marian

mindnlp.transformers.models.marian.configuration_marian

Marian model configuration

mindnlp.transformers.models.marian.configuration_marian.MarianConfig

Bases: PretrainedConfig

This is the configuration class to store the configuration of a [MarianModel]. It is used to instantiate an Marian model according to the specified arguments, defining the model architecture. Instantiating a configuration with the defaults will yield a similar configuration to that of the Marian Helsinki-NLP/opus-mt-en-de architecture.

Configuration objects inherit from [PretrainedConfig] and can be used to control the model outputs. Read the documentation from [PretrainedConfig] for more information.

PARAMETER	DESCRIPTION
vocab_size	Vocabulary size of the Marian model. Defines the number of different tokens that can be represented by the inputs_ids passed when calling [MarianModel] or [TFMarianModel]. TYPE: `int`, *optional*, defaults to 58101 DEFAULT: 58101
d_model	Dimensionality of the layers and the pooler layer. TYPE: `int`, *optional*, defaults to 1024 DEFAULT: 1024
encoder_layers	Number of encoder layers. TYPE: `int`, *optional*, defaults to 12 DEFAULT: 12
decoder_layers	Number of decoder layers. TYPE: `int`, *optional*, defaults to 12 DEFAULT: 12
encoder_attention_heads	Number of attention heads for each attention layer in the Transformer encoder. TYPE: `int`, *optional*, defaults to 16 DEFAULT: 16
decoder_attention_heads	Number of attention heads for each attention layer in the Transformer decoder. TYPE: `int`, *optional*, defaults to 16 DEFAULT: 16
decoder_ffn_dim	Dimensionality of the "intermediate" (often named feed-forward) layer in decoder. TYPE: `int`, *optional*, defaults to 4096 DEFAULT: 4096
encoder_ffn_dim	Dimensionality of the "intermediate" (often named feed-forward) layer in decoder. TYPE: `int`, *optional*, defaults to 4096 DEFAULT: 4096
activation_function	The non-linear activation function (function or string) in the encoder and pooler. If string, "gelu", "relu", "silu" and "gelu_new" are supported. TYPE: `str` or `function`, *optional*, defaults to `"gelu"` DEFAULT: 'gelu'
dropout	The dropout probability for all fully connected layers in the embeddings, encoder, and pooler. TYPE: `float`, *optional*, defaults to 0.1 DEFAULT: 0.1
attention_dropout	The dropout ratio for the attention probabilities. TYPE: `float`, *optional*, defaults to 0.0 DEFAULT: 0.0
activation_dropout	The dropout ratio for activations inside the fully connected layer. TYPE: `float`, *optional*, defaults to 0.0 DEFAULT: 0.0
max_position_embeddings	The maximum sequence length that this model might ever be used with. Typically set this to something large just in case (e.g., 512 or 1024 or 2048). TYPE: `int`, *optional*, defaults to 1024 DEFAULT: 1024
init_std	The standard deviation of the truncated_normal_initializer for initializing all weight matrices. TYPE: `float`, *optional*, defaults to 0.02 DEFAULT: 0.02
encoder_layerdrop	The LayerDrop probability for the encoder. See the LayerDrop paper for more details. TYPE: `float`, *optional*, defaults to 0.0 DEFAULT: 0.0
decoder_layerdrop	The LayerDrop probability for the decoder. See the LayerDrop paper for more details. TYPE: `float`, *optional*, defaults to 0.0 DEFAULT: 0.0
scale_embedding	Scale embeddings by diving by sqrt(d_model). TYPE: `bool`, *optional*, defaults to `False` DEFAULT: False
use_cache	Whether or not the model should return the last key/values attentions (not used by all models) TYPE: `bool`, *optional*, defaults to `True` DEFAULT: True
forced_eos_token_id	The id of the token to force as the last generated token when max_length is reached. Usually set to eos_token_id. TYPE: `int`, *optional*, defaults to 0 DEFAULT: 0

Example

>>> from transformers import MarianModel, MarianConfig
...
>>> # Initializing a Marian Helsinki-NLP/opus-mt-en-de style configuration
>>> configuration = MarianConfig()
...
>>> # Initializing a model from the Helsinki-NLP/opus-mt-en-de style configuration
>>> model = MarianModel(configuration)
...
>>> # Accessing the model configuration
>>> configuration = model.config

Source code in mindnlp/transformers/models/marian/configuration_marian.py

class MarianConfig(PretrainedConfig):
    r"""
    This is the configuration class to store the configuration of a [`MarianModel`]. It is used to instantiate an
    Marian model according to the specified arguments, defining the model architecture. Instantiating a configuration
    with the defaults will yield a similar configuration to that of the Marian
    [Helsinki-NLP/opus-mt-en-de](https://huggingface.co/Helsinki-NLP/opus-mt-en-de) architecture.

    Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the
    documentation from [`PretrainedConfig`] for more information.


    Args:
        vocab_size (`int`, *optional*, defaults to 58101):
            Vocabulary size of the Marian model. Defines the number of different tokens that can be represented by the
            `inputs_ids` passed when calling [`MarianModel`] or [`TFMarianModel`].
        d_model (`int`, *optional*, defaults to 1024):
            Dimensionality of the layers and the pooler layer.
        encoder_layers (`int`, *optional*, defaults to 12):
            Number of encoder layers.
        decoder_layers (`int`, *optional*, defaults to 12):
            Number of decoder layers.
        encoder_attention_heads (`int`, *optional*, defaults to 16):
            Number of attention heads for each attention layer in the Transformer encoder.
        decoder_attention_heads (`int`, *optional*, defaults to 16):
            Number of attention heads for each attention layer in the Transformer decoder.
        decoder_ffn_dim (`int`, *optional*, defaults to 4096):
            Dimensionality of the "intermediate" (often named feed-forward) layer in decoder.
        encoder_ffn_dim (`int`, *optional*, defaults to 4096):
            Dimensionality of the "intermediate" (often named feed-forward) layer in decoder.
        activation_function (`str` or `function`, *optional*, defaults to `"gelu"`):
            The non-linear activation function (function or string) in the encoder and pooler. If string, `"gelu"`,
            `"relu"`, `"silu"` and `"gelu_new"` are supported.
        dropout (`float`, *optional*, defaults to 0.1):
            The dropout probability for all fully connected layers in the embeddings, encoder, and pooler.
        attention_dropout (`float`, *optional*, defaults to 0.0):
            The dropout ratio for the attention probabilities.
        activation_dropout (`float`, *optional*, defaults to 0.0):
            The dropout ratio for activations inside the fully connected layer.
        max_position_embeddings (`int`, *optional*, defaults to 1024):
            The maximum sequence length that this model might ever be used with. Typically set this to something large
            just in case (e.g., 512 or 1024 or 2048).
        init_std (`float`, *optional*, defaults to 0.02):
            The standard deviation of the truncated_normal_initializer for initializing all weight matrices.
        encoder_layerdrop (`float`, *optional*, defaults to 0.0):
            The LayerDrop probability for the encoder. See the [LayerDrop paper](see https://arxiv.org/abs/1909.11556)
            for more details.
        decoder_layerdrop (`float`, *optional*, defaults to 0.0):
            The LayerDrop probability for the decoder. See the [LayerDrop paper](see https://arxiv.org/abs/1909.11556)
            for more details.
        scale_embedding (`bool`, *optional*, defaults to `False`):
            Scale embeddings by diving by sqrt(d_model).
        use_cache (`bool`, *optional*, defaults to `True`):
            Whether or not the model should return the last key/values attentions (not used by all models)
        forced_eos_token_id (`int`, *optional*, defaults to 0):
            The id of the token to force as the last generated token when `max_length` is reached. Usually set to
            `eos_token_id`.

    Example:
        ```python
        >>> from transformers import MarianModel, MarianConfig
        ...
        >>> # Initializing a Marian Helsinki-NLP/opus-mt-en-de style configuration
        >>> configuration = MarianConfig()
        ...
        >>> # Initializing a model from the Helsinki-NLP/opus-mt-en-de style configuration
        >>> model = MarianModel(configuration)
        ...
        >>> # Accessing the model configuration
        >>> configuration = model.config
        ```
    """

    model_type = "marian"
    keys_to_ignore_at_inference = ["past_key_values"]
    attribute_map = {"num_attention_heads": "encoder_attention_heads", "hidden_size": "d_model"}

    def __init__(
        self,
        vocab_size=58101,
        decoder_vocab_size=None,
        max_position_embeddings=1024,
        encoder_layers=12,
        encoder_ffn_dim=4096,
        encoder_attention_heads=16,
        decoder_layers=12,
        decoder_ffn_dim=4096,
        decoder_attention_heads=16,
        encoder_layerdrop=0.0,
        decoder_layerdrop=0.0,
        use_cache=True,
        is_encoder_decoder=True,
        activation_function="gelu",
        d_model=1024,
        dropout=0.1,
        attention_dropout=0.0,
        activation_dropout=0.0,
        init_std=0.02,
        decoder_start_token_id=58100,
        scale_embedding=False,
        pad_token_id=58100,
        eos_token_id=0,
        forced_eos_token_id=0,
        share_encoder_decoder_embeddings=True,
        **kwargs,
    ):
        self.vocab_size = vocab_size
        self.decoder_vocab_size = decoder_vocab_size or vocab_size
        self.max_position_embeddings = max_position_embeddings
        self.d_model = d_model
        self.encoder_ffn_dim = encoder_ffn_dim
        self.encoder_layers = encoder_layers
        self.encoder_attention_heads = encoder_attention_heads
        self.decoder_ffn_dim = decoder_ffn_dim
        self.decoder_layers = decoder_layers
        self.decoder_attention_heads = decoder_attention_heads
        self.dropout = dropout
        self.attention_dropout = attention_dropout
        self.activation_dropout = activation_dropout
        self.activation_function = activation_function
        self.init_std = init_std
        self.encoder_layerdrop = encoder_layerdrop
        self.decoder_layerdrop = decoder_layerdrop
        self.use_cache = use_cache
        self.num_hidden_layers = encoder_layers
        self.scale_embedding = scale_embedding  # scale factor will be sqrt(d_model) if True
        self.share_encoder_decoder_embeddings = share_encoder_decoder_embeddings
        super().__init__(
            pad_token_id=pad_token_id,
            eos_token_id=eos_token_id,
            is_encoder_decoder=is_encoder_decoder,
            decoder_start_token_id=decoder_start_token_id,
            forced_eos_token_id=forced_eos_token_id,
            **kwargs,
        )

mindnlp.transformers.models.marian.modeling_marian

Mindspore MarianMTModel model, ported from the Marian C++ repo.

mindnlp.transformers.models.marian.modeling_marian.MarianAttention

Bases: Module

Multi-headed attention from 'Attention Is All You Need' paper

Source code in mindnlp/transformers/models/marian/modeling_marian.py

class MarianAttention(nn.Module):
    """Multi-headed attention from 'Attention Is All You Need' paper"""
    def __init__(
        self,
        embed_dim: int,
        num_heads: int,
        dropout: float = 0.0,
        is_decoder: bool = False,
        bias: bool = True,
        is_causal: bool = False,
        config: Optional[MarianConfig] = None,
    ):
        """
        Initializes an instance of the BartAttention class.

        Args:
            embed_dim (int): The dimensionality of the input embeddings.
            num_heads (int): The number of attention heads.
            dropout (float, optional): The dropout probability. Defaults to 0.0.
            is_decoder (bool, optional): Whether the attention layer is used as a decoder. Defaults to False.
            bias (bool, optional): Whether to include bias terms in the linear projections. Defaults to True.
            is_causal (bool, optional): Whether the attention layer is causal. Defaults to False.
            config (Optional[BartConfig], optional): An optional BART configuration object. Defaults to None.

        Returns:
            None

        Raises:
            ValueError: If `embed_dim` is not divisible by `num_heads`.
        """
        super().__init__()
        self.embed_dim = embed_dim
        self.num_heads = num_heads
        self.dropout = dropout
        self.head_dim = embed_dim // num_heads
        self.config = config

        if (self.head_dim * num_heads) != self.embed_dim:
            raise ValueError(
                f"embed_dim must be divisible by num_heads (got `embed_dim`: {self.embed_dim}"
                f" and `num_heads`: {num_heads})."
            )
        self.scaling = self.head_dim**-0.5
        self.is_decoder = is_decoder
        self.is_causal = is_causal

        self.k_proj = nn.Linear(embed_dim, embed_dim, bias=bias)
        self.v_proj = nn.Linear(embed_dim, embed_dim, bias=bias)
        self.q_proj = nn.Linear(embed_dim, embed_dim, bias=bias)
        self.out_proj = nn.Linear(embed_dim, embed_dim, bias=bias)

    def _shape(self, tensor: mindspore.Tensor, seq_len: int, bsz: int):
        """
        This method _shape is defined within the class BartAttention.

        Args:
            self: The instance of the BartAttention class.
            tensor (mindspore.Tensor): The input tensor to be reshaped. It should be a multi-dimensional tensor.
            seq_len (int): The length of the sequence. It should be a positive integer.
            bsz (int): The batch size. It should be a positive integer.

        Returns:
            None.

        Raises:
            None
        """
        return tensor.view(bsz, seq_len, self.num_heads, self.head_dim).swapaxes(1, 2)

    def forward(
        self,
        hidden_states: mindspore.Tensor,
        key_value_states: Optional[mindspore.Tensor] = None,
        past_key_value: Optional[Tuple[mindspore.Tensor]] = None,
        attention_mask: Optional[mindspore.Tensor] = None,
        layer_head_mask: Optional[mindspore.Tensor] = None,
        output_attentions: bool = False,
    ) -> Tuple[mindspore.Tensor, Optional[mindspore.Tensor], Optional[Tuple[mindspore.Tensor]]]:
        """Input shape: Batch x Time x Channel"""
        # if key_value_states are provided this layer is used as a cross-attention layer
        # for the decoder
        is_cross_attention = key_value_states is not None

        bsz, tgt_len, _ = hidden_states.shape

        # get query proj
        query_states = self.q_proj(hidden_states) * self.scaling
        # get key, value proj
        # `past_key_value[0].shape[2] == key_value_states.shape[1]`
        # is checking that the `sequence_length` of the `past_key_value` is the same as
        # the provided `key_value_states` to support prefix tuning
        if (
            is_cross_attention
            and past_key_value is not None
            and past_key_value[0].shape[2] == key_value_states.shape[1]
        ):
            # reuse k,v, cross_attentions
            key_states = past_key_value[0]
            value_states = past_key_value[1]
        elif is_cross_attention:
            # cross_attentions
            key_states = self._shape(self.k_proj(key_value_states), -1, bsz)
            value_states = self._shape(self.v_proj(key_value_states), -1, bsz)
        elif past_key_value is not None:
            # reuse k, v, self_attention
            key_states = self._shape(self.k_proj(hidden_states), -1, bsz)
            value_states = self._shape(self.v_proj(hidden_states), -1, bsz)
            key_states = ops.cat([past_key_value[0], key_states], axis=2)
            value_states = ops.cat([past_key_value[1], value_states], axis=2)
        else:
            # self_attention
            key_states = self._shape(self.k_proj(hidden_states), -1, bsz)
            value_states = self._shape(self.v_proj(hidden_states), -1, bsz)

        if self.is_decoder:
            # if cross_attention save Tuple(mindspore.Tensor, mindspore.Tensor) of all cross attention key/value_states.
            # Further calls to cross_attention layer can then reuse all cross-attention
            # key/value_states (first "if" case)
            # if uni-directional self-attention (decoder) save Tuple(mindspore.Tensor, mindspore.Tensor) of
            # all previous decoder key/value_states. Further calls to uni-directional self-attention
            # can concat previous decoder key/value_states to current projected key/value_states (third "elif" case)
            # if encoder bi-directional self-attention `past_key_value` is always `None`
            past_key_value = (key_states, value_states)

