Skip to content

visual_bert

mindnlp.transformers.models.visual_bert.configuration_visual_bert

VisualBERT model configuration

mindnlp.transformers.models.visual_bert.configuration_visual_bert.VisualBertConfig

Bases: PretrainedConfig

This is the configuration class to store the configuration of a [VisualBertModel]. It is used to instantiate an VisualBERT 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 VisualBERT uclanlp/visualbert-vqa-coco-pre 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 VisualBERT model. Defines the number of different tokens that can be represented by the inputs_ids passed when calling [VisualBertModel]. Vocabulary size of the model. Defines the different tokens that can be represented by the inputs_ids passed to the forward method of [VisualBertModel].

TYPE: `int`, *optional*, defaults to 30522 DEFAULT: 30522

hidden_size

Dimensionality of the encoder layers and the pooler layer.

TYPE: `int`, *optional*, defaults to 768 DEFAULT: 768

visual_embedding_dim

Dimensionality of the visual embeddings to be passed to the model.

TYPE: `int`, *optional*, defaults to 512 DEFAULT: 512

num_hidden_layers

Number of hidden layers in the Transformer encoder.

TYPE: `int`, *optional*, defaults to 12 DEFAULT: 12

num_attention_heads

Number of attention heads for each attention layer in the Transformer encoder.

TYPE: `int`, *optional*, defaults to 12 DEFAULT: 12

intermediate_size

Dimensionality of the "intermediate" (i.e., feed-forward) layer in the Transformer encoder.

TYPE: `int`, *optional*, defaults to 3072 DEFAULT: 3072

hidden_act

The non-linear activation function (function or string) in the encoder and pooler. If string, "gelu", "relu", "selu" and "gelu_new" are supported.

TYPE: `str` or `function`, *optional*, defaults to `"gelu"` DEFAULT: 'gelu'

hidden_dropout_prob

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_probs_dropout_prob

The dropout ratio for the attention probabilities.

TYPE: `float`, *optional*, defaults to 0.1 DEFAULT: 0.1

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 512 DEFAULT: 512

type_vocab_size

The vocabulary size of the token_type_ids passed when calling [VisualBertModel].

TYPE: `int`, *optional*, defaults to 2 DEFAULT: 2

initializer_range

The standard deviation of the truncated_normal_initializer for initializing all weight matrices.

TYPE: `float`, *optional*, defaults to 0.02 DEFAULT: 0.02

layer_norm_eps

The epsilon used by the layer normalization layers.

TYPE: `float`, *optional*, defaults to 1e-12 DEFAULT: 1e-12

bypass_transformer

Whether or not the model should bypass the transformer for the visual embeddings. If set to True, the model directly concatenates the visual embeddings from [VisualBertEmbeddings] with text output from transformers, and then pass it to a self-attention layer.

TYPE: `bool`, *optional*, defaults to `False` DEFAULT: False

special_visual_initialize

Whether or not the visual token type and position type embedding weights should be initialized the same as the textual token type and positive type embeddings. When set to True, the weights of the textual token type and position type embeddings are copied to the respective visual embedding layers.

TYPE: `bool`, *optional*, defaults to `True` DEFAULT: True

Example 
>>> from transformers import VisualBertConfig, VisualBertModel
...
>>> # Initializing a VisualBERT visualbert-vqa-coco-pre style configuration
>>> configuration = VisualBertConfig.from_pretrained("uclanlp/visualbert-vqa-coco-pre")
...
>>> # Initializing a model (with random weights) from the visualbert-vqa-coco-pre style configuration
>>> model = VisualBertModel(configuration)
...
>>> # Accessing the model configuration
>>> configuration = model.config
Source code in mindnlp/transformers/models/visual_bert/configuration_visual_bert.py 
 25
 26
 27
 28
 29
 30
 31
 32
 33
 34
 35
 36
 37
 38
 39
 40
 41
 42
 43
 44
 45
 46
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
class VisualBertConfig(PretrainedConfig):
    r"""
    This is the configuration class to store the configuration of a [`VisualBertModel`]. It is used to instantiate an
    VisualBERT 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 VisualBERT
    [uclanlp/visualbert-vqa-coco-pre](https://huggingface.co/uclanlp/visualbert-vqa-coco-pre) 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 30522):
            Vocabulary size of the VisualBERT model. Defines the number of different tokens that can be represented by
            the `inputs_ids` passed when calling [`VisualBertModel`]. Vocabulary size of the model. Defines the
            different tokens that can be represented by the `inputs_ids` passed to the forward method of
            [`VisualBertModel`].
        hidden_size (`int`, *optional*, defaults to 768):
            Dimensionality of the encoder layers and the pooler layer.
        visual_embedding_dim (`int`, *optional*, defaults to 512):
            Dimensionality of the visual embeddings to be passed to the model.
        num_hidden_layers (`int`, *optional*, defaults to 12):
            Number of hidden layers in the Transformer encoder.
        num_attention_heads (`int`, *optional*, defaults to 12):
            Number of attention heads for each attention layer in the Transformer encoder.
        intermediate_size (`int`, *optional*, defaults to 3072):
            Dimensionality of the "intermediate" (i.e., feed-forward) layer in the Transformer encoder.
        hidden_act (`str` or `function`, *optional*, defaults to `"gelu"`):
            The non-linear activation function (function or string) in the encoder and pooler. If string, `"gelu"`,
            `"relu"`, `"selu"` and `"gelu_new"` are supported.
        hidden_dropout_prob (`float`, *optional*, defaults to 0.1):
            The dropout probability for all fully connected layers in the embeddings, encoder, and pooler.
        attention_probs_dropout_prob (`float`, *optional*, defaults to 0.1):
            The dropout ratio for the attention probabilities.
        max_position_embeddings (`int`, *optional*, defaults to 512):
            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_vocab_size (`int`, *optional*, defaults to 2):
            The vocabulary size of the `token_type_ids` passed when calling [`VisualBertModel`].
        initializer_range (`float`, *optional*, defaults to 0.02):
            The standard deviation of the truncated_normal_initializer for initializing all weight matrices.
        layer_norm_eps (`float`, *optional*, defaults to 1e-12):
            The epsilon used by the layer normalization layers.
        bypass_transformer (`bool`, *optional*, defaults to `False`):
            Whether or not the model should bypass the transformer for the visual embeddings. If set to `True`, the
            model directly concatenates the visual embeddings from [`VisualBertEmbeddings`] with text output from
            transformers, and then pass it to a self-attention layer.
        special_visual_initialize (`bool`, *optional*, defaults to `True`):
            Whether or not the visual token type and position type embedding weights should be initialized the same as
            the textual token type and positive type embeddings. When set to `True`, the weights of the textual token
            type and position type embeddings are copied to the respective visual embedding layers.

    Example:
        ```python
        >>> from transformers import VisualBertConfig, VisualBertModel
        ...
        >>> # Initializing a VisualBERT visualbert-vqa-coco-pre style configuration
        >>> configuration = VisualBertConfig.from_pretrained("uclanlp/visualbert-vqa-coco-pre")
        ...
        >>> # Initializing a model (with random weights) from the visualbert-vqa-coco-pre style configuration
        >>> model = VisualBertModel(configuration)
        ...
        >>> # Accessing the model configuration
        >>> configuration = model.config
        ```
    """

    model_type = "visual_bert"

    def __init__(
        self,
        vocab_size=30522,
        hidden_size=768,
        visual_embedding_dim=512,
        num_hidden_layers=12,
        num_attention_heads=12,
        intermediate_size=3072,
        hidden_act="gelu",
        hidden_dropout_prob=0.1,
        attention_probs_dropout_prob=0.1,
        max_position_embeddings=512,
        type_vocab_size=2,
        initializer_range=0.02,
        layer_norm_eps=1e-12,
        bypass_transformer=False,
        special_visual_initialize=True,
        pad_token_id=1,
        bos_token_id=0,
        eos_token_id=2,
        **kwargs,
    ):
        super().__init__(pad_token_id=pad_token_id, bos_token_id=bos_token_id, eos_token_id=eos_token_id, **kwargs)

        self.vocab_size = vocab_size
        self.max_position_embeddings = max_position_embeddings
        self.hidden_size = hidden_size
        self.visual_embedding_dim = visual_embedding_dim
        self.num_hidden_layers = num_hidden_layers
        self.num_attention_heads = num_attention_heads
        self.intermediate_size = intermediate_size
        self.hidden_act = hidden_act
        self.hidden_dropout_prob = hidden_dropout_prob
        self.attention_probs_dropout_prob = attention_probs_dropout_prob
        self.initializer_range = initializer_range
        self.type_vocab_size = type_vocab_size
        self.layer_norm_eps = layer_norm_eps
        self.bypass_transformer = bypass_transformer
        self.special_visual_initialize = special_visual_initialize

mindnlp.transformers.models.visual_bert.modeling_visual_bert

MindSpore VisualBERT model.

mindnlp.transformers.models.visual_bert.modeling_visual_bert.VisualBertEmbeddings

Bases: Module

Construct the embeddings from word, position and token_type embeddings and visual embeddings.

