MNIST Dataset is used for training the model.
- Optimizer - Adam
- 3 types of loss trackers
- Reconstruction Loss
- KL divergence Loss
- Total Loss
- Epochs for training - 30
- Latent Dimension Size - 2
- Callbacks used - Early Stopping, Reduce LR
- Model Weights saved to
model_weights.h5
after training.
For retraining, load the previous weights to the model and train your new model on top of the previous one.
Run the main.py
file to visualize the following:
- Latent Space Representations of the Dataset as encoded by the Encoder network.
- Newly generated images formed by sampling random noise from the latent space and feeding it to the Decoder network.