        proj_shape = (bsz * self.num_heads, -1, self.head_dim)
        query_states = self._shape(query_states, tgt_len, bsz).view(*proj_shape)
        key_states = key_states.reshape(*proj_shape)
        value_states = value_states.reshape(*proj_shape)

        src_len = key_states.shape[1]
        attn_weights = ops.bmm(query_states, key_states.swapaxes(1, 2))

        if attn_weights.shape != (bsz * self.num_heads, tgt_len, src_len):
            raise ValueError(
                f"Attention weights should be of size {(bsz * self.num_heads, tgt_len, src_len)}, but is"
                f" {attn_weights.shape}"
            )

        if attention_mask is not None:
            if attention_mask.shape != (bsz, 1, tgt_len, src_len):
                raise ValueError(
                    f"Attention mask should be of size {(bsz, 1, tgt_len, src_len)}, but is {attention_mask.shape}"
                )
            attn_weights = attn_weights.view(bsz, self.num_heads, tgt_len, src_len) + attention_mask
            attn_weights = attn_weights.view(bsz * self.num_heads, tgt_len, src_len)

        attn_weights = ops.softmax(attn_weights, axis=-1)

        if layer_head_mask is not None:
            if layer_head_mask.shape != (self.num_heads,):
                raise ValueError(
                    f"Head mask for a single layer should be of size {(self.num_heads,)}, but is"
                    f" {layer_head_mask.shape}"
                )
            attn_weights = layer_head_mask.view(1, -1, 1, 1) * attn_weights.view(bsz, self.num_heads, tgt_len, src_len)
            attn_weights = attn_weights.view(bsz * self.num_heads, tgt_len, src_len)

        if output_attentions:
            # this operation is a bit awkward, but it's required to
            # make sure that attn_weights keeps its gradient.
            # In order to do so, attn_weights have to be reshaped
            # twice and have to be reused in the following
            attn_weights_reshaped = attn_weights.view(bsz, self.num_heads, tgt_len, src_len)
            attn_weights = attn_weights_reshaped.view(bsz * self.num_heads, tgt_len, src_len)
        else:
            attn_weights_reshaped = None

        attn_probs = ops.dropout(attn_weights, p=self.dropout, training=self.training)

        attn_output = ops.bmm(attn_probs, value_states)

        if attn_output.shape != (bsz * self.num_heads, tgt_len, self.head_dim):
            raise ValueError(
                f"`attn_output` should be of size {(bsz * self.num_heads, tgt_len, self.head_dim)}, but is"
                f" {attn_output.shape}"
            )

        attn_output = attn_output.view(bsz, self.num_heads, tgt_len, self.head_dim)
        attn_output = attn_output.swapaxes(1, 2)

        # Use the `embed_dim` from the config (stored in the class) rather than `hidden_state` because `attn_output` can be
        # partitioned across GPUs when using tensor-parallelism.
        attn_output = attn_output.reshape(bsz, tgt_len, self.embed_dim)

        attn_output = self.out_proj(attn_output)

        return attn_output, attn_weights_reshaped, past_key_value

mindnlp.transformers.models.marian.modeling_marian.MarianAttention.__init__(embed_dim, num_heads, dropout=0.0, is_decoder=False, bias=True, is_causal=False, config=None)

Initializes an instance of the BartAttention class.

PARAMETER	DESCRIPTION
embed_dim	The dimensionality of the input embeddings. TYPE: int
num_heads	The number of attention heads. TYPE: int
dropout	The dropout probability. Defaults to 0.0. TYPE: float DEFAULT: 0.0
is_decoder	Whether the attention layer is used as a decoder. Defaults to False. TYPE: bool DEFAULT: False
bias	Whether to include bias terms in the linear projections. Defaults to True. TYPE: bool DEFAULT: True
is_causal	Whether the attention layer is causal. Defaults to False. TYPE: bool DEFAULT: False
config	An optional BART configuration object. Defaults to None. TYPE: Optional[BartConfig] DEFAULT: None

RETURNS	DESCRIPTION
	None

RAISES	DESCRIPTION
ValueError	If embed_dim is not divisible by num_heads.

Source code in mindnlp/transformers/models/marian/modeling_marian.py

def __init__(
    self,
    embed_dim: int,
    num_heads: int,
    dropout: float = 0.0,
    is_decoder: bool = False,
    bias: bool = True,
    is_causal: bool = False,
    config: Optional[MarianConfig] = None,
):
    """
    Initializes an instance of the BartAttention class.

    Args:
        embed_dim (int): The dimensionality of the input embeddings.
        num_heads (int): The number of attention heads.
        dropout (float, optional): The dropout probability. Defaults to 0.0.
        is_decoder (bool, optional): Whether the attention layer is used as a decoder. Defaults to False.
        bias (bool, optional): Whether to include bias terms in the linear projections. Defaults to True.
        is_causal (bool, optional): Whether the attention layer is causal. Defaults to False.
        config (Optional[BartConfig], optional): An optional BART configuration object. Defaults to None.

    Returns:
        None

    Raises:
        ValueError: If `embed_dim` is not divisible by `num_heads`.
    """
    super().__init__()
    self.embed_dim = embed_dim
    self.num_heads = num_heads
    self.dropout = dropout
    self.head_dim = embed_dim // num_heads
    self.config = config

    if (self.head_dim * num_heads) != self.embed_dim:
        raise ValueError(
            f"embed_dim must be divisible by num_heads (got `embed_dim`: {self.embed_dim}"
            f" and `num_heads`: {num_heads})."
        )
    self.scaling = self.head_dim**-0.5
    self.is_decoder = is_decoder
    self.is_causal = is_causal

    self.k_proj = nn.Linear(embed_dim, embed_dim, bias=bias)
    self.v_proj = nn.Linear(embed_dim, embed_dim, bias=bias)
    self.q_proj = nn.Linear(embed_dim, embed_dim, bias=bias)
    self.out_proj = nn.Linear(embed_dim, embed_dim, bias=bias)

mindnlp.transformers.models.marian.modeling_marian.MarianAttention.forward(hidden_states, key_value_states=None, past_key_value=None, attention_mask=None, layer_head_mask=None, output_attentions=False)

Input shape: Batch x Time x Channel

Source code in mindnlp/transformers/models/marian/modeling_marian.py

def forward(
    self,
    hidden_states: mindspore.Tensor,
    key_value_states: Optional[mindspore.Tensor] = None,
    past_key_value: Optional[Tuple[mindspore.Tensor]] = None,
    attention_mask: Optional[mindspore.Tensor] = None,
    layer_head_mask: Optional[mindspore.Tensor] = None,
    output_attentions: bool = False,
) -> Tuple[mindspore.Tensor, Optional[mindspore.Tensor], Optional[Tuple[mindspore.Tensor]]]:
    """Input shape: Batch x Time x Channel"""
    # if key_value_states are provided this layer is used as a cross-attention layer
    # for the decoder
    is_cross_attention = key_value_states is not None

    bsz, tgt_len, _ = hidden_states.shape

    # get query proj
    query_states = self.q_proj(hidden_states) * self.scaling
    # get key, value proj
    # `past_key_value[0].shape[2] == key_value_states.shape[1]`
    # is checking that the `sequence_length` of the `past_key_value` is the same as
    # the provided `key_value_states` to support prefix tuning
    if (
        is_cross_attention
        and past_key_value is not None
        and past_key_value[0].shape[2] == key_value_states.shape[1]
    ):
        # reuse k,v, cross_attentions
        key_states = past_key_value[0]
        value_states = past_key_value[1]
    elif is_cross_attention:
        # cross_attentions
        key_states = self._shape(self.k_proj(key_value_states), -1, bsz)
        value_states = self._shape(self.v_proj(key_value_states), -1, bsz)
    elif past_key_value is not None:
        # reuse k, v, self_attention
        key_states = self._shape(self.k_proj(hidden_states), -1, bsz)
        value_states = self._shape(self.v_proj(hidden_states), -1, bsz)
        key_states = ops.cat([past_key_value[0], key_states], axis=2)
        value_states = ops.cat([past_key_value[1], value_states], axis=2)
    else:
        # self_attention
        key_states = self._shape(self.k_proj(hidden_states), -1, bsz)
        value_states = self._shape(self.v_proj(hidden_states), -1, bsz)

    if self.is_decoder:
        # if cross_attention save Tuple(mindspore.Tensor, mindspore.Tensor) of all cross attention key/value_states.
        # Further calls to cross_attention layer can then reuse all cross-attention
        # key/value_states (first "if" case)
        # if uni-directional self-attention (decoder) save Tuple(mindspore.Tensor, mindspore.Tensor) of
        # all previous decoder key/value_states. Further calls to uni-directional self-attention
        # can concat previous decoder key/value_states to current projected key/value_states (third "elif" case)
        # if encoder bi-directional self-attention `past_key_value` is always `None`
        past_key_value = (key_states, value_states)

    proj_shape = (bsz * self.num_heads, -1, self.head_dim)
    query_states = self._shape(query_states, tgt_len, bsz).view(*proj_shape)
    key_states = key_states.reshape(*proj_shape)
    value_states = value_states.reshape(*proj_shape)

    src_len = key_states.shape[1]
    attn_weights = ops.bmm(query_states, key_states.swapaxes(1, 2))

    if attn_weights.shape != (bsz * self.num_heads, tgt_len, src_len):
        raise ValueError(
            f"Attention weights should be of size {(bsz * self.num_heads, tgt_len, src_len)}, but is"
            f" {attn_weights.shape}"
        )

    if attention_mask is not None:
        if attention_mask.shape != (bsz, 1, tgt_len, src_len):
            raise ValueError(
                f"Attention mask should be of size {(bsz, 1, tgt_len, src_len)}, but is {attention_mask.shape}"
            )
        attn_weights = attn_weights.view(bsz, self.num_heads, tgt_len, src_len) + attention_mask
        attn_weights = attn_weights.view(bsz * self.num_heads, tgt_len, src_len)

    attn_weights = ops.softmax(attn_weights, axis=-1)

    if layer_head_mask is not None:
        if layer_head_mask.shape != (self.num_heads,):
            raise ValueError(
                f"Head mask for a single layer should be of size {(self.num_heads,)}, but is"
                f" {layer_head_mask.shape}"
            )
        attn_weights = layer_head_mask.view(1, -1, 1, 1) * attn_weights.view(bsz, self.num_heads, tgt_len, src_len)
        attn_weights = attn_weights.view(bsz * self.num_heads, tgt_len, src_len)

    if output_attentions:
        # this operation is a bit awkward, but it's required to
        # make sure that attn_weights keeps its gradient.
        # In order to do so, attn_weights have to be reshaped
        # twice and have to be reused in the following
        attn_weights_reshaped = attn_weights.view(bsz, self.num_heads, tgt_len, src_len)
        attn_weights = attn_weights_reshaped.view(bsz * self.num_heads, tgt_len, src_len)
    else:
        attn_weights_reshaped = None

    attn_probs = ops.dropout(attn_weights, p=self.dropout, training=self.training)

    attn_output = ops.bmm(attn_probs, value_states)

    if attn_output.shape != (bsz * self.num_heads, tgt_len, self.head_dim):
        raise ValueError(
            f"`attn_output` should be of size {(bsz * self.num_heads, tgt_len, self.head_dim)}, but is"
            f" {attn_output.shape}"
        )

    attn_output = attn_output.view(bsz, self.num_heads, tgt_len, self.head_dim)
    attn_output = attn_output.swapaxes(1, 2)

    # Use the `embed_dim` from the config (stored in the class) rather than `hidden_state` because `attn_output` can be
    # partitioned across GPUs when using tensor-parallelism.
    attn_output = attn_output.reshape(bsz, tgt_len, self.embed_dim)

    attn_output = self.out_proj(attn_output)

    return attn_output, attn_weights_reshaped, past_key_value

mindnlp.transformers.models.marian.modeling_marian.MarianDecoder

Bases: MarianPreTrainedModel

Transformer decoder consisting of config.decoder_layers layers. Each layer is a [MarianDecoderLayer]

PARAMETER	DESCRIPTION
config	MarianConfig TYPE: MarianConfig
embed_tokens	output embedding TYPE: Embedding DEFAULT: None

Source code in mindnlp/transformers/models/marian/modeling_marian.py

class MarianDecoder(MarianPreTrainedModel):
    """
    Transformer decoder consisting of *config.decoder_layers* layers. Each layer is a [`MarianDecoderLayer`]

    Args:
        config: MarianConfig
        embed_tokens (nn.Embedding): output embedding
    """

    def __init__(self, config: MarianConfig, embed_tokens: Optional[nn.Embedding] = None):
        super().__init__(config)
        self.dropout = config.dropout
        self.layerdrop = config.decoder_layerdrop
        self.padding_idx = config.pad_token_id
        self.max_target_positions = config.max_position_embeddings
        self.embed_scale = math.sqrt(config.d_model) if config.scale_embedding else 1.0

        if embed_tokens is not None:
            self.embed_tokens = embed_tokens
        else:
            self.embed_tokens = nn.Embedding(config.decoder_vocab_size, config.d_model, padding_idx=self.padding_idx)

        self.embed_positions = MarianSinusoidalPositionalEmbedding(
            config.max_position_embeddings, config.d_model, self.padding_idx
        )
        self.layers = nn.ModuleList([MarianDecoderLayer(config) for _ in range(config.decoder_layers)])

        self.gradient_checkpointing = False
        # Initialize weights and apply final processing
        self.post_init()

    def get_input_embeddings(self):
        return self.embed_tokens

    def set_input_embeddings(self, value):
        self.embed_tokens = value

    def forward(
        self,
        input_ids: mindspore.Tensor = None,
        attention_mask: Optional[mindspore.Tensor] = None,
        encoder_hidden_states: Optional[mindspore.Tensor] = None,
        encoder_attention_mask: Optional[mindspore.Tensor] = None,
        head_mask: Optional[mindspore.Tensor] = None,
        cross_attn_head_mask: Optional[mindspore.Tensor] = None,
        past_key_values: Optional[Tuple[Tuple[mindspore.Tensor]]] = None,
        inputs_embeds: Optional[mindspore.Tensor] = None,
        use_cache: Optional[bool] = None,
        output_attentions: Optional[bool] = None,
        output_hidden_states: Optional[bool] = None,
        return_dict: Optional[bool] = None,
    ) -> Union[Tuple[mindspore.Tensor], BaseModelOutputWithPastAndCrossAttentions]:
        r"""
        Args:
            input_ids (`torch.LongTensor` of shape `(batch_size, sequence_length)`):
                Indices of input sequence tokens in the vocabulary. Padding will be ignored by default should you
                provide it.

                Indices can be obtained using [`AutoTokenizer`]. See [`PreTrainedTokenizer.encode`] and
                [`PreTrainedTokenizer.__call__`] for details.

                [What are input IDs?](../glossary#input-ids)
            attention_mask (`torch.Tensor` of shape `(batch_size, sequence_length)`, *optional*):
                Mask to avoid performing attention on padding token indices. Mask values selected in `[0, 1]`:

               - 1 for tokens that are **not masked**,
               - 0 for tokens that are **masked**.