Source code in mindnlp/transformers/models/visual_bert/modeling_visual_bert.py 
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
class VisualBertEmbeddings(nn.Module):
    """Construct the embeddings from word, position and token_type embeddings and visual embeddings."""

    def __init__(self, config):
        super().__init__()
        self.word_embeddings = nn.Embedding(config.vocab_size, config.hidden_size, padding_idx=config.pad_token_id)
        self.position_embeddings = nn.Embedding(config.max_position_embeddings, config.hidden_size)
        self.token_type_embeddings = nn.Embedding(config.type_vocab_size, config.hidden_size)

        # self.LayerNorm is not snake-cased to stick with TensorFlow model variable name and be able to load
        # any TensorFlow checkpoint file

        self.LayerNorm = nn.LayerNorm([config.hidden_size], eps=config.layer_norm_eps)
        self.dropout = nn.Dropout(p=config.hidden_dropout_prob)

        # position_ids (1, len position emb) is contiguous in memory and exported when serialized
        self.position_ids = ops.arange(config.max_position_embeddings).reshape((1, -1))
        # For Visual Features
        # Token type and position embedding for image features
        self.visual_token_type_embeddings = nn.Embedding(config.type_vocab_size, config.hidden_size)
        self.visual_position_embeddings = nn.Embedding(config.max_position_embeddings, config.hidden_size)

        if config.special_visual_initialize:
            self.visual_token_type_embeddings.weight = Parameter(
                self.token_type_embeddings.weight.clone(), requires_grad=True
            )
            self.visual_position_embeddings.weight = Parameter(
                self.position_embeddings.weight.clone(), requires_grad=True
            )

        self.visual_projection = nn.Linear(config.visual_embedding_dim, config.hidden_size)

    def forward(
        self,
        input_ids=None,
        token_type_ids=None,
        position_ids=None,
        inputs_embeds=None,
        visual_embeds=None,
        visual_token_type_ids=None,
        image_text_alignment=None,
    ):
        if input_ids is not None:
            input_shape = input_ids.shape
        else:
            input_shape = inputs_embeds.shape[:-1]

        seq_length = input_shape[1]

        if position_ids is None:
            position_ids = self.position_ids[:, :seq_length]

        if inputs_embeds is None:
            inputs_embeds = self.word_embeddings(input_ids)

        if token_type_ids is None:
            token_type_ids = ops.zeros(input_shape, dtype=mindspore.int64)

        token_type_embeddings = self.token_type_embeddings(token_type_ids)

        embeddings = inputs_embeds + token_type_embeddings

        # Absolute Position Embeddings
        position_embeddings = self.position_embeddings(position_ids)
        embeddings += position_embeddings

        if visual_embeds is not None:
            if visual_token_type_ids is None:
                visual_token_type_ids = ops.ones(
                    visual_embeds.shape[:-1], dtype=mindspore.int64
                )

            visual_embeds = self.visual_projection(visual_embeds)
            visual_token_type_embeddings = self.visual_token_type_embeddings(visual_token_type_ids)

            if image_text_alignment is not None:
                # image_text_alignment = Batch x image_length x alignment_number.
                # Each element denotes the position of the word corresponding to the image feature. -1 is the padding value.

                dtype = token_type_embeddings.dtype
                image_text_alignment_mask = (image_text_alignment != -1).long()
                # Get rid of the -1.
                image_text_alignment = image_text_alignment_mask * image_text_alignment

                # Batch x image_length x alignment length x dim
                visual_position_embeddings = self.position_embeddings(image_text_alignment)
                visual_position_embeddings *= image_text_alignment_mask.to(dtype=dtype).unsqueeze(-1)
                visual_position_embeddings = visual_position_embeddings.sum(2)

                # We want to averge along the alignment_number dimension.
                image_text_alignment_mask = image_text_alignment_mask.to(dtype=dtype).sum(2)

                if (image_text_alignment_mask == 0).sum() != 0:
                    image_text_alignment_mask[image_text_alignment_mask == 0] = 1  # Avoid divide by zero error
                    logger.warning(
                        "Found 0 values in `image_text_alignment_mask`. Setting them to 1 to avoid divide-by-zero"
                        " error."
                    )
                visual_position_embeddings = visual_position_embeddings / image_text_alignment_mask.unsqueeze(-1)

                visual_position_ids = ops.zeros(
                    *visual_embeds.shape[:-1], dtype=mindspore.int64
                )

                # When fine-tuning the detector , the image_text_alignment is sometimes padded too long.
                if visual_position_embeddings.shape[1] != visual_embeds.shape[1]:
                    if visual_position_embeddings.shape[1] < visual_embeds.shape[1]:
                        raise ValueError(
                            f"Visual position embeddings length: {visual_position_embeddings.shape[1]} "
                            f"should be the same as `visual_embeds` length: {visual_embeds.shape[1]}"
                        )
                    visual_position_embeddings = visual_position_embeddings[:, : visual_embeds.shape[1], :]

                visual_position_embeddings = visual_position_embeddings + self.visual_position_embeddings(
                    visual_position_ids
                )
            else:
                visual_position_ids = ops.zeros(
                    *visual_embeds.shape[:-1], dtype=mindspore.int64
                )
                visual_position_embeddings = self.visual_position_embeddings(visual_position_ids)

            visual_embeddings = visual_embeds + visual_position_embeddings + visual_token_type_embeddings

            embeddings = ops.cat((embeddings, visual_embeddings), axis=1)

        embeddings = self.LayerNorm(embeddings)
        embeddings = self.dropout(embeddings)
        return embeddings

mindnlp.transformers.models.visual_bert.modeling_visual_bert.VisualBertForMultipleChoice

Bases: VisualBertPreTrainedModel

Source code in mindnlp/transformers/models/visual_bert/modeling_visual_bert.py 
 890
 891
 892
 893
 894
 895
 896
 897
 898
 899
 900
 901
 902
 903
 904
 905
 906
 907
 908
 909
 910
 911
 912
 913
 914
 915
 916
 917
 918
 919
 920
 921
 922
 923
 924
 925
 926
 927
 928
 929
 930
 931
 932
 933
 934
 935
 936
 937
 938
 939
 940
 941
 942
 943
 944
 945
 946
 947
 948
 949
 950
 951
 952
 953
 954
 955
 956
 957
 958
 959
 960
 961
 962
 963
 964
 965
 966
 967
 968
 969
 970
 971
 972
 973
 974
 975
 976
 977
 978
 979
 980
 981
 982
 983
 984
 985
 986
 987
 988
 989
 990
 991
 992
 993
 994
 995
 996
 997
 998
 999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
class VisualBertForMultipleChoice(VisualBertPreTrainedModel):
    def __init__(self, config):
        super().__init__(config)

        self.visual_bert = VisualBertModel(config)
        self.dropout = nn.Dropout(p=config.hidden_dropout_prob)
        self.cls = nn.Linear(config.hidden_size, 1)

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

    def forward(
        self,
        input_ids: Optional[mindspore.Tensor] = None,
        attention_mask: Optional[mindspore.Tensor] = None,
        token_type_ids: Optional[mindspore.Tensor] = None,
        position_ids: Optional[mindspore.Tensor] = None,
        head_mask: Optional[mindspore.Tensor] = None,
        inputs_embeds: Optional[mindspore.Tensor] = None,
        visual_embeds: Optional[mindspore.Tensor] = None,
        visual_attention_mask: Optional[mindspore.Tensor] = None,
        visual_token_type_ids: Optional[mindspore.Tensor] = None,
        image_text_alignment: Optional[mindspore.Tensor] = None,
        output_attentions: Optional[bool] = None,
        output_hidden_states: Optional[bool] = None,
        return_dict: Optional[bool] = None,
        labels: Optional[mindspore.Tensor] = None,
    ) -> Union[Tuple[mindspore.Tensor], MultipleChoiceModelOutput]:
        r"""
        Args:
            labels (`mindspore.Tensor` of shape `(batch_size,)`, *optional*):
                Labels for computing the multiple choice classification loss. Indices should be in `[0, ...,
                num_choices-1]` where `num_choices` is the size of the second dimension of the input tensors. (See
                `input_ids` above)

        Returns:
            `Union[Tuple[mindspore.Tensor], MultipleChoiceModelOutput]`

        Example:
            ```python
            >>> # Assumption: *get_visual_embeddings(image)* gets the visual embeddings of the image in the batch.
            >>> from transformers import AutoTokenizer, VisualBertForMultipleChoice
            >>> import mindspore, ops
            ...
            >>> tokenizer = AutoTokenizer.from_pretrained("google-bert/bert-base-uncased")
            >>> model = VisualBertForMultipleChoice.from_pretrained("uclanlp/visualbert-vcr")
            ...
            >>> prompt = "In Italy, pizza served in formal settings, such as at a restaurant, is presented unsliced."
            >>> choice0 = "It is eaten with a fork and a knife."
            >>> choice1 = "It is eaten while held in the hand."
            ...
            >>> visual_embeds = get_visual_embeddings(image)
            >>> # (batch_size, num_choices, visual_seq_length, visual_embedding_dim)
            >>> visual_embeds = visual_embeds.expand(1, 2, *visual_embeds.shape)
            >>> visual_token_type_ids = ops.ones(visual_embeds.shape[:-1], dtype=mindspore.int64)
            >>> visual_attention_mask = ops.ones(visual_embeds.shape[:-1], dtype=mindspore.float32)
            ...
            >>> labels = mindspore.Tensor(0).unsqueeze(0)  # choice0 is correct (according to Wikipedia ;)), batch size 1
            ...
            >>> encoding = tokenizer([[prompt, prompt], [choice0, choice1]], return_tensors="pt", padding=True)
            >>> # batch size is 1
            >>> inputs_dict = {k: v.unsqueeze(0) for k, v in encoding.items()}
            >>> inputs_dict.update(
            ...     {
            ...         "visual_embeds": visual_embeds,
            ...         "visual_attention_mask": visual_attention_mask,
            ...         "visual_token_type_ids": visual_token_type_ids,
            ...         "labels": labels,
            ...     }
            ... )
            >>> outputs = model(**inputs_dict)
            ...
            >>> loss = outputs.loss
            >>> logits = outputs.logits
            ```
        """
        return_dict = return_dict if return_dict is not None else self.config.use_return_dict
        num_choices = input_ids.shape[1] if input_ids is not None else inputs_embeds.shape[1]

        input_ids = input_ids.view(-1, input_ids.shape[-1]) if input_ids is not None else None
        attention_mask = attention_mask.view(-1, attention_mask.shape[-1]) if attention_mask is not None else None
        token_type_ids = token_type_ids.view(-1, token_type_ids.shape[-1]) if token_type_ids is not None else None
        position_ids = position_ids.view(-1, position_ids.shape[-1]) if position_ids is not None else None
        inputs_embeds = (
            inputs_embeds.view(-1, inputs_embeds.shape[-2], inputs_embeds.shape[-1])
            if inputs_embeds is not None
            else None
        )

        visual_embeds = (
            visual_embeds.view(-1, visual_embeds.shape[-2], visual_embeds.shape[-1])
            if visual_embeds is not None
            else None
        )
        visual_attention_mask = (
            visual_attention_mask.view(-1, visual_attention_mask.shape[-1])
            if visual_attention_mask is not None
            else None
        )
        visual_token_type_ids = (
            visual_token_type_ids.view(-1, visual_token_type_ids.shape[-1])
            if visual_token_type_ids is not None
            else None
        )

        outputs = self.visual_bert(
            input_ids,
            attention_mask=attention_mask,
            token_type_ids=token_type_ids,
            position_ids=position_ids,
            head_mask=head_mask,
            inputs_embeds=inputs_embeds,
            visual_embeds=visual_embeds,
            visual_attention_mask=visual_attention_mask,
            visual_token_type_ids=visual_token_type_ids,
            image_text_alignment=image_text_alignment,
            output_attentions=output_attentions,
            output_hidden_states=output_hidden_states,
            return_dict=return_dict,
        )

