Machine Learning and Perception Lab, Georgia Tech visdial Issues

batra-mlp-lab/visdial Issues

THCudaCheck FAIL error=2 : out of memory... MultiGPU training
Updated 4 years ago
How did you generate Figure 5 in your paper?
Closed 5 years ago1
Generate the new data_img.h5 from prepro_img_pool5.lua
Closed 6 years ago10
The chat interface of visual dialog on this page: https://visualdialog.org/ is not working.
Closed 6 years ago1
Support for extracting conv layer / spatial features for ResNets (for attention-based models)
Updated 6 years ago1
Invalid permutation error reading image feature you gave
Closed 6 years ago2
Wrong filename for visdial_data_{}.h5 in VisDial v1.0
Closed 6 years ago1
Issue on extracting image feature
Closed 6 years ago6
Is there any particular reason why you add +1
Closed 6 years ago1
dimension doesn't match during evalution
Closed 6 years ago7
Pytorch starter code
Closed 6 years ago2
Error with command "python prepro.py -download 1 -image_root /path/to/coco/images
Closed 6 years ago1
why 80,000 visual feature? not 82,783
Closed 6 years ago
feature extraction error
Closed 7 years ago2
Issue with prepro.py at line 150.
Closed 6 years ago3
bad argument #2 to 'add' (sizes do not match at /torch-addons/cutorch/lib/THC/generated/../generic/THCTensorMathPointwise.cu:269)
Closed 6 years ago3
Why is the dimension of input image features in attention case 14x14x4096?
Closed 6 years ago1
Update pretrained checkpoints
Closed 6 years ago1
Inconsistent tensor size, training with mn-att-ques-im-hist
Closed 6 years ago1
Evaluation of pretrained late fusion model fails
Closed 6 years ago1
Error in forwardbackward() function while training the data
Closed 7 years ago1
maybe an error in dataloader.lua
Closed 7 years ago1
Where is the code concatenating the t-rounds of history for lf-encoder?
Closed 7 years ago4
Why use "CAddTable" instead of concatenating features in the memory network?
Closed 7 years ago2
Models for Android Use?
Closed 7 years ago3
Beam search with length normalized log likelihood?
Closed 7 years ago1
About sampling
Closed 7 years ago2
Do we need to use early stopping to prevent over-fitting?
Closed 7 years ago6
prepro.py: TypeError: slice indices must be integers or None or have an __index__ method
Closed 7 years ago2

batra-mlp-lab / visdial

batra-mlp-lab/visdial Issues

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

How did you generate Figure 5 in your paper?

Generate the new data_img.h5 from prepro_img_pool5.lua

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

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

Invalid permutation error reading image feature you gave

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

Issue on extracting image feature

Is there any particular reason why you add +1

dimension doesn't match during evalution

Pytorch starter code

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

why 80,000 visual feature? not 82,783

feature extraction error

Issue with prepro.py at line 150.

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

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

Update pretrained checkpoints

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

Evaluation of pretrained late fusion model fails

Error in forwardbackward() function while training the data

maybe an error in dataloader.lua

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

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

Models for Android Use?

Beam search with length normalized log likelihood?

About sampling

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

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