Concept

Depth-Adaptive BERT Models

Depth-adaptive models are a type of dynamic network that improves BERT's inference efficiency. The core principle is to dynamically determine the optimal number of layers required to process a given token. The model can then exit early from an intermediate layer, skipping the remaining layers in the stack to reduce computation.

0

1

Updated 2026-04-18

Contributors are:

Who are from:

Tags

Ch.1 Pre-training - Foundations of Large Language Models

Foundations of Large Language Models

Foundations of Large Language Models Course

Computing Sciences