search

greentfrapp / lucent

Lucid library adapted for PyTorch
Apache License 2.0
597 stars 89 forks source link

Neuron has grey optimal input #34

Open TomFrederik opened 2 years ago

TomFrederik commented 2 years ago

For some neurons the optimized image is simply grey. This does not happen when choosing objectives.channel

Minimal example for colab:

!pip install torch-lucent
import torch
import torchvision as tv
from lucent.optvis import render, objectives, transform, param

device = 'cuda' if torch.cuda.is_available() else 'cpu'
model = tv.models.googlenet(pretrained=True).to(device)
model.eval()

param_f = lambda: param.image(64, decorrelate=False, fft=True)
objs = [objectives.neuron('inception3b_branch2_1', i) for i in range(3)]

for i, obj in enumerate(objs):
    _ = render.render_vis(model, obj, param_f, thresholds=(256,), show_inline=True)
    # for i = 1 the image will be grey, if fft=False, then i=2 will also be grey.

image