614c41e7f05d274991d95271c49a87af3fa2cf49,onmt/Loss.py,NMTLossCompute,compute_loss,#NMTLossCompute#Any#Any#Any#,150

Before Change

        if self.label_smoothing > 0:
            mask = target_feed.unsqueeze(1).eq(self.padding_idx) \
                              .repeat(1, scores.size(1))
            target_ = Variable(self.one_hot.repeat(target_feed.size(0), 1),
                               requires_grad=False)
            target_.scatter_(1, target_feed.unsqueeze(1),
                             1 - self.label_smoothing)
            target_.masked_fill_(mask, 0)
            target_feed = target_

After Change

        gtruth = target.view(-1)
        if self.confidence < 1:
            tdata = gtruth.data
            mask = torch.nonzero(tdata.eq(self.padding_idx)).squeeze()
            likelihood = torch.gather(scores.data, 1, tdata.unsqueeze(1))
            tmp_ = self.one_hot.repeat(gtruth.size(0), 1)
            tmp_.scatter_(1, tdata.unsqueeze(1), self.confidence)
            if mask.dim() > 0:
                likelihood.index_fill_(0, mask, 0)
                tmp_.index_fill_(0, mask, 0)
            gtruth = Variable(tmp_, requires_grad=False)

        loss = self.criterion(scores, gtruth)
        if self.confidence < 1:
            loss_data = - likelihood.sum(0)
        else:
            loss_data = loss.data.clone()

        stats = self.stats(loss_data, scores.data, target.view(-1).data)

Instances