---

SmartGPT is an experimental program designed to empower LLMs (particularly GPT-3.5 and GPT-4) to tackle complex tasks independently. It does this by breaking down these tasks into smaller, manageable problems and efficiently gathering information from the internet and other external sources.

Why SmartGPT?

In the rapidly evolving world of LLMs, several solutions exist for enhancing their capabilities. Among these are Auto-GPT and BabyAGI. However, SmartGPT stands out for several reasons:

• Modularity: SmartGPT allows for first-class plugin support, enabling the composition of Autos tailored to your specific project needs.
• Flexibility: One config.yml file, generated automatically, allows for comprehensive configuration options.
• Consistency: A sophisticated system of dynamic action execution and static tool-chaining ensures highly reliable outputs.

However, it’s essential to acknowledge its limitations:

• Ecosystem: Popularity has granted AutoGPT a more polished, refined toolset, alongside extensive tools and memory integration.
• Memory Management: As a nascent project, SmartGPT currently has a basic and limited memory system, which is expected to evolve over time.

Supporting Development

Currently, SmartGPT primarily tests with GPT-3.5, with occasional forays into GPT-4, due to costs involved. The vision for the project is to experiment with multiple agents concurrently and fully exploit GPT-4’s capabilities, which requires financial backing. If you’re interested in contributing to the development, consider joining our Patreon.

Quickstart Guide

Getting started with SmartGPT involves a few simple steps:

1. Install cargo, preferably the latest stable version.
2. Clone the repository: git clone https://github.com/Cormanz/smartgpt.git and navigate into the directory with cd smartgpt.
3. Run the application in release mode with: cargo run --release. This command generates a config.yml file for you.
4. Customize the configuration to suit your needs and execute the application again.

If you want more information or would like to use SmartGPT as a crate in your own projects, read the documentation.

Understanding SmartGPT through an Analogy

Imagine SmartGPT as a well-equipped workshop where complex projects are tackled. The Autos are your specialized tools:

• Runner: This tool focuses on completing a specific task, like a hammer driving a nail into wood.
• Assistants: Like having a colleague who can brainstorm or provide advice during the project; however, they are still experimental and less reliable than the Runner.

In the workshop, the Dynamic Agent acts like a project manager, deciding the next steps and coordinating the workflow. Think of it as someone who thinks creatively, delegates tasks, and ensures everything runs smoothly.

On the other hand, the Static Agent is the worker who meticulously follows the plan laid out by the manager, completing each task like assembling various pieces of furniture. It also remembers what tools were used before—much like keeping the workshop organized so you can find things later.

Troubleshooting

Should you face challenges while using SmartGPT, consider the following troubleshooting steps:

• Ensure that Cargo is properly installed and updated to the latest version.
• Revisit your config.yml file to check for any misconfigurations.
• If the program crashes unexpectedly, check for the latest updates or known issues on the GitHub repository.
• For additional insights, updates, or to collaborate on AI development projects, stay connected with fxis.ai.

Disclaimer

SmartGPT is an incredibly experimental application wherein the goal is to unlock the maximum potential of LLMs, often at the cost of stability. While backwards compatibility may seem like a daunting task, advancements are continually being made—some ideas are groundbreaking, yielding promising results.

At fxis.ai, we believe that such advancements are crucial for the future of AI, as they enable more comprehensive and effective solutions. Our team is continually exploring new methodologies to push the envelope in artificial intelligence, ensuring that our clients benefit from the latest technological innovations.