Python utils.load_embedding方法代碼示例

# 需要導入模塊: from fairseq import utils [as 別名]
# 或者: from fairseq.utils import load_embedding [as 別名]
def load_embedding(embedding, dictionary, pretrained_embed):
    """Loads pretrained embeddings.

    Loads pretrained embeddings into a nn.Embedding layer. pretrained_embed
    can either be a nn.Embedding layer, in which case the embedding is set
    to the pretrained_embed argument, or a path to an embedding file.

    Arguments:
        embedding (pytorch_translate.common_layers.Embedding):
            Embedding layer whose weights are to be set.
        dictionary (fairseq.data.dictionary.Dictionary): dictionary with the
            same vocabulary size as the embedding argument.
        pretrained_embed (Union(string, nn.Embedding)): source of the
            weights to be loaded.
    """
    if pretrained_embed is None:
        return

    if isinstance(pretrained_embed, torch.nn.Embedding):
        embedding.weight = pretrained_embed.weight
    else:
        embed_dict = utils.parse_embedding(pretrained_embed)
        utils.load_embedding(embed_dict, dictionary, embedding)

    embedding.init_normalization_if_needed() 

# 需要導入模塊: from fairseq import utils [as 別名]
# 或者: from fairseq.utils import load_embedding [as 別名]
def build_embedding(cls, args, dictionary, embed_dim, path=None):
        num_embeddings = len(dictionary)
        padding_idx = dictionary.pad()

        emb = Embedding(num_embeddings, embed_dim, padding_idx)
        # if provided, load from preloaded dictionaries
        if path:
            embed_dict = utils.parse_embedding(path)
            utils.load_embedding(embed_dict, dictionary, emb)
        return emb 

# 需要導入模塊: from fairseq import utils [as 別名]
# 或者: from fairseq.utils import load_embedding [as 別名]
def build_embedding(dictionary, embed_dim, path=None, freeze=False):
    num_embeddings = len(dictionary)
    padding_idx = dictionary.pad()
    emb = TransformerTokenEmbedding(num_embeddings, embed_dim, padding_idx, freeze)
    # if provided, load from preloaded dictionaries
    if path:
        embed_dict = utils.parse_embedding(path)
        utils.load_embedding(embed_dict, dictionary, emb)
    return emb 

# 需要導入模塊: from fairseq import utils [as 別名]
# 或者: from fairseq.utils import load_embedding [as 別名]
def build_embedding(dictionary, embed_dim, path=None):
    num_embeddings = len(dictionary)
    padding_idx = dictionary.pad()
    emb = Embedding(num_embeddings, embed_dim, padding_idx)
    # if provided, load from preloaded dictionaries
    if path:
        embed_dict = utils.parse_embedding(path)
        utils.load_embedding(embed_dict, dictionary, emb)
    return emb 

# 需要導入模塊: from fairseq import utils [as 別名]
# 或者: from fairseq.utils import load_embedding [as 別名]
def build_model(cls, args, task):
        """ Build a new model instance. """
        base_architecture(args)

        if not hasattr(args, 'max_source_positions'):
            args.max_source_positions = 1024
        if not hasattr(args, 'max_target_positions'):
            args.max_target_positions = 1024

        src_dict, tgt_dict = task.source_dictionary, task.target_dictionary

        def build_embedding(dictionary, embed_dim, path=None):
            num_embeddings = len(dictionary)
            padding_idx = dictionary.pad()
            emb = Embedding(num_embeddings, embed_dim, padding_idx)
            # if provided, load from preloaded dictionaries
            if path:
                embed_dict = utils.parse_embedding(path)
                utils.load_embedding(embed_dict, dictionary, emb)
            return emb

        # NOT sharing encoder-decoder embeddings
        encoder_embed_tokens = build_embedding(
            src_dict, args.encoder_embed_dim, args.encoder_embed_path
        )
        decoder_embed_tokens = build_embedding(
            tgt_dict, args.decoder_embed_dim, args.decoder_embed_path
        )
        ctrl_encoder_embed_tokens = build_embedding(
            src_dict, args.encoder_embed_dim, args.encoder_embed_path
        )
        ctrl_decoder_embed_tokens = build_embedding(
            tgt_dict, args.decoder_embed_dim, args.decoder_embed_path
        )

        encoder = Attn2dEncoder(args, src_dict, encoder_embed_tokens, ctrl_encoder_embed_tokens)
        decoder = Attn2dDecoder(args, tgt_dict, decoder_embed_tokens, ctrl_decoder_embed_tokens)

        return cls(encoder, decoder) 

