Model-Agent 🤖

A sophisticated mathematical modeling assistant that leverages advanced language models to iteratively solve, evaluate, and improve mathematical modeling problems.

🌟 Features

• Iterative Problem Solving: Multi-round approach combining solution generation, evaluation, and improvement
• Rigorous Mathematical Validation: Strict verification of mathematical correctness and completeness
• LaTeX Support: Full support for mathematical expressions using TeX notation
• Structured Output: Well-organized solution format with summaries and detailed derivations
• Continuous Improvement: Progressive refinement through automated review and enhancement cycles

🏗️ Project Structure

Model-Agent/
├── api_client.py      # API client for model interactions
├── prompt.py         # Core system prompts and instructions
├── reasoning.py      # Main iterative solving logic
├── api.ENV          # API configuration (create from template)
├── problem.txt      # Input problem statement
└── extra.txt        # Additional requirements/constraints

🚀 Getting Started

Prerequisites

• Python 3.9+
• Access to a compatible language model API

Installation

1. Clone the repository:

git clone https://github.com/pancacake/Model-Agent.git
cd Model-Agent

2. Create and configure your API environment file:

cp api.ENV.template api.ENV
# Edit api.ENV with your API credentials

Usage

1. Place your mathematical modeling problem in problem.txt
2. (Optional) Add additional requirements in extra.txt
3. Run the solver:

python reasoning.py --rounds 3 --max_tokens 15000

Command Line Arguments

• --problem_txt: Problem statement file (default: problem.txt)
• --extra_txt: Additional requirements file (default: extra.txt)
• --env: API configuration file (default: api.ENV)
• --rounds: Number of improvement iterations (default: 3)
• --outdir: Output directory (default: outputs)
• --run_name: Name for this run (default: run)
• --max_tokens: Maximum tokens per response (default: 15000)
• --temperature: Response randomness (default: 0.2)

📤 Outputs

The system generates three main outputs in the specified output directory:

1. trace.md: Detailed record of each iteration, including:

   • SOLVE: Initial solution attempt
   • JUDGE: Validation and error identification
   • IMPROVE: Enhanced solution based on feedback

2. final.md: The final, polished solution

3. memory.json: Complete execution record including:

   • Metadata (rounds, parameters)
   • Context (problem, requirements)
   • Full iteration history

💡 Solution Process

1. SOLVE Phase

   • Constructs rigorous mathematical model
   • Provides structured solution with clear assumptions
   • Ensures LaTeX formatting for mathematical content

2. JUDGE Phase

   • Validates mathematical correctness
   • Identifies critical errors and gaps
   • Provides detailed verification log

3. IMPROVE Phase

   • Addresses identified issues
   • Refines mathematical arguments
   • Maintains solution coherence

🔍 Quality Assurance

The system enforces:

• Rigorous mathematical reasoning
• Complete step-by-step derivations
• Clear documentation of assumptions
• Proper LaTeX formatting
• Systematic error detection
• Progressive solution improvement

📖 Examples

Example run with a basic optimization problem:

python reasoning.py --problem_txt examples/optimization.txt --rounds 2

The solution will be generated in outputs/run-TIMESTAMP/ with full iteration history and final solution.

📝 License

This project is licensed under the MIT License - see the LICENSE file for details.

✨ Contributing

Contributions are welcome! Please feel free to submit a Pull Request. For major changes, please open an issue first to discuss what you would like to change.

1. Fork the Project
2. Create your Feature Branch (git checkout -b feature/AmazingFeature)
3. Commit your Changes (git commit -m 'Add some AmazingFeature')
4. Push to the Branch (git push origin feature/AmazingFeature)
5. Open a Pull Request

🤝 Support

If you encounter any problems or have suggestions, please open an issue on GitHub.