WLTech.Ai ARC Challenge Solver

This repository contains a Python implementation designed to solve the ARC Challenge (2024). The script integrates with APIs and libraries to process and analyze problem sets. The original notebook can be viewed and executed directly on Google Colab here.

Solution Workflow

Below is a diagram illustrating how the solution process works:

What This Code Does

This script is designed to solve ARC challenges by leveraging AI capabilities through the following steps:

1. Task Loading: Loads challenge tasks and corresponding solutions from JSON files.
2. AI Integration: Uses Anthropic's Claude and OpenAI's GPT models for reasoning and solution generation.
3. Result Processing: Generates solutions to the given challenges, which can then be validated or further analyzed.

Features

• Automated solving of ARC tasks using advanced AI models.
• Flexible task loading from user-provided JSON files.
• Seamless integration with Anthropic and OpenAI APIs.

Requirements

To run this script, ensure the following dependencies are installed:

• Python 3.8+
• Libraries:
  • anthropic
  • pydantic
  • openai

Install them using:

pip install -U anthropic pydantic openai

Usage

1. Clone the repository:

   git clone https://github.com/your-repo/WLTech.Ai-ARC-Challenge.git
   cd WLTech.Ai-ARC-Challenge

2. Prepare your API keys:

   • Store ANTHROPIC_API_KEY and OPENAI_API_KEY in a secure location. You can use the google.colab.userdata method or set them as environment variables.

3. Run the script:

   python WLTech.Ai-ARC-Challenge.py

4. Input your challenge and solution files as JSON:

   load_tasks('path/to/challenges.json', 'path/to/solutions.json')

Project Structure

• WLTech.Ai-ARC-Challenge.py: Main script to solve ARC challenges.
• Dependencies: All required libraries are listed in the script.

Contributions

Contributions are welcome! Feel free to submit issues or pull requests for improvements.

License

This project is licensed under the MIT License. See LICENSE for more details.