mogicianxd/ksr

Reimplementation of KSR in pytorch version. The original is written in theano (https://github.com/BetsyHJ/KSR).

ksr's Issues

confused about `model_step`, fixed-length or not?

Hi MogicianXD,
I am confused about these code.
self.pos_len = train_items.shape[1]
and

for t in range(0, X.shape[1]):
           y, h, MN = self.model_step(Sx[:, t, :], X[:, t], h, MN)

Are you set fixed-length for every session by padding and truncate, because I don't find any preprocessed code?

Waiting for your answer and thank you for sharing!