# 需要導入模塊: from fairseq import utils [as 別名]
# 或者: from fairseq.utils import load_embedding [as 別名]
def __init__(
        self, dictionary, embed_dim=512, embed_dict=None, max_positions=1024,
        convolutions=((512, 3),) * 20, dropout=0.1, normalization_constant=0.5,
        left_pad=True,
    ):
        super().__init__(dictionary)
        self.dropout = dropout
        self.normalization_constant = normalization_constant
        self.left_pad = left_pad
        self.num_attention_layers = None

        num_embeddings = len(dictionary)
        self.padding_idx = dictionary.pad()
        self.embed_tokens = Embedding(num_embeddings, embed_dim, self.padding_idx)
        if embed_dict:
            self.embed_tokens = utils.load_embedding(embed_dict, self.dictionary, self.embed_tokens)

        self.embed_positions = PositionalEmbedding(
            max_positions,
            embed_dim,
            self.padding_idx,
            left_pad=self.left_pad,
        )

        convolutions = extend_conv_spec(convolutions)
        in_channels = convolutions[0][0]
        self.fc1 = Linear(embed_dim, in_channels, dropout=dropout)
        self.projections = nn.ModuleList()
        self.convolutions = nn.ModuleList()
        self.residuals = []

        layer_in_channels = [in_channels]
        for i, (out_channels, kernel_size, residual) in enumerate(convolutions):
            if residual == 0:
                residual_dim = out_channels
            else:
                residual_dim = layer_in_channels[-residual]
            self.projections.append(Linear(residual_dim, out_channels)
                                    if residual_dim != out_channels else None)
            if kernel_size % 2 == 1:
                padding = kernel_size // 2
            else:
                padding = 0
            self.convolutions.append(
                ConvTBC(in_channels, out_channels * 2, kernel_size,
                        dropout=dropout, padding=padding)
            )
            self.residuals.append(residual)
            in_channels = out_channels
            layer_in_channels.append(out_channels)
        self.fc2 = Linear(in_channels, embed_dim) 

# 需要導入模塊: from fairseq import utils [as 別名]
# 或者: from fairseq.utils import load_embedding [as 別名]
def __init__(
        self, dictionary, embed_dim=512, embed_dict=None, max_positions=1024,
        convolutions=((512, 3),) * 20, dropout=0.1, normalization_constant=0.5,
        left_pad=True, token_dropout=0.0
    ):
        super().__init__(dictionary)
        self.dropout = dropout
        self.token_dropout = token_dropout
        self.normalization_constant = normalization_constant
        self.left_pad = left_pad
        self.num_attention_layers = None

        num_embeddings = len(dictionary)
        self.padding_idx = dictionary.pad()
        self.embed_tokens = Embedding(num_embeddings, embed_dim, self.padding_idx)
        if embed_dict:
            self.embed_tokens = utils.load_embedding(embed_dict, self.dictionary, self.embed_tokens)

        self.embed_positions = PositionalEmbedding(
            max_positions,
            embed_dim,
            self.padding_idx,
            left_pad=self.left_pad,
        )

        convolutions = extend_conv_spec(convolutions)
        in_channels = convolutions[0][0]
        self.fc1 = Linear(embed_dim, in_channels, dropout=dropout)
        self.projections = nn.ModuleList()
        self.convolutions = nn.ModuleList()
        self.residuals = []

        layer_in_channels = [in_channels]
        for i, (out_channels, kernel_size, residual) in enumerate(convolutions):
            if residual == 0:
                residual_dim = out_channels
            else:
                residual_dim = layer_in_channels[-residual]
            self.projections.append(Linear(residual_dim, out_channels)
                                    if residual_dim != out_channels else None)
            if kernel_size % 2 == 1:
                padding = kernel_size // 2
            else:
                padding = 0
            self.convolutions.append(
                ConvTBC(in_channels, out_channels * 2, kernel_size,
                        dropout=dropout, padding=padding)
            )
            self.residuals.append(residual)
            in_channels = out_channels
            layer_in_channels.append(out_channels)
        self.fc2 = Linear(in_channels, embed_dim) 