        _, pooled_output = outputs[0], outputs[1]

        pooled_output = self.dropout(pooled_output)
        logits = self.cls(pooled_output)
        reshaped_logits = logits.view(-1, num_choices)

        loss = None
        if labels is not None:
            loss = ops.cross_entropy(reshaped_logits, labels)

        if not return_dict:
            output = (reshaped_logits,) + outputs[2:]
            return ((loss,) + output) if loss is not None else output

        return MultipleChoiceModelOutput(
            loss=loss,
            logits=reshaped_logits,
            hidden_states=outputs.hidden_states,
            attentions=outputs.attentions,
        )

mindnlp.transformers.models.visual_bert.modeling_visual_bert.VisualBertForMultipleChoice.forward(input_ids=None, attention_mask=None, token_type_ids=None, position_ids=None, head_mask=None, inputs_embeds=None, visual_embeds=None, visual_attention_mask=None, visual_token_type_ids=None, image_text_alignment=None, output_attentions=None, output_hidden_states=None, return_dict=None, labels=None)

PARAMETER DESCRIPTION
labels

Labels for computing the multiple choice classification loss. Indices should be in [0, ..., num_choices-1] where num_choices is the size of the second dimension of the input tensors. (See input_ids above)

TYPE: `mindspore.Tensor` of shape `(batch_size,)`, *optional* DEFAULT: None

RETURNS DESCRIPTION
Union[Tuple[Tensor], MultipleChoiceModelOutput]

Union[Tuple[mindspore.Tensor], MultipleChoiceModelOutput]
Example 
>>> # Assumption: *get_visual_embeddings(image)* gets the visual embeddings of the image in the batch.
>>> from transformers import AutoTokenizer, VisualBertForMultipleChoice
>>> import mindspore, ops
...
>>> tokenizer = AutoTokenizer.from_pretrained("google-bert/bert-base-uncased")
>>> model = VisualBertForMultipleChoice.from_pretrained("uclanlp/visualbert-vcr")
...
>>> prompt = "In Italy, pizza served in formal settings, such as at a restaurant, is presented unsliced."
>>> choice0 = "It is eaten with a fork and a knife."
>>> choice1 = "It is eaten while held in the hand."
...
>>> visual_embeds = get_visual_embeddings(image)
>>> # (batch_size, num_choices, visual_seq_length, visual_embedding_dim)
>>> visual_embeds = visual_embeds.expand(1, 2, *visual_embeds.shape)
>>> visual_token_type_ids = ops.ones(visual_embeds.shape[:-1], dtype=mindspore.int64)
>>> visual_attention_mask = ops.ones(visual_embeds.shape[:-1], dtype=mindspore.float32)
...
>>> labels = mindspore.Tensor(0).unsqueeze(0)  # choice0 is correct (according to Wikipedia ;)), batch size 1
...
>>> encoding = tokenizer([[prompt, prompt], [choice0, choice1]], return_tensors="pt", padding=True)
>>> # batch size is 1
>>> inputs_dict = {k: v.unsqueeze(0) for k, v in encoding.items()}
>>> inputs_dict.update(
...     {
...         "visual_embeds": visual_embeds,
...         "visual_attention_mask": visual_attention_mask,
...         "visual_token_type_ids": visual_token_type_ids,
...         "labels": labels,
...     }
... )
>>> outputs = model(**inputs_dict)
...
>>> loss = outputs.loss
>>> logits = outputs.logits
Source code in mindnlp/transformers/models/visual_bert/modeling_visual_bert.py 
 901
 902
 903
 904
 905
 906
 907
 908
 909
 910
 911
 912
 913
 914
 915
 916
 917
 918
 919
 920
 921
 922
 923
 924
 925
 926
 927
 928
 929
 930
 931
 932
 933
 934
 935
 936
 937
 938
 939
 940
 941
 942
 943
 944
 945
 946
 947
 948
 949
 950
 951
 952
 953
 954
 955
 956
 957
 958
 959
 960
 961
 962
 963
 964
 965
 966
 967
 968
 969
 970
 971
 972
 973
 974
 975
 976
 977
 978
 979
 980
 981
 982
 983
 984
 985
 986
 987
 988
 989
 990
 991
 992
 993
 994
 995
 996
 997
 998
 999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
def forward(
    self,
    input_ids: Optional[mindspore.Tensor] = None,
    attention_mask: Optional[mindspore.Tensor] = None,
    token_type_ids: Optional[mindspore.Tensor] = None,
    position_ids: Optional[mindspore.Tensor] = None,
    head_mask: Optional[mindspore.Tensor] = None,
    inputs_embeds: Optional[mindspore.Tensor] = None,
    visual_embeds: Optional[mindspore.Tensor] = None,
    visual_attention_mask: Optional[mindspore.Tensor] = None,
    visual_token_type_ids: Optional[mindspore.Tensor] = None,
    image_text_alignment: Optional[mindspore.Tensor] = None,
    output_attentions: Optional[bool] = None,
    output_hidden_states: Optional[bool] = None,
    return_dict: Optional[bool] = None,
    labels: Optional[mindspore.Tensor] = None,
) -> Union[Tuple[mindspore.Tensor], MultipleChoiceModelOutput]:
    r"""
    Args:
        labels (`mindspore.Tensor` of shape `(batch_size,)`, *optional*):
            Labels for computing the multiple choice classification loss. Indices should be in `[0, ...,
            num_choices-1]` where `num_choices` is the size of the second dimension of the input tensors. (See
            `input_ids` above)

    Returns:
        `Union[Tuple[mindspore.Tensor], MultipleChoiceModelOutput]`

    Example:
        ```python
        >>> # Assumption: *get_visual_embeddings(image)* gets the visual embeddings of the image in the batch.
        >>> from transformers import AutoTokenizer, VisualBertForMultipleChoice
        >>> import mindspore, ops
        ...
        >>> tokenizer = AutoTokenizer.from_pretrained("google-bert/bert-base-uncased")
        >>> model = VisualBertForMultipleChoice.from_pretrained("uclanlp/visualbert-vcr")
        ...
        >>> prompt = "In Italy, pizza served in formal settings, such as at a restaurant, is presented unsliced."
        >>> choice0 = "It is eaten with a fork and a knife."
        >>> choice1 = "It is eaten while held in the hand."
        ...
        >>> visual_embeds = get_visual_embeddings(image)
        >>> # (batch_size, num_choices, visual_seq_length, visual_embedding_dim)
        >>> visual_embeds = visual_embeds.expand(1, 2, *visual_embeds.shape)
        >>> visual_token_type_ids = ops.ones(visual_embeds.shape[:-1], dtype=mindspore.int64)
        >>> visual_attention_mask = ops.ones(visual_embeds.shape[:-1], dtype=mindspore.float32)
        ...
        >>> labels = mindspore.Tensor(0).unsqueeze(0)  # choice0 is correct (according to Wikipedia ;)), batch size 1
        ...
        >>> encoding = tokenizer([[prompt, prompt], [choice0, choice1]], return_tensors="pt", padding=True)
        >>> # batch size is 1
        >>> inputs_dict = {k: v.unsqueeze(0) for k, v in encoding.items()}
        >>> inputs_dict.update(
        ...     {
        ...         "visual_embeds": visual_embeds,
        ...         "visual_attention_mask": visual_attention_mask,
        ...         "visual_token_type_ids": visual_token_type_ids,
        ...         "labels": labels,
        ...     }
        ... )
        >>> outputs = model(**inputs_dict)
        ...
        >>> loss = outputs.loss
        >>> logits = outputs.logits
        ```
    """
    return_dict = return_dict if return_dict is not None else self.config.use_return_dict
    num_choices = input_ids.shape[1] if input_ids is not None else inputs_embeds.shape[1]

    input_ids = input_ids.view(-1, input_ids.shape[-1]) if input_ids is not None else None
    attention_mask = attention_mask.view(-1, attention_mask.shape[-1]) if attention_mask is not None else None
    token_type_ids = token_type_ids.view(-1, token_type_ids.shape[-1]) if token_type_ids is not None else None
    position_ids = position_ids.view(-1, position_ids.shape[-1]) if position_ids is not None else None
    inputs_embeds = (
        inputs_embeds.view(-1, inputs_embeds.shape[-2], inputs_embeds.shape[-1])
        if inputs_embeds is not None
        else None
    )

    visual_embeds = (
        visual_embeds.view(-1, visual_embeds.shape[-2], visual_embeds.shape[-1])
        if visual_embeds is not None
        else None
    )
    visual_attention_mask = (
        visual_attention_mask.view(-1, visual_attention_mask.shape[-1])
        if visual_attention_mask is not None
        else None
    )
    visual_token_type_ids = (
        visual_token_type_ids.view(-1, visual_token_type_ids.shape[-1])
        if visual_token_type_ids is not None
        else None
    )

    outputs = self.visual_bert(
        input_ids,
        attention_mask=attention_mask,
        token_type_ids=token_type_ids,
        position_ids=position_ids,
        head_mask=head_mask,
        inputs_embeds=inputs_embeds,
        visual_embeds=visual_embeds,
        visual_attention_mask=visual_attention_mask,
        visual_token_type_ids=visual_token_type_ids,
        image_text_alignment=image_text_alignment,
        output_attentions=output_attentions,
        output_hidden_states=output_hidden_states,
        return_dict=return_dict,
    )