                [What are attention masks?](../glossary#attention-mask)
            encoder_hidden_states (`torch.FloatTensor` of shape `(batch_size, encoder_sequence_length, hidden_size)`, *optional*):
                Sequence of hidden-states at the output of the last layer of the encoder. Used in the cross-attention
                of the decoder.
            encoder_attention_mask (`torch.LongTensor` of shape `(batch_size, encoder_sequence_length)`, *optional*):
                Mask to avoid performing cross-attention on padding tokens indices of encoder input_ids. Mask values
                selected in `[0, 1]`:

                - 1 for tokens that are **not masked**,
                - 0 for tokens that are **masked**.

                [What are attention masks?](../glossary#attention-mask)
            head_mask (`torch.Tensor` of shape `(decoder_layers, decoder_attention_heads)`, *optional*):
                Mask to nullify selected heads of the attention modules. Mask values selected in `[0, 1]`:

                - 1 indicates the head is **not masked**,
                - 0 indicates the head is **masked**.

            cross_attn_head_mask (`torch.Tensor` of shape `(decoder_layers, decoder_attention_heads)`, *optional*):
                Mask to nullify selected heads of the cross-attention modules in the decoder to avoid performing
                cross-attention on hidden heads. Mask values selected in `[0, 1]`:

                - 1 indicates the head is **not masked**,
                - 0 indicates the head is **masked**.

            past_key_values (`tuple(tuple(torch.FloatTensor))`, *optional*, returned when `use_cache=True` is passed or when `config.use_cache=True`):
                Tuple of `tuple(torch.FloatTensor)` of length `config.n_layers`, with each tuple having 2 tensors of
                shape `(batch_size, num_heads, sequence_length, embed_size_per_head)`) and 2 additional tensors of
                shape `(batch_size, num_heads, encoder_sequence_length, embed_size_per_head)`.

                Contains pre-computed hidden-states (key and values in the self-attention blocks and in the
                cross-attention blocks) that can be used (see `past_key_values` input) to speed up sequential decoding.

                If `past_key_values` are used, the user can optionally input only the last `decoder_input_ids` (those
                that don't have their past key value states given to this model) of shape `(batch_size, 1)` instead of
                all `decoder_input_ids` of shape `(batch_size, sequence_length)`.
            inputs_embeds (`torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`, *optional*):
                Optionally, instead of passing `input_ids` you can choose to directly pass an embedded representation.
                This is useful if you want more control over how to convert `input_ids` indices into associated vectors
                than the model's internal embedding lookup matrix.
            output_attentions (`bool`, *optional*):
                Whether or not to return the attentions tensors of all attention layers. See `attentions` under
                returned tensors for more detail.
            output_hidden_states (`bool`, *optional*):
                Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors
                for more detail.
            return_dict (`bool`, *optional*):
                Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple.
        """
        output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions
        output_hidden_states = (
            output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
        )
        use_cache = use_cache if use_cache is not None else self.config.use_cache
        return_dict = return_dict if return_dict is not None else self.config.use_return_dict

        # retrieve input_ids and inputs_embeds
        if input_ids is not None and inputs_embeds is not None:
            raise ValueError("You cannot specify both decoder_input_ids and decoder_inputs_embeds at the same time")
        elif input_ids is not None:
            input_shape = input_ids.shape
            input_ids = input_ids.view(-1, input_shape[-1])
        elif inputs_embeds is not None:
            input_shape = inputs_embeds.shape[:-1]
        else:
            raise ValueError("You have to specify either decoder_input_ids or decoder_inputs_embeds")

        # past_key_values_length
        past_key_values_length = past_key_values[0][0].shape[2] if past_key_values is not None else 0

        if inputs_embeds is None:
            inputs_embeds = self.embed_tokens(input_ids) * self.embed_scale

        attention_mask = _prepare_4d_causal_attention_mask(
            attention_mask, input_shape, inputs_embeds, past_key_values_length
        )

        # expand encoder attention mask
        if encoder_hidden_states is not None and encoder_attention_mask is not None:
            # [bsz, seq_len] -> [bsz, 1, tgt_seq_len, src_seq_len]
            encoder_attention_mask = _prepare_4d_attention_mask(
                encoder_attention_mask, inputs_embeds.dtype, tgt_len=input_shape[-1]
            )

        # embed positions
        positions = self.embed_positions(input_shape, past_key_values_length)

        hidden_states = inputs_embeds + positions

        hidden_states = ops.dropout(hidden_states, p=self.dropout, training=self.training)

        if self.gradient_checkpointing and self.training:
            if use_cache:
                logger.warning_once(
                    "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`..."
                )
                use_cache = False

        # decoder layers
        all_hidden_states = () if output_hidden_states else None
        all_self_attns = () if output_attentions else None
        all_cross_attentions = () if (output_attentions and encoder_hidden_states is not None) else None
        next_decoder_cache = () if use_cache else None

        # check if head_mask/cross_attn_head_mask has a correct number of layers specified if desired
        for attn_mask, mask_name in zip([head_mask, cross_attn_head_mask], ["head_mask", "cross_attn_head_mask"]):
            if attn_mask is not None:
                assert attn_mask.shape[0] == (len(self.layers)), (
                    f"The `{mask_name}` should be specified for {len(self.layers)} layers, but it is for"
                    f" {head_mask.shape[0]}."
                )
        for idx, decoder_layer in enumerate(self.layers):
            # add LayerDrop (see https://arxiv.org/abs/1909.11556 for description)
            if output_hidden_states:
                all_hidden_states += (hidden_states,)
            if self.training:
                dropout_probability = np.random.uniform(0, 1)
                if dropout_probability < self.layerdrop:
                    continue

            past_key_value = past_key_values[idx] if past_key_values is not None else None

            if self.gradient_checkpointing and self.training:
                layer_outputs = self._gradient_checkpointing_func(
                    decoder_layer.__call__,
                    hidden_states,
                    attention_mask,
                    encoder_hidden_states,
                    encoder_attention_mask,
                    head_mask[idx] if head_mask is not None else None,
                    cross_attn_head_mask[idx] if cross_attn_head_mask is not None else None,
                    None,
                    output_attentions,
                    use_cache,
                )
            else:
                layer_outputs = decoder_layer(
                    hidden_states,
                    attention_mask=attention_mask,
                    encoder_hidden_states=encoder_hidden_states,
                    encoder_attention_mask=encoder_attention_mask,
                    layer_head_mask=(head_mask[idx] if head_mask is not None else None),
                    cross_attn_layer_head_mask=(
                        cross_attn_head_mask[idx] if cross_attn_head_mask is not None else None
                    ),
                    past_key_value=past_key_value,
                    output_attentions=output_attentions,
                    use_cache=use_cache,
                )
            hidden_states = layer_outputs[0]

            if use_cache:
                next_decoder_cache += (layer_outputs[3 if output_attentions else 1],)

            if output_attentions:
                all_self_attns += (layer_outputs[1],)

                if encoder_hidden_states is not None:
                    all_cross_attentions += (layer_outputs[2],)

        # add hidden states from the last decoder layer
        if output_hidden_states:
            all_hidden_states += (hidden_states,)

        next_cache = next_decoder_cache if use_cache else None
        if not return_dict:
            return tuple(
                v
                for v in [hidden_states, next_cache, all_hidden_states, all_self_attns, all_cross_attentions]
                if v is not None
            )
        return BaseModelOutputWithPastAndCrossAttentions(
            last_hidden_state=hidden_states,
            past_key_values=next_cache,
            hidden_states=all_hidden_states,
            attentions=all_self_attns,
            cross_attentions=all_cross_attentions,
        )

mindnlp.transformers.models.marian.modeling_marian.MarianDecoder.forward(input_ids=None, attention_mask=None, encoder_hidden_states=None, encoder_attention_mask=None, head_mask=None, cross_attn_head_mask=None, past_key_values=None, inputs_embeds=None, use_cache=None, output_attentions=None, output_hidden_states=None, return_dict=None)

PARAMETER	DESCRIPTION
input_ids	Indices of input sequence tokens in the vocabulary. Padding will be ignored by default should you provide it. Indices can be obtained using [AutoTokenizer]. See [PreTrainedTokenizer.encode] and [PreTrainedTokenizer.__call__] for details. What are input IDs? TYPE: `torch.LongTensor` of shape `(batch_size, sequence_length)` DEFAULT: None
attention_mask	Mask to avoid performing attention on padding token indices. Mask values selected in [0, 1]: 1 for tokens that are not masked, 0 for tokens that are masked. What are attention masks? TYPE: `torch.Tensor` of shape `(batch_size, sequence_length)`, *optional* DEFAULT: None
encoder_hidden_states	Sequence of hidden-states at the output of the last layer of the encoder. Used in the cross-attention of the decoder. TYPE: `torch.FloatTensor` of shape `(batch_size, encoder_sequence_length, hidden_size)`, *optional* DEFAULT: None
encoder_attention_mask	Mask to avoid performing cross-attention on padding tokens indices of encoder input_ids. Mask values selected in [0, 1]: 1 for tokens that are not masked, 0 for tokens that are masked. What are attention masks? TYPE: `torch.LongTensor` of shape `(batch_size, encoder_sequence_length)`, *optional* DEFAULT: None
head_mask	Mask to nullify selected heads of the attention modules. Mask values selected in [0, 1]: 1 indicates the head is not masked, 0 indicates the head is masked. TYPE: `torch.Tensor` of shape `(decoder_layers, decoder_attention_heads)`, *optional* DEFAULT: None
cross_attn_head_mask	Mask to nullify selected heads of the cross-attention modules in the decoder to avoid performing cross-attention on hidden heads. Mask values selected in [0, 1]: 1 indicates the head is not masked, 0 indicates the head is masked. TYPE: `torch.Tensor` of shape `(decoder_layers, decoder_attention_heads)`, *optional* DEFAULT: None
past_key_values	Tuple of tuple(torch.FloatTensor) of length config.n_layers, with each tuple having 2 tensors of shape (batch_size, num_heads, sequence_length, embed_size_per_head)) and 2 additional tensors of shape (batch_size, num_heads, encoder_sequence_length, embed_size_per_head). Contains pre-computed hidden-states (key and values in the self-attention blocks and in the cross-attention blocks) that can be used (see past_key_values input) to speed up sequential decoding. If past_key_values are used, the user can optionally input only the last decoder_input_ids (those that don't have their past key value states given to this model) of shape (batch_size, 1) instead of all decoder_input_ids of shape (batch_size, sequence_length). TYPE: `tuple(tuple(torch.FloatTensor))`, *optional*, returned when `use_cache=True` is passed or when `config.use_cache=True` DEFAULT: None
inputs_embeds	Optionally, instead of passing input_ids you can choose to directly pass an embedded representation. This is useful if you want more control over how to convert input_ids indices into associated vectors than the model's internal embedding lookup matrix. TYPE: `torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`, *optional* DEFAULT: None
output_attentions	Whether or not to return the attentions tensors of all attention layers. See attentions under returned tensors for more detail. TYPE: `bool`, *optional* DEFAULT: None
output_hidden_states	Whether or not to return the hidden states of all layers. See hidden_states under returned tensors for more detail. TYPE: `bool`, *optional* DEFAULT: None
return_dict	Whether or not to return a [~utils.ModelOutput] instead of a plain tuple. TYPE: `bool`, *optional* DEFAULT: None

Source code in mindnlp/transformers/models/marian/modeling_marian.py

def forward(
    self,
    input_ids: mindspore.Tensor = None,
    attention_mask: Optional[mindspore.Tensor] = None,
    encoder_hidden_states: Optional[mindspore.Tensor] = None,
    encoder_attention_mask: Optional[mindspore.Tensor] = None,
    head_mask: Optional[mindspore.Tensor] = None,
    cross_attn_head_mask: Optional[mindspore.Tensor] = None,
    past_key_values: Optional[Tuple[Tuple[mindspore.Tensor]]] = None,
    inputs_embeds: Optional[mindspore.Tensor] = None,
    use_cache: Optional[bool] = None,
    output_attentions: Optional[bool] = None,
    output_hidden_states: Optional[bool] = None,
    return_dict: Optional[bool] = None,
) -> Union[Tuple[mindspore.Tensor], BaseModelOutputWithPastAndCrossAttentions]:
    r"""
    Args:
        input_ids (`torch.LongTensor` of shape `(batch_size, sequence_length)`):
            Indices of input sequence tokens in the vocabulary. Padding will be ignored by default should you
            provide it.

            Indices can be obtained using [`AutoTokenizer`]. See [`PreTrainedTokenizer.encode`] and
            [`PreTrainedTokenizer.__call__`] for details.

            [What are input IDs?](../glossary#input-ids)
        attention_mask (`torch.Tensor` of shape `(batch_size, sequence_length)`, *optional*):
            Mask to avoid performing attention on padding token indices. Mask values selected in `[0, 1]`:

           - 1 for tokens that are **not masked**,
           - 0 for tokens that are **masked**.

            [What are attention masks?](../glossary#attention-mask)
        encoder_hidden_states (`torch.FloatTensor` of shape `(batch_size, encoder_sequence_length, hidden_size)`, *optional*):
            Sequence of hidden-states at the output of the last layer of the encoder. Used in the cross-attention
            of the decoder.
        encoder_attention_mask (`torch.LongTensor` of shape `(batch_size, encoder_sequence_length)`, *optional*):
            Mask to avoid performing cross-attention on padding tokens indices of encoder input_ids. Mask values
            selected in `[0, 1]`:

            - 1 for tokens that are **not masked**,
            - 0 for tokens that are **masked**.

            [What are attention masks?](../glossary#attention-mask)
        head_mask (`torch.Tensor` of shape `(decoder_layers, decoder_attention_heads)`, *optional*):
            Mask to nullify selected heads of the attention modules. Mask values selected in `[0, 1]`:

            - 1 indicates the head is **not masked**,
            - 0 indicates the head is **masked**.

        cross_attn_head_mask (`torch.Tensor` of shape `(decoder_layers, decoder_attention_heads)`, *optional*):
            Mask to nullify selected heads of the cross-attention modules in the decoder to avoid performing
            cross-attention on hidden heads. Mask values selected in `[0, 1]`:

            - 1 indicates the head is **not masked**,
            - 0 indicates the head is **masked**.

        past_key_values (`tuple(tuple(torch.FloatTensor))`, *optional*, returned when `use_cache=True` is passed or when `config.use_cache=True`):
            Tuple of `tuple(torch.FloatTensor)` of length `config.n_layers`, with each tuple having 2 tensors of
            shape `(batch_size, num_heads, sequence_length, embed_size_per_head)`) and 2 additional tensors of
            shape `(batch_size, num_heads, encoder_sequence_length, embed_size_per_head)`.

            Contains pre-computed hidden-states (key and values in the self-attention blocks and in the
            cross-attention blocks) that can be used (see `past_key_values` input) to speed up sequential decoding.