# 需要導入模塊: from fairseq import utils [as 別名]
# 或者: from fairseq.utils import load_embedding [as 別名]
def build_model(cls, args, task):
        """Build a new model instance."""
        # make sure that all args are properly defaulted (in case there are any new ones)
        base_architecture(args)

        def load_pretrained_embedding_from_file(embed_path, dictionary, embed_dim):
            num_embeddings = len(dictionary)
            padding_idx = dictionary.pad()
            embed_tokens = Embedding(num_embeddings, embed_dim, padding_idx)
            embed_dict = utils.parse_embedding(embed_path)
            utils.print_embed_overlap(embed_dict, dictionary)
            return utils.load_embedding(embed_dict, dictionary, embed_tokens)

        pretrained_encoder_embed = None
        if args.encoder_embed_path:
            pretrained_encoder_embed = load_pretrained_embedding_from_file(
                args.encoder_embed_path, task.source_dictionary, args.encoder_embed_dim)
        pretrained_decoder_embed = None
        if args.decoder_embed_path:
            pretrained_decoder_embed = load_pretrained_embedding_from_file(
                args.decoder_embed_path, task.target_dictionary, args.decoder_embed_dim)

        encoder = LSTMEncoder(
            dictionary=task.source_dictionary,
            embed_dim=args.encoder_embed_dim,
            hidden_size=args.encoder_hidden_size,
            num_layers=args.encoder_layers,
            dropout_in=args.encoder_dropout_in,
            dropout_out=args.encoder_dropout_out,
            bidirectional=args.encoder_bidirectional,
            pretrained_embed=pretrained_encoder_embed,
        )
        decoder = LSTMDecoder(
            dictionary=task.target_dictionary,
            embed_dim=args.decoder_embed_dim,
            hidden_size=args.decoder_hidden_size,
            out_embed_dim=args.decoder_out_embed_dim,
            num_layers=args.decoder_layers,
            dropout_in=args.decoder_dropout_in,
            dropout_out=args.decoder_dropout_out,
            attention=options.eval_bool(args.decoder_attention),
            encoder_embed_dim=args.encoder_embed_dim,
            encoder_output_units=encoder.output_units,
            pretrained_embed=pretrained_decoder_embed,
        )
        return cls(encoder, decoder) 

# 需要導入模塊: from fairseq import utils [as 別名]
# 或者: from fairseq.utils import load_embedding [as 別名]
def __init__(
        self, dictionary, embed_dim=512, embed_dict=None, max_positions=1024,
        convolutions=((512, 3),) * 20, dropout=0.1,
    ):
        super().__init__(dictionary)
        self.dropout = dropout
        self.num_attention_layers = None

        num_embeddings = len(dictionary)
        self.padding_idx = dictionary.pad()
        self.embed_tokens = Embedding(num_embeddings, embed_dim, self.padding_idx)
        if embed_dict:
            self.embed_tokens = utils.load_embedding(embed_dict, self.dictionary, self.embed_tokens)

        self.embed_positions = PositionalEmbedding(
            max_positions,
            embed_dim,
            self.padding_idx,
        )

        convolutions = extend_conv_spec(convolutions)
        in_channels = convolutions[0][0]
        self.fc1 = Linear(embed_dim, in_channels, dropout=dropout)
        self.projections = nn.ModuleList()
        self.convolutions = nn.ModuleList()
        self.residuals = []

        layer_in_channels = [in_channels]
        for _, (out_channels, kernel_size, residual) in enumerate(convolutions):
            if residual == 0:
                residual_dim = out_channels
            else:
                residual_dim = layer_in_channels[-residual]
            self.projections.append(Linear(residual_dim, out_channels)
                                    if residual_dim != out_channels else None)
            if kernel_size % 2 == 1:
                padding = kernel_size // 2
            else:
                padding = 0
            self.convolutions.append(
                ConvTBC(in_channels, out_channels * 2, kernel_size,
                        dropout=dropout, padding=padding)
            )
            self.residuals.append(residual)
            in_channels = out_channels
            layer_in_channels.append(out_channels)
        self.fc2 = Linear(in_channels, embed_dim) 

# 需要導入模塊: from fairseq import utils [as 別名]
# 或者: from fairseq.utils import load_embedding [as 別名]
def build_model(cls, args, task):
        """Build a new model instance."""

