dno 0.1.1

Dynamic Neural Organism (DNO): A self-evolving, growing, and pruning neural network framework.

DNO: Dynamic Neural Organism 🧬

DNO (formerly DeeDe) is a PyTorch-based framework for creating biological neural networks that grow, think, and evolve at runtime.

Unlike static deep learning models, a DNO is an organism that starts small and evolves its architecture based on the problem complexity.

Key Features

• 🧠 Dynamic Growth: Starts with a seed and adds layers via mitosis when "confused" (High Entropy).
• 🗡️ Survival of the Fittest: Prunes inefficient neurons and layers using Information Gain (KL-Divergence).
• 🧬 Fluid Serialization: Save the entire organism (Weights + Topology + History) into a single .dno file.
• 📺 Live Dashboard: Visualize the organism's anatomy and health in the terminal.

Installation

pip install dno

Quick Start

import torch
import torch.nn as nn
from dno.core.organism import OrganismManager, BaseEvolvableModule
from dno.core.network import DynamicNetwork
from dno.config import DnoConfig
from dno.utils.dashboard import print_organism_status

# 1. Create the Organism
manager = OrganismManager()
config = DnoConfig(entropy_threshold=0.6, growth_alpha=0.5)
network = DynamicNetwork(manager, config)

# 2. Add a Seed Layer
seed = BaseEvolvableModule(nn.Linear(10, 10))
seed.dynamic_id = "seed_layer"
network.add_layer(seed)

# 3. Live Life (Forward Pass)
input_data = torch.randn(1, 10)
output = network(input_data)

# 4. Check Status
print_organism_status(manager)

License

MIT