            If `past_key_values` are used, the user can optionally input only the last `decoder_input_ids` (those
            that don't have their past key value states given to this model) of shape `(batch_size, 1)` instead of
            all `decoder_input_ids` of shape `(batch_size, sequence_length)`.
        inputs_embeds (`torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`, *optional*):
            Optionally, instead of passing `input_ids` you can choose to directly pass an embedded representation.
            This is useful if you want more control over how to convert `input_ids` indices into associated vectors
            than the model's internal embedding lookup matrix.
        output_attentions (`bool`, *optional*):
            Whether or not to return the attentions tensors of all attention layers. See `attentions` under
            returned tensors for more detail.
        output_hidden_states (`bool`, *optional*):
            Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors
            for more detail.
        return_dict (`bool`, *optional*):
            Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple.
    """
    output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions
    output_hidden_states = (
        output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
    )
    use_cache = use_cache if use_cache is not None else self.config.use_cache
    return_dict = return_dict if return_dict is not None else self.config.use_return_dict

    # retrieve input_ids and inputs_embeds
    if input_ids is not None and inputs_embeds is not None:
        raise ValueError("You cannot specify both decoder_input_ids and decoder_inputs_embeds at the same time")
    elif input_ids is not None:
        input_shape = input_ids.shape
        input_ids = input_ids.view(-1, input_shape[-1])
    elif inputs_embeds is not None:
        input_shape = inputs_embeds.shape[:-1]
    else:
        raise ValueError("You have to specify either decoder_input_ids or decoder_inputs_embeds")

    # past_key_values_length
    past_key_values_length = past_key_values[0][0].shape[2] if past_key_values is not None else 0

    if inputs_embeds is None:
        inputs_embeds = self.embed_tokens(input_ids) * self.embed_scale

    attention_mask = _prepare_4d_causal_attention_mask(
        attention_mask, input_shape, inputs_embeds, past_key_values_length
    )

    # expand encoder attention mask
    if encoder_hidden_states is not None and encoder_attention_mask is not None:
        # [bsz, seq_len] -> [bsz, 1, tgt_seq_len, src_seq_len]
        encoder_attention_mask = _prepare_4d_attention_mask(
            encoder_attention_mask, inputs_embeds.dtype, tgt_len=input_shape[-1]
        )

    # embed positions
    positions = self.embed_positions(input_shape, past_key_values_length)

    hidden_states = inputs_embeds + positions

    hidden_states = ops.dropout(hidden_states, p=self.dropout, training=self.training)

    if self.gradient_checkpointing and self.training:
        if use_cache:
            logger.warning_once(
                "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`..."
            )
            use_cache = False

    # decoder layers
    all_hidden_states = () if output_hidden_states else None
    all_self_attns = () if output_attentions else None
    all_cross_attentions = () if (output_attentions and encoder_hidden_states is not None) else None
    next_decoder_cache = () if use_cache else None

    # check if head_mask/cross_attn_head_mask has a correct number of layers specified if desired
    for attn_mask, mask_name in zip([head_mask, cross_attn_head_mask], ["head_mask", "cross_attn_head_mask"]):
        if attn_mask is not None:
            assert attn_mask.shape[0] == (len(self.layers)), (
                f"The `{mask_name}` should be specified for {len(self.layers)} layers, but it is for"
                f" {head_mask.shape[0]}."
            )
    for idx, decoder_layer in enumerate(self.layers):
        # add LayerDrop (see https://arxiv.org/abs/1909.11556 for description)
        if output_hidden_states:
            all_hidden_states += (hidden_states,)
        if self.training:
            dropout_probability = np.random.uniform(0, 1)
            if dropout_probability < self.layerdrop:
                continue

        past_key_value = past_key_values[idx] if past_key_values is not None else None

        if self.gradient_checkpointing and self.training:
            layer_outputs = self._gradient_checkpointing_func(
                decoder_layer.__call__,
                hidden_states,
                attention_mask,
                encoder_hidden_states,
                encoder_attention_mask,
                head_mask[idx] if head_mask is not None else None,
                cross_attn_head_mask[idx] if cross_attn_head_mask is not None else None,
                None,
                output_attentions,
                use_cache,
            )
        else:
            layer_outputs = decoder_layer(
                hidden_states,
                attention_mask=attention_mask,
                encoder_hidden_states=encoder_hidden_states,
                encoder_attention_mask=encoder_attention_mask,
                layer_head_mask=(head_mask[idx] if head_mask is not None else None),
                cross_attn_layer_head_mask=(
                    cross_attn_head_mask[idx] if cross_attn_head_mask is not None else None
                ),
                past_key_value=past_key_value,
                output_attentions=output_attentions,
                use_cache=use_cache,
            )
        hidden_states = layer_outputs[0]

        if use_cache:
            next_decoder_cache += (layer_outputs[3 if output_attentions else 1],)

        if output_attentions:
            all_self_attns += (layer_outputs[1],)

            if encoder_hidden_states is not None:
                all_cross_attentions += (layer_outputs[2],)

    # add hidden states from the last decoder layer
    if output_hidden_states:
        all_hidden_states += (hidden_states,)

    next_cache = next_decoder_cache if use_cache else None
    if not return_dict:
        return tuple(
            v
            for v in [hidden_states, next_cache, all_hidden_states, all_self_attns, all_cross_attentions]
            if v is not None
        )
    return BaseModelOutputWithPastAndCrossAttentions(
        last_hidden_state=hidden_states,
        past_key_values=next_cache,
        hidden_states=all_hidden_states,
        attentions=all_self_attns,
        cross_attentions=all_cross_attentions,
    )

mindnlp.transformers.models.marian.modeling_marian.MarianDecoderWrapper

Bases: MarianPreTrainedModel

This wrapper class is a helper class to correctly load pretrained checkpoints when the causal language model is used in combination with the [EncoderDecoderModel] framework.

Source code in mindnlp/transformers/models/marian/modeling_marian.py

class MarianDecoderWrapper(MarianPreTrainedModel):
    """
    This wrapper class is a helper class to correctly load pretrained checkpoints when the causal language model is
    used in combination with the [`EncoderDecoderModel`] framework.
    """

    def __init__(self, config):
        super().__init__(config)
        self.decoder = MarianDecoder(config)

    def forward(self, *args, **kwargs):
        return self.decoder(*args, **kwargs)

mindnlp.transformers.models.marian.modeling_marian.MarianEncoder

Bases: MarianPreTrainedModel

Transformer encoder consisting of config.encoder_layers self attention layers. Each layer is a [MarianEncoderLayer].

PARAMETER	DESCRIPTION
config	MarianConfig TYPE: MarianConfig
embed_tokens	output embedding TYPE: Embedding DEFAULT: None

Source code in mindnlp/transformers/models/marian/modeling_marian.py

class MarianEncoder(MarianPreTrainedModel):
    """
    Transformer encoder consisting of *config.encoder_layers* self attention layers. Each layer is a
    [`MarianEncoderLayer`].

    Args:
        config: MarianConfig
        embed_tokens (nn.Embedding): output embedding
    """

    def __init__(self, config: MarianConfig, embed_tokens: Optional[nn.Embedding] = None):
        super().__init__(config)

        self.dropout = config.dropout
        self.layerdrop = config.encoder_layerdrop

        embed_dim = config.d_model
        self.padding_idx = config.pad_token_id
        self.max_source_positions = config.max_position_embeddings
        self.embed_scale = math.sqrt(embed_dim) if config.scale_embedding else 1.0

        if embed_tokens is not None:
            self.embed_tokens = embed_tokens
        else:
            self.embed_tokens = nn.Embedding(config.vocab_size, embed_dim, padding_idx=self.padding_idx)

        self.embed_positions = MarianSinusoidalPositionalEmbedding(
            config.max_position_embeddings, embed_dim, self.padding_idx
        )
        self.layers = nn.ModuleList([MarianEncoderLayer(config) for _ in range(config.encoder_layers)])

        self.gradient_checkpointing = False
        # Initialize weights and apply final processing
        self.post_init()

    def get_input_embeddings(self):
        return self.embed_tokens

    def set_input_embeddings(self, value):
        self.embed_tokens = value

    def forward(
        self,
        input_ids: mindspore.Tensor = None,
        attention_mask: Optional[mindspore.Tensor] = None,
        head_mask: Optional[mindspore.Tensor] = None,
        inputs_embeds: Optional[mindspore.Tensor] = None,
        output_attentions: Optional[bool] = None,
        output_hidden_states: Optional[bool] = None,
        return_dict: Optional[bool] = None,
    ) -> Union[Tuple[mindspore.Tensor], BaseModelOutput]:
        r"""
        Args:
            input_ids (`torch.LongTensor` of shape `(batch_size, sequence_length)`):
                Indices of input sequence tokens in the vocabulary. Padding will be ignored by default should you
                provide it.

                Indices can be obtained using [`AutoTokenizer`]. See [`PreTrainedTokenizer.encode`] and
                [`PreTrainedTokenizer.__call__`] for details.

                [What are input IDs?](../glossary#input-ids)
            attention_mask (`torch.Tensor` of shape `(batch_size, sequence_length)`, *optional*):
                Mask to avoid performing attention on padding token indices. Mask values selected in `[0, 1]`:

                - 1 for tokens that are **not masked**,
                - 0 for tokens that are **masked**.

                [What are attention masks?](../glossary#attention-mask)
            head_mask (`torch.Tensor` of shape `(encoder_layers, encoder_attention_heads)`, *optional*):
                Mask to nullify selected heads of the attention modules. Mask values selected in `[0, 1]`:

                - 1 indicates the head is **not masked**,
                - 0 indicates the head is **masked**.
            inputs_embeds (`torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`, *optional*):
                Optionally, instead of passing `input_ids` you can choose to directly pass an embedded representation.
                This is useful if you want more control over how to convert `input_ids` indices into associated vectors
                than the model's internal embedding lookup matrix.
            output_attentions (`bool`, *optional*):
                Whether or not to return the attentions tensors of all attention layers. See `attentions` under
                returned tensors for more detail.
            output_hidden_states (`bool`, *optional*):
                Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors
                for more detail.
            return_dict (`bool`, *optional*):
                Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple.
        """
        output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions
        output_hidden_states = (
            output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
        )
        return_dict = return_dict if return_dict is not None else self.config.use_return_dict

        # retrieve input_ids and inputs_embeds
        if input_ids is not None and inputs_embeds is not None:
            raise ValueError("You cannot specify both input_ids and inputs_embeds at the same time")
        elif input_ids is not None:
            self.warn_if_padding_and_no_attention_mask(input_ids, attention_mask)
            input_shape = input_ids.shape
            input_ids = input_ids.view(-1, input_shape[-1])
        elif inputs_embeds is not None:
            input_shape = inputs_embeds.shape[:-1]
        else:
            raise ValueError("You have to specify either input_ids or inputs_embeds")

        if inputs_embeds is None:
            inputs_embeds = self.embed_tokens(input_ids) * self.embed_scale

        embed_pos = self.embed_positions(input_shape)

        hidden_states = inputs_embeds + embed_pos
        hidden_states = ops.dropout(hidden_states, p=self.dropout, training=self.training)

        # expand attention_mask
        if attention_mask is not None:
            # [bsz, seq_len] -> [bsz, 1, tgt_seq_len, src_seq_len]
            attention_mask = _prepare_4d_attention_mask(attention_mask, inputs_embeds.dtype)

        encoder_states = () if output_hidden_states else None
        all_attentions = () if output_attentions else None

        # check if head_mask has a correct number of layers specified if desired
        if head_mask is not None:
            assert head_mask.shape[0] == (
                len(self.layers)
            ), f"The head_mask should be specified for {len(self.layers)} layers, but it is for {head_mask.shape[0]}."
        for idx, encoder_layer in enumerate(self.layers):
            if output_hidden_states:
                encoder_states = encoder_states + (hidden_states,)
            # add LayerDrop (see https://arxiv.org/abs/1909.11556 for description)
            to_drop = False
            if self.training:
                dropout_probability = np.random.uniform(0, 1)
                if dropout_probability < self.layerdrop:  # skip the layer
                    to_drop = True

            if to_drop:
                layer_outputs = (None, None)
            else:
                if self.gradient_checkpointing and self.training:
                    layer_outputs = self._gradient_checkpointing_func(
                        encoder_layer.__call__,
                        hidden_states,
                        attention_mask,
                        (head_mask[idx] if head_mask is not None else None),
                        output_attentions,
                    )
                else:
                    layer_outputs = encoder_layer(
                        hidden_states,
                        attention_mask,
                        layer_head_mask=(head_mask[idx] if head_mask is not None else None),
                        output_attentions=output_attentions,
                    )

                hidden_states = layer_outputs[0]

            if output_attentions:
                all_attentions = all_attentions + (layer_outputs[1],)

        if output_hidden_states:
            encoder_states = encoder_states + (hidden_states,)

        if not return_dict:
            return tuple(v for v in [hidden_states, encoder_states, all_attentions] if v is not None)
        return BaseModelOutput(
            last_hidden_state=hidden_states, hidden_states=encoder_states, attentions=all_attentions
        )

mindnlp.transformers.models.marian.modeling_marian.MarianEncoder.forward(input_ids=None, attention_mask=None, head_mask=None, inputs_embeds=None, output_attentions=None, output_hidden_states=None, return_dict=None)

PARAMETER	DESCRIPTION
input_ids	Indices of input sequence tokens in the vocabulary. Padding will be ignored by default should you provide it. Indices can be obtained using [AutoTokenizer]. See [PreTrainedTokenizer.encode] and [PreTrainedTokenizer.__call__] for details. What are input IDs? TYPE: `torch.LongTensor` of shape `(batch_size, sequence_length)` DEFAULT: None
attention_mask	Mask to avoid performing attention on padding token indices. Mask values selected in [0, 1]: 1 for tokens that are not masked, 0 for tokens that are masked. What are attention masks? TYPE: `torch.Tensor` of shape `(batch_size, sequence_length)`, *optional* DEFAULT: None
head_mask	Mask to nullify selected heads of the attention modules. Mask values selected in [0, 1]: 1 indicates the head is not masked, 0 indicates the head is masked. TYPE: `torch.Tensor` of shape `(encoder_layers, encoder_attention_heads)`, *optional* DEFAULT: None
inputs_embeds	Optionally, instead of passing input_ids you can choose to directly pass an embedded representation. This is useful if you want more control over how to convert input_ids indices into associated vectors than the model's internal embedding lookup matrix. TYPE: `torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`, *optional* DEFAULT: None
output_attentions	Whether or not to return the attentions tensors of all attention layers. See attentions under returned tensors for more detail. TYPE: `bool`, *optional* DEFAULT: None
output_hidden_states	Whether or not to return the hidden states of all layers. See hidden_states under returned tensors for more detail. TYPE: `bool`, *optional* DEFAULT: None
return_dict	Whether or not to return a [~utils.ModelOutput] instead of a plain tuple. TYPE: `bool`, *optional* DEFAULT: None

Source code in mindnlp/transformers/models/marian/modeling_marian.py

def forward(
    self,
    input_ids: mindspore.Tensor = None,
    attention_mask: Optional[mindspore.Tensor] = None,
    head_mask: Optional[mindspore.Tensor] = None,
    inputs_embeds: Optional[mindspore.Tensor] = None,
    output_attentions: Optional[bool] = None,
    output_hidden_states: Optional[bool] = None,
    return_dict: Optional[bool] = None,
) -> Union[Tuple[mindspore.Tensor], BaseModelOutput]:
    r"""
    Args:
        input_ids (`torch.LongTensor` of shape `(batch_size, sequence_length)`):
            Indices of input sequence tokens in the vocabulary. Padding will be ignored by default should you
            provide it.

            Indices can be obtained using [`AutoTokenizer`]. See [`PreTrainedTokenizer.encode`] and
            [`PreTrainedTokenizer.__call__`] for details.

            [What are input IDs?](../glossary#input-ids)
        attention_mask (`torch.Tensor` of shape `(batch_size, sequence_length)`, *optional*):
            Mask to avoid performing attention on padding token indices. Mask values selected in `[0, 1]`:

            - 1 for tokens that are **not masked**,
            - 0 for tokens that are **masked**.

