# DaFucV2 AI - Dynamic AI Model

This repository hosts the model for **DaFucV2 AI**, a dynamic AI architecture built using the **Fractal Universe Chocolate Wafer Model (FUCWM)**. The model is designed to integrate with the **DaFucV2 app**, offering interactive conversational capabilities and adaptive thinking loops. 

## Model Overview

- **Model Architecture**: Combines a **Variational Autoencoder (VAE)** with fractal-like expanding layers based on complexity, using a **FractalNode** structure for dynamic growth.
- **Self-Thinking and Feedback**: Incorporates an iterative feedback mechanism allowing the model to send its own thoughts back into itself for further refinement.
- **Applications**: Optimized for conversational agents, adaptive feedback systems, and deeper multi-layered reasoning.
- **Attention Mechanism**: The model dynamically adjusts attention across fractal layers to modulate responses based on the complexity of the input.

## DaFucV2 App Integration

The **DaFucV2 AI** model is designed to work seamlessly with the **DaFucV2 app**, available on [GitHub](https://github.com/anttiluode/DaFucV2/tree/main). You can use the app to interact with the model, send queries, and explore its capabilities in real time.

### Demo Video

Watch a video demonstration of me talking to the DaFucV2 AI [here on YouTube](https://www.youtube.com/watch?v=-PQ-rTkqwQ8).

## Usage

To load and use the model within the app:

1. **Download the app** from the [DaFucV2 GitHub repository](https://github.com/anttiluode/DaFucV2/tree/main).
2. **Place the model** (`model.pth`) in the appropriate directory.
3. Run the app by following the instructions in the repository.

To manually load the model in PyTorch:

```python
import torch
from model import DynamicAI

# Load the saved model
model = DynamicAI(vocab_size=50000, embed_dim=256, latent_dim=256, output_dim=256, max_depth=7)
model.load_state_dict(torch.load("model.pth"))

# Set model to evaluation mode
model.eval()

# Example usage with input text
input_text = "Hello, how are you?"
response = model.chat(input_text)
print(response)