🧪 SIMLAB_GENERATOR

📌 Overview

SIMLAB_GENERATOR is a Python-based application that provides an innovative graphical user interface for simulation and laboratory tasks. It leverages the power of multiple AI models to generate and refine scientific simulations through an interactive process.

🔄 Process Flow

1. Main Process

flowchart TB
    Start([Launch Streamlit App]) --> ApiKeys[Enter API Keys]
    ApiKeys --> Prompt[Enter Simulation Prompt]
    Prompt --> Research[Initial Research Phase]
    Research --> Code[Code Generation Phase]
    Code --> Refine[Code Refinement Phase]
    Refine --> Review[Final Review Phase]
    Review --> Save[Save Results]
    Save --> End([End])

    style Start fill:#f9f9f9,stroke:#333
    style End fill:#f9f9f9,stroke:#333
    style Research fill:#FFE6E6,stroke:#FF9999
    style Code fill:#E6F3FF,stroke:#99CCFF
    style Refine fill:#FFE6E6,stroke:#FF9999
    style Review fill:#E6F3FF,stroke:#99CCFF

Loading

2. Research Phase

flowchart TB
    subgraph ResearchPhase [Research Phase - Qwen/Qwen2.5-72B-Instruct]
        direction TB
        Start([Start Research]) --> Analysis[Analyze Requirements]
        Analysis --> Libraries[Identify Required Libraries]
        Libraries --> Methods[Define Implementation Methods]
        Methods --> Tech[Select Technologies]
        Tech --> Plan[Create Implementation Plan]
        Plan --> End([Complete Research])
    end

    style ResearchPhase fill:#FFE6E6,stroke:#FF9999
    style Start fill:#f9f9f9,stroke:#333
    style End fill:#f9f9f9,stroke:#333

Loading

3. Code Generation and Refinement

flowchart TB
    subgraph InitialCode [Initial Code - Claude-3.5-sonnet]
        direction TB
        Start([Start Generation]) --> Base[Generate Base Code]
        Base --> Doc[Create Documentation]
        Doc --> Suggest[Suggest Improvements]
    end

    subgraph Refinement [Code Refinement - Qwen/Qwen2.5-Coder-32B-Instruct]
        direction TB
        Iter1[1st Iteration<br>Basic Functionality] --> Iter2[2nd Iteration<br>Performance]
        Iter2 --> Iter3[3rd Iteration<br>UX Enhancement]
    end

    InitialCode --> Refinement

    style InitialCode fill:#E6F3FF,stroke:#99CCFF
    style Refinement fill:#FFE6E6,stroke:#FF9999
    style Start fill:#f9f9f9,stroke:#333

Loading

4. Final Review Phase

flowchart TB
    subgraph FinalReview [Final Review - Claude-3.5-sonnet]
        Start([Start Review]) --> CodeReview[Review Code]
        CodeReview --> Check{Error Check}
        Check -->|Errors Found| Fix[Fix Errors]
        Fix --> Check
        Check -->|No Errors| Validate[Validate Code]
        Validate --> Final[Generate Final Version]
        Final --> Save[Save Results]
        Save --> End([Complete Review])
    end

    style FinalReview fill:#E6F3FF,stroke:#99CCFF
    style Start fill:#f9f9f9,stroke:#333
    style End fill:#f9f9f9,stroke:#333
    style Check fill:#f0f0f0,stroke:#333

Loading

🚀 Quick Start

Prerequisites

• Python 3.x
• pip (Python package installer)

Installation

1️⃣ Clone the repository

git clone https://github.com/[your-username]/SIMLAB_GENERATOR.git
cd SIMLAB_GENERATOR

2️⃣ Install dependencies

You have several options for installing the required dependencies:

Option 1: Using requirements.txt (Recommended)

pip install -r requirements.txt

Option 2: Manual Installation If you encounter any issues with requirements.txt, you can install packages individually:

# Install Streamlit
pip install streamlit
# Install OpenAI and Anthropic packages
pip install openai anthropic

Note: If you encounter any permission errors during installation, try:

pip install --user -r requirements.txt
# or
pip install --user streamlit openai anthropic

3️⃣ Run the application

streamlit run main/innovate_gui.py

After running the application, you can enter your OpenAI and Anthropic API keys directly in the Streamlit user interface.

📁 Project Structure

SIMLAB_GENERATOR/
├── main/
│   ├── api_calls.py          # API integration
│   ├── innovate_gui.py       # Main GUI interface
│   └── time_to_innovate.py   # Core functionality
├── .gitattributes
├── .gitignore
└── requirements.txt          # Project dependencies

🔧 Configuration

Before running the application, make sure you have all the required dependencies installed using the requirements.txt file. If you face any dependency conflicts, try creating a virtual environment:

# Create a virtual environment
python -m venv venv

# Activate the virtual environment
# On Windows:
venv\Scripts\activate
# On macOS/Linux:
source venv/bin/activate

# Install dependencies
pip install -r requirements.txt

📝 Example Prompt Here's an educational simulation designed to visually demonstrate the learning process of the self-attention algorithm:

1. Step-by-Step Visualization Step 1: Input Sentence Tokenization and Embedding

Display sentence separation into tokens Animate the process of converting each token into embedding vectors Visualize embedding vector dimensions using colorful bar graphs

Step 2: Q, K, V Vector Generation Process

Matrix multiplication animation showing embeddings being multiplied with WQ, WK, WV matrices Display generated Q, K, V vectors in distinct colors Popup shows actual vector values on mouse hover

Step 3: Attention Score Calculation

Visual matrix representation of dot product calculations between Query and Key vectors Darker colors indicate higher scores Line thickness between token pairs represents relationship strength

Step 4: Scaling and Softmax

Animate value adjustment during scaling process Visualize probability conversion after Softmax using pie charts or bar graphs Display attention weights between tokens using heatmap

Step 5: Final Output Generation

Visualize the summation process of weighted Value vectors Display final output vectors alongside original tokens

2. Interactive Elements Token Selection Features

Click tokens to view detailed attention relationships from that token's perspective Display detailed Q, K, V vector values for selected tokens

Visualization Controls

Play/pause/rewind controls for each step Animation speed adjustment slider Toggle buttons for detailed explanations of each step

Parameter Adjustments

Embedding dimension size control Attention head count adjustment (for Multi-head attention extension) Scaling factor adjustment

3. Additional Features Explanation Panel

Display formulas and explanations for each step in right panel Highlight current active step

Comparison Features

Side-by-side view to compare attention patterns across sentences Provide examples showing similar patterns in different sentences

Performance Metrics

Display statistical information of calculated attention scores Graph showing distribution of token relationship strengths

4. Educational Elements Step-by-Step Quizzes

Brief comprehension checks after each step Prediction challenges (predict next step results)

Guide Mode

Step-by-step tutorial for first-time users Key concept explanation popups

Example Sets

Provide diverse example sentences Explain key learning points for each example

5. Visualization Design Color Scheme

Consistent color coding for Q, K, V vectors Gradients to represent attention intensity Colorblind-friendly accessible design

Layout

Natural left-to-right process flow Central placement for key information, side panels for additional info Responsive design supporting various screen sizes

This visualization enables users to intuitively understand each step of self-attention while observing its operation with real data. The interactive elements facilitate active learning, and the step-by-step visualization makes it easier to understand complex algorithms.

--> output: https://simlabapp.com/simulation/AKioPu8cmjcCw9bL3OOK

🤝 Contributing

Contributions are always welcome! Please feel free to submit a Pull Request.

📫 Contact

• GitHub: https://github.com/Royaltyprogram/SimLab_simualtions
• Email: edulens43@gmail.com

---