            [What are attention masks?](../glossary#attention-mask)
        head_mask (`torch.Tensor` of shape `(encoder_layers, encoder_attention_heads)`, *optional*):
            Mask to nullify selected heads of the attention modules. Mask values selected in `[0, 1]`:

            - 1 indicates the head is **not masked**,
            - 0 indicates the head is **masked**.
        inputs_embeds (`torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`, *optional*):
            Optionally, instead of passing `input_ids` you can choose to directly pass an embedded representation.
            This is useful if you want more control over how to convert `input_ids` indices into associated vectors
            than the model's internal embedding lookup matrix.
        output_attentions (`bool`, *optional*):
            Whether or not to return the attentions tensors of all attention layers. See `attentions` under
            returned tensors for more detail.
        output_hidden_states (`bool`, *optional*):
            Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors
            for more detail.
        return_dict (`bool`, *optional*):
            Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple.
    """
    output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions
    output_hidden_states = (
        output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
    )
    return_dict = return_dict if return_dict is not None else self.config.use_return_dict

    # retrieve input_ids and inputs_embeds
    if input_ids is not None and inputs_embeds is not None:
        raise ValueError("You cannot specify both input_ids and inputs_embeds at the same time")
    elif input_ids is not None:
        self.warn_if_padding_and_no_attention_mask(input_ids, attention_mask)
        input_shape = input_ids.shape
        input_ids = input_ids.view(-1, input_shape[-1])
    elif inputs_embeds is not None:
        input_shape = inputs_embeds.shape[:-1]
    else:
        raise ValueError("You have to specify either input_ids or inputs_embeds")

    if inputs_embeds is None:
        inputs_embeds = self.embed_tokens(input_ids) * self.embed_scale

    embed_pos = self.embed_positions(input_shape)

    hidden_states = inputs_embeds + embed_pos
    hidden_states = ops.dropout(hidden_states, p=self.dropout, training=self.training)

    # expand attention_mask
    if attention_mask is not None:
        # [bsz, seq_len] -> [bsz, 1, tgt_seq_len, src_seq_len]
        attention_mask = _prepare_4d_attention_mask(attention_mask, inputs_embeds.dtype)

    encoder_states = () if output_hidden_states else None
    all_attentions = () if output_attentions else None

    # check if head_mask has a correct number of layers specified if desired
    if head_mask is not None:
        assert head_mask.shape[0] == (
            len(self.layers)
        ), f"The head_mask should be specified for {len(self.layers)} layers, but it is for {head_mask.shape[0]}."
    for idx, encoder_layer in enumerate(self.layers):
        if output_hidden_states:
            encoder_states = encoder_states + (hidden_states,)
        # add LayerDrop (see https://arxiv.org/abs/1909.11556 for description)
        to_drop = False
        if self.training:
            dropout_probability = np.random.uniform(0, 1)
            if dropout_probability < self.layerdrop:  # skip the layer
                to_drop = True

        if to_drop:
            layer_outputs = (None, None)
        else:
            if self.gradient_checkpointing and self.training:
                layer_outputs = self._gradient_checkpointing_func(
                    encoder_layer.__call__,
                    hidden_states,
                    attention_mask,
                    (head_mask[idx] if head_mask is not None else None),
                    output_attentions,
                )
            else:
                layer_outputs = encoder_layer(
                    hidden_states,
                    attention_mask,
                    layer_head_mask=(head_mask[idx] if head_mask is not None else None),
                    output_attentions=output_attentions,
                )

            hidden_states = layer_outputs[0]

        if output_attentions:
            all_attentions = all_attentions + (layer_outputs[1],)

    if output_hidden_states:
        encoder_states = encoder_states + (hidden_states,)

    if not return_dict:
        return tuple(v for v in [hidden_states, encoder_states, all_attentions] if v is not None)
    return BaseModelOutput(
        last_hidden_state=hidden_states, hidden_states=encoder_states, attentions=all_attentions
    )

mindnlp.transformers.models.marian.modeling_marian.MarianMTModel

Bases: MarianPreTrainedModel

Source code in mindnlp/transformers/models/marian/modeling_marian.py

class MarianMTModel(MarianPreTrainedModel):
    base_model_prefix = "model"
    # _keys_to_ignore_on_load_missing = [
    #     "final_logits_bias",
    #     "encoder.embed_positions.weight",
    #     "decoder.embed_positions.weight",
    # ]
    #_keys_to_ignore_on_save = ["model.encoder.embed_positions.weight", "model.decoder.embed_positions.weight"]
    _tied_weights_keys = ["model.encoder.embed_tokens.weight", "model.decoder.embed_tokens.weight", "lm_head.weight"]

    def __init__(self, config: MarianConfig):
        super().__init__(config)
        self.model = MarianModel(config)

        target_vocab_size = config.vocab_size if config.share_encoder_decoder_embeddings else config.decoder_vocab_size
        self.final_logits_bias=mindspore.Parameter(ops.zeros((1, target_vocab_size)),requires_grad=False)
        self.lm_head = nn.Linear(config.d_model, target_vocab_size, bias=False)

        # Initialize weights and apply final processing
        self.post_init()

    def get_encoder(self):
        return self.model.get_encoder()

    def get_decoder(self):
        return self.model.get_decoder()

    def resize_token_embeddings(self, new_num_tokens: int, pad_to_multiple_of: Optional[int] = None) -> nn.Embedding:
        new_embeddings = super().resize_token_embeddings(new_num_tokens, pad_to_multiple_of)
        if self.config.share_encoder_decoder_embeddings:
            self._resize_final_logits_bias(new_num_tokens)
        return new_embeddings

    def _resize_token_embeddings(self, new_num_tokens: int, pad_to_multiple_of=None) -> nn.Embedding:
        old_embeddings = self.get_input_embeddings()
        new_embeddings = self._get_resized_embeddings(old_embeddings, new_num_tokens, pad_to_multiple_of)
        self.set_input_embeddings(new_embeddings)

        new_num_tokens = new_embeddings.weight.shape[0]
        # update config.decoder_vocab_size if embeddings are tied
        if self.config.share_encoder_decoder_embeddings:
            self.config.decoder_vocab_size = new_num_tokens

        # if word embeddings are not tied, make sure that lm head is resized as well
        if (
            self.config.share_encoder_decoder_embeddings
            and self.get_output_embeddings() is not None
            and not self.config.tie_word_embeddings
        ):
            old_lm_head = self.get_output_embeddings()
            new_lm_head = self._get_resized_lm_head(old_lm_head, new_num_tokens)
            self.set_output_embeddings(new_lm_head)

        return self.get_input_embeddings()

    def resize_decoder_token_embeddings(self, new_num_tokens):
        if self.config.share_encoder_decoder_embeddings:
            raise ValueError(
                "`resize_decoder_token_embeddings` should not be called if `config.share_encoder_decoder_embeddings` "
                "is `True`. Please use `resize_token_embeddings` instead."
            )

        old_embeddings = self.model.get_decoder_input_embeddings()
        new_embeddings = self._get_resized_embeddings(old_embeddings, new_num_tokens)
        self.model.set_decoder_input_embeddings(new_embeddings)

        # if word embeddings are not tied, make sure that lm head is resized as well
        if self.get_output_embeddings() is not None and not self.config.tie_word_embeddings:
            old_lm_head = self.get_output_embeddings()
            new_lm_head = self._get_resized_lm_head(old_lm_head, new_num_tokens)
            self.set_output_embeddings(new_lm_head)

        model_embeds = self.model.get_decoder_input_embeddings()

        if new_num_tokens is None:
            return model_embeds

        # Update base model and current model config
        self.config.decoder_vocab_size = new_num_tokens

        # Tie weights again if needed
        self.tie_weights()

        self._resize_final_logits_bias(new_num_tokens)

        return model_embeds

    def _resize_final_logits_bias(self, new_num_tokens: int) -> None:
        old_num_tokens = self.final_logits_bias.shape[-1]
        if new_num_tokens <= old_num_tokens:
            new_bias = self.final_logits_bias[:, :new_num_tokens]
        else:
            extra_bias = ops.zeros((1, new_num_tokens - old_num_tokens))
            new_bias = ops.cat([self.final_logits_bias, extra_bias], axis=1)
        self.final_logits_bias=mindspore.Parameter(new_bias)
    def get_output_embeddings(self):
        return self.lm_head

    def set_output_embeddings(self, new_embeddings: nn.Embedding):
        self.lm_head = new_embeddings

    def tie_weights(self):
        """
        Tie the weights between the input embeddings and the output embeddings.

        If the `torchscript` flag is set in the configuration, can't handle parameter sharing so we are cloning the
        weights instead.
        """
        output_embeddings = self.get_output_embeddings()
        if output_embeddings is not None and getattr(self.config, "tie_word_embeddings", True):
            # if embeddings are shared this will return shared embeddings otherwise decoder embed_tokens
            word_embeddings = self.get_decoder().get_input_embeddings()
            self._tie_or_clone_weights(output_embeddings, word_embeddings)

        if getattr(self.config, "is_encoder_decoder", False) and getattr(self.config, "tie_encoder_decoder", False):
            tied_weights = self._tie_encoder_decoder_weights(
                self.encoder, self.decoder, self.base_model_prefix, "encoder"
            )
            # Setting a dynamic variable instead of `_tied_weights_keys` because it's a class
            # attributed not an instance member, therefore modifying it will modify the entire class
            # Leading to issues on subsequent calls by different tests or subsequent calls.
            self._dynamic_tied_weights_keys = tied_weights

        for _, cell in self.cells_and_names():
            if hasattr(cell, "_tie_weights"):
                cell._tie_weights()

    def forward(
        self,
        input_ids: mindspore.Tensor = None,
        attention_mask: Optional[mindspore.Tensor] = None,
        decoder_input_ids: Optional[mindspore.Tensor] = None,
        decoder_attention_mask: Optional[mindspore.Tensor] = None,
        head_mask: Optional[mindspore.Tensor] = None,
        decoder_head_mask: Optional[mindspore.Tensor] = None,
        cross_attn_head_mask: Optional[mindspore.Tensor] = None,
        encoder_outputs: Optional[Union[Tuple[mindspore.Tensor], BaseModelOutput]] = None,
        past_key_values: Optional[Tuple[Tuple[mindspore.Tensor]]] = None,
        inputs_embeds: Optional[mindspore.Tensor] = None,
        decoder_inputs_embeds: Optional[mindspore.Tensor] = None,
        labels: Optional[mindspore.Tensor] = None,
        use_cache: Optional[bool] = None,
        output_attentions: Optional[bool] = None,
        output_hidden_states: Optional[bool] = None,
        return_dict: Optional[bool] = None,
    ) -> Seq2SeqLMOutput:
        r"""
        Args:
            labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*):
                Labels for computing the masked language modeling loss. Indices should either be in `[0, ...,
                config.vocab_size]` or -100 (see `input_ids` docstring). Tokens with indices set to `-100` are ignored
                (masked), the loss is only computed for the tokens with labels in `[0, ..., config.vocab_size]`.

        Returns:
            `Seq2SeqLMOutput`
        """
        return_dict = return_dict if return_dict is not None else self.config.use_return_dict

        if labels is not None:
            if use_cache:
                logger.warning("The `use_cache` argument is changed to `False` since `labels` is provided.")
            use_cache = False
            if decoder_input_ids is None and decoder_inputs_embeds is None:
                decoder_input_ids = shift_tokens_right(
                    labels, self.config.pad_token_id, self.config.decoder_start_token_id
                )

        outputs = self.model(
            input_ids,
            attention_mask=attention_mask,
            decoder_input_ids=decoder_input_ids,
            encoder_outputs=encoder_outputs,
            decoder_attention_mask=decoder_attention_mask,
            head_mask=head_mask,
            decoder_head_mask=decoder_head_mask,
            cross_attn_head_mask=cross_attn_head_mask,
            past_key_values=past_key_values,
            inputs_embeds=inputs_embeds,
            decoder_inputs_embeds=decoder_inputs_embeds,
            use_cache=use_cache,
            output_attentions=output_attentions,
            output_hidden_states=output_hidden_states,
            return_dict=return_dict,
        )
        lm_logits = self.lm_head(outputs[0]) + self.final_logits_bias

        masked_lm_loss = None
        if labels is not None:
            masked_lm_loss = ops.cross_entropy(lm_logits.view(-1, self.config.decoder_vocab_size), labels.view(-1))

        if not return_dict:
            output = (lm_logits,) + outputs[1:]
            return ((masked_lm_loss,) + output) if masked_lm_loss is not None else output

        return Seq2SeqLMOutput(
            loss=masked_lm_loss,
            logits=lm_logits,
            past_key_values=outputs.past_key_values,
            decoder_hidden_states=outputs.decoder_hidden_states,
            decoder_attentions=outputs.decoder_attentions,
            cross_attentions=outputs.cross_attentions,
            encoder_last_hidden_state=outputs.encoder_last_hidden_state,
            encoder_hidden_states=outputs.encoder_hidden_states,
            encoder_attentions=outputs.encoder_attentions,
        )

    def prepare_inputs_for_generation(
        self,
        decoder_input_ids: mindspore.Tensor,
        past_key_values: Optional[Tuple[Tuple[mindspore.Tensor]]] = None,
        attention_mask: Optional[mindspore.Tensor] = None,
        head_mask: Optional[mindspore.Tensor] = None,
        decoder_head_mask: Optional[mindspore.Tensor] = None,
        cross_attn_head_mask: Optional[mindspore.Tensor] = None,
        use_cache: Optional[bool] = None,
        encoder_outputs: Optional[Union[Tuple[mindspore.Tensor], BaseModelOutput]] = None,
        **kwargs,
    ) -> Dict:
        # cut decoder_input_ids if past is used
        if past_key_values is not None:
            past_length = past_key_values[0][0].shape[2]

            # Some generation methods already pass only the last input ID
            if decoder_input_ids.shape[1] > past_length:
                remove_prefix_length = past_length
            else:
                # Default to old behavior: keep only final ID
                remove_prefix_length = decoder_input_ids.shape[1] - 1

            decoder_input_ids = decoder_input_ids[:, remove_prefix_length:]

        return {
            "input_ids": None,  # encoder_outputs is defined. input_ids not needed
            "encoder_outputs": encoder_outputs,
            "past_key_values": past_key_values,
            "decoder_input_ids": decoder_input_ids,
            "attention_mask": attention_mask,
            "head_mask": head_mask,
            "decoder_head_mask": decoder_head_mask,
            "cross_attn_head_mask": cross_attn_head_mask,
            "use_cache": use_cache,  # change this to avoid caching (presumably for debugging)
        }

    def prepare_decoder_input_ids_from_labels(self, labels: mindspore.Tensor):
        return shift_tokens_right(labels, self.config.pad_token_id, self.config.decoder_start_token_id)

    @staticmethod
    def _reorder_cache(past_key_values, beam_idx):
        reordered_past = ()
        for layer_past in past_key_values:
            # cached cross_attention states don't have to be reordered -> they are always the same
            reordered_past += (
                tuple(past_state.index_select(0, beam_idx) for past_state in layer_past[:2])
                + layer_past[2:],
            )
        return reordered_past

mindnlp.transformers.models.marian.modeling_marian.MarianMTModel.forward(input_ids=None, attention_mask=None, decoder_input_ids=None, decoder_attention_mask=None, head_mask=None, decoder_head_mask=None, cross_attn_head_mask=None, encoder_outputs=None, past_key_values=None, inputs_embeds=None, decoder_inputs_embeds=None, labels=None, use_cache=None, output_attentions=None, output_hidden_states=None, return_dict=None)

PARAMETER	DESCRIPTION
labels	Labels for computing the masked language modeling loss. Indices should either be in [0, ..., config.vocab_size] or -100 (see input_ids docstring). Tokens with indices set to -100 are ignored (masked), the loss is only computed for the tokens with labels in [0, ..., config.vocab_size]. TYPE: `torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional* DEFAULT: None

RETURNS	DESCRIPTION
Seq2SeqLMOutput	Seq2SeqLMOutput

Source code in mindnlp/transformers/models/marian/modeling_marian.py

def forward(
    self,
    input_ids: mindspore.Tensor = None,
    attention_mask: Optional[mindspore.Tensor] = None,
    decoder_input_ids: Optional[mindspore.Tensor] = None,
    decoder_attention_mask: Optional[mindspore.Tensor] = None,
    head_mask: Optional[mindspore.Tensor] = None,
    decoder_head_mask: Optional[mindspore.Tensor] = None,
    cross_attn_head_mask: Optional[mindspore.Tensor] = None,
    encoder_outputs: Optional[Union[Tuple[mindspore.Tensor], BaseModelOutput]] = None,
    past_key_values: Optional[Tuple[Tuple[mindspore.Tensor]]] = None,
    inputs_embeds: Optional[mindspore.Tensor] = None,
    decoder_inputs_embeds: Optional[mindspore.Tensor] = None,
    labels: Optional[mindspore.Tensor] = None,
    use_cache: Optional[bool] = None,
    output_attentions: Optional[bool] = None,
    output_hidden_states: Optional[bool] = None,
    return_dict: Optional[bool] = None,
) -> Seq2SeqLMOutput:
    r"""
    Args:
        labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*):
            Labels for computing the masked language modeling loss. Indices should either be in `[0, ...,
            config.vocab_size]` or -100 (see `input_ids` docstring). Tokens with indices set to `-100` are ignored
            (masked), the loss is only computed for the tokens with labels in `[0, ..., config.vocab_size]`.

