AndersonStra/MuKEA

Issues

• 请问vqa_v2_pretrain/prepare_img.py脚本中所需的npy和npz文件是如何生成的

  #38 opened 4 months ago by tanshaosuannai
  0

• 请问krvqa-ans_dic.pickle文件是如何生成的，有什么用

  #37 opened 4 months ago by tanshaosuannai
  0

• 加载预训练模型出错

  #30 opened 2 years ago by DoubleSsh
  2

• 下游任务okvqa、krvqa的batch_size、学习率分别是多少？

  #36 opened a year ago by linxi1158
  8

• 关于预训练过程的问题

  #18 opened 2 years ago by yjmlaile
  2

• 关于在OK-VQA模型上答案词表不对应的问题

  #28 opened 2 years ago by wgada
  8

• Fine-tuning时遇到的问题

  #29 opened 2 years ago by ningkekeke
  2

• 您好，您在KRVQA数据集上微调有出现权重维度不匹配的问题嘛？类似这样的RuntimeError: Error(s) in loading state_dict for KgPreModel: size mismatch for tail_decode.weight: copying a param with shape torch.Size([146348, 300]) from checkpoint, the shape in current model is torch.Size([22875, 300]).> > Hello, is the result of your pre-training on VQAV2 normal? There's something wrong with my accuracy. It's all 0

  #7 opened 3 years ago by wangfengjuan
  3

• 关于vqa_img_feature_test_dev.pickle文件的问题

  #15 opened 2 years ago by yanhan111
  8

• vqa_v2_pretrain下的prepare_img.py与tsv2feature_objects.py

  #35 opened a year ago by ProudZhangXiaolong
  0

• Hello,can you tell me the order of use for vqa_v2_pretrain files？

  #3 opened 2 years ago by peizeli96
  69

• 附录中图1积累的图谱是如何得到的呢？

  #34 opened a year ago by beng11133
  0

• pretrain部分处理其他数据

  #33 opened a year ago by beng11133
  0

• 您好 关于数据集的一些问题

  #11 opened 2 years ago by Su1008
  14

• 如何自定义数据集

  #32 opened 2 years ago by zml110120
  0

• 如何仅进行测试？

  #31 opened 2 years ago by zml110120
  0

• 损失函数问题

  #21 opened 2 years ago by Su1008
  1

• 损失函数问题

  #16 opened 2 years ago by yuemingPAN
  1

• 关于可视化问题

  #25 opened 2 years ago by liu23ff
  2

• 关于prepare_annotation.py的问题

  #14 opened 2 years ago by a670642151
  2

• i am a poor man who loves computer

  #27 opened 2 years ago by haorui-harry
  0

• KR-VQA image data

  #26 opened 2 years ago by ebarezi
  0

• krisp+Mukea

  #23 opened 2 years ago by ebarezi
  2

• will you release the trained model

  #10 opened 3 years ago by xixihawokao
  1

• not able to find the "trainval2014_36/trainval2014_resnet101_faster_rcnn_genome_36.tsv"

  #24 opened 2 years ago by AxithChoudhary
  1

• 数据预处理问题

  #22 opened 2 years ago by a670642151
  0

• KRVQA验证集图片特征

  #20 opened 2 years ago by yuemingPAN
  0

• KRVQA数据集

  #19 opened 2 years ago by hit-syh
  0

• 预训练数据集使用问题

  #17 opened 2 years ago by yuemingPAN
  3

• 关于预训练问题

  #13 opened 2 years ago by cmcv2022
  2

• ok-vqa 的一些疑惑

  #12 opened 2 years ago by ZHUZHU1995
  3

• Hello, Why is the training time of this project so long?

  #4 opened 2 years ago by zhongfansun
  14

• what is "Look-Up Table" in this paper?

  #8 opened 2 years ago by xixihawokao
  1

• hallo，How to acquire the file of“cocobu_att”and“cocobu_box” in the prepare_img.py？thank you！

  #2 opened 2 years ago by wangfengjuan
  13

• How to acquire the file of ' vqa_img_feature_train.pickle'?

  #1 opened 2 years ago by coral3333
  3

• Why is it always forbidden to download this form halfway

  #9 opened 3 years ago by decode12
  0

• OSError: Can't load weights for './premodel'.

  #5 opened 3 years ago by PamelaDDD
  2