Why we need to expand and reshape the tensor to [(slot_dim*ds*ts), bert_seq, hid_size]

[yixuan004]

Why we need to expand and reshape the tensor to [(slot_dimdsts), bert_seq, hid_size], is that mean every utterance need to mapping into slot_dim times prediction

SUMBT/code/BeliefTrackerSlotQueryMultiSlot.py

Line 214 in 011127b

hidden = torch.mul(hidden, attention_mask.view(-1, self.max_seq_length, 1).expand(hidden.size()).float())

[yixuan004]

My code base is weak, hope to get your help! Thanks!