moabitcoin / adversarial-video-summarization-pytorch Goto Github PK

Hi,
I want to train this model now, But I found the dataset is too large (1.8TB), where can I get the data with several GB?

Hello,
thanks for sharing your code. It seems good to have the pre-trained LSTM autoencoder model in your code.
I tried to train this model using different code, but it is so difficult to train and test the model.
how about your result? Are the losses converged? If possible, can you upload the results of your experiment?
And in the s_p part, it seems to multiply every element of the feature by a uniform sample. Shouldn't s_p have one scalar value for each frame?

Thank you