    _, pooled_output = outputs[0], outputs[1]

    pooled_output = self.dropout(pooled_output)
    logits = self.cls(pooled_output)
    reshaped_logits = logits.view(-1, num_choices)

    loss = None
    if labels is not None:
        loss = ops.cross_entropy(reshaped_logits, labels)

    if not return_dict:
        output = (reshaped_logits,) + outputs[2:]
        return ((loss,) + output) if loss is not None else output

    return MultipleChoiceModelOutput(
        loss=loss,
        logits=reshaped_logits,
        hidden_states=outputs.hidden_states,
        attentions=outputs.attentions,
    )

mindnlp.transformers.models.visual_bert.modeling_visual_bert.VisualBertForPreTraining

Bases: VisualBertPreTrainedModel

Source code in mindnlp/transformers/models/visual_bert/modeling_visual_bert.py 
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
class VisualBertForPreTraining(VisualBertPreTrainedModel):
    _tied_weights_keys = ["cls.predictions.decoder.weight", "cls.predictions.decoder.bias"]

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

        self.visual_bert = VisualBertModel(config)
        self.cls = VisualBertPreTrainingHeads(config)

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

    def get_output_embeddings(self):
        return self.cls.predictions.decoder

    def set_output_embeddings(self, new_embeddings):
        self.cls.predictions.decoder = new_embeddings
        self.cls.predictions.bias = new_embeddings.bias

    def forward(
        self,
        input_ids: Optional[mindspore.Tensor] = None,
        attention_mask: Optional[mindspore.Tensor] = None,
        token_type_ids: Optional[mindspore.Tensor] = None,
        position_ids: Optional[mindspore.Tensor] = None,
        head_mask: Optional[mindspore.Tensor] = None,
        inputs_embeds: Optional[mindspore.Tensor] = None,
        visual_embeds: Optional[mindspore.Tensor] = None,
        visual_attention_mask: Optional[mindspore.Tensor] = None,
        visual_token_type_ids: Optional[mindspore.Tensor] = None,
        image_text_alignment: Optional[mindspore.Tensor] = None,
        output_attentions: Optional[bool] = None,
        output_hidden_states: Optional[bool] = None,
        return_dict: Optional[bool] = None,
        labels: Optional[mindspore.Tensor] = None,
        sentence_image_labels: Optional[mindspore.Tensor] = None,
    ) -> Union[Tuple[mindspore.Tensor], VisualBertForPreTrainingOutput]:
        r"""
        Args:
            labels (`mindspore.Tensor` of shape `(batch_size, total_sequence_length)`, *optional*):
                Labels for computing the masked language modeling loss. Indices should be in `[-100, 0, ...,
                config.vocab_size]` (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]`
            sentence_image_labels (`mindspore.Tensor` of shape `(batch_size,)`, *optional*):
                Labels for computing the sentence-image prediction (classification) loss. Input should be a sequence pair
                (see `input_ids` docstring) Indices should be in `[0, 1]`:

                - 0 indicates sequence B is a matching pair of sequence A for the given image,
                - 1 indicates sequence B is a random sequence w.r.t A for the given image.

        Returns:
            `Union[Tuple[mindspore.Tensor], VisualBertForPreTrainingOutput]`

        Example:
            ```python
            >>> # Assumption: *get_visual_embeddings(image)* gets the visual embeddings of the image in the batch.
            >>> from transformers import AutoTokenizer, VisualBertForPreTraining
            ...
            >>> tokenizer = AutoTokenizer.from_pretrained("google-bert/bert-base-uncased")
            >>> model = VisualBertForPreTraining.from_pretrained("uclanlp/visualbert-vqa-coco-pre")
            ...
            >>> inputs = tokenizer("The capital of France is [MASK].", return_tensors="pt")
            >>> visual_embeds = get_visual_embeddings(image).unsqueeze(0)
            >>> visual_token_type_ids = ops.ones(visual_embeds.shape[:-1], dtype=mindspore.int64)
            >>> visual_attention_mask = ops.ones(visual_embeds.shape[:-1], dtype=mindspore.float32)
            ...
            >>> inputs.update(
            ...     {
            ...         "visual_embeds": visual_embeds,
            ...         "visual_token_type_ids": visual_token_type_ids,
            ...         "visual_attention_mask": visual_attention_mask,
            ...     }
            ... )
            >>> max_length = inputs["input_ids"].shape[-1] + visual_embeds.shape[-2]
            >>> labels = tokenizer(
            ...     "The capital of France is Paris.", return_tensors="pt", padding="max_length", max_length=max_length
            ... )["input_ids"]
            >>> sentence_image_labels = mindspore.Tensor(1).unsqueeze(0)  # Batch_size
            ...
            >>> outputs = model(**inputs, labels=labels, sentence_image_labels=sentence_image_labels)
            >>> loss = outputs.loss
            >>> prediction_logits = outputs.prediction_logits
            >>> seq_relationship_logits = outputs.seq_relationship_logits
            ```
        """
        return_dict = return_dict if return_dict is not None else self.config.use_return_dict

        outputs = self.visual_bert(
            input_ids,
            attention_mask=attention_mask,
            token_type_ids=token_type_ids,
            position_ids=position_ids,
            head_mask=head_mask,
            inputs_embeds=inputs_embeds,
            visual_embeds=visual_embeds,
            visual_attention_mask=visual_attention_mask,
            visual_token_type_ids=visual_token_type_ids,
            image_text_alignment=image_text_alignment,
            output_attentions=output_attentions,
            output_hidden_states=output_hidden_states,
            return_dict=return_dict,
        )

        sequence_output, pooled_output = outputs[:2]
        prediction_scores, seq_relationship_score = self.cls(sequence_output, pooled_output)

        total_loss = None
        if labels is not None and sentence_image_labels is not None:
            total_size = attention_mask.shape[-1] + visual_attention_mask.shape[-1]
            if labels.shape[-1] != total_size:
                raise ValueError(
                    "The labels provided should have same sequence length as total attention mask. "
                    f"Found labels with sequence length {labels.shape[-1]}, expected {total_size}."
                )

            masked_lm_loss = ops.cross_entropy(prediction_scores.view(-1, self.config.vocab_size), labels.view(-1))
            sentence_image_loss = ops.cross_entropy(seq_relationship_score.view(-1, 2), sentence_image_labels.view(-1))
            total_loss = masked_lm_loss + sentence_image_loss

        if labels is not None and sentence_image_labels is None:
            total_size = attention_mask.shape[-1] + visual_attention_mask.shape[-1]
            if labels.shape[-1] != total_size:
                raise ValueError(
                    "The labels provided should have same sequence length as total attention mask. "
                    f"Found labels with sequence length {labels.shape[-1]}, expected {total_size}."
                )

            total_loss = ops.cross_entropy(prediction_scores.view(-1, self.config.vocab_size), labels.view(-1))

        if not return_dict:
            output = (prediction_scores, seq_relationship_score) + outputs[2:]
            return ((total_loss,) + output) if total_loss is not None else output

        return VisualBertForPreTrainingOutput(
            loss=total_loss,
            prediction_logits=prediction_scores,
            seq_relationship_logits=seq_relationship_score,
            hidden_states=outputs.hidden_states,
            attentions=outputs.attentions,
        )

mindnlp.transformers.models.visual_bert.modeling_visual_bert.VisualBertForPreTraining.forward(input_ids=None, attention_mask=None, token_type_ids=None, position_ids=None, head_mask=None, inputs_embeds=None, visual_embeds=None, visual_attention_mask=None, visual_token_type_ids=None, image_text_alignment=None, output_attentions=None, output_hidden_states=None, return_dict=None, labels=None, sentence_image_labels=None)

PARAMETER DESCRIPTION
labels

Labels for computing the masked language modeling loss. Indices should be in [-100, 0, ..., config.vocab_size] (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: `mindspore.Tensor` of shape `(batch_size, total_sequence_length)`, *optional* DEFAULT: None

sentence_image_labels

Labels for computing the sentence-image prediction (classification) loss. Input should be a sequence pair (see input_ids docstring) Indices should be in [0, 1]:

  • 0 indicates sequence B is a matching pair of sequence A for the given image,
  • 1 indicates sequence B is a random sequence w.r.t A for the given image.

