shouldn’t this code:
test_iter = gluon.data.DataLoader(np.array(train_inter_mat), shuffle=False,
last_batch=“keep”, batch_size=1024,
num_workers=d2l.get_dataloader_workers())
be like
test_iter = gluon.data.DataLoader(np.array(test_inter_mat), shuffle=False,
last_batch=“keep”, batch_size=1024,
num_workers=d2l.get_dataloader_workers())
I think the example is the user-based AutoRec, not item-basd as mentioned:
" For brevity, here we only introduce the item-based AutoRec."
The sentence is correct. It’s item-based AutoRec.
The input size of the model is num_users, which means all users’ (incomplete) ratings to one single item.
The example model takes incomplete feedbacks of one item from all users and predicts what the complete feedbacks of that exact item should be.
Please have a look at the reference: [Sedhain et al., 2015].
I think getting to know what m & n mean in Figure 1 would be helpful.