        # make sure all arguments are present in older models
        base_architecture(args)

        if getattr(args, 'max_target_positions', None) is not None:
            max_target_positions = args.max_target_positions
        else:
            max_target_positions = getattr(args, 'tokens_per_sample', DEFAULT_MAX_TARGET_POSITIONS)

        def load_pretrained_embedding_from_file(embed_path, dictionary, embed_dim):
            num_embeddings = len(dictionary)
            padding_idx = dictionary.pad()
            embed_tokens = Embedding(num_embeddings, embed_dim, padding_idx)
            embed_dict = utils.parse_embedding(embed_path)
            utils.print_embed_overlap(embed_dict, dictionary)
            return utils.load_embedding(embed_dict, dictionary, embed_tokens)

        pretrained_decoder_embed = None
        if args.decoder_embed_path:
            pretrained_decoder_embed = load_pretrained_embedding_from_file(
                args.decoder_embed_path,
                task.target_dictionary,
                args.decoder_embed_dim
            )

        if args.share_decoder_input_output_embed:
            # double check all parameters combinations are valid
            if task.source_dictionary != task.target_dictionary:
                raise ValueError('--share-decoder-input-output-embeddings requires a joint dictionary')

            if args.decoder_embed_dim != args.decoder_out_embed_dim:
                raise ValueError(
                    '--share-decoder-input-output-embeddings requires '
                    '--decoder-embed-dim to match --decoder-out-embed-dim'
                    )

        decoder = LSTMDecoder(
            dictionary=task.dictionary,
            embed_dim=args.decoder_embed_dim,
            hidden_size=args.decoder_hidden_size,
            out_embed_dim=args.decoder_out_embed_dim,
            num_layers=args.decoder_layers,
            dropout_in=args.decoder_dropout_in,
            dropout_out=args.decoder_dropout_out,
            attention=False,  # decoder-only language model doesn't support attention
            encoder_output_units=0,
            pretrained_embed=pretrained_decoder_embed,
            share_input_output_embed=args.share_decoder_input_output_embed,
            adaptive_softmax_cutoff=(
                options.eval_str_list(args.adaptive_softmax_cutoff, type=int)
                if args.criterion == 'adaptive_loss' else None
            ),
            max_target_positions=max_target_positions,
            residuals=args.residuals
        )

        return cls(decoder) 

# 需要導入模塊: from fairseq import utils [as 別名]
# 或者: from fairseq.utils import load_embedding [as 別名]
def build_model(cls, args, task):
        """Build a new model instance."""

        # make sure all arguments are present in older models
        base_architecture(args)

        if not hasattr(args, 'max_source_positions'):
            args.max_source_positions = 1024
        if not hasattr(args, 'max_target_positions'):
            args.max_target_positions = 1024

        src_dict, tgt_dict = task.source_dictionary, task.target_dictionary

        def build_embedding(dictionary, embed_dim, path=None):
            num_embeddings = len(dictionary)
            padding_idx = dictionary.pad()
            emb = Embedding(num_embeddings, embed_dim, padding_idx)
            # if provided, load from preloaded dictionaries
            if path:
                embed_dict = utils.parse_embedding(path)
                utils.load_embedding(embed_dict, dictionary, emb)
            return emb

        if args.share_all_embeddings:
            if src_dict != tgt_dict:
                raise RuntimeError('--share-all-embeddings requires a joined dictionary')
            if args.encoder_embed_dim != args.decoder_embed_dim:
                raise RuntimeError(
                    '--share-all-embeddings requires --encoder-embed-dim to match --decoder-embed-dim')
            if args.decoder_embed_path and (
                    args.decoder_embed_path != args.encoder_embed_path):
                raise RuntimeError('--share-all-embeddings not compatible with --decoder-embed-path')
            encoder_embed_tokens = build_embedding(
                src_dict, args.encoder_embed_dim, args.encoder_embed_path
            )
            decoder_embed_tokens = encoder_embed_tokens
            args.share_decoder_input_output_embed = True
        else:
            encoder_embed_tokens = build_embedding(
                src_dict, args.encoder_embed_dim, args.encoder_embed_path
            )
            decoder_embed_tokens = build_embedding(
                tgt_dict, args.decoder_embed_dim, args.decoder_embed_path
            )

        encoder = LightConvEncoder(args, src_dict, encoder_embed_tokens)
        decoder = LightConvDecoder(args, tgt_dict, decoder_embed_tokens)
        return LightConvModel(encoder, decoder) 

# 需要導入模塊: from fairseq import utils [as 別名]
# 或者: from fairseq.utils import load_embedding [as 別名]
def build_model(cls, args, task):
        """Build a new model instance."""

