v1.2

New in v.1.2

• Saving and loading a trained model using safetensors format
• Digit recognition from a hand drawn image
• Continuous model training with reporting of the current losses and accuracy
• Layer visualizations