Skip to content

Code for "An Information-theoretic Progressive Framework for Interpretation"

Notifications You must be signed in to change notification settings

hezq06/progressive_interpretation

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

4 Commits
 
 
 
 
 
 
 
 

Repository files navigation

Code for reproducing results of paper: "An Information-theoretic Progressive Framework for Interpretation" https://arxiv.org/abs/2101.02879 Authors: Zhengqi He @ NCA Lab, CBS, RIKEN, Japan Taro Toyoizumi @ NCA Lab, CBS, RIKEN & UTokyo, Japan

Dependencies: python (3.0) jupyter notebook numpy matplotlib pytorch (1.2.0) pycocotools

Dataset: We use standard CLEVR dataset in this project: https://cs.stanford.edu/people/jcjohns/clevr/ Download CLEVR v1.0 (18 GB) Extra data like calculated object mask and pretrained model is available at: https://riken-share.box.com/s/rc95iet4af680o3pll2sqywc6kevbduk Put folders "dataset" and "pretrained_models" inside the progressive_interpretation main folder

Run the code: To reproduce results in the paper, check out the jupyter notebooks: CLEVRTask1SuperviseLearning.ipynb CLEVRTask2MultipleChoice.ipynb

About

Code for "An Information-theoretic Progressive Framework for Interpretation"

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published