What Is Hyperparameter Tuning?
Hyperparameter tuning compares training configurations that are chosen outside the model’s ordinary parameter updates.
Don't Just Read it – Explore It
Hyperparameter tuning compares training configurations that are chosen outside the model’s ordinary parameter updates.
Regularization discourages a model from fitting training data with more complexity than the task appears to require.
Data augmentation creates altered training examples that preserve the label or behavior the model should learn.
Overfitting occurs when a model learns the training data too narrowly and performs poorly on new examples.
A batch is a subset used for one training update, while an epoch is one pass through the training dataset.
Gradient descent updates model parameters in the direction that is expected to reduce a loss function.
Backpropagation efficiently calculates how each model parameter contributed to the training loss.
Logits are a model’s raw scores, while softmax converts a set of scores into values that sum to one.
A loss function converts the difference between a model’s output and a training target into a value to minimize.
Tokenization converts text or other input into numbered units that a model can process.
An AI hallucination is a fluent output that is unsupported by evidence, inconsistent with the input, or factually wrong.
AI model drift occurs when changes in real-world data or relationships reduce how well a deployed model performs compared with its original evaluation.