    Returns:
        `Seq2SeqLMOutput`
    """
    return_dict = return_dict if return_dict is not None else self.config.use_return_dict

    if labels is not None:
        if use_cache:
            logger.warning("The `use_cache` argument is changed to `False` since `labels` is provided.")
        use_cache = False
        if decoder_input_ids is None and decoder_inputs_embeds is None:
            decoder_input_ids = shift_tokens_right(
                labels, self.config.pad_token_id, self.config.decoder_start_token_id
            )

    outputs = self.model(
        input_ids,
        attention_mask=attention_mask,
        decoder_input_ids=decoder_input_ids,
        encoder_outputs=encoder_outputs,
        decoder_attention_mask=decoder_attention_mask,
        head_mask=head_mask,
        decoder_head_mask=decoder_head_mask,
        cross_attn_head_mask=cross_attn_head_mask,
        past_key_values=past_key_values,
        inputs_embeds=inputs_embeds,
        decoder_inputs_embeds=decoder_inputs_embeds,
        use_cache=use_cache,
        output_attentions=output_attentions,
        output_hidden_states=output_hidden_states,
        return_dict=return_dict,
    )
    lm_logits = self.lm_head(outputs[0]) + self.final_logits_bias

    masked_lm_loss = None
    if labels is not None:
        masked_lm_loss = ops.cross_entropy(lm_logits.view(-1, self.config.decoder_vocab_size), labels.view(-1))

    if not return_dict:
        output = (lm_logits,) + outputs[1:]
        return ((masked_lm_loss,) + output) if masked_lm_loss is not None else output

    return Seq2SeqLMOutput(
        loss=masked_lm_loss,
        logits=lm_logits,
        past_key_values=outputs.past_key_values,
        decoder_hidden_states=outputs.decoder_hidden_states,
        decoder_attentions=outputs.decoder_attentions,
        cross_attentions=outputs.cross_attentions,
        encoder_last_hidden_state=outputs.encoder_last_hidden_state,
        encoder_hidden_states=outputs.encoder_hidden_states,
        encoder_attentions=outputs.encoder_attentions,
    )

mindnlp.transformers.models.marian.modeling_marian.MarianMTModel.tie_weights()

Tie the weights between the input embeddings and the output embeddings.

If the torchscript flag is set in the configuration, can't handle parameter sharing so we are cloning the weights instead.

Source code in mindnlp/transformers/models/marian/modeling_marian.py

def tie_weights(self):
    """
    Tie the weights between the input embeddings and the output embeddings.

    If the `torchscript` flag is set in the configuration, can't handle parameter sharing so we are cloning the
    weights instead.
    """
    output_embeddings = self.get_output_embeddings()
    if output_embeddings is not None and getattr(self.config, "tie_word_embeddings", True):
        # if embeddings are shared this will return shared embeddings otherwise decoder embed_tokens
        word_embeddings = self.get_decoder().get_input_embeddings()
        self._tie_or_clone_weights(output_embeddings, word_embeddings)

    if getattr(self.config, "is_encoder_decoder", False) and getattr(self.config, "tie_encoder_decoder", False):
        tied_weights = self._tie_encoder_decoder_weights(
            self.encoder, self.decoder, self.base_model_prefix, "encoder"
        )
        # Setting a dynamic variable instead of `_tied_weights_keys` because it's a class
        # attributed not an instance member, therefore modifying it will modify the entire class
        # Leading to issues on subsequent calls by different tests or subsequent calls.
        self._dynamic_tied_weights_keys = tied_weights

    for _, cell in self.cells_and_names():
        if hasattr(cell, "_tie_weights"):
            cell._tie_weights()

mindnlp.transformers.models.marian.modeling_marian.MarianModel

Bases: MarianPreTrainedModel

Source code in mindnlp/transformers/models/marian/modeling_marian.py

class MarianModel(MarianPreTrainedModel):
    _tied_weights_keys = ["encoder.embed_tokens.weight", "decoder.embed_tokens.weight"]

    def __init__(self, config: MarianConfig):
        super().__init__(config)

        padding_idx, vocab_size = config.pad_token_id, config.vocab_size

        # We always use self.shared for token embeddings to ensure compatibility with all marian models
        self.shared = nn.Embedding(vocab_size, config.d_model, padding_idx=padding_idx)
        if self.config.share_encoder_decoder_embeddings:
            encoder_embed_tokens = decoder_embed_tokens = self.shared
        else:
            # Since the embeddings are not shared, deepcopy the embeddings here for encoder
            # and decoder to make sure they are not tied.
            encoder_embed_tokens = copy.deepcopy(self.shared)
            decoder_embed_tokens = copy.deepcopy(self.shared)
            self.shared = None

        self.encoder = MarianEncoder(config, encoder_embed_tokens)
        self.decoder = MarianDecoder(config, decoder_embed_tokens)

        # Initialize weights and apply final processing
        self.post_init()

    def get_input_embeddings(self):
        # This will return shared embeddings if they are shared else specific to encoder.
        return self.get_encoder().get_input_embeddings()

    def set_input_embeddings(self, value):
        if self.config.share_encoder_decoder_embeddings:
            self.shared = value
            self.encoder.embed_tokens = self.shared
            self.decoder.embed_tokens = self.shared
        else:  # if not shared only set encoder embeedings
            self.encoder.embed_tokens = value

    def get_decoder_input_embeddings(self):
        if self.config.share_encoder_decoder_embeddings:
            raise ValueError(
                "`get_decoder_input_embeddings` should not be called if `config.share_encoder_decoder_embeddings` "
                "is `True`. Please use `get_input_embeddings` instead."
            )
        return self.get_decoder().get_input_embeddings()

    def set_decoder_input_embeddings(self, value):
        if self.config.share_encoder_decoder_embeddings:
            raise ValueError(
                "`config.share_encoder_decoder_embeddings` is set to `True` meaning the decoder input embeddings "
                "are shared with the encoder. In order to set the decoder input embeddings, you should simply set "
                "the encoder input embeddings by calling `set_input_embeddings` with the appropriate embeddings."
            )
        self.decoder.embed_tokens = value

    def get_encoder(self):
        return self.encoder

    def get_decoder(self):
        return self.decoder

    def resize_decoder_token_embeddings(self, new_num_tokens: int) -> nn.Embedding:
        if self.config.share_encoder_decoder_embeddings:
            raise ValueError(
                "`resize_decoder_token_embeddings` should not be called if `config.share_encoder_decoder_embeddings` "
                "is `True`. Please use `resize_token_embeddings` instead."
            )

        old_embeddings = self.get_decoder_input_embeddings()
        new_embeddings = self._get_resized_embeddings(old_embeddings, new_num_tokens)
        self.set_decoder_input_embeddings(new_embeddings)

        model_embeds = self.get_decoder_input_embeddings()

        if new_num_tokens is None:
            return model_embeds

        # Update base model and current model config
        self.config.decoder_vocab_size = new_num_tokens

        # Tie weights again if needed
        self.tie_weights()

        return model_embeds

    def forward(
        self,
        input_ids: mindspore.Tensor = None,
        attention_mask: Optional[mindspore.Tensor] = None,
        decoder_input_ids: Optional[mindspore.Tensor] = None,
        decoder_attention_mask: Optional[mindspore.Tensor] = None,
        head_mask: Optional[mindspore.Tensor] = None,
        decoder_head_mask: Optional[mindspore.Tensor] = None,
        cross_attn_head_mask: Optional[mindspore.Tensor] = None,
        encoder_outputs: Optional[Union[Tuple[mindspore.Tensor], BaseModelOutput]] = None,
        past_key_values: Optional[Tuple[Tuple[mindspore.Tensor]]] = None,
        inputs_embeds: Optional[mindspore.Tensor] = None,
        decoder_inputs_embeds: Optional[mindspore.Tensor] = None,
        use_cache: Optional[bool] = None,
        output_attentions: Optional[bool] = None,
        output_hidden_states: Optional[bool] = None,
        return_dict: Optional[bool] = None,
    ) -> Seq2SeqModelOutput:
        r"""

        Returns:
            `Seq2SeqModelOutput`

        Example:
            ```python
            >>> from transformers import AutoTokenizer, MarianModel
            ...
            >>> tokenizer = AutoTokenizer.from_pretrained("Helsinki-NLP/opus-mt-en-de")
            >>> model = MarianModel.from_pretrained("Helsinki-NLP/opus-mt-en-de")
            ...
            >>> inputs = tokenizer("Studies have been shown that owning a dog is good for you", return_tensors="pt")
            >>> decoder_inputs = tokenizer(
            ...     "<pad> Studien haben gezeigt dass es hilfreich ist einen Hund zu besitzen",
            ...     return_tensors="pt",
            ...     add_special_tokens=False,
            ... )
            >>> outputs = model(input_ids=inputs.input_ids, decoder_input_ids=decoder_inputs.input_ids)
            ...
            >>> last_hidden_states = outputs.last_hidden_state
            >>> list(last_hidden_states.shape)
            [1, 26, 512]
            ```
        """
        output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions
        output_hidden_states = (
            output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
        )
        use_cache = use_cache if use_cache is not None else self.config.use_cache
        return_dict = return_dict if return_dict is not None else self.config.use_return_dict

        if encoder_outputs is None:
            encoder_outputs = self.encoder(
                input_ids=input_ids,
                attention_mask=attention_mask,
                head_mask=head_mask,
                inputs_embeds=inputs_embeds,
                output_attentions=output_attentions,
                output_hidden_states=output_hidden_states,
                return_dict=return_dict,
            )
        # If the user passed a tuple for encoder_outputs, we wrap it in a BaseModelOutput when return_dict=True
        elif return_dict and not isinstance(encoder_outputs, BaseModelOutput):
            encoder_outputs = BaseModelOutput(
                last_hidden_state=encoder_outputs[0],
                hidden_states=encoder_outputs[1] if len(encoder_outputs) > 1 else None,
                attentions=encoder_outputs[2] if len(encoder_outputs) > 2 else None,
            )

        # decoder outputs consists of (dec_features, past_key_value, dec_hidden, dec_attn)
        decoder_outputs = self.decoder(
            input_ids=decoder_input_ids,
            attention_mask=decoder_attention_mask,
            encoder_hidden_states=encoder_outputs[0],
            encoder_attention_mask=attention_mask,
            head_mask=decoder_head_mask,
            cross_attn_head_mask=cross_attn_head_mask,
            past_key_values=past_key_values,
            inputs_embeds=decoder_inputs_embeds,
            use_cache=use_cache,
            output_attentions=output_attentions,
            output_hidden_states=output_hidden_states,
            return_dict=return_dict,
        )

        if not return_dict:
            return decoder_outputs + encoder_outputs

        return Seq2SeqModelOutput(
            last_hidden_state=decoder_outputs.last_hidden_state,
            past_key_values=decoder_outputs.past_key_values,
            decoder_hidden_states=decoder_outputs.hidden_states,
            decoder_attentions=decoder_outputs.attentions,
            cross_attentions=decoder_outputs.cross_attentions,
            encoder_last_hidden_state=encoder_outputs.last_hidden_state,
            encoder_hidden_states=encoder_outputs.hidden_states,
            encoder_attentions=encoder_outputs.attentions,
        )

mindnlp.transformers.models.marian.modeling_marian.MarianModel.forward(input_ids=None, attention_mask=None, decoder_input_ids=None, decoder_attention_mask=None, head_mask=None, decoder_head_mask=None, cross_attn_head_mask=None, encoder_outputs=None, past_key_values=None, inputs_embeds=None, decoder_inputs_embeds=None, use_cache=None, output_attentions=None, output_hidden_states=None, return_dict=None)

RETURNS	DESCRIPTION
Seq2SeqModelOutput	Seq2SeqModelOutput

Example

>>> from transformers import AutoTokenizer, MarianModel
...
>>> tokenizer = AutoTokenizer.from_pretrained("Helsinki-NLP/opus-mt-en-de")
>>> model = MarianModel.from_pretrained("Helsinki-NLP/opus-mt-en-de")
...
>>> inputs = tokenizer("Studies have been shown that owning a dog is good for you", return_tensors="pt")
>>> decoder_inputs = tokenizer(
...     "<pad> Studien haben gezeigt dass es hilfreich ist einen Hund zu besitzen",
...     return_tensors="pt",
...     add_special_tokens=False,
... )
>>> outputs = model(input_ids=inputs.input_ids, decoder_input_ids=decoder_inputs.input_ids)
...
>>> last_hidden_states = outputs.last_hidden_state
>>> list(last_hidden_states.shape)
[1, 26, 512]

Source code in mindnlp/transformers/models/marian/modeling_marian.py

def forward(
    self,
    input_ids: mindspore.Tensor = None,
    attention_mask: Optional[mindspore.Tensor] = None,
    decoder_input_ids: Optional[mindspore.Tensor] = None,
    decoder_attention_mask: Optional[mindspore.Tensor] = None,
    head_mask: Optional[mindspore.Tensor] = None,
    decoder_head_mask: Optional[mindspore.Tensor] = None,
    cross_attn_head_mask: Optional[mindspore.Tensor] = None,
    encoder_outputs: Optional[Union[Tuple[mindspore.Tensor], BaseModelOutput]] = None,
    past_key_values: Optional[Tuple[Tuple[mindspore.Tensor]]] = None,
    inputs_embeds: Optional[mindspore.Tensor] = None,
    decoder_inputs_embeds: Optional[mindspore.Tensor] = None,
    use_cache: Optional[bool] = None,
    output_attentions: Optional[bool] = None,
    output_hidden_states: Optional[bool] = None,
    return_dict: Optional[bool] = None,
) -> Seq2SeqModelOutput:
    r"""

