What Is an Adapter Layer in AI?
An adapter layer is a small trainable module inserted into a pretrained network while most original model weights stay frozen.
Don't Just Read it – Explore It
An adapter layer is a small trainable module inserted into a pretrained network while most original model weights stay frozen.
Prompt tuning adapts a frozen language model by learning a small set of continuous prompt vectors for a task.
Prefix tuning adapts a frozen language model by learning continuous vectors that its layers attend to as task-specific context.
Chain-of-thought prompting encourages a language model to generate intermediate steps before giving a final answer.
Triplet loss trains an embedding by comparing an anchor with one related example and one unrelated example.
In-context learning is a language model’s ability to follow patterns or examples supplied in its prompt without updating its weights.
A Siamese neural network processes two inputs with shared weights and compares the representations it produces.
Representation learning lets a model discover useful features from data instead of relying entirely on hand-designed measurements.
Metric learning trains a distance or similarity function so related examples are close and unrelated examples are separated.
Contrastive learning trains a representation by drawing related examples closer together and separating examples treated as different.
Time-series forecasting uses ordered historical observations to estimate future values while respecting temporal structure.
An autoregressive model predicts a current or future value from earlier values in the same ordered sequence.