3795eabbdc08e59fe43622a0e1bffe3916332f0f,onmt/modules/ImageEncoder.py,ImageEncoder,forward,#ImageEncoder#Any#Any#,52

Before Change

        for row in range(input.size(2)):
            inp = input[:, :, row, :].transpose(0, 2)\
                                     .transpose(1, 2)
            if input.is_cuda:
                pos_emb = self.pos_lut(
                    Variable(torch.cuda.LongTensor(batchSize).fill_(row)))
            else:
                pos_emb = self.pos_lut(
                    Variable(torch.LongTensor(batchSize).fill_(row)))
            with_pos = torch.cat(
                (pos_emb.view(1, pos_emb.size(0), pos_emb.size(1)), inp), 0)
            outputs, hidden_t = self.rnn(with_pos)
            all_outputs.append(outputs)

After Change

        for row in range(input.size(2)):
            inp = input[:, :, row, :].transpose(0, 2)\
                                     .transpose(1, 2)
            row_vec = torch.Tensor(batchSize).type_as(input).long().fill_(row)
            pos_emb = self.pos_lut(Variable(row_vec))
            with_pos = torch.cat(
                (pos_emb.view(1, pos_emb.size(0), pos_emb.size(1)), inp), 0)
            outputs, hidden_t = self.rnn(with_pos)
            all_outputs.append(outputs)

Instances