        # make sure all arguments are present in older models
        base_architecture(args)

        if not hasattr(args, "max_source_positions"):
            args.max_source_positions = DEFAULT_MAX_SOURCE_POSITIONS
        if not hasattr(args, "max_target_positions"):
            args.max_target_positions = DEFAULT_MAX_TARGET_POSITIONS

        src_dict, tgt_dict = task.source_dictionary, task.target_dictionary

        def build_embedding(dictionary, embed_dim, path=None):
            num_embeddings = len(dictionary)
            padding_idx = dictionary.pad()
            emb = Embedding(num_embeddings, embed_dim, padding_idx)
            # if provided, load from preloaded dictionaries
            if path:
                embed_dict = utils.parse_embedding(path)
                utils.load_embedding(embed_dict, dictionary, emb)
            return emb

        if args.share_all_embeddings:
            if src_dict != tgt_dict:
                raise ValueError("--share-all-embeddings requires a joined dictionary")
            if args.encoder_embed_dim != args.decoder_embed_dim:
                raise ValueError(
                    "--share-all-embeddings requires --encoder-embed-dim "
                    "to match --decoder-embed-dim"
                )
            if args.decoder_embed_path and (
                args.decoder_embed_path != args.encoder_embed_path
            ):
                raise ValueError(
                    "--share-all-embeddings not compatible with --decoder-embed-path"
                )
            encoder_embed_tokens = build_embedding(
                src_dict, args.encoder_embed_dim, args.encoder_embed_path
            )
            decoder_embed_tokens = encoder_embed_tokens
            args.share_decoder_input_output_embed = True
        else:
            encoder_embed_tokens = build_embedding(
                src_dict, args.encoder_embed_dim, args.encoder_embed_path
            )
            decoder_embed_tokens = build_embedding(
                tgt_dict, args.decoder_embed_dim, args.decoder_embed_path
            )

        encoder = cls.build_encoder(args, src_dict, encoder_embed_tokens)
        decoder = cls.build_decoder(args, tgt_dict, decoder_embed_tokens)
        return TwoPhaseTransformerModel(encoder, decoder) 

# 需要導入模塊: from fairseq import utils [as 別名]
# 或者: from fairseq.utils import load_embedding [as 別名]
def build_model(cls, args, task):
        """Build a new model instance."""

        # make sure all arguments are present in older models
        base_architecture(args)

        if not hasattr(args, "max_source_positions"):
            args.max_source_positions = 1024
        if not hasattr(args, "max_target_positions"):
            args.max_target_positions = 1024

        src_dict, tgt_dict = task.source_dictionary, task.target_dictionary

        def build_embedding(dictionary, embed_dim, path=None):
            num_embeddings = len(dictionary)
            padding_idx = dictionary.pad()
            emb = Embedding(num_embeddings, embed_dim, padding_idx)
            # if provided, load from preloaded dictionaries
            if path:
                embed_dict = utils.parse_embedding(path)
                utils.load_embedding(embed_dict, dictionary, emb)
            return emb

        if args.share_all_embeddings:
            if src_dict != tgt_dict:
                raise RuntimeError(
                    "--share-all-embeddings requires a joined dictionary"
                )
            if args.encoder_embed_dim != args.decoder_embed_dim:
                raise RuntimeError(
                    """--share-all-embeddings requires --encoder-embed-dim \
                    to match --decoder-embed-dim"""
                )
            if args.decoder_embed_path and (
                args.decoder_embed_path != args.encoder_embed_path
            ):
                raise RuntimeError(
                    "--share-all-embeddings not compatible with --decoder-embed-path"
                )
            encoder_embed_tokens = build_embedding(
                src_dict, args.encoder_embed_dim, args.encoder_embed_path
            )
            decoder_embed_tokens = encoder_embed_tokens
            args.share_decoder_input_output_embed = True
        else:
            encoder_embed_tokens = build_embedding(
                src_dict, args.encoder_embed_dim, args.encoder_embed_path
            )
            decoder_embed_tokens = build_embedding(
                tgt_dict, args.decoder_embed_dim, args.decoder_embed_path
            )

        encoder = pytorch_translate_transformer.TransformerEncoder(
            args, src_dict, encoder_embed_tokens
        )
        decoder = TransformerAANDecoder(args, src_dict, tgt_dict, decoder_embed_tokens)
        return TransformerAANModel(task, encoder, decoder) 

# 需要導入模塊: from fairseq import utils [as 別名]
# 或者: from fairseq.utils import load_embedding [as 別名]
def build_model(cls, args, task):
        """Build a new model instance."""