TYPE: `mindspore.Tensor` of shape `(batch_size,)`, *optional* DEFAULT: None

RETURNS DESCRIPTION
Union[Tuple[Tensor], VisualBertForPreTrainingOutput]

Union[Tuple[mindspore.Tensor], VisualBertForPreTrainingOutput]
Example 
>>> # Assumption: *get_visual_embeddings(image)* gets the visual embeddings of the image in the batch.
>>> from transformers import AutoTokenizer, VisualBertForPreTraining
...
>>> tokenizer = AutoTokenizer.from_pretrained("google-bert/bert-base-uncased")
>>> model = VisualBertForPreTraining.from_pretrained("uclanlp/visualbert-vqa-coco-pre")
...
>>> inputs = tokenizer("The capital of France is [MASK].", return_tensors="pt")
>>> visual_embeds = get_visual_embeddings(image).unsqueeze(0)
>>> visual_token_type_ids = ops.ones(visual_embeds.shape[:-1], dtype=mindspore.int64)
>>> visual_attention_mask = ops.ones(visual_embeds.shape[:-1], dtype=mindspore.float32)
...
>>> inputs.update(
...     {
...         "visual_embeds": visual_embeds,
...         "visual_token_type_ids": visual_token_type_ids,
...         "visual_attention_mask": visual_attention_mask,
...     }
... )
>>> max_length = inputs["input_ids"].shape[-1] + visual_embeds.shape[-2]
>>> labels = tokenizer(
...     "The capital of France is Paris.", return_tensors="pt", padding="max_length", max_length=max_length
... )["input_ids"]
>>> sentence_image_labels = mindspore.Tensor(1).unsqueeze(0)  # Batch_size
...
>>> outputs = model(**inputs, labels=labels, sentence_image_labels=sentence_image_labels)
>>> loss = outputs.loss
>>> prediction_logits = outputs.prediction_logits
>>> seq_relationship_logits = outputs.seq_relationship_logits
Source code in mindnlp/transformers/models/visual_bert/modeling_visual_bert.py 
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
def forward(
    self,
    input_ids: Optional[mindspore.Tensor] = None,
    attention_mask: Optional[mindspore.Tensor] = None,
    token_type_ids: Optional[mindspore.Tensor] = None,
    position_ids: Optional[mindspore.Tensor] = None,
    head_mask: Optional[mindspore.Tensor] = None,
    inputs_embeds: Optional[mindspore.Tensor] = None,
    visual_embeds: Optional[mindspore.Tensor] = None,
    visual_attention_mask: Optional[mindspore.Tensor] = None,
    visual_token_type_ids: Optional[mindspore.Tensor] = None,
    image_text_alignment: Optional[mindspore.Tensor] = None,
    output_attentions: Optional[bool] = None,
    output_hidden_states: Optional[bool] = None,
    return_dict: Optional[bool] = None,
    labels: Optional[mindspore.Tensor] = None,
    sentence_image_labels: Optional[mindspore.Tensor] = None,
) -> Union[Tuple[mindspore.Tensor], VisualBertForPreTrainingOutput]:
    r"""
    Args:
        labels (`mindspore.Tensor` of shape `(batch_size, total_sequence_length)`, *optional*):
            Labels for computing the masked language modeling loss. Indices should be in `[-100, 0, ...,
            config.vocab_size]` (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]`
        sentence_image_labels (`mindspore.Tensor` of shape `(batch_size,)`, *optional*):
            Labels for computing the sentence-image prediction (classification) loss. Input should be a sequence pair
            (see `input_ids` docstring) Indices should be in `[0, 1]`:

            - 0 indicates sequence B is a matching pair of sequence A for the given image,
            - 1 indicates sequence B is a random sequence w.r.t A for the given image.

    Returns:
        `Union[Tuple[mindspore.Tensor], VisualBertForPreTrainingOutput]`

    Example:
        ```python
        >>> # Assumption: *get_visual_embeddings(image)* gets the visual embeddings of the image in the batch.
        >>> from transformers import AutoTokenizer, VisualBertForPreTraining
        ...
        >>> tokenizer = AutoTokenizer.from_pretrained("google-bert/bert-base-uncased")
        >>> model = VisualBertForPreTraining.from_pretrained("uclanlp/visualbert-vqa-coco-pre")
        ...
        >>> inputs = tokenizer("The capital of France is [MASK].", return_tensors="pt")
        >>> visual_embeds = get_visual_embeddings(image).unsqueeze(0)
        >>> visual_token_type_ids = ops.ones(visual_embeds.shape[:-1], dtype=mindspore.int64)
        >>> visual_attention_mask = ops.ones(visual_embeds.shape[:-1], dtype=mindspore.float32)
        ...
        >>> inputs.update(
        ...     {
        ...         "visual_embeds": visual_embeds,
        ...         "visual_token_type_ids": visual_token_type_ids,
        ...         "visual_attention_mask": visual_attention_mask,
        ...     }
        ... )
        >>> max_length = inputs["input_ids"].shape[-1] + visual_embeds.shape[-2]
        >>> labels = tokenizer(
        ...     "The capital of France is Paris.", return_tensors="pt", padding="max_length", max_length=max_length
        ... )["input_ids"]
        >>> sentence_image_labels = mindspore.Tensor(1).unsqueeze(0)  # Batch_size
        ...
        >>> outputs = model(**inputs, labels=labels, sentence_image_labels=sentence_image_labels)
        >>> loss = outputs.loss
        >>> prediction_logits = outputs.prediction_logits
        >>> seq_relationship_logits = outputs.seq_relationship_logits
        ```
    """
    return_dict = return_dict if return_dict is not None else self.config.use_return_dict

    outputs = self.visual_bert(
        input_ids,
        attention_mask=attention_mask,
        token_type_ids=token_type_ids,
        position_ids=position_ids,
        head_mask=head_mask,
        inputs_embeds=inputs_embeds,
        visual_embeds=visual_embeds,
        visual_attention_mask=visual_attention_mask,
        visual_token_type_ids=visual_token_type_ids,
        image_text_alignment=image_text_alignment,
        output_attentions=output_attentions,
        output_hidden_states=output_hidden_states,
        return_dict=return_dict,
    )

    sequence_output, pooled_output = outputs[:2]
    prediction_scores, seq_relationship_score = self.cls(sequence_output, pooled_output)

    total_loss = None
    if labels is not None and sentence_image_labels is not None:
        total_size = attention_mask.shape[-1] + visual_attention_mask.shape[-1]
        if labels.shape[-1] != total_size:
            raise ValueError(
                "The labels provided should have same sequence length as total attention mask. "
                f"Found labels with sequence length {labels.shape[-1]}, expected {total_size}."
            )

        masked_lm_loss = ops.cross_entropy(prediction_scores.view(-1, self.config.vocab_size), labels.view(-1))
        sentence_image_loss = ops.cross_entropy(seq_relationship_score.view(-1, 2), sentence_image_labels.view(-1))
        total_loss = masked_lm_loss + sentence_image_loss

    if labels is not None and sentence_image_labels is None:
        total_size = attention_mask.shape[-1] + visual_attention_mask.shape[-1]
        if labels.shape[-1] != total_size:
            raise ValueError(
                "The labels provided should have same sequence length as total attention mask. "
                f"Found labels with sequence length {labels.shape[-1]}, expected {total_size}."
            )

        total_loss = ops.cross_entropy(prediction_scores.view(-1, self.config.vocab_size), labels.view(-1))

    if not return_dict:
        output = (prediction_scores, seq_relationship_score) + outputs[2:]
        return ((total_loss,) + output) if total_loss is not None else output

    return VisualBertForPreTrainingOutput(
        loss=total_loss,
        prediction_logits=prediction_scores,
        seq_relationship_logits=seq_relationship_score,
        hidden_states=outputs.hidden_states,
        attentions=outputs.attentions,
    )

mindnlp.transformers.models.visual_bert.modeling_visual_bert.VisualBertForPreTrainingOutput dataclass

Bases: ModelOutput

Output type of [VisualBertForPreTraining].

PARAMETER DESCRIPTION
loss

Total loss as the sum of the masked language modeling loss and the sentence-image prediction (classification) loss.

TYPE: *optional*, returned when `labels` is provided, `mindspore.Tensor` of shape `(1,)` DEFAULT: None

prediction_logits

Prediction scores of the language modeling head (scores for each vocabulary token before SoftMax).

TYPE: `mindspore.Tensor` of shape `(batch_size, sequence_length, config.vocab_size)` DEFAULT: None

seq_relationship_logits

Prediction scores of the sentence-image prediction (classification) head (scores of True/False continuation before SoftMax).

TYPE: `mindspore.Tensor` of shape `(batch_size, 2)` DEFAULT: None

Source code in mindnlp/transformers/models/visual_bert/modeling_visual_bert.py 
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
@dataclass
class VisualBertForPreTrainingOutput(ModelOutput):
    """
    Output type of [`VisualBertForPreTraining`].

    Args:
        loss (*optional*, returned when `labels` is provided, `mindspore.Tensor` of shape `(1,)`):
            Total loss as the sum of the masked language modeling loss and the sentence-image prediction
            (classification) loss.
        prediction_logits (`mindspore.Tensor` of shape `(batch_size, sequence_length, config.vocab_size)`):
            Prediction scores of the language modeling head (scores for each vocabulary token before SoftMax).
        seq_relationship_logits (`mindspore.Tensor` of shape `(batch_size, 2)`):
            Prediction scores of the sentence-image prediction (classification) head (scores of True/False continuation
            before SoftMax).
        hidden_states (`tuple(mindspore.Tensor)`, *optional*, returned when `output_hidden_states=True` is passed
            or when `config.output_hidden_states=True`):
            Tuple of `mindspore.Tensor` (one for the output of the embeddings + one for the output of each layer) of
            shape `(batch_size, sequence_length, hidden_size)`.

            Hidden-states of the model at the output of each layer plus the initial embedding outputs.
        attentions (`tuple(mindspore.Tensor)`, *optional*, returned when `output_attentions=True` is passed or when
            `config.output_attentions=True`):
            Tuple of `mindspore.Tensor` (one for each layer) of shape `(batch_size, num_heads, sequence_length,
            sequence_length)`.

