src/mlm/models/__init__.py [188:210]:
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
            kwargs = {
                'dataset_name': dataset_prefix + dataset_suffix,
                'pretrained': True,
                'ctx': ctxs,
                'use_pooler': False,
                'use_decoder': False,
                'use_classifier': False
            }
            if finetune or regression:
                kwargs['use_pooler'] = True
            else:
                kwargs['use_decoder'] = True
            # Override GluonNLP's default location?
            if root is not None:
                kwargs['root'] = str(root)
            model, vocab = get_model(model_fullname, **kwargs)

            # Freeze initial layers if needed
            for i in range(freeze):
                model.encoder.transformer_cells[i].collect_params().setattr('grad_req', 'null')

            # Wrapper if appropriate
            if regression:
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -



src/mlm/models/__init__.py [241:263]:
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
            kwargs = {
                'dataset_name': dataset_prefix + dataset_suffix,
                'pretrained': True,
                'ctx': ctxs,
                'use_pooler': False,
                'use_decoder': False,
                'use_classifier': False
            }
            if finetune or regression:
                kwargs['use_pooler'] = True
            else:
                kwargs['use_decoder'] = True
            # Override GluonNLP's default location?
            if root is not None:
                kwargs['root'] = str(root)
            model, vocab = get_model(model_fullname, **kwargs)

            # Freeze initial layers if needed
            for i in range(freeze):
                model.encoder.transformer_cells[i].collect_params().setattr('grad_req', 'null')

            # Wrapper if appropriate
            if regression:
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -