alexatartaglini/developmental-shape-bias

Issues

• enable a verbose option

  #20 opened 2 years ago by alexatartaglini
  0

• implement supervised vision transformer

  #21 opened 2 years ago by alexatartaglini
  1

• rewrite classification code

  #19 opened 2 years ago by alexatartaglini
  0

• Visualize a subset of the stimuli and the model responses

  #18 opened 3 years ago by alexatartaglini
  1

• try running simulations separately for each distance metric

  #14 opened 3 years ago by alexatartaglini
  0

• write a unit test for the model initialization step

  #10 opened 3 years ago by alexatartaglini
  0

• Improve the histogram

  #17 opened 3 years ago by alexatartaglini
  0

• check if the Geirhos github resizes images

  #16 opened 3 years ago by alexatartaglini
  1

• try removing some of the resizing code/transforms

  #11 opened 3 years ago by alexatartaglini
  0

• change the name of "fake" dataset

  #13 opened 3 years ago by alexatartaglini
  0

• retrieve correct transforms

  #8 opened 3 years ago by alexatartaglini
  0

• normalize CLIP feature vectors

  #9 opened 3 years ago by alexatartaglini
  0

• dataloader loop in main.py

  #6 opened 3 years ago by wkvong
  0

• instead of calling __getitem__ directly, wrap the dataset in a Dataloader object

  #15 opened 3 years ago by alexatartaglini
  1

• move penult_model initializations to the initialize_model function

  #7 opened 3 years ago by alexatartaglini
  0

• delete the penult_model parameters loop

  #12 opened 3 years ago by alexatartaglini
  0

• replicate original shape bias results using ImageNet trained ResNet-50

  #2 opened 4 years ago by wkvong
  2

• improvements to dataloader.py

  #5 opened 4 years ago by wkvong
  2

• rename repo

  #1 opened 4 years ago by wkvong
  0

• download full dataset of cue-conflict images

  #4 opened 4 years ago by wkvong
  0

• create separate dataset folder with stimuli grouped by texture

  #3 opened 4 years ago by wkvong
  0