How to use the fairseq.utils.item function in fairseq

            result.alignments.append('A\t{}'.format(' '.join(map(lambda x: str(utils.item(x)), alignment))))
        return result

def upgrade_state_dict_named(self, state_dict, name):
        """Upgrade a (possibly old) state dict for new versions of fairseq."""
        for i in range(len(self.layers)):
            # update layer norms
            self.layers[i].upgrade_state_dict_named(state_dict, "{}.layers.{}".format(name, i))

        version_key = '{}.version'.format(name)
        if utils.item(state_dict.get(version_key, torch.Tensor([1]))[0]) < 2:
            # earlier checkpoints did not normalize after the stack of layers
            self.layer_norm = None
            self.normalize = False
            state_dict[version_key] = torch.Tensor([1])
        return state_dict

def upgrade_state_dict(self, state_dict):
        if utils.item(state_dict.get('decoder.version', torch.Tensor([1]))[0]) < 2:
            # old models use incorrect weight norm dimension
            for i, conv in enumerate(self.convolutions):
                # reconfigure weight norm
                nn.utils.remove_weight_norm(conv)
                self.convolutions[i] = nn.utils.weight_norm(conv, dim=0)
            state_dict['decoder.version'] = torch.Tensor([1])
        return state_dict

            print('A\t{}'.format(' '.join(map(lambda x: str(utils.item(x)), alignment))))

def forward(self, model, sample, reduce=True):
        """Compute the loss for the given sample.

        Returns a tuple with three elements:
        1) the loss
        2) the sample size, which is used as the denominator for the gradient
        3) logging outputs to display while training
        """
        net_output = model(**sample['net_input'])
        loss, nll_loss = self.compute_loss(model, net_output, sample, reduce=reduce)
        sample_size = sample['target'].size(0) if self.args.sentence_avg else sample['ntokens']
        logging_output = {
            'loss': utils.item(loss.data) if reduce else loss.data,
            'nll_loss': utils.item(nll_loss.data) if reduce else nll_loss.data,
            'ntokens': sample['ntokens'],
            'nsentences': sample['target'].size(0),
            'sample_size': sample_size,
        }
        return loss, sample_size, logging_output

                                ' '.join(map(lambda x: str(utils.item(x)), alignment))
                            ))

                'A\t{}'.format(' '.join(map(lambda x: str(utils.item(x)), alignment)))
                if args.print_alignment else None

# is set to true in the BERT model. Ideally we should set
            # masked_lm_only to true in this case, but that requires some
            # refactor in the BERT model.
            if sentence_logits is not None:
                sentence_loss = compute_cross_entropy_loss(
                    sentence_logits, sentence_targets)

                loss += self.args.nsp_loss_weight * (sentence_loss / nsentences)

        # NOTE: as we are summing up per token mlm loss and per sentence nsp loss
        # we don't need to use sample_size as denominator for the gradient
        # here sample_size is just used for logging
        sample_size = 1
        logging_output = {
            'loss': utils.item(loss.data) if reduce else loss.data,
            'lm_loss': utils.item(lm_loss.data) if reduce else lm_loss.data,
            # sentence loss is not always computed
            'sentence_loss': (
                (
                    utils.item(sentence_loss.data) if reduce
                    else sentence_loss.data
                ) if sentence_loss is not None else 0.0
            ),
            'ntokens': ntokens,
            'nsentences': nsentences,
            'sample_size': sample_size,
        }
        return loss, sample_size, logging_output

"""Compute the loss for the given sample.

        Returns a tuple with three elements:
        1) the loss, as a Variable
        2) the sample size, which is used as the denominator for the gradient
        3) logging outputs to display while training
        """
        net_output = model(**sample['net_input'])
        lprobs = model.get_normalized_probs(net_output, log_probs=True)
        lprobs = lprobs.view(-1, lprobs.size(-1))
        target = model.get_targets(sample, net_output).view(-1)
        loss = F.nll_loss(lprobs, target, size_average=False, ignore_index=self.padding_idx,
                          reduce=reduce)
        sample_size = sample['target'].size(0) if self.args.sentence_avg else sample['ntokens']
        logging_output = {
            'loss': utils.item(loss.data) if reduce else loss.data,
            'ntokens': sample['ntokens'],
            'sample_size': sample_size,
        }
        return loss, sample_size, logging_output