AI Binding Mechanism

Creator
Creator
Seonglae ChoSeonglae Cho
Created
Created
2024 Oct 25 14:57
Editor
Edited
Edited
2024 Oct 25 15:0

To correctly use in-context information, language models (LMs) must bind entities to their attributes.

 
 
 
 
LMs’ internal activations represent binding information by attaching binding ID vectors to corresponding entities and attributes. We further show that binding ID vectors form a continuous subspace, in which distances between binding ID vectors reflect their discernability.
 
 

Recommendations