# mic-dkfz/dynamic-network-architectures architecture

> Architecture diagram of the GitHub repository mic-dkfz/dynamic-network-architectures, generated by GitDiagram from its file tree, README and a sample of source files (last updated 2025-05-11). It is an AI-made overview: check details against the source.

- Interactive diagram: https://gitdiagram.com/mic-dkfz/dynamic-network-architectures
- Repository: https://github.com/mic-dkfz/dynamic-network-architectures

## Overview

##### Project type and purpose
- This is an open-source Python library (“dynamic_network_architectures”) providing dynamically configurable PyTorch model architectures (ResNet, VGG, U-Net, Primus) that adapt to 1D/2D/3D inputs and variable channel counts.

##### File structure
- Top-level package:
- dynamic_network_architectures/
– architectures/
· abstract_arch.py (base class)
· resnet.py, vgg.py, unet.py, primus.py (concrete model classes)
– building_blocks/
· helper modules (eva.py, helper.py, patch_encode_decode.py…)
· core building blocks (plain_conv_encoder.py, residual.py, unet_decoder.py, etc.)
– initialization/
· weight_init.py (weight‐initialization strategies)
- setup.py drives installation.
- imgs/Logos holds documentation assets.

##### README for additional insights
- Supported architectures: ResNet (including bottleneck & CIFAR variants), VGG variants, U-Net (plain and residual).
- Regularization options: Stochastic Depth, Squeeze & Excitation, Final Layer Dropout.
- Target use: research/deployment in varying dimensionalities.

a. Main components:
– Client code / downstream application (training scripts, inference pipelines)
– dynamic_network_architectures package
- API layer (constructor/factory functions)
- Architectures module (abstract base + concrete models)
- Building blocks module (reusable layers & utilities)
- Initialization module (weight_init)
– External dependencies: PyTorch (torch.nn, torch.functional), NumPy, etc.
– Data sources (datasets, data loaders) and training loop (optimizers, schedulers) can be shown as external.
b. Relationships and interactions:
– Client code → imports API classes/factories → creates an instance of AbstractArch subclass (e.g., ResNet3D)
– Model constructor calls building_blocks submodules to assemble layers
– After construction, initialization.weight_init is applied
– The resulting torch.nn.Module flows into training/inference, leveraging PyTorch’s engine, optimizers, and data loaders
c. Architectural patterns & design principles:
– Modular layering: clear separation between “architecture definitions” and “building blocks”
– Inheritance/polymorphism: abstract_arch defines common interface, concrete classes extend it
– Configuration-driven: number of dimensions, channels, depth, regularization toggles
– Single Responsibility: weight initialization lives in its own subpackage
d. Key technologies/frameworks:
– Python package structure, setuptools (setup.py)
– PyTorch (torch.nn.Module, layers, dropout, etc.)
– Custom regularization techniques (Stochastic Depth, SE blocks)

##### Elements
– Clear, descriptive labels for each package/subpackage and major class (e.g., AbstractArch, ResNet, UnetDecoder)
- import/call relationships (API → architectures → building_blocks → initialization)
- data flow (input data → model → output predictions)
- Modules (blue rectangles)
- External dependencies (green clouds)

##### Detailed capture of architectural elements
- API/factory layer
- Abstract architecture interface
- Concrete models (ResNet, VGG, U-Net, Primus)
- Building blocks (plain conv, residual, patch encoder/decoder, regularization helpers)
- Initialization routines
- External PyTorch engine
- User application/entry point

