model.py

import torch
import torchvision
from torchvision import models
from torch import nn

def create_effnetb2_model(
    num_classes:int = 3, # Default output classes = 3 (pizza, steak, sushi)
    seed: int = 42
):
  # 1. Setup the pretrained weights of EffNetB2
  effnetb2_weights = torchvision.models.EfficientNet_B2_Weights.DEFAULT

  # 2. Get EffNetB2 transforms to make sure we utilize the same transform as the model that was trained on.
  effnetb2_transforms = effnetb2_weights.transforms()

  # 3. Setup pretrained model instance
  model = torchvision.models.efficientnet_b2(weights=effnetb2_weights) # could also use weights="DEFAULT"

  # 4. Freeze the base layers in the model (this will stop all layers from training)
  for param in model.parameters():
    param.requires_grad = False

  # 5. Adjust the classifier head of the pretrained model to suit our use case
  # Random seed for reproducibility
  torch.manual_seed(seed)
  model.classifier = nn.Sequential(
    nn.Dropout(p=0.3, inplace=True),
    nn.Linear(in_features=1408, out_features=num_classes, bias=True)
  )
  return model, effnetb2_transforms