    Returns:
        `Seq2SeqModelOutput`

    Example:
        ```python
        >>> from transformers import AutoTokenizer, MarianModel
        ...
        >>> tokenizer = AutoTokenizer.from_pretrained("Helsinki-NLP/opus-mt-en-de")
        >>> model = MarianModel.from_pretrained("Helsinki-NLP/opus-mt-en-de")
        ...
        >>> inputs = tokenizer("Studies have been shown that owning a dog is good for you", return_tensors="pt")
        >>> decoder_inputs = tokenizer(
        ...     "<pad> Studien haben gezeigt dass es hilfreich ist einen Hund zu besitzen",
        ...     return_tensors="pt",
        ...     add_special_tokens=False,
        ... )
        >>> outputs = model(input_ids=inputs.input_ids, decoder_input_ids=decoder_inputs.input_ids)
        ...
        >>> last_hidden_states = outputs.last_hidden_state
        >>> list(last_hidden_states.shape)
        [1, 26, 512]
        ```
    """
    output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions
    output_hidden_states = (
        output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
    )
    use_cache = use_cache if use_cache is not None else self.config.use_cache
    return_dict = return_dict if return_dict is not None else self.config.use_return_dict

    if encoder_outputs is None:
        encoder_outputs = self.encoder(
            input_ids=input_ids,
            attention_mask=attention_mask,
            head_mask=head_mask,
            inputs_embeds=inputs_embeds,
            output_attentions=output_attentions,
            output_hidden_states=output_hidden_states,
            return_dict=return_dict,
        )
    # If the user passed a tuple for encoder_outputs, we wrap it in a BaseModelOutput when return_dict=True
    elif return_dict and not isinstance(encoder_outputs, BaseModelOutput):
        encoder_outputs = BaseModelOutput(
            last_hidden_state=encoder_outputs[0],
            hidden_states=encoder_outputs[1] if len(encoder_outputs) > 1 else None,
            attentions=encoder_outputs[2] if len(encoder_outputs) > 2 else None,
        )

    # decoder outputs consists of (dec_features, past_key_value, dec_hidden, dec_attn)
    decoder_outputs = self.decoder(
        input_ids=decoder_input_ids,
        attention_mask=decoder_attention_mask,
        encoder_hidden_states=encoder_outputs[0],
        encoder_attention_mask=attention_mask,
        head_mask=decoder_head_mask,
        cross_attn_head_mask=cross_attn_head_mask,
        past_key_values=past_key_values,
        inputs_embeds=decoder_inputs_embeds,
        use_cache=use_cache,
        output_attentions=output_attentions,
        output_hidden_states=output_hidden_states,
        return_dict=return_dict,
    )

    if not return_dict:
        return decoder_outputs + encoder_outputs

    return Seq2SeqModelOutput(
        last_hidden_state=decoder_outputs.last_hidden_state,
        past_key_values=decoder_outputs.past_key_values,
        decoder_hidden_states=decoder_outputs.hidden_states,
        decoder_attentions=decoder_outputs.attentions,
        cross_attentions=decoder_outputs.cross_attentions,
        encoder_last_hidden_state=encoder_outputs.last_hidden_state,
        encoder_hidden_states=encoder_outputs.hidden_states,
        encoder_attentions=encoder_outputs.attentions,
    )

mindnlp.transformers.models.marian.modeling_marian.MarianSinusoidalPositionalEmbedding

Bases: Embedding

This module produces sinusoidal positional embeddings of any length.

Source code in mindnlp/transformers/models/marian/modeling_marian.py

class MarianSinusoidalPositionalEmbedding(nn.Embedding):
    """This module produces sinusoidal positional embeddings of any length."""

    def __init__(self, num_positions: int, embedding_dim: int, padding_idx: Optional[int] = None) -> None:
        super().__init__(num_positions, embedding_dim)
        self.weight = self._init_weight(self.weight)

    @staticmethod
    def _init_weight(out: mindspore.Parameter) -> mindspore.Parameter:
        """
        Identical to the XLM create_sinusoidal_embeddings except features are not interleaved. The cos features are in
        the 2nd half of the vector. [dim // 2:]
        """
        n_pos, dim = out.shape
        position_enc = np.array(
            [[pos / np.power(10000, 2 * (j // 2) / dim) for j in range(dim)] for pos in range(n_pos)]
        )
        out.requires_grad = False  # set early to avoid an error in pytorch-1.8+
        sentinel = dim // 2 if dim % 2 == 0 else (dim // 2) + 1
        out[:, 0:sentinel] = Tensor(np.sin(position_enc[:, 0::2]))
        out[:, sentinel:] = Tensor(np.cos(position_enc[:, 1::2]))
        out.stop_gradient = True
        return out

    @mindspore._no_grad()
    def forward(self, input_ids_shape, past_key_values_length: int = 0) -> mindspore.Tensor:
        """`input_ids_shape` is expected to be [bsz x seqlen]."""
        bsz, seq_len = input_ids_shape[:2]
        positions = ops.arange(
            past_key_values_length, past_key_values_length + seq_len, dtype=mindspore.int64
        )
        return super().forward(positions)

mindnlp.transformers.models.marian.modeling_marian.MarianSinusoidalPositionalEmbedding.forward(input_ids_shape, past_key_values_length=0)

input_ids_shape is expected to be [bsz x seqlen].

Source code in mindnlp/transformers/models/marian/modeling_marian.py

@mindspore._no_grad()
def forward(self, input_ids_shape, past_key_values_length: int = 0) -> mindspore.Tensor:
    """`input_ids_shape` is expected to be [bsz x seqlen]."""
    bsz, seq_len = input_ids_shape[:2]
    positions = ops.arange(
        past_key_values_length, past_key_values_length + seq_len, dtype=mindspore.int64
    )
    return super().forward(positions)

mindnlp.transformers.models.marian.modeling_marian.shift_tokens_right(input_ids, pad_token_id, decoder_start_token_id)

Shift input ids one token to the right.

Source code in mindnlp/transformers/models/marian/modeling_marian.py

def shift_tokens_right(input_ids: mindspore.Tensor, pad_token_id: int, decoder_start_token_id: int):
    """
    Shift input ids one token to the right.
    """
    shifted_input_ids = input_ids.new_zeros(input_ids.shape)
    shifted_input_ids[:, 1:] = input_ids[:, :-1].copy()
    shifted_input_ids[:, 0] = decoder_start_token_id

    if pad_token_id is None:
        raise ValueError("self.model.config.pad_token_id has to be defined.")
    # replace possible -100 values in labels by `pad_token_id`
    shifted_input_ids = shifted_input_ids.masked_fill(shifted_input_ids == -100, pad_token_id)

    return shifted_input_ids

mindnlp.transformers.models.marian.tokenization_marian

Tokenization classes for Marian.

mindnlp.transformers.models.marian.tokenization_marian.MarianTokenizer

Bases: PreTrainedTokenizer

Construct a Marian tokenizer. Based on SentencePiece.

This tokenizer inherits from [PreTrainedTokenizer] which contains most of the main methods. Users should refer to this superclass for more information regarding those methods.

PARAMETER	DESCRIPTION
source_spm	SentencePiece file (generally has a .spm extension) that contains the vocabulary for the source language. TYPE: `str`
target_spm	SentencePiece file (generally has a .spm extension) that contains the vocabulary for the target language. TYPE: `str`
source_lang	A string representing the source language. TYPE: `str`, *optional* DEFAULT: None
target_lang	A string representing the target language. TYPE: `str`, *optional* DEFAULT: None
unk_token	The unknown token. A token that is not in the vocabulary cannot be converted to an ID and is set to be this token instead. TYPE: `str`, *optional*, defaults to `"<unk>"` DEFAULT: '<unk>'
eos_token	The end of sequence token. TYPE: `str`, *optional*, defaults to `"</s>"` DEFAULT: '</s>'
pad_token	The token used for padding, for example when batching sequences of different lengths. TYPE: `str`, *optional*, defaults to `"<pad>"` DEFAULT: '<pad>'
model_max_length	The maximum sentence length the model accepts. TYPE: `int`, *optional*, defaults to 512 DEFAULT: 512
additional_special_tokens	Additional special tokens used by the tokenizer. TYPE: `List[str]`, *optional*, defaults to `["<eop>", "<eod>"]`
sp_model_kwargs	Will be passed to the SentencePieceProcessor.__init__() method. The Python wrapper for SentencePiece can be used, among other things, to set: enable_sampling: Enable subword regularization. nbest_size: Sampling parameters for unigram. Invalid for BPE-Dropout. nbest_size = {0,1}: No sampling is performed. nbest_size > 1: samples from the nbest_size results. nbest_size < 0: assuming that nbest_size is infinite and samples from the all hypothesis (lattice) using forward-filtering-and-backward-sampling algorithm. alpha: Smoothing parameter for unigram sampling, and dropout probability of merge operations for BPE-dropout. TYPE: `dict`, *optional* DEFAULT: None

Example

>>> from transformers import MarianForCausalLM, MarianTokenizer
...
>>> model = MarianForCausalLM.from_pretrained("Helsinki-NLP/opus-mt-en-de")
>>> tokenizer = MarianTokenizer.from_pretrained("Helsinki-NLP/opus-mt-en-de")
>>> src_texts = ["I am a small frog.", "Tom asked his teacher for advice."]
>>> tgt_texts = ["Ich bin ein kleiner Frosch.", "Tom bat seinen Lehrer um Rat."]  # optional
>>> inputs = tokenizer(src_texts, text_target=tgt_texts, return_tensors="pt", padding=True)
...
>>> outputs = model(**inputs)  # should work

Source code in mindnlp/transformers/models/marian/tokenization_marian.py

class MarianTokenizer(PreTrainedTokenizer):
    r"""
    Construct a Marian tokenizer. Based on [SentencePiece](https://github.com/google/sentencepiece).

    This tokenizer inherits from [`PreTrainedTokenizer`] which contains most of the main methods. Users should refer to
    this superclass for more information regarding those methods.

    Args:
        source_spm (`str`):
            [SentencePiece](https://github.com/google/sentencepiece) file (generally has a .spm extension) that
            contains the vocabulary for the source language.
        target_spm (`str`):
            [SentencePiece](https://github.com/google/sentencepiece) file (generally has a .spm extension) that
            contains the vocabulary for the target language.
        source_lang (`str`, *optional*):
            A string representing the source language.
        target_lang (`str`, *optional*):
            A string representing the target language.
        unk_token (`str`, *optional*, defaults to `"<unk>"`):
            The unknown token. A token that is not in the vocabulary cannot be converted to an ID and is set to be this
            token instead.
        eos_token (`str`, *optional*, defaults to `"</s>"`):
            The end of sequence token.
        pad_token (`str`, *optional*, defaults to `"<pad>"`):
            The token used for padding, for example when batching sequences of different lengths.
        model_max_length (`int`, *optional*, defaults to 512):
            The maximum sentence length the model accepts.
        additional_special_tokens (`List[str]`, *optional*, defaults to `["<eop>", "<eod>"]`):
            Additional special tokens used by the tokenizer.
        sp_model_kwargs (`dict`, *optional*):
            Will be passed to the `SentencePieceProcessor.__init__()` method. The [Python wrapper for
            SentencePiece](https://github.com/google/sentencepiece/tree/master/python) can be used, among other things,
            to set:

            - `enable_sampling`: Enable subword regularization.
            - `nbest_size`: Sampling parameters for unigram. Invalid for BPE-Dropout.

                - `nbest_size = {0,1}`: No sampling is performed.
                - `nbest_size > 1`: samples from the nbest_size results.
                - `nbest_size < 0`: assuming that nbest_size is infinite and samples from the all hypothesis (lattice)
                using forward-filtering-and-backward-sampling algorithm.
            - `alpha`: Smoothing parameter for unigram sampling, and dropout probability of merge operations for
            BPE-dropout.

    Example:
        ```python
        >>> from transformers import MarianForCausalLM, MarianTokenizer
        ...
        >>> model = MarianForCausalLM.from_pretrained("Helsinki-NLP/opus-mt-en-de")
        >>> tokenizer = MarianTokenizer.from_pretrained("Helsinki-NLP/opus-mt-en-de")
        >>> src_texts = ["I am a small frog.", "Tom asked his teacher for advice."]
        >>> tgt_texts = ["Ich bin ein kleiner Frosch.", "Tom bat seinen Lehrer um Rat."]  # optional
        >>> inputs = tokenizer(src_texts, text_target=tgt_texts, return_tensors="pt", padding=True)
        ...
        >>> outputs = model(**inputs)  # should work
        ```
    """

    vocab_files_names = VOCAB_FILES_NAMES
    model_input_names = ["input_ids", "attention_mask"]
    language_code_re = re.compile(">>.+<<")  # type: re.Pattern

    def __init__(
        self,
        source_spm,
        target_spm,
        vocab,
        target_vocab_file=None,
        source_lang=None,
        target_lang=None,
        unk_token="<unk>",
        eos_token="</s>",
        pad_token="<pad>",
        model_max_length=512,
        sp_model_kwargs: Optional[Dict[str, Any]] = None,
        separate_vocabs=False,
        **kwargs,
    ) -> None:
        self.sp_model_kwargs = {} if sp_model_kwargs is None else sp_model_kwargs

        assert Path(source_spm).exists(), f"cannot find spm source {source_spm}"

        self.separate_vocabs = separate_vocabs
        self.encoder = load_json(vocab)
        if str(unk_token) not in self.encoder:
            raise KeyError("<unk> token must be in the vocab")
        assert str(pad_token) in self.encoder

        if separate_vocabs:
            self.target_encoder = load_json(target_vocab_file)
            self.decoder = {v: k for k, v in self.target_encoder.items()}
            self.supported_language_codes = []
        else:
            self.decoder = {v: k for k, v in self.encoder.items()}
            self.supported_language_codes: list = [k for k in self.encoder if k.startswith(">>") and k.endswith("<<")]

        self.source_lang = source_lang
        self.target_lang = target_lang
        self.spm_files = [source_spm, target_spm]

        # load SentencePiece model for pre-processing
        self.spm_source = load_spm(source_spm, self.sp_model_kwargs)
        self.spm_target = load_spm(target_spm, self.sp_model_kwargs)
        self.current_spm = self.spm_source
        self.current_encoder = self.encoder

        # Multilingual target side: default to using first supported language code.

        self._setup_normalizer()

        super().__init__(
            # bos_token=bos_token,  unused. Start decoding with config.decoder_start_token_id
            source_lang=source_lang,
            target_lang=target_lang,
            unk_token=unk_token,
            eos_token=eos_token,
            pad_token=pad_token,
            model_max_length=model_max_length,
            sp_model_kwargs=self.sp_model_kwargs,
            target_vocab_file=target_vocab_file,
            separate_vocabs=separate_vocabs,
            **kwargs,
        )

    def _setup_normalizer(self):
        try:
            from sacremoses import MosesPunctNormalizer

            self.punc_normalizer = MosesPunctNormalizer(self.source_lang).normalize
        except (ImportError, FileNotFoundError):
            warnings.warn("Recommended: pip install sacremoses.")
            self.punc_normalizer = lambda x: x

    def normalize(self, x: str) -> str:
        """Cover moses empty string edge case. They return empty list for '' input!"""
        return self.punc_normalizer(x) if x else ""

    def _convert_token_to_id(self, token):
        return self.current_encoder.get(token, self.current_encoder[self.unk_token])

    def remove_language_code(self, text: str):
        """Remove language codes like >>fr<< before sentencepiece"""
        match = self.language_code_re.match(text)
        code: list = [match.group(0)] if match else []
        return code, self.language_code_re.sub("", text)

    def _tokenize(self, text: str) -> List[str]:
        code, text = self.remove_language_code(text)
        pieces = self.current_spm.encode(text, out_type=str)
        return code + pieces

    def _convert_id_to_token(self, index: int) -> str:
        """Converts an index (integer) in a token (str) using the decoder."""
        return self.decoder.get(index, self.unk_token)

    def batch_decode(self, sequences, **kwargs):
        """
        Convert a list of lists of token ids into a list of strings by calling decode.