            Attentions weights after the attention softmax, used to compute the weighted average in the self-attention
            heads.
    """

    loss: Optional[mindspore.Tensor] = None
    prediction_logits: mindspore.Tensor = None
    seq_relationship_logits: mindspore.Tensor = None
    hidden_states: Optional[Tuple[mindspore.Tensor]] = None
    attentions: Optional[Tuple[mindspore.Tensor]] = None

mindnlp.transformers.models.visual_bert.modeling_visual_bert.VisualBertForQuestionAnswering

Bases: VisualBertPreTrainedModel

Source code in mindnlp/transformers/models/visual_bert/modeling_visual_bert.py 
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
class VisualBertForQuestionAnswering(VisualBertPreTrainedModel):
    def __init__(self, config):
        super().__init__(config)
        self.num_labels = config.num_labels

        self.visual_bert = VisualBertModel(config)
        self.dropout = nn.Dropout(p=config.hidden_dropout_prob)
        self.cls = nn.Linear(config.hidden_size, config.num_labels)

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

    def forward(
        self,
        input_ids: Optional[mindspore.Tensor] = None,
        attention_mask: Optional[mindspore.Tensor] = None,
        token_type_ids: Optional[mindspore.Tensor] = None,
        position_ids: Optional[mindspore.Tensor] = None,
        head_mask: Optional[mindspore.Tensor] = None,
        inputs_embeds: Optional[mindspore.Tensor] = None,
        visual_embeds: Optional[mindspore.Tensor] = None,
        visual_attention_mask: Optional[mindspore.Tensor] = None,
        visual_token_type_ids: Optional[mindspore.Tensor] = None,
        image_text_alignment: Optional[mindspore.Tensor] = None,
        output_attentions: Optional[bool] = None,
        output_hidden_states: Optional[bool] = None,
        return_dict: Optional[bool] = None,
        labels: Optional[mindspore.Tensor] = None,
    ) -> Union[Tuple[mindspore.Tensor], SequenceClassifierOutput]:
        r"""
        Args:
            labels (`mindspore.Tensor` of shape `(batch_size, total_sequence_length)`, *optional*):
                Labels for computing the sequence classification/regression loss. Indices should be in `[0, ...,
                config.num_labels - 1]`. A KLDivLoss is computed between the labels and the returned logits.

        Returns:
            `Union[Tuple[mindspore.Tensor], SequenceClassifierOutput]`

        Example:
            ```python
            >>> # Assumption: *get_visual_embeddings(image)* gets the visual embeddings of the image in the batch.
            >>> from transformers import AutoTokenizer, VisualBertForQuestionAnswering
            >>> import mindspore, ops
            ...
            >>> tokenizer = AutoTokenizer.from_pretrained("google-bert/bert-base-uncased")
            >>> model = VisualBertForQuestionAnswering.from_pretrained("uclanlp/visualbert-vqa")
            ...
            >>> text = "Who is eating the apple?"
            >>> inputs = tokenizer(text, return_tensors="pt")
            >>> visual_embeds = get_visual_embeddings(image).unsqueeze(0)
            >>> visual_token_type_ids = ops.ones(visual_embeds.shape[:-1], dtype=mindspore.int64)
            >>> visual_attention_mask = ops.ones(visual_embeds.shape[:-1], dtype=mindspore.float32)
            ...
            >>> inputs.update(
            ...     {
            ...         "visual_embeds": visual_embeds,
            ...         "visual_token_type_ids": visual_token_type_ids,
            ...         "visual_attention_mask": visual_attention_mask,
            ...     }
            ... )
            ...
            >>> labels = mindspore.Tensor([[0.0, 1.0]]).unsqueeze(0)  # Batch size 1, Num labels 2
            ...
            >>> outputs = model(**inputs, labels=labels)
            >>> loss = outputs.loss
            >>> scores = outputs.logits
            ```
        """
        return_dict = return_dict if return_dict is not None else self.config.use_return_dict

        # Get the index of the last text token
        index_to_gather = attention_mask.sum(1) - 2  # as in original code

        outputs = self.visual_bert(
            input_ids,
            attention_mask=attention_mask,
            token_type_ids=token_type_ids,
            position_ids=position_ids,
            head_mask=head_mask,
            inputs_embeds=inputs_embeds,
            visual_embeds=visual_embeds,
            visual_attention_mask=visual_attention_mask,
            visual_token_type_ids=visual_token_type_ids,
            image_text_alignment=image_text_alignment,
            output_attentions=output_attentions,
            output_hidden_states=output_hidden_states,
            return_dict=return_dict,
        )

        sequence_output = outputs[0]

        # TO-CHECK: From the original code
        index_to_gather = (
            index_to_gather.unsqueeze(-1).unsqueeze(-1).broadcast_to((index_to_gather.shape[0], 1, sequence_output.shape[-1]))
        )
        pooled_output = ops.gather_elements(sequence_output,1 , index_to_gather)

        pooled_output = self.dropout(pooled_output)
        logits = self.cls(pooled_output)
        reshaped_logits = logits.view(-1, self.num_labels)

        loss = None
        if labels is not None:
            loss_fct = ops.KLDivLoss(reduction="batchmean")
            reshaped_logits = ops.log_softmax(reshaped_logits,axis=-1)
            loss = loss_fct(reshaped_logits, labels)
        if not return_dict:
            output = (reshaped_logits,) + outputs[2:]
            return ((loss,) + output) if loss is not None else output

        return SequenceClassifierOutput(
            loss=loss,
            logits=reshaped_logits,
            hidden_states=outputs.hidden_states,
            attentions=outputs.attentions,
        )

mindnlp.transformers.models.visual_bert.modeling_visual_bert.VisualBertForQuestionAnswering.forward(input_ids=None, attention_mask=None, token_type_ids=None, position_ids=None, head_mask=None, inputs_embeds=None, visual_embeds=None, visual_attention_mask=None, visual_token_type_ids=None, image_text_alignment=None, output_attentions=None, output_hidden_states=None, return_dict=None, labels=None)

PARAMETER DESCRIPTION
labels

Labels for computing the sequence classification/regression loss. Indices should be in [0, ..., config.num_labels - 1]. A KLDivLoss is computed between the labels and the returned logits.

TYPE: `mindspore.Tensor` of shape `(batch_size, total_sequence_length)`, *optional* DEFAULT: None

RETURNS DESCRIPTION
Union[Tuple[Tensor], SequenceClassifierOutput]

Union[Tuple[mindspore.Tensor], SequenceClassifierOutput]
Example 
>>> # Assumption: *get_visual_embeddings(image)* gets the visual embeddings of the image in the batch.
>>> from transformers import AutoTokenizer, VisualBertForQuestionAnswering
>>> import mindspore, ops
...
>>> tokenizer = AutoTokenizer.from_pretrained("google-bert/bert-base-uncased")
>>> model = VisualBertForQuestionAnswering.from_pretrained("uclanlp/visualbert-vqa")
...
>>> text = "Who is eating the apple?"
>>> inputs = tokenizer(text, return_tensors="pt")
>>> visual_embeds = get_visual_embeddings(image).unsqueeze(0)
>>> visual_token_type_ids = ops.ones(visual_embeds.shape[:-1], dtype=mindspore.int64)
>>> visual_attention_mask = ops.ones(visual_embeds.shape[:-1], dtype=mindspore.float32)
...
>>> inputs.update(
...     {
...         "visual_embeds": visual_embeds,
...         "visual_token_type_ids": visual_token_type_ids,
...         "visual_attention_mask": visual_attention_mask,
...     }
... )
...
>>> labels = mindspore.Tensor([[0.0, 1.0]]).unsqueeze(0)  # Batch size 1, Num labels 2
...
>>> outputs = model(**inputs, labels=labels)
>>> loss = outputs.loss
>>> scores = outputs.logits
Source code in mindnlp/transformers/models/visual_bert/modeling_visual_bert.py 
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
def forward(
    self,
    input_ids: Optional[mindspore.Tensor] = None,
    attention_mask: Optional[mindspore.Tensor] = None,
    token_type_ids: Optional[mindspore.Tensor] = None,
    position_ids: Optional[mindspore.Tensor] = None,
    head_mask: Optional[mindspore.Tensor] = None,
    inputs_embeds: Optional[mindspore.Tensor] = None,
    visual_embeds: Optional[mindspore.Tensor] = None,
    visual_attention_mask: Optional[mindspore.Tensor] = None,
    visual_token_type_ids: Optional[mindspore.Tensor] = None,
    image_text_alignment: Optional[mindspore.Tensor] = None,
    output_attentions: Optional[bool] = None,
    output_hidden_states: Optional[bool] = None,
    return_dict: Optional[bool] = None,
    labels: Optional[mindspore.Tensor] = None,
) -> Union[Tuple[mindspore.Tensor], SequenceClassifierOutput]:
    r"""
    Args:
        labels (`mindspore.Tensor` of shape `(batch_size, total_sequence_length)`, *optional*):
            Labels for computing the sequence classification/regression loss. Indices should be in `[0, ...,
            config.num_labels - 1]`. A KLDivLoss is computed between the labels and the returned logits.

