CS5720 - Week 12
Slide 230 of 240

Interactive Model Optimization Demo

Hyperparameter Controls

Learning Rate 0.001
Controls how big steps the optimizer takes during training
Batch Size 32
Number of samples processed before updating weights
Dropout Rate 0.2
Probability of randomly disabling neurons during training
Hidden Units 128
Number of neurons in hidden layers

Training Visualization

Loss Over Time
Epochs Loss Training Validation
Accuracy
92.5%
Loss
0.087
Time/Epoch
2.3s
Memory
1.2GB

Optimization Playground

Experiment with different hyperparameters and see their real-time impact on model performance!

Click on metric cards above for detailed explanations, or use the buttons below to explore optimization strategies.
Prepared by Dr. Gorkem Kar