AutoChain – lightweight and extensible framework designed to simplify the process of building generative agents using Large Language Models (LLMs)

AutoChain is a lightweight and extensible framework designed to simplify the process of building generative agents using Large Language Models (LLMs). Unlike existing frameworks, AutoChain allows developers to easily customize and evaluate generative agents for specific use cases. It draws inspiration from LangChain and AutoGPT but offers simpler concepts. The framework supports OpenAI function calling and provides automated multi-turn conversation evaluation through simulated conversations. AutoChain’s goal is to enable rapid iteration on generative agents by streamlining agent customization and evaluation. The evaluation process involves running conversations between the generative agent and LLM-simulated test users, facilitating the addition of new test cases and fast evaluation.

Introduce AutoChain

AutoChain is an alternative to LangChain, and the concept is basically universal. If you already know LangChain, you can basically get started with AutoChain directly. However, AutoChain simplifies many unnecessary levels of abstraction, and it is very convenient for automatic evaluation, which will be much more convenient for developing LLM applications. Its philosophy is: Build lightweight, extensible, and testable LLM Agents

Large language models (LLMs) have achieved great success in different text generation tasks and enable developers to build generative agents based on goals expressed in natural language.

However, most build agents require extensive customization for specific purposes, and supporting different use cases with existing tools and frameworks can sometimes be overwhelming. Therefore, building custom generation agents is still very challenging.

Furthermore, evaluating such generative agents is usually done by manually trying different scenarios, a very manual, repetitive and expensive task.

Drawing inspiration from LangChain and AutoGPT, AutoChain aims to solve these two problems by providing a lightweight and extensible framework for developers to build their own agents using LLM and custom tools, and simulate dialogue Automatically evaluate different user scenarios. Experienced users of LangChain will find AutoChain easy to navigate as they share similar but simpler concepts.

The goal is to enable rapid iteration of generated agents by simplifying agent customization and evaluation.

Feature

🚀 Lightweight and extensible build agent pipeline.
🔗 Proxy, can use different custom tools and support OpenAI function calls
💾 Simple memory tracking of dialog history and tool output
🤖 Automated agent multi-round dialogue evaluation through simulated dialogue

How does AutoChain simplify building agents?

AutoChain aims to provide a lightweight framework and simplifies the agent building process in a few ways, as compared to existing frameworks

1. Easy prompt update
   Engineering and iterating over prompts is a crucial part of building generative agent. AutoChain makes it very easy to update prompts and visualize prompt outputs. Run with -v flag to output verbose prompt and outputs in console.
2. Up to 2 layers of abstraction
   As part of enabling rapid iteration, AutoChain chooses to remove most of the abstraction layers from alternative frameworks
3. Automated multi-turn evaluation
   Evaluation is the most painful and undefined part of building generative agents. Updating the agent to better perform in one scenario often causes regression in other use cases. AutoChain provides a testing framework to automatically evaluate agent’s ability under different user scenarios.