## Diagram (Mermaid)

```mermaid
flowchart TD
    %% External User Application
    Client["User Application / Entry Point"]:::external

    %% Main Package
    subgraph "dynamic_network_architectures"
        %% API Layer
        API["API / Factory Layer"]:::module

        %% Architectures Module
        subgraph "Architectures Module"
            abstract["AbstractArch"]:::module
            resnet["ResNet"]:::module
            vgg["VGG"]:::module
            unet["U-Net"]:::module
            primus["Primus"]:::module
        end

        %% Building Blocks Module
        subgraph "Building Blocks Module"
            eva["EVA Helpers"]:::module
            helper["General Helpers"]:::module
            patch["Patch Encoder/Decoder"]:::module
            plain_conv["Plain Conv Encoder"]:::module
            regularization["Regularization Utilities"]:::module
            residual["Residual Blocks"]:::module
            residual_enc["Residual Encoders"]:::module
            simple_conv["Simple Conv Blocks"]:::module
            unet_dec["U-Net Decoder"]:::module
            unet_res_dec["U-Net Residual Decoder"]:::module
        end

        %% Initialization Module
        subgraph "Initialization Module"
            init["Weight Initialization"]:::module
        end
    end

    %% External Dependencies
    subgraph "External Dependencies"
        torch["PyTorch Engine"]:::external
        numpy["NumPy"]:::external
    end

    %% Data & Training
    subgraph "Data & Training"
        data["Data Sources / Loaders"]:::data
        train["Training Loop / Optimizers"]:::data
    end

    %% Relationships
    Client -->|imports| API
    API -->|constructs| abstract
    abstract -.->|inherits| resnet
    abstract -.->|inherits| vgg
    abstract -.->|inherits| unet
    abstract -.->|inherits| primus

    resnet -->|uses| plain_conv
    resnet -->|uses| residual
    resnet -->|uses| regularization

    vgg -->|uses| simple_conv
    vgg -->|uses| regularization

    unet -->|uses| plain_conv
    unet -->|uses| unet_dec
    unet -->|uses| patch
    unet -->|uses| regularization

    primus -->|uses| residual_enc
    primus -->|uses| patch
    primus -->|uses| regularization

    abstract -->|applies| init
    init -->|initializes| resnet
    init -->|initializes| vgg
    init -->|initializes| unet
    init -->|initializes| primus

    %% Model Execution Flow
    abstract -->|Module as torch.nn.Module| torch
    torch -->|runs on| data
    data -->|feeds| train
    train -->|updates| torch

    %% Click Events
    click API "https://github.com/mic-dkfz/dynamic-network-architectures/blob/main/dynamic_network_architectures/__init__.py"
    click abstract "https://github.com/mic-dkfz/dynamic-network-architectures/blob/main/dynamic_network_architectures/architectures/abstract_arch.py"
    click resnet "https://github.com/mic-dkfz/dynamic-network-architectures/blob/main/dynamic_network_architectures/architectures/resnet.py"
    click vgg "https://github.com/mic-dkfz/dynamic-network-architectures/blob/main/dynamic_network_architectures/architectures/vgg.py"
    click unet "https://github.com/mic-dkfz/dynamic-network-architectures/blob/main/dynamic_network_architectures/architectures/unet.py"
    click primus "https://github.com/mic-dkfz/dynamic-network-architectures/blob/main/dynamic_network_architectures/architectures/primus.py"
    click eva "https://github.com/mic-dkfz/dynamic-network-architectures/blob/main/dynamic_network_architectures/building_blocks/eva.py"
    click helper "https://github.com/mic-dkfz/dynamic-network-architectures/blob/main/dynamic_network_architectures/building_blocks/helper.py"
    click patch "https://github.com/mic-dkfz/dynamic-network-architectures/blob/main/dynamic_network_architectures/building_blocks/patch_encode_decode.py"
    click plain_conv "https://github.com/mic-dkfz/dynamic-network-architectures/blob/main/dynamic_network_architectures/building_blocks/plain_conv_encoder.py"
    click regularization "https://github.com/mic-dkfz/dynamic-network-architectures/blob/main/dynamic_network_architectures/building_blocks/regularization.py"
    click residual "https://github.com/mic-dkfz/dynamic-network-architectures/blob/main/dynamic_network_architectures/building_blocks/residual.py"
    click residual_enc "https://github.com/mic-dkfz/dynamic-network-architectures/blob/main/dynamic_network_architectures/building_blocks/residual_encoders.py"
    click simple_conv "https://github.com/mic-dkfz/dynamic-network-architectures/blob/main/dynamic_network_architectures/building_blocks/simple_conv_blocks.py"
    click unet_dec "https://github.com/mic-dkfz/dynamic-network-architectures/blob/main/dynamic_network_architectures/building_blocks/unet_decoder.py"
    click unet_res_dec "https://github.com/mic-dkfz/dynamic-network-architectures/blob/main/dynamic_network_architectures/building_blocks/unet_residual_decoder.py"
    click init "https://github.com/mic-dkfz/dynamic-network-architectures/blob/main/dynamic_network_architectures/initialization/weight_init.py"
    click setup "https://github.com/mic-dkfz/dynamic-network-architectures/blob/main/setup.py"

    %% Styles
    classDef module fill:#ccf,stroke:#333,stroke-width:1px
    classDef external fill:#cfc,stroke:#333,stroke-width:1px
    classDef data fill:#fc9,stroke:#333,stroke-width:1px
```
