Epoch

Definition

An epoch represents one full cycle through all training data during model training.

Training Process: - Multiple epochs typically needed - Each epoch: Model sees all examples once - Parameters updated many times per epoch

How Many Epochs? - Depends on dataset size and complexity - Too few: Underfitting - Too many: Overfitting - Use validation loss to decide

Typical Ranges: - Image classification: 50-200 epochs - LLM pre-training: 1-2 epochs (data is huge) - Fine-tuning: 3-10 epochs

Early Stopping: - Monitor validation loss - Stop when it starts increasing - Prevents overfitting

Training for 100 epochs means the model sees each training image 100 times.

The process of teaching an AI model to perform tasks by exposing it to data.

Number of training examples processed together before updating model weights.

When a model learns training data too well and fails to generalize to new data.