llambda 0.1.0a1

No project description provided

LLambda

LLambda is a library that makes it easy to integrate Large Language Models (LLM) and Python. It enables calling multiple pre-defined functions using natural language and executing them with appropriate arguments.

Installation

You can easily install LLambda using pip:

pip install llambda

Usage

Here is a basic usage example. The following code defines two functions and calls them using natural language.

Registering Functions with Decorators

from llambda import register

@register
def make_coffee(n_cups: int) -> str:
    """
    Make coffee of the amount you want.
    Can't make tea.
    """
    return "☕" * n_cups


@register
def send_email(recipient_name: str, subject: str, content: str, your_name: str) -> str:
    """
    Send an email to someone.
    """
    return f"Sent email to {recipient_name} with subject '{subject}'\nand content '{content}' from {your_name}"

Now we have two functions. Let's call them using natural language:

from llambda import create_llambda, set_openai_api_key

set_openai_api_key("Your OpenAI API Key")

llambda = create_llambda()

print(llambda("I want to drink 3 cups of coffee"))
# ☕☕☕
print(llambda("Write an email in which GPT shares the method for training neural networks with Adam."))
# Sent email to Adam with subject 'Training Neural Networks'
# and content'Hey there! I wanted to share my method for training neural networks. Let's discuss soon!' from GPT

LLambda automatically recognizes the functions and their arguments, and calls the appropriate function with the arguments inferred from the instruction.

print(llambda("I want to drink 3 cups of tea"))
# NotImplementedError: Can't make tea.

LLambda raises NotImplementedError when it cannot find a function that matches the instruction.

Directly Passing Functions

In addition to using the @register decorator, you can also directly pass the functions to create_llambda.

llambda = create_llambda([make_coffee, send_email])

print(llambda("I want to drink 3 cups of coffee"))
# ☕☕☕

Custom Context

Using ContextVars Class

Functions

can access context variables by defining a subclass of ContextVars. The class variables of this subclass will be automatically recognized as context variables:

class Context(ContextVars):
    your_name: str = "GPT"
    your_friends: list[str] = ["Adam", "Bert"]

llambda = create_llambda(context=Context)

print(llambda("Share with your friend about your method for training neural networks"))
# Sent email to Adam with subject 'Training Neural Networks'
# and content'Hey there! I wanted to share my method for training neural networks. Let's discuss soon!' from GPT

Directly Passing ContextVar Instances

You can also pass the context directly as a list of ContextVar instances:

context_vars = [
    ContextVar("your_name", "str", "GPT"),
    ContextVar("your_friends", "list[str]", ["Adam", "Bert"])
]

llambda = create_llambda(context=context_vars)