batra-mlp-lab/visdial

Issues

• THCudaCheck FAIL error=2 : out of memory... MultiGPU training

  #42 opened 4 years ago by eugene8568
  0

• How did you generate Figure 5 in your paper?

  #41 opened 5 years ago by MottoX
  1

• Generate the new data_img.h5 from prepro_img_pool5.lua

  #12 opened 6 years ago by loripandka
  10

• The chat interface of visual dialog on this page: https://visualdialog.org/ is not working.

  #40 opened 6 years ago by divyanshj16
  1

• Support for extracting conv layer / spatial features for ResNets (for attention-based models)

  #38 opened 6 years ago by abhshkdz
  1

• Invalid permutation error reading image feature you gave

  #37 opened 6 years ago by gicheonkang
  2

• Wrong filename for visdial_data_{}.h5 in VisDial v1.0

  #35 opened 6 years ago by DJJune
  1

• Issue on extracting image feature

  #31 opened 6 years ago by yellowjs0304
  6

• Is there any particular reason why you add +1

  #34 opened 6 years ago by gicheonkang
  1

• dimension doesn't match during evalution

  #27 opened 6 years ago by DJJune
  7

• Pytorch starter code

  #28 opened 6 years ago by shubhamagarwal92
  2

• Error with command "python prepro.py -download 1 -image_root /path/to/coco/images

  #30 opened 6 years ago by yellowjs0304
  1

• why 80,000 visual feature? not 82,783

  #29 opened 6 years ago by gicheonkang
  0

• feature extraction error

  #9 opened 7 years ago by sxjzwq
  2

• Issue with prepro.py at line 150.

  #26 opened 6 years ago by tarun005
  3

• bad argument #2 to 'add' (sizes do not match at /torch-addons/cutorch/lib/THC/generated/../generic/THCTensorMathPointwise.cu:269)

  #24 opened 6 years ago by krsandeep98
  3

• Why is the dimension of input image features in attention case 14x14x4096?

  #25 opened 6 years ago by krsandeep98
  1

• Update pretrained checkpoints

  #21 opened 6 years ago by abhshkdz
  1

• Inconsistent tensor size, training with mn-att-ques-im-hist

  #13 opened 6 years ago by enidvx
  1

• Evaluation of pretrained late fusion model fails

  #15 opened 7 years ago by kracwarlock
  1

• Error in forwardbackward() function while training the data

  #14 opened 7 years ago by hardik2396
  1

• maybe an error in dataloader.lua

  #8 opened 7 years ago by hehefan
  1

• Where is the code concatenating the t-rounds of history for lf-encoder?

  #6 opened 7 years ago by ruizhaogit
  4

• Why use "CAddTable" instead of concatenating features in the memory network?

  #7 opened 7 years ago by ruizhaogit
  2

• Models for Android Use?

  #2 opened 7 years ago by vyse8
  3

• Beam search with length normalized log likelihood?

  #5 opened 7 years ago by ruizhaogit
  1

• About sampling

  #4 opened 7 years ago by ruizhaogit
  2

• Do we need to use early stopping to prevent over-fitting?

  #3 opened 7 years ago by ruizhaogit
  6

• prepro.py: TypeError: slice indices must be integers or None or have an __index__ method

  #1 opened 7 years ago by ruizhaogit
  2