ASI-GO-3

Artificial SuperIntelligence GO type

ASI-GO-3: Autonomous System Intelligence - General Optimizer

A simplified implementation of the ASI-ARCH architecture for general problem-solving tasks. This is the upgraded version according to chapters 3-7 in the book.

Overview

ASI-GO-3 adapts the four-component architecture from ASI-ARCH:

• Cognition Base: Stores problem-solving strategies and patterns
• Researcher: Proposes solutions based on goals
• Engineer: Implements and tests solutions
• Analyst: Analyzes results and extracts insights

Setup

1. Install Dependencies:

   pip install -r requirements.txt
   

Configure API Keys:

Copy .env.example to .env Add your API key for your chosen provider:

OpenAI (ChatGPT) Google (Gemini) Anthropic (Claude)

Select LLM Provider:

Edit .env and set LLM_PROVIDER to one of:

openai (default) google anthropic

Usage

Interactive Mode bashpython main.py Command Line Mode python main.py "Find the first 100 prime numbers" Features

Multi-LLM Support: Works with ChatGPT, Gemini, or Claude

Iterative Refinement: Automatically improves solutions based on feedback Knowledge Persistence: Learns from each problem-solving session

To use:

1. Install dependencies: pip install -r requirements.txt
2. Set up your .env file with appropriate API keys
3. Run: python main.py

The system will guide you through solving any computational problem you provide!

Checkpoint System: Saves progress at each step

Comprehensive Logging: Detailed logs for debugging Windows Optimized: Colored output and Windows-friendly paths

Example Goals

"Find the first 40 prime numbers" "Generate the Fibonacci sequence up to 1000" "Create a function to check if a string is a palindrome" "Solve the Tower of Hanoi problem with 4 disks" "Find all perfect numbers less than 10000"

Testing ASI-GO-3's Learning

After running several problems, try these to test if the system has learned:

"Create an efficient function to check if a number is prime"

After solving prime-related problems, this tests if ASI-GO-3 has learned optimization patterns.

"Implement the most efficient sorting algorithm you know"

Tests if the system has accumulated knowledge about sorting algorithms.

"Solve a string manipulation problem using the best approach you've learned"

Tests transfer learning and pattern application.

Architecture

The system follows a cyclic process:

Researcher proposes a solution Engineer tests the implementation Analyst evaluates results Cognition Base stores insights Loop continues until success or max iterations

Configuration

Edit .env to customize:

MAX_ITERATIONS: Maximum solving attempts (default: 5) TEMPERATURE: LLM creativity level (default: 0.7) LOG_LEVEL: Logging detail (INFO/DEBUG)

Troubleshooting

API Key Errors: Ensure your API key is correctly set in .env Import Errors: Run pip install -r requirements.txt Execution Timeouts: Complex problems may need increased timeout in engineer.py

LLM Library Compatibility

OpenAI Library Version Conflicts

When installing ASI-GO, you may encounter a TypeError: Client.__init__() got an unexpected keyword argument 'proxies' error. This occurs due to version conflicts between the OpenAI, Anthropic, and httpx libraries. The issue arises because different LLM providers require different versions of the shared httpx dependency. To resolve this, uninstall all LLM-related packages first (pip uninstall openai anthropic google-generativeai httpx httpcore anyio -y), then install httpx version 0.24.1 specifically (pip install httpx==0.24.1), followed by the OpenAI library (pip install openai==1.12.0). If you need multiple LLM providers, install them in this specific order to maintain compatibility. Alternatively, consider using separate conda environments for different LLM providers to avoid conflicts entirely.

License

MIT License - Feel free to modify and extend!

This implementation provides:

1. Modular Architecture: Each component (Cognition Base, Researcher, Engineer, Analyst) is separate and focused
2. Multi-LLM Support: Works with ChatGPT, Gemini, or Claude (Grok uses OpenAI's API)
3. Robust Error Handling: Comprehensive try-catch blocks and logging
4. Checkpoint System: Saves progress at each step
5. Interactive Interface: Easy-to-use command line interface with colored output
6. Knowledge Persistence: Learns from each session
7. Windows Optimization: Uses appropriate paths and libraries for Windows