Skip to content

Adding Custom Models

Guide to creating and integrating custom model architectures.

For Torchvision Models

No code needed! Just update config:

model:
  type: 'base'
  architecture: 'efficientnet_b0'  # Any torchvision model
  num_classes: 10
  weights: 'DEFAULT'

All torchvision models are automatically supported.

For Custom Models

Step 1: Define Model Class

Edit ml_src/core/network/custom.py:

import torch.nn as nn

class MyCustomModel(nn.Module):
    def __init__(self, num_classes, input_size=224, **kwargs):
        super().__init__()

        # Your architecture here
        self.features = nn.Sequential(
            nn.Conv2d(3, 64, kernel_size=3, padding=1),
            nn.ReLU(),
            nn.MaxPool2d(2),
            # ... more layers
        )

        self.classifier = nn.Sequential(
            nn.Linear(64 * 112 * 112, 256),
            nn.ReLU(),
            nn.Linear(256, num_classes)
        )

    def forward(self, x):
        x = self.features(x)
        x = x.view(x.size(0), -1)
        x = self.classifier(x)
        return x

Step 2: Register Model

Add to MODEL_REGISTRY inside the get_custom_model() function in ml_src/core/network/custom.py (around line 172):

def get_custom_model(model_name, num_classes, input_size=224, device="cpu", **kwargs):
    # Registry of available custom models
    MODEL_REGISTRY = {
        "simple_cnn": SimpleCNN,
        "tiny_net": TinyNet,
        "my_custom_model": MyCustomModel,  # Add here
    }
    # ... rest of function

Note: The registry is function-scoped, not module-level.

Step 3: Use in Config

model:
  type: 'custom'
  custom_architecture: 'my_custom_model'
  num_classes: 10
  input_size: 224

Step 4: Train

ml-train

Example: ResNet-like Model

class MyResNet(nn.Module):
    def __init__(self, num_classes, **kwargs):
        super().__init__()

        # Stem
        self.conv1 = nn.Conv2d(3, 64, 7, stride=2, padding=3)
        self.bn1 = nn.BatchNorm2d(64)
        self.relu = nn.ReLU()
        self.maxpool = nn.MaxPool2d(3, stride=2, padding=1)

        # Residual blocks
        self.layer1 = self._make_layer(64, 64, 2)
        self.layer2 = self._make_layer(64, 128, 2, stride=2)
        self.layer3 = self._make_layer(128, 256, 2, stride=2)

        # Classifier
        self.avgpool = nn.AdaptiveAvgPool2d((1, 1))
        self.fc = nn.Linear(256, num_classes)

    def _make_layer(self, in_channels, out_channels, blocks, stride=1):
        layers = []
        layers.append(ResidualBlock(in_channels, out_channels, stride))
        for _ in range(1, blocks):
            layers.append(ResidualBlock(out_channels, out_channels))
        return nn.Sequential(*layers)

    def forward(self, x):
        x = self.conv1(x)
        x = self.bn1(x)
        x = self.relu(x)
        x = self.maxpool(x)

        x = self.layer1(x)
        x = self.layer2(x)
        x = self.layer3(x)

        x = self.avgpool(x)
        x = torch.flatten(x, 1)
        x = self.fc(x)
        return x

Best Practices

  1. Inherit from nn.Module
  2. Accept num_classes parameter
  3. Return logits (no softmax)
  4. Test forward pass before training
  5. Document architecture

Testing Your Model

# Test script
from ml_src.core.network import get_model
import torch

config = {
    'model': {
        'type': 'custom',
        'custom_architecture': 'my_custom_model',
        'num_classes': 10
    }
}

model = get_model(config, 'cpu')
x = torch.randn(2, 3, 224, 224)
y = model(x)
print(f"Output shape: {y.shape}")  # Should be (2, 10)