Concept icon
Concept

Long-Distance Dependency Tracking in Self-Attention Heads

Encoder self-attention heads can track long-distance syntactic dependencies across a sentence. As demonstrated in layer 5 of a 6-layer Transformer encoder, multiple attention heads attend across intervening words to link separated components of a phrase, such as connecting the verb "making" to "more difficult" across distant token positions.

0

1

Concept icon
Updated 2026-09-07

Tags

Prep Sessions

Transformer Architecture and Large Language Model Capabilities @ University of Michigan - Ann Arbor

Ch.1 Transformer Architecture Fundamentals - Transformer Architecture and Large Language Model Capabilities @ University of Michigan - Ann Arbor

Attention Visualizations and Linguistic Structure Resolution - Transformer Architecture and Large Language Model Capabilities @ University of Michigan - Ann Arbor