Concept
Adapter layers
Adapter layers are small trainable modules inserted into a frozen pre-trained network, letting you specialize a large model for a new task without updating its original weights. Think of fine-tuning, except instead of adjusting all the model's parameters you leave them…
The rest of “Adapter layers” is a premium feature: every concept in the library gets a precise, practitioner-focused write-up like this one, cross-linked straight from the paper summaries that use it.
Log in to unlock→