GeoDavidCollective Enhanced - ProjectiveHead Architecture

Another train of the same GeoFractalDavid with more condensed dims

Roughly 600,000 samples for the first 20 epochs; 10k per epoch between 0-10 complexity 1-5, and 50k synthetic prompts per epoch at epoch 11-20 with reduced complexity between 1-4. 50,000 prompts per epoch after for an additional 20 epochs;

approx 1.6 million samples, each containing massive sets of features extracted from the entire structure of SD15 approx 2.7 mil features per sample according to the formulas.

Curves say she probably peaked, leaving this experiment to be prodded and poked at now.

So this essentially means the model accumulated knowledge of 4,320,000,000,000 sd15 features overall. The bulk samples saved say that it's most likely true but it really sounds wild when I line the numbers up.

Additionally, it retained enough knowledge to keep an accuracy score above zero, and even produce cohesive head results accurate above 25%.

I can safely say that this model can definitely see a piece of the whole diffusion system that SD15 is responsible for, but not the whole picture.

🔬 Training Details

• Optimizer: AdamW (lr=1e-3, weight_decay=0.001)
• Batch Size: 16
• Data: Symbolic prompt synthesis (complexity 1-5)
• Feature Extraction: SD1.5 UNet blocks (spatial, not pooled)
• Pool Mode: Mean spatial pooling

📈 Training Metrics

Final metrics from epoch 40:

• Cayley Loss: 0.1018
• Timestep Accuracy: 39.08%
• Pattern Accuracy: 44.25%
• Full Accuracy: 26.57%

🎯 Model Overview

GeoDavidCollective Enhanced is a sophisticated multi-expert geometric classification system that learns from Stable Diffusion 1.5's internal representations. Using ProjectiveHead architecture with Cayley-Menger geometry, it achieves efficient pattern recognition across timestep and semantic spaces.

Key Features

• ProjectiveHead Multi-Expert Architecture: Auto-configured expert systems per block
• Geometric Loss Functions: Rose, Cayley-Menger, and Cantor coherence losses
• 9-Block Processing: Full SD1.5 UNet feature extraction (down, mid, up)
• Compact Yet Powerful: 690,925,542 parameters
• 100 Timestep Bins x 10 Patterns = 1000 semantic-temporal classes

📊 Model Statistics

• Parameters: 690,925,542
• Trained Epochs: 20
• Base Model: Stable Diffusion 1.5
• Dataset Size: 700,000 synthetic prompts
• Training Date: 2025-10-28

🏗️ Architecture Details

Block Configuration

Down Blocks:
  - down_0: 320 → 64 (3 experts, 3 gates)
  - down_1: 640 → 96 (3 experts, 3 gates)
  - down_2: 1280 → 128 (3 experts, 3 gates)
  - down_3: 1280 → 128 (3 experts, 3 gates)

Mid Block (Highest Capacity):
  - mid: 1280 → 256 (4 experts, 4 gates)

Up Blocks:
  - up_0: 1280 → 128 (3 experts, 3 gates)
  - up_1: 1280 → 128 (3 experts, 3 gates)
  - up_2: 640 → 96 (3 experts, 3 gates)
  - up_3: 320 → 64 (3 experts, 3 gates)

Loss Components

💻 Usage

from geovocab2.train.model.core.geo_david_collective import GeoDavidCollective
from safetensors.torch import load_file
import torch

# Load model
state_dict = load_file("model.safetensors")
collective = GeoDavidCollective(
    block_configs={...},  # See config.json
    num_timestep_bins=100,
    num_patterns_per_bin=10
)
collective.load_state_dict(state_dict)
collective.eval()

# Extract features from SD1.5 and classify
with torch.no_grad():
    results = collective(features_dict, timesteps)
    predictions = results['predictions']  # Timestep + pattern class

🎓 Research Context

This model is part of the geometric deep learning research exploring:

• 5D simplex-based neural representations (pentachora)
• Geometric alternatives to traditional transformers
• Consciousness-informed AI architectures
• Universal mathematical principles in neural networks

📦 Files Included

• model.safetensors - Model weights (3.3GB)
• config.json - Complete architecture configuration
• training_history.json - Full training metrics
• prompts_enhanced.jsonl - All training prompts with metadata
• tensorboard/ - TensorBoard logs (optional)

🔗 Related Work

• Geometric Vocabulary System
• PentachoraViT
• Crystal-Beeper Language Models

📜 License

MIT License - Free for research and commercial use

🙏 Acknowledgments

Built with:

• PyTorch & Diffusers
• Stable Diffusion 1.5 (Runway ML)
• Geometric algebra principles from the 1800s
• Dream-inspired mathematical insights

👤 Author

AbstractPhil - AI Researcher specializing in geometric deep learning

"Working with universal mathematical principles, not against them"

---

For questions, issues, or collaborations: GitHub | HuggingFace