MILVLG/mcan-vqa

Issues

• test-dev or test-std

  #43 opened a year ago by QA-x
  3

• VQA CPv2

  #29 opened 4 years ago by zizheng777
  2

• help!

  #47 opened 9 months ago by Rooike111
  0

• Co-Attention?

  #44 opened a year ago by dreichCSL
  1

• Why our reproduced accuracy is so much higher than in the paper.

  #45 opened a year ago by 805934132
  1

• Why does the answer dictionary need to be filtered with less than 8 occurrences

  #42 opened a year ago by Su1008
  1

• mcan_encoder_decoder

  #41 opened a year ago by Su1008
  0

• About initialize the GloVe

  #46 opened a year ago by DLrook
  1

• box is xyxy or xywh?

  #39 opened 2 years ago by Originofamonia
  1

• Can I run code with CPU?

  #40 opened 2 years ago by msg67
  1

• single question test

  #38 opened 3 years ago by miao404
  3

• How did you create "npz" files from tsv files

  #37 opened 3 years ago by ifmaq1
  1

• Features' file loading in the code

  #36 opened 3 years ago by ifmaq1
  1

• Why use only image guided-attention rather than both image and question guided-attention?

  #35 opened 4 years ago
  1

• improved results in val set but decrease results in online evaluation

  #34 opened 4 years ago by trahman8
  1

• How to use pretrained models

  #33 opened 4 years ago by Dishank99
  1

• Visualizations of the learned attention maps

  #32 opened 4 years ago by clytze0216
  1

• Code for learned image and question attentions

  #19 opened 4 years ago by Karol-G
  2

• linear fusion model

  #30 opened 4 years ago by clytze0216
  1

• the problem about result jons file

  #28 opened 4 years ago by 1055032529
  2

• log file

  #25 opened 4 years ago by muaz1994
  1

• net.train() not called again after evaluation finishes

  #26 opened 4 years ago by fawazsammani
  2

• question regarding test split

  #24 opened 4 years ago by fawazsammani
  2

• question about answer processing

  #23 opened 4 years ago by muaz1994
  2

• MCAN

  #22 opened 4 years ago by RainyMoo
  0

• Direct link to the Bottom-up features

  #20 opened 4 years ago by gkv91
  1

• size mismatch for embedding.weight

  #18 opened 5 years ago by kayoyin
  1

• Are the provided bottomup feature same as the it in the original repo?

  #17 opened 5 years ago by Wangt-CN
  0

• about val score

  #16 opened 5 years ago by cikers
  5

• Imbalance of GPU comsumption.

  #14 opened 5 years ago by Zzmonica
  3

• Overfitting on val dataset

  #15 opened 5 years ago by cikers
  2

• Bert encoding

  #13 opened 5 years ago by shubhamagarwal92
  1

• # of params for the small model

  #12 opened 5 years ago by jnhwkim
  2

• where is image feature about .npy?

  #10 opened 5 years ago by Archershi
  2

• How long did you take on training with ssd?

  #8 opened 5 years ago by orchardchang
  11

• How much impovement does VG dataset bring?

  #6 opened 5 years ago by eustcPL
  1

• Feature file download failed

  #4 opened 5 years ago by AIstudentSH
  1

• pretrained frcnn and network

  #2 opened 5 years ago by nithinraok
  2