpt-4-turbo', 'gpt-3.5-turbo-0125', 'gpt-4o-mini', 'o1-preview', 'o1-mini'], 'anthropic': ['claude-3-opus-20240229', 'claude-3-sonnet-20240229', 'claude-3-haiku-20240307', 'claude-3-5-sonnet-20240620'], 'google': ['gemini-1.5-pro-latest', 'gemini-1.5-flash-latest', 'gemini-1.0-pro-latest'], 'groq': ['llama3-8b-8192', 'llama3-70b-8192', 'mixtral-8x7b-32768', 'gemma-7b-it']}
# Simple GPT chain
prompt = Prompt("What is the capital of France?")
model = Model("gpt")
chain = Chain(prompt, model)
response = chain.run()
print(response)
# Response(content='The capital of France is Paris.', status='success', prompt='What is the capitol of France?', model='gpt-4o', duration=0.9836819171905518, messages=[], variables=<built-in function input>)
print(response.content)
# "The capital of France is Paris."Chain uses Jinja templating to allow for dynamic prompts. You can pass variables to the prompt and model using the input_variables parameter in the Chain and Prompt classes.
prompt_string = "Name ten {{things}} that are red."
prompt = Prompt(prompt_string)
model = Model("claude-3-5-sonnet-20240620")
chain = Chain(prompt, model)
response = chain.run(input_variables={"things": "frogs"})
'Here are ten different types of frogs:\n\n1. Red-Eyed Tree Frog\n2. Poison Dart Frog\n3. American Bullfrog\n4. African Clawed Frog\n5. Green Tree Frog\n6. Glass Frog\n7. Tomato Frog\n8. Pacman Frog (Horned Frog)\n9. Northern Leopard Frog\n10. Goliath Frog\n\nThese frogs represent a diverse range of species from various habitats around the world, each with unique characteristics and adaptations.'Chain uses the excellent Instructor library to get pydantic objects from LLM calls. You can use the Parser class to parse the output of a chain into a pydantic object.
from pydantic import BaseModel
from Chain import Parser
class Frog(BaseModel):
name: str
color: str
habitat: str
prompt = Prompt("Come up with a frog that is red and lives in the rainforest")
model = Model("claude-3-5-sonnet-20240620")
parser = Parser(Frog)
chain = Chain(prompt, model, parser) # Note that the parser is passed to the Chain class in initialization
response = chain.run()
print(response.content)
# Frog(name='Fred', color='red', habitat='rainforest')from Chain import AsyncChain, ModelAsync
prompt_strings = ["name 10 mammals", "name ten birds"]
model = ModelAsync("llama3.1:latest")
chain = AsyncChain(model=model)
results = chain.run(prompt_strings=prompt_strings)- You can create a messagestore to either: save response objects in a Message format throughout a script; save a persistent chat conversation between sessions; log prompt flows.
- You can set system prompts with the create_messages method in the MessageStore class.
- Every client has an AsyncClient implementation for async calls.
- Chatbot class
- HuggingFace/unsloth implementation for working with finetuned models
- Agentic support for tools, resources, using MCP protocol