This repo contains code and articles on PAIR interpretability projects.
Patchscopes: A Unifying Framework for Inspecting Hidden Representations of Language Models
The Patchscopes mini-site contains a brief introduction to the longer paper, and will have the code uploaded once it is a little cleaner also.
bert-tree and context-atlas are repos for two interactive blogposts/visualizations for the paper Visualizing and Measuring the Geometry of BERT :
-
Language, trees, and geometry in neural networks explores the geometry of syntactic information in BERT (bert-tree)
-
Language, Context, and Geometry in Neural Network explores semantics and context in BERT. See the accompanying tool, Context Atlas, for more details (context-atlas)
text-dream contains different experiments and tools to work with deep dreaming for text.
data-synth-syntax contains LinguisticLens, a tool for visualizing generated text data.