taoxugit
Question about data split in CUB dataset + confirming code snippet
Dear Authors,
I want to ask about how you split the dataset of CUB into train, dev, and test, as I only see train and test dataset in CUB.
I also want to confirm if you optimize your model on test dataset, instead of dev dataset in this code snippet: https://github.com/taoxugit/AttnGAN/blob/0d000e652b407e976cb88fab299e8566f3de8a37/code/pretrain_DAMSM.py#L252-L254 is it related to the same problem?
If yes, in MSCOCO, when dev dataset is available, did you change this so that the training is optimized into dev dataset?
Best regards, Johanes Effendi
Dear Authors,
I want to ask about how you split the dataset of CUB into train, dev, and test, as I only see train and test dataset in CUB.
I also want to confirm if you optimize your model on test dataset, instead of dev dataset in this code snippet: https://github.com/taoxugit/AttnGAN/blob/0d000e652b407e976cb88fab299e8566f3de8a37/code/pretrain_DAMSM.py#L252-L254
is it related to the same problem? If yes, in MSCOCO, when dev dataset is available, did you change this so that the training is optimized into dev dataset?
Best regards, Johanes Effendi
Have you figured out this problem? I am puzzled with this problem for a long time~~/(ㄒoㄒ)/~~