        # make sure all arguments are present in older models
        base_architecture(args)

        if not hasattr(args, 'max_source_positions'):
            args.max_source_positions = 1024
        if not hasattr(args, 'max_target_positions'):
            args.max_target_positions = 1024
        transformer_print(key=mlperf_log.INPUT_MAX_LENGTH, value=args.max_source_positions)
        src_dict, tgt_dict = task.source_dictionary, task.target_dictionary

        def build_embedding(dictionary, embed_dim, path=None):
            num_embeddings = len(dictionary)
            padding_idx = dictionary.pad()
            emb = Embedding(num_embeddings, embed_dim, padding_idx)
            # if provided, load from preloaded dictionaries
            if path:
                embed_dict = utils.parse_embedding(path)
                utils.load_embedding(embed_dict, dictionary, emb)
            return emb

        if args.share_all_embeddings:
            if src_dict != tgt_dict:
                raise RuntimeError('--share-all-embeddings requires a joined dictionary')
            if args.encoder_embed_dim != args.decoder_embed_dim:
                raise RuntimeError(
                    '--share-all-embeddings requires --encoder-embed-dim to match --decoder-embed-dim')
            if args.decoder_embed_path and (
                    args.decoder_embed_path != args.encoder_embed_path):
                raise RuntimeError('--share-all-embeddings not compatible with --decoder-embed-path')
            transformer_print(key=mlperf_log.MODEL_HP_EMBEDDING_SHARED_WEIGHTS,
                    value={'hidden_size':args.encoder_embed_dim, 'vocab_size':len(src_dict)})
            encoder_embed_tokens = build_embedding(
                src_dict, args.encoder_embed_dim, args.encoder_embed_path
            )
            decoder_embed_tokens = encoder_embed_tokens
            args.share_decoder_input_output_embed = True
        else:
            encoder_embed_tokens = build_embedding(
                src_dict, args.encoder_embed_dim, args.encoder_embed_path
            )
            decoder_embed_tokens = build_embedding(
                tgt_dict, args.decoder_embed_dim, args.decoder_embed_path
            )

        encoder = TransformerEncoder(args, src_dict, encoder_embed_tokens)
        decoder = TransformerDecoder(args, tgt_dict, decoder_embed_tokens)
        return TransformerModel(encoder, decoder) 

# 需要導入模塊: from fairseq import utils [as 別名]
# 或者: from fairseq.utils import load_embedding [as 別名]
def build_model(cls, args, task):
        """Build a new model instance."""

        # make sure all arguments are present in older models
        base_architecture(args)

        if not hasattr(args, 'max_source_positions'):
            args.max_source_positions = 1024
        if not hasattr(args, 'max_target_positions'):
            args.max_target_positions = 1024

        src_dict, tgt_dict = task.source_dictionary, task.target_dictionary

        def build_embedding(dictionary, embed_dim, path=None):
            num_embeddings = len(dictionary)
            padding_idx = dictionary.pad()
            emb = Embedding(num_embeddings, embed_dim, padding_idx)
            # if provided, load from preloaded dictionaries
            if path:
                embed_dict = utils.parse_embedding(path)
                utils.load_embedding(embed_dict, dictionary, emb)
            return emb

        if args.share_all_embeddings:
            if src_dict != tgt_dict:
                raise RuntimeError('--share-all-embeddings requires a joined dictionary')
            if args.encoder_embed_dim != args.decoder_embed_dim:
                raise RuntimeError(
                    '--share-all-embeddings requires --encoder-embed-dim to match --decoder-embed-dim')
            if args.decoder_embed_path and (
                    args.decoder_embed_path != args.encoder_embed_path):
                raise RuntimeError('--share-all-embeddings not compatible with --decoder-embed-path')
            encoder_embed_tokens = build_embedding(
                src_dict, args.encoder_embed_dim, args.encoder_embed_path
            )
            decoder_embed_tokens = encoder_embed_tokens
            args.share_decoder_input_output_embed = True
        else:
            encoder_embed_tokens = build_embedding(
                src_dict, args.encoder_embed_dim, args.encoder_embed_path
            )
            decoder_embed_tokens = build_embedding(
                tgt_dict, args.decoder_embed_dim, args.decoder_embed_path
            )

        encoder = TransformerEncoder(args, src_dict, encoder_embed_tokens)
        decoder = TransformerDecoder(args, tgt_dict, decoder_embed_tokens)
        return TransformerModel(encoder, decoder)