    Returns:
        `Union[Tuple[mindspore.Tensor], SequenceClassifierOutput]`

    Example:
        ```python
        >>> # Assumption: *get_visual_embeddings(image)* gets the visual embeddings of the image in the batch.
        >>> from transformers import AutoTokenizer, VisualBertForQuestionAnswering
        >>> import mindspore, ops
        ...
        >>> tokenizer = AutoTokenizer.from_pretrained("google-bert/bert-base-uncased")
        >>> model = VisualBertForQuestionAnswering.from_pretrained("uclanlp/visualbert-vqa")
        ...
        >>> text = "Who is eating the apple?"
        >>> inputs = tokenizer(text, return_tensors="pt")
        >>> visual_embeds = get_visual_embeddings(image).unsqueeze(0)
        >>> visual_token_type_ids = ops.ones(visual_embeds.shape[:-1], dtype=mindspore.int64)
        >>> visual_attention_mask = ops.ones(visual_embeds.shape[:-1], dtype=mindspore.float32)
        ...
        >>> inputs.update(
        ...     {
        ...         "visual_embeds": visual_embeds,
        ...         "visual_token_type_ids": visual_token_type_ids,
        ...         "visual_attention_mask": visual_attention_mask,
        ...     }
        ... )
        ...
        >>> labels = mindspore.Tensor([[0.0, 1.0]]).unsqueeze(0)  # Batch size 1, Num labels 2
        ...
        >>> outputs = model(**inputs, labels=labels)
        >>> loss = outputs.loss
        >>> scores = outputs.logits
        ```
    """
    return_dict = return_dict if return_dict is not None else self.config.use_return_dict

    # Get the index of the last text token
    index_to_gather = attention_mask.sum(1) - 2  # as in original code

    outputs = self.visual_bert(
        input_ids,
        attention_mask=attention_mask,
        token_type_ids=token_type_ids,
        position_ids=position_ids,
        head_mask=head_mask,
        inputs_embeds=inputs_embeds,
        visual_embeds=visual_embeds,
        visual_attention_mask=visual_attention_mask,
        visual_token_type_ids=visual_token_type_ids,
        image_text_alignment=image_text_alignment,
        output_attentions=output_attentions,
        output_hidden_states=output_hidden_states,
        return_dict=return_dict,
    )

    sequence_output = outputs[0]

    # TO-CHECK: From the original code
    index_to_gather = (
        index_to_gather.unsqueeze(-1).unsqueeze(-1).broadcast_to((index_to_gather.shape[0], 1, sequence_output.shape[-1]))
    )
    pooled_output = ops.gather_elements(sequence_output,1 , index_to_gather)

    pooled_output = self.dropout(pooled_output)
    logits = self.cls(pooled_output)
    reshaped_logits = logits.view(-1, self.num_labels)

    loss = None
    if labels is not None:
        loss_fct = ops.KLDivLoss(reduction="batchmean")
        reshaped_logits = ops.log_softmax(reshaped_logits,axis=-1)
        loss = loss_fct(reshaped_logits, labels)
    if not return_dict:
        output = (reshaped_logits,) + outputs[2:]
        return ((loss,) + output) if loss is not None else output

    return SequenceClassifierOutput(
        loss=loss,
        logits=reshaped_logits,
        hidden_states=outputs.hidden_states,
        attentions=outputs.attentions,
    )

mindnlp.transformers.models.visual_bert.modeling_visual_bert.VisualBertForRegionToPhraseAlignment

Bases: VisualBertPreTrainedModel

Source code in mindnlp/transformers/models/visual_bert/modeling_visual_bert.py 
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
class VisualBertForRegionToPhraseAlignment(VisualBertPreTrainedModel):
    _tied_weights_keys = ["cls.predictions.decoder.bias"]

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

        self.visual_bert = VisualBertModel(config)
        self.dropout = nn.Dropout(p=config.hidden_dropout_prob)
        self.cls = VisualBertPreTrainingHeads(config)
        self.attention = VisualBertRegionToPhraseAttention(config)

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

    def forward(
        self,
        input_ids: Optional[mindspore.Tensor] = None,
        attention_mask: Optional[mindspore.Tensor] = None,
        token_type_ids: Optional[mindspore.Tensor] = None,
        position_ids: Optional[mindspore.Tensor] = None,
        head_mask: Optional[mindspore.Tensor] = None,
        inputs_embeds: Optional[mindspore.Tensor] = None,
        visual_embeds: Optional[mindspore.Tensor] = None,
        visual_attention_mask: Optional[mindspore.Tensor] = None,
        visual_token_type_ids: Optional[mindspore.Tensor] = None,
        image_text_alignment: Optional[mindspore.Tensor] = None,
        output_attentions: Optional[bool] = None,
        output_hidden_states: Optional[bool] = None,
        return_dict: Optional[bool] = None,
        region_to_phrase_position: Optional[mindspore.Tensor] = None,
        labels: Optional[mindspore.Tensor] = None,
    ) -> Union[Tuple[mindspore.Tensor], SequenceClassifierOutput]:
        r"""
        Args:
            region_to_phrase_position (`mindspore.Tensor` of shape `(batch_size, total_sequence_length)`, *optional*):
                The positions depicting the position of the image embedding corresponding to the textual tokens.

            labels (`mindspore.Tensor` of shape `(batch_size, total_sequence_length, visual_sequence_length)`, *optional*):
                Labels for computing the masked language modeling loss. KLDivLoss is computed against these labels and the
                outputs from the attention layer.

        Returns:
            `Union[Tuple[mindspore.Tensor], SequenceClassifierOutput]`

        Example:
            ```python
            >>> # Assumption: *get_visual_embeddings(image)* gets the visual embeddings of the image in the batch.
            >>> from transformers import AutoTokenizer, VisualBertForRegionToPhraseAlignment
            >>> import mindspore
            ...
            >>> tokenizer = AutoTokenizer.from_pretrained("google-bert/bert-base-uncased")
            >>> model = VisualBertForRegionToPhraseAlignment.from_pretrained("uclanlp/visualbert-vqa-coco-pre")
            ...
            >>> text = "Who is eating the apple?"
            >>> inputs = tokenizer(text, return_tensors="ms")
            >>> visual_embeds = get_visual_embeddings(image).unsqueeze(0)
            >>> visual_token_type_ids = ops.ones(visual_embeds.shape[:-1], dtype=mindspore.int64)
            >>> visual_attention_mask = ops.ones(visual_embeds.shape[:-1], dtype=mindspore.float32)
            >>> region_to_phrase_position = ops.ones((1, inputs["input_ids"].shape[-1] + visual_embeds.shape[-2]))
            ...
            >>> inputs.update(
            ...     {
            ...         "region_to_phrase_position": region_to_phrase_position,
            ...         "visual_embeds": visual_embeds,
            ...         "visual_token_type_ids": visual_token_type_ids,
            ...         "visual_attention_mask": visual_attention_mask,
            ...     }
            ... )
            ...
            >>> labels = ops.ones(
            ...     (1, inputs["input_ids"].shape[-1] + visual_embeds.shape[-2], visual_embeds.shape[-2])
            ... )  # Batch size 1
            ...
            >>> outputs = model(**inputs, labels=labels)
            >>> loss = outputs.loss
            >>> scores = outputs.logits
            ```
        """
        if region_to_phrase_position is None:
            raise ValueError("`region_to_phrase_position` should not be None when using Flickr Model.")

        return_dict = return_dict if return_dict is not None else self.config.use_return_dict

        outputs = self.visual_bert(
            input_ids,
            attention_mask=attention_mask,
            token_type_ids=token_type_ids,
            position_ids=position_ids,
            head_mask=head_mask,
            inputs_embeds=inputs_embeds,
            visual_embeds=visual_embeds,
            visual_attention_mask=visual_attention_mask,
            visual_token_type_ids=visual_token_type_ids,
            image_text_alignment=image_text_alignment,
            output_attentions=output_attentions,
            output_hidden_states=output_hidden_states,
            return_dict=return_dict,
        )

        sequence_output = outputs[0]
        region_to_phrase_position_mask = (region_to_phrase_position != -1).long()

        # Make the -1 become 0
        region_to_phrase_position = region_to_phrase_position * region_to_phrase_position_mask

        # Selected_positions = batch x selected position x dim
        expanded_region_to_phrase_positions = region_to_phrase_position.unsqueeze(2).broadcast_to((
            region_to_phrase_position.shape[0], region_to_phrase_position.shape[1], sequence_output.shape[2]
        ))
        selected_positions = sequence_output.gather_elements(1,expanded_region_to_phrase_positions)

        # Visual Features = batch x visual_feature_length x dim
        # This will need separate image and visual masks.
        visual_features = sequence_output[:, attention_mask.shape[1] :]

        if visual_features.shape[1] != visual_attention_mask.shape[1]:
            raise ValueError(
                f"Visual features length :{visual_features.shape[1]} should be the same"
                f" as visual attention mask length: {visual_attention_mask.shape[1]}."
            )

        logits = self.attention(selected_positions, visual_features, visual_attention_mask)

        loss = None

        if labels is not None:
            # scores = batch x selected position x visual_feature
            # scores = selected_positions.bmm(visual_features.transpose(1,2))
            # label = batch x selected_postion x needed position
            loss_fct = ops.KLDivLoss(reduction="batchmean")
            scores = ops.log_softmax(logits,axis=-1)
            labels = labels.contiguous()
            loss = loss_fct(scores, labels)

        if not return_dict:
            output = (logits,) + outputs[2:]
            return ((loss,) + output) if loss is not None else output

        return SequenceClassifierOutput(
            loss=loss,
            logits=logits,
            hidden_states=outputs.hidden_states,
            attentions=outputs.attentions,
        )

mindnlp.transformers.models.visual_bert.modeling_visual_bert.VisualBertForRegionToPhraseAlignment.forward(input_ids=None, attention_mask=None, token_type_ids=None, position_ids=None, head_mask=None, inputs_embeds=None, visual_embeds=None, visual_attention_mask=None, visual_token_type_ids=None, image_text_alignment=None, output_attentions=None, output_hidden_states=None, return_dict=None, region_to_phrase_position=None, labels=None)

PARAMETER DESCRIPTION
region_to_phrase_position

The positions depicting the position of the image embedding corresponding to the textual tokens.

TYPE: `mindspore.Tensor` of shape `(batch_size, total_sequence_length)`, *optional* DEFAULT: None

labels

Labels for computing the masked language modeling loss. KLDivLoss is computed against these labels and the outputs from the attention layer.

TYPE: `mindspore.Tensor` of shape `(batch_size, total_sequence_length, visual_sequence_length)`, *optional* DEFAULT: None

RETURNS DESCRIPTION
Union[Tuple[Tensor], SequenceClassifierOutput]

Union[Tuple[mindspore.Tensor], SequenceClassifierOutput]
Example 
>>> # Assumption: *get_visual_embeddings(image)* gets the visual embeddings of the image in the batch.
>>> from transformers import AutoTokenizer, VisualBertForRegionToPhraseAlignment
>>> import mindspore
...
>>> tokenizer = AutoTokenizer.from_pretrained("google-bert/bert-base-uncased")
>>> model = VisualBertForRegionToPhraseAlignment.from_pretrained("uclanlp/visualbert-vqa-coco-pre")
...
>>> text = "Who is eating the apple?"
>>> inputs = tokenizer(text, return_tensors="ms")
>>> visual_embeds = get_visual_embeddings(image).unsqueeze(0)
>>> visual_token_type_ids = ops.ones(visual_embeds.shape[:-1], dtype=mindspore.int64)
>>> visual_attention_mask = ops.ones(visual_embeds.shape[:-1], dtype=mindspore.float32)
>>> region_to_phrase_position = ops.ones((1, inputs["input_ids"].shape[-1] + visual_embeds.shape[-2]))
...
>>> inputs.update(
...     {
...         "region_to_phrase_position": region_to_phrase_position,
...         "visual_embeds": visual_embeds,
...         "visual_token_type_ids": visual_token_type_ids,
...         "visual_attention_mask": visual_attention_mask,
...     }
... )
...
>>> labels = ops.ones(
...     (1, inputs["input_ids"].shape[-1] + visual_embeds.shape[-2], visual_embeds.shape[-2])
... )  # Batch size 1
...
>>> outputs = model(**inputs, labels=labels)
>>> loss = outputs.loss
>>> scores = outputs.logits
Source code in mindnlp/transformers/models/visual_bert/modeling_visual_bert.py 
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
def forward(
    self,
    input_ids: Optional[mindspore.Tensor] = None,
    attention_mask: Optional[mindspore.Tensor] = None,
    token_type_ids: Optional[mindspore.Tensor] = None,
    position_ids: Optional[mindspore.Tensor] = None,
    head_mask: Optional[mindspore.Tensor] = None,
    inputs_embeds: Optional[mindspore.Tensor] = None,
    visual_embeds: Optional[mindspore.Tensor] = None,
    visual_attention_mask: Optional[mindspore.Tensor] = None,
    visual_token_type_ids: Optional[mindspore.Tensor] = None,
    image_text_alignment: Optional[mindspore.Tensor] = None,
    output_attentions: Optional[bool] = None,
    output_hidden_states: Optional[bool] = None,
    return_dict: Optional[bool] = None,
    region_to_phrase_position: Optional[mindspore.Tensor] = None,
    labels: Optional[mindspore.Tensor] = None,
) -> Union[Tuple[mindspore.Tensor], SequenceClassifierOutput]:
    r"""
    Args:
        region_to_phrase_position (`mindspore.Tensor` of shape `(batch_size, total_sequence_length)`, *optional*):
            The positions depicting the position of the image embedding corresponding to the textual tokens.

        labels (`mindspore.Tensor` of shape `(batch_size, total_sequence_length, visual_sequence_length)`, *optional*):
            Labels for computing the masked language modeling loss. KLDivLoss is computed against these labels and the
            outputs from the attention layer.

    Returns:
        `Union[Tuple[mindspore.Tensor], SequenceClassifierOutput]`

    Example:
        ```python
        >>> # Assumption: *get_visual_embeddings(image)* gets the visual embeddings of the image in the batch.
        >>> from transformers import AutoTokenizer, VisualBertForRegionToPhraseAlignment
        >>> import mindspore
        ...
        >>> tokenizer = AutoTokenizer.from_pretrained("google-bert/bert-base-uncased")
        >>> model = VisualBertForRegionToPhraseAlignment.from_pretrained("uclanlp/visualbert-vqa-coco-pre")
        ...
        >>> text = "Who is eating the apple?"
        >>> inputs = tokenizer(text, return_tensors="ms")
        >>> visual_embeds = get_visual_embeddings(image).unsqueeze(0)
        >>> visual_token_type_ids = ops.ones(visual_embeds.shape[:-1], dtype=mindspore.int64)
        >>> visual_attention_mask = ops.ones(visual_embeds.shape[:-1], dtype=mindspore.float32)
        >>> region_to_phrase_position = ops.ones((1, inputs["input_ids"].shape[-1] + visual_embeds.shape[-2]))
        ...
        >>> inputs.update(
        ...     {
        ...         "region_to_phrase_position": region_to_phrase_position,
        ...         "visual_embeds": visual_embeds,
        ...         "visual_token_type_ids": visual_token_type_ids,
        ...         "visual_attention_mask": visual_attention_mask,
        ...     }
        ... )
        ...
        >>> labels = ops.ones(
        ...     (1, inputs["input_ids"].shape[-1] + visual_embeds.shape[-2], visual_embeds.shape[-2])
        ... )  # Batch size 1
        ...
        >>> outputs = model(**inputs, labels=labels)
        >>> loss = outputs.loss
        >>> scores = outputs.logits
        ```
    """
    if region_to_phrase_position is None:
        raise ValueError("`region_to_phrase_position` should not be None when using Flickr Model.")

    return_dict = return_dict if return_dict is not None else self.config.use_return_dict

    outputs = self.visual_bert(
        input_ids,
        attention_mask=attention_mask,
        token_type_ids=token_type_ids,
        position_ids=position_ids,
        head_mask=head_mask,
        inputs_embeds=inputs_embeds,
        visual_embeds=visual_embeds,
        visual_attention_mask=visual_attention_mask,
        visual_token_type_ids=visual_token_type_ids,
        image_text_alignment=image_text_alignment,
        output_attentions=output_attentions,
        output_hidden_states=output_hidden_states,
        return_dict=return_dict,
    )

    sequence_output = outputs[0]
    region_to_phrase_position_mask = (region_to_phrase_position != -1).long()

    # Make the -1 become 0
    region_to_phrase_position = region_to_phrase_position * region_to_phrase_position_mask

    # Selected_positions = batch x selected position x dim
    expanded_region_to_phrase_positions = region_to_phrase_position.unsqueeze(2).broadcast_to((
        region_to_phrase_position.shape[0], region_to_phrase_position.shape[1], sequence_output.shape[2]
    ))
    selected_positions = sequence_output.gather_elements(1,expanded_region_to_phrase_positions)

    # Visual Features = batch x visual_feature_length x dim
    # This will need separate image and visual masks.
    visual_features = sequence_output[:, attention_mask.shape[1] :]

    if visual_features.shape[1] != visual_attention_mask.shape[1]:
        raise ValueError(
            f"Visual features length :{visual_features.shape[1]} should be the same"
            f" as visual attention mask length: {visual_attention_mask.shape[1]}."
        )

    logits = self.attention(selected_positions, visual_features, visual_attention_mask)

    loss = None

    if labels is not None:
        # scores = batch x selected position x visual_feature
        # scores = selected_positions.bmm(visual_features.transpose(1,2))
        # label = batch x selected_postion x needed position
        loss_fct = ops.KLDivLoss(reduction="batchmean")
        scores = ops.log_softmax(logits,axis=-1)
        labels = labels.contiguous()
        loss = loss_fct(scores, labels)

    if not return_dict:
        output = (logits,) + outputs[2:]
        return ((loss,) + output) if loss is not None else output

    return SequenceClassifierOutput(
        loss=loss,
        logits=logits,
        hidden_states=outputs.hidden_states,
        attentions=outputs.attentions,
    )

mindnlp.transformers.models.visual_bert.modeling_visual_bert.VisualBertForVisualReasoning

Bases: VisualBertPreTrainedModel

Source code in mindnlp/transformers/models/visual_bert/modeling_visual_bert.py 
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
class VisualBertForVisualReasoning(VisualBertPreTrainedModel):
    def __init__(self, config):
        super().__init__(config)
        self.num_labels = config.num_labels

        self.visual_bert = VisualBertModel(config)
        self.dropout = nn.Dropout(p=config.hidden_dropout_prob)
        self.cls = nn.Linear(config.hidden_size, config.num_labels)  # 2

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

    def forward(
        self,
        input_ids: Optional[mindspore.Tensor] = None,
        attention_mask: Optional[mindspore.Tensor] = None,
        token_type_ids: Optional[mindspore.Tensor] = None,
        position_ids: Optional[mindspore.Tensor] = None,
        head_mask: Optional[mindspore.Tensor] = None,
        inputs_embeds: Optional[mindspore.Tensor] = None,
        visual_embeds: Optional[mindspore.Tensor] = None,
        visual_attention_mask: Optional[mindspore.Tensor] = None,
        visual_token_type_ids: Optional[mindspore.Tensor] = None,
        image_text_alignment: Optional[mindspore.Tensor] = None,
        output_attentions: Optional[bool] = None,
        output_hidden_states: Optional[bool] = None,
        return_dict: Optional[bool] = None,
        labels: Optional[mindspore.Tensor] = None,
    ) -> Union[Tuple[mindspore.Tensor], SequenceClassifierOutput]:
        r"""
        Args:
            labels (`mindspore.Tensor` of shape `(batch_size,)`, *optional*):
                Labels for computing the sequence classification/regression loss. Indices should be in `[0, ...,
                config.num_labels - 1]`. A classification loss is computed (Cross-Entropy) against these labels.

        Returns:
            `Union[Tuple[mindspore.Tensor], SequenceClassifierOutput]`

        Example:
            ```python
            >>> # Assumption: *get_visual_embeddings(image)* gets the visual embeddings of the image in the batch.
            >>> from transformers import AutoTokenizer, VisualBertForVisualReasoning
            >>> import mindspore
            ...
            >>> tokenizer = AutoTokenizer.from_pretrained("google-bert/bert-base-uncased")
            >>> model = VisualBertForVisualReasoning.from_pretrained("uclanlp/visualbert-nlvr2")
            ...
            >>> text = "Who is eating the apple?"
            >>> inputs = tokenizer(text, return_tensors="pt")
            >>> visual_embeds = get_visual_embeddings(image).unsqueeze(0)
            >>> visual_token_type_ids = ops.ones(visual_embeds.shape[:-1], dtype=mindspore.int64)
            >>> visual_attention_mask = ops.ones(visual_embeds.shape[:-1], dