- Epochs — Number of times the full dataset is used during training. Higher values can improve learning but may increase training time.
- Batch size — Number of samples processed at once during training. Affects training speed and memory usage.
- Epochs without improvement — Controls early stopping. Defines how many epochs to wait before stopping if the validation metric does not improve.
- Learning rate — Controls the step size during learning. Small steps result in slower but more careful learning, while large steps are faster but riskier. The value must be between 0 and 1.