        Args:
            sequences (`Union[List[int], List[List[int]], np.ndarray, torch.Tensor, tf.Tensor]`):
                List of tokenized input ids. Can be obtained using the `__call__` method.
            skip_special_tokens (`bool`, *optional*, defaults to `False`):
                Whether or not to remove special tokens in the decoding.
            clean_up_tokenization_spaces (`bool`, *optional*):
                Whether or not to clean up the tokenization spaces. If `None`, will default to
                `self.clean_up_tokenization_spaces` (available in the `tokenizer_config`).
            use_source_tokenizer (`bool`, *optional*, defaults to `False`):
                Whether or not to use the source tokenizer to decode sequences (only applicable in sequence-to-sequence
                problems).
            kwargs (additional keyword arguments, *optional*):
                Will be passed to the underlying model specific decode method.

        Returns:
            `List[str]`: The list of decoded sentences.
        """
        return super().batch_decode(sequences, **kwargs)

    def decode(self, token_ids, **kwargs):
        """
        Converts a sequence of ids in a string, using the tokenizer and vocabulary with options to remove special
        tokens and clean up tokenization spaces.

        Similar to doing `self.convert_tokens_to_string(self.convert_ids_to_tokens(token_ids))`.

        Args:
            token_ids (`Union[int, List[int], np.ndarray, torch.Tensor, tf.Tensor]`):
                List of tokenized input ids. Can be obtained using the `__call__` method.
            skip_special_tokens (`bool`, *optional*, defaults to `False`):
                Whether or not to remove special tokens in the decoding.
            clean_up_tokenization_spaces (`bool`, *optional*):
                Whether or not to clean up the tokenization spaces. If `None`, will default to
                `self.clean_up_tokenization_spaces` (available in the `tokenizer_config`).
            use_source_tokenizer (`bool`, *optional*, defaults to `False`):
                Whether or not to use the source tokenizer to decode sequences (only applicable in sequence-to-sequence
                problems).
            kwargs (additional keyword arguments, *optional*):
                Will be passed to the underlying model specific decode method.

        Returns:
            `str`: The decoded sentence.
        """
        return super().decode(token_ids, **kwargs)

    def convert_tokens_to_string(self, tokens: List[str]) -> str:
        """Uses source spm if _decode_use_source_tokenizer is True, and target spm otherwise"""
        sp_model = self.spm_source if self._decode_use_source_tokenizer else self.spm_target
        current_sub_tokens = []
        out_string = ""
        for token in tokens:
            # make sure that special tokens are not decoded using sentencepiece model
            if token in self.all_special_tokens:
                out_string += sp_model.decode_pieces(current_sub_tokens) + token + " "
                current_sub_tokens = []
            else:
                current_sub_tokens.append(token)
        out_string += sp_model.decode_pieces(current_sub_tokens)
        out_string = out_string.replace(SPIECE_UNDERLINE, " ")
        return out_string.strip()

    def build_inputs_with_special_tokens(self, token_ids_0, token_ids_1=None) -> List[int]:
        """Build model inputs from a sequence by appending eos_token_id."""
        if token_ids_1 is None:
            return token_ids_0 + [self.eos_token_id]
        # We don't expect to process pairs, but leave the pair logic for API consistency
        return token_ids_0 + token_ids_1 + [self.eos_token_id]

    def _switch_to_input_mode(self):
        self.current_spm = self.spm_source
        self.current_encoder = self.encoder

    def _switch_to_target_mode(self):
        self.current_spm = self.spm_target
        if self.separate_vocabs:
            self.current_encoder = self.target_encoder

    @property
    def vocab_size(self) -> int:
        return len(self.encoder)

    def save_vocabulary(self, save_directory: str, filename_prefix: Optional[str] = None) -> Tuple[str]:
        if not os.path.isdir(save_directory):
            logger.error(f"Vocabulary path ({save_directory}) should be a directory")
            return
        saved_files = []

        if self.separate_vocabs:
            out_src_vocab_file = os.path.join(
                save_directory,
                (filename_prefix + "-" if filename_prefix else "") + VOCAB_FILES_NAMES["vocab"],
            )
            out_tgt_vocab_file = os.path.join(
                save_directory,
                (filename_prefix + "-" if filename_prefix else "") + VOCAB_FILES_NAMES["target_vocab_file"],
            )
            save_json(self.encoder, out_src_vocab_file)
            save_json(self.target_encoder, out_tgt_vocab_file)
            saved_files.append(out_src_vocab_file)
            saved_files.append(out_tgt_vocab_file)
        else:
            out_vocab_file = os.path.join(
                save_directory, (filename_prefix + "-" if filename_prefix else "") + VOCAB_FILES_NAMES["vocab"]
            )
            save_json(self.encoder, out_vocab_file)
            saved_files.append(out_vocab_file)

        for spm_save_filename, spm_orig_path, spm_model in zip(
            [VOCAB_FILES_NAMES["source_spm"], VOCAB_FILES_NAMES["target_spm"]],
            self.spm_files,
            [self.spm_source, self.spm_target],
        ):
            spm_save_path = os.path.join(
                save_directory, (filename_prefix + "-" if filename_prefix else "") + spm_save_filename
            )
            if os.path.abspath(spm_orig_path) != os.path.abspath(spm_save_path) and os.path.isfile(spm_orig_path):
                copyfile(spm_orig_path, spm_save_path)
                saved_files.append(spm_save_path)
            elif not os.path.isfile(spm_orig_path):
                with open(spm_save_path, "wb") as fi:
                    content_spiece_model = spm_model.serialized_model_proto()
                    fi.write(content_spiece_model)
                saved_files.append(spm_save_path)

        return tuple(saved_files)

    def get_vocab(self) -> Dict:
        return self.get_src_vocab()

    def get_src_vocab(self):
        return dict(self.encoder, **self.added_tokens_encoder)

    def get_tgt_vocab(self):
        return dict(self.target_encoder, **self.added_tokens_decoder)

    def __getstate__(self) -> Dict:
        state = self.__dict__.copy()
        state.update(
            {k: None for k in ["spm_source", "spm_target", "current_spm", "punc_normalizer", "target_vocab_file"]}
        )
        return state

    def __setstate__(self, d: Dict) -> None:
        self.__dict__ = d

        # for backward compatibility
        if not hasattr(self, "sp_model_kwargs"):
            self.sp_model_kwargs = {}

        self.spm_source, self.spm_target = (load_spm(f, self.sp_model_kwargs) for f in self.spm_files)
        self.current_spm = self.spm_source
        self._setup_normalizer()

    def num_special_tokens_to_add(self, *args, **kwargs):
        """Just EOS"""
        return 1

    def _special_token_mask(self, seq):
        all_special_ids = set(self.all_special_ids)  # call it once instead of inside list comp
        all_special_ids.remove(self.unk_token_id)  # <unk> is only sometimes special
        return [1 if x in all_special_ids else 0 for x in seq]

    def get_special_tokens_mask(
        self, token_ids_0: List, token_ids_1: Optional[List] = None, already_has_special_tokens: bool = False
    ) -> List[int]:
        """Get list where entries are [1] if a token is [eos] or [pad] else 0."""
        if already_has_special_tokens:
            return self._special_token_mask(token_ids_0)
        elif token_ids_1 is None:
            return self._special_token_mask(token_ids_0) + [1]
        else:
            return self._special_token_mask(token_ids_0 + token_ids_1) + [1]

mindnlp.transformers.models.marian.tokenization_marian.MarianTokenizer.batch_decode(sequences, **kwargs)

Convert a list of lists of token ids into a list of strings by calling decode.

PARAMETER	DESCRIPTION
sequences	List of tokenized input ids. Can be obtained using the __call__ method. TYPE: `Union[List[int], List[List[int]], np.ndarray, torch.Tensor, tf.Tensor]`
skip_special_tokens	Whether or not to remove special tokens in the decoding. TYPE: `bool`, *optional*, defaults to `False`
clean_up_tokenization_spaces	Whether or not to clean up the tokenization spaces. If None, will default to self.clean_up_tokenization_spaces (available in the tokenizer_config). TYPE: `bool`, *optional*
use_source_tokenizer	Whether or not to use the source tokenizer to decode sequences (only applicable in sequence-to-sequence problems). TYPE: `bool`, *optional*, defaults to `False`
kwargs	Will be passed to the underlying model specific decode method. TYPE: additional keyword arguments, *optional* DEFAULT: {}

RETURNS	DESCRIPTION
	List[str]: The list of decoded sentences.

Source code in mindnlp/transformers/models/marian/tokenization_marian.py

def batch_decode(self, sequences, **kwargs):
    """
    Convert a list of lists of token ids into a list of strings by calling decode.

    Args:
        sequences (`Union[List[int], List[List[int]], np.ndarray, torch.Tensor, tf.Tensor]`):
            List of tokenized input ids. Can be obtained using the `__call__` method.
        skip_special_tokens (`bool`, *optional*, defaults to `False`):
            Whether or not to remove special tokens in the decoding.
        clean_up_tokenization_spaces (`bool`, *optional*):
            Whether or not to clean up the tokenization spaces. If `None`, will default to
            `self.clean_up_tokenization_spaces` (available in the `tokenizer_config`).
        use_source_tokenizer (`bool`, *optional*, defaults to `False`):
            Whether or not to use the source tokenizer to decode sequences (only applicable in sequence-to-sequence
            problems).
        kwargs (additional keyword arguments, *optional*):
            Will be passed to the underlying model specific decode method.

    Returns:
        `List[str]`: The list of decoded sentences.
    """
    return super().batch_decode(sequences, **kwargs)

mindnlp.transformers.models.marian.tokenization_marian.MarianTokenizer.build_inputs_with_special_tokens(token_ids_0, token_ids_1=None)

Build model inputs from a sequence by appending eos_token_id.

Source code in mindnlp/transformers/models/marian/tokenization_marian.py

def build_inputs_with_special_tokens(self, token_ids_0, token_ids_1=None) -> List[int]:
    """Build model inputs from a sequence by appending eos_token_id."""
    if token_ids_1 is None:
        return token_ids_0 + [self.eos_token_id]
    # We don't expect to process pairs, but leave the pair logic for API consistency
    return token_ids_0 + token_ids_1 + [self.eos_token_id]

mindnlp.transformers.models.marian.tokenization_marian.MarianTokenizer.convert_tokens_to_string(tokens)

Uses source spm if _decode_use_source_tokenizer is True, and target spm otherwise

Source code in mindnlp/transformers/models/marian/tokenization_marian.py

def convert_tokens_to_string(self, tokens: List[str]) -> str:
    """Uses source spm if _decode_use_source_tokenizer is True, and target spm otherwise"""
    sp_model = self.spm_source if self._decode_use_source_tokenizer else self.spm_target
    current_sub_tokens = []
    out_string = ""
    for token in tokens:
        # make sure that special tokens are not decoded using sentencepiece model
        if token in self.all_special_tokens:
            out_string += sp_model.decode_pieces(current_sub_tokens) + token + " "
            current_sub_tokens = []
        else:
            current_sub_tokens.append(token)
    out_string += sp_model.decode_pieces(current_sub_tokens)
    out_string = out_string.replace(SPIECE_UNDERLINE, " ")
    return out_string.strip()

mindnlp.transformers.models.marian.tokenization_marian.MarianTokenizer.decode(token_ids, **kwargs)

Converts a sequence of ids in a string, using the tokenizer and vocabulary with options to remove special tokens and clean up tokenization spaces.

Similar to doing self.convert_tokens_to_string(self.convert_ids_to_tokens(token_ids)).

PARAMETER	DESCRIPTION
token_ids	List of tokenized input ids. Can be obtained using the __call__ method. TYPE: `Union[int, List[int], np.ndarray, torch.Tensor, tf.Tensor]`
skip_special_tokens	Whether or not to remove special tokens in the decoding. TYPE: `bool`, *optional*, defaults to `False`
clean_up_tokenization_spaces	Whether or not to clean up the tokenization spaces. If None, will default to self.clean_up_tokenization_spaces (available in the tokenizer_config). TYPE: `bool`, *optional*
use_source_tokenizer	Whether or not to use the source tokenizer to decode sequences (only applicable in sequence-to-sequence problems). TYPE: `bool`, *optional*, defaults to `False`
kwargs	Will be passed to the underlying model specific decode method. TYPE: additional keyword arguments, *optional* DEFAULT: {}

RETURNS	DESCRIPTION
	str: The decoded sentence.

Source code in mindnlp/transformers/models/marian/tokenization_marian.py

def decode(self, token_ids, **kwargs):
    """
    Converts a sequence of ids in a string, using the tokenizer and vocabulary with options to remove special
    tokens and clean up tokenization spaces.

    Similar to doing `self.convert_tokens_to_string(self.convert_ids_to_tokens(token_ids))`.

    Args:
        token_ids (`Union[int, List[int], np.ndarray, torch.Tensor, tf.Tensor]`):
            List of tokenized input ids. Can be obtained using the `__call__` method.
        skip_special_tokens (`bool`, *optional*, defaults to `False`):
            Whether or not to remove special tokens in the decoding.
        clean_up_tokenization_spaces (`bool`, *optional*):
            Whether or not to clean up the tokenization spaces. If `None`, will default to
            `self.clean_up_tokenization_spaces` (available in the `tokenizer_config`).
        use_source_tokenizer (`bool`, *optional*, defaults to `False`):
            Whether or not to use the source tokenizer to decode sequences (only applicable in sequence-to-sequence
            problems).
        kwargs (additional keyword arguments, *optional*):
            Will be passed to the underlying model specific decode method.

    Returns:
        `str`: The decoded sentence.
    """
    return super().decode(token_ids, **kwargs)

mindnlp.transformers.models.marian.tokenization_marian.MarianTokenizer.get_special_tokens_mask(token_ids_0, token_ids_1=None, already_has_special_tokens=False)

Get list where entries are [1] if a token is [eos] or [pad] else 0.

Source code in mindnlp/transformers/models/marian/tokenization_marian.py

def get_special_tokens_mask(
    self, token_ids_0: List, token_ids_1: Optional[List] = None, already_has_special_tokens: bool = False
) -> List[int]:
    """Get list where entries are [1] if a token is [eos] or [pad] else 0."""
    if already_has_special_tokens:
        return self._special_token_mask(token_ids_0)
    elif token_ids_1 is None:
        return self._special_token_mask(token_ids_0) + [1]
    else:
        return self._special_token_mask(token_ids_0 + token_ids_1) + [1]

mindnlp.transformers.models.marian.tokenization_marian.MarianTokenizer.normalize(x)

Cover moses empty string edge case. They return empty list for '' input!

Source code in mindnlp/transformers/models/marian/tokenization_marian.py

def normalize(self, x: str) -> str:
    """Cover moses empty string edge case. They return empty list for '' input!"""
    return self.punc_normalizer(x) if x else ""

mindnlp.transformers.models.marian.tokenization_marian.MarianTokenizer.num_special_tokens_to_add(*args, **kwargs)

Just EOS

Source code in mindnlp/transformers/models/marian/tokenization_marian.py

def num_special_tokens_to_add(self, *args, **kwargs):
    """Just EOS"""
    return 1

mindnlp.transformers.models.marian.tokenization_marian.MarianTokenizer.remove_language_code(text)

Remove language codes like >>fr<< before sentencepiece

Source code in mindnlp/transformers/models/marian/tokenization_marian.py

def remove_language_code(self, text: str):
    """Remove language codes like >>fr<< before sentencepiece"""
    match = self.language_code_re.match(text)
    code: list = [match.group(0)] if match else []
    return code, self.language_code_re.sub("", text)