dissect

dissect copied to clipboard

Add two concept categories "texture" and "scene"

1

Is it possible to add more concept categories? Like what you did in "Network Dissection: Quantifying Interpretability of Deep Visual Representations" We really need another two concept categories. Thanks

clalanliu

Error while running the notebook dissect_classifier_experiment.ipynb

1

I encountered the following error while running dissect_classifier_experiment.ipynb `compute_acts() takes 1 positional argument but 2 were given` while executing the following cell `unit_images = iv.masked_images_for_topk( compute_acts, dataset, topk, k=5, num_workers=30,...

Vikramank

  Hi I encountered this error when I am trying to run the experiment dissect_experiment.py The Cuda Driver that I am using is cuda-10.1 (cuda-10.2 in this repository). What...

DanWijaya

djech2021

Hello, I wonder if you publish dissect results on models other than VGG16 somewhere?

AhmadM-DL

Is it actually reasonable to use single-class accuracy?

1

Hi, I'd like to ask a question regarding your paper. I'm interested in Fig 2.C and Fig S4.a,b. If I understand correctly, there you remove 256 neurons least relevant to...

AwesomeLemon

ImportError: No module named '_prroi_pooling'

1

Hi, I am trying to setup the codebase and run the cells part of dissect_classifier_experiment.ipynb However, I am running into the following error: ImportError: No module named '_prroi_pooling' Please find...

adithya-samavedhi

Hello and thanks for your fascinating work, I was using the Broden dataset and stumbled upon images (e.g. the 30th image) that contained the scene label 0 instead of nan...

ThomasNorr