lobsang 0.1.0

A simple framework to interact with conversational LLMs

Lobsang

Welcome to Lobsang 🧘‍♂️

Lobsang is a framework to interact with conversational LLMs in the simplest way possible.
It's easy to get started, yet designed to scale up to complex use cases. Enjoy the ride! 🚀

Note: This project is still in early development. Expect breaking changes.

Installation

Requires Python 3.11 or higher.

pip install lobsang

Core Concepts

Lobsang is built around a few core concepts:

• Chat is the main class of Lobsang. It represents a conversation between a user and a LLM. It is the main entry point to the framework and stores the chat history for a single conversation. After creating a Chat instance, you can interact with it by calling itself with one or more messages:

  from lobsang import Chat, UserMessage
  from lobsang.answers import TextAnswer
  from lobsang.llms import FakeLLM
  chat = Chat(llm=FakeLLM()) 
  #                 👆 FakeLLM returns dummy responses
  # Call the chat with one message and a corresponding answer (placeholder for the LLM's response)
  chat([UserMessage("Hello"), TextAnswer()])
  print("Chat history:")  
  print(chat)
  
  # Call the chat with multiple messages
  res = chat([
      UserMessage("What is 1+1?"),
      TextAnswer(),
      UserMessage("What is 2+2?"),
      TextAnswer()
  ])
  
  print("Conversation:")
  print(res)
  

  • Answers are used to guide the LLM to generate a specific response by modifying user messages. For example, you can use the JSONAnswerto instruct the LLM to generate a JSON response (see examples/2_directives.ipynb for more details).

Examples

We provide a few examples to get you started. You can find them in the examples folder. The examples use the openai package, make sure to install it before running the examples (pip install openai). You will also need an OpenAI API key, which you can get here: https://platform.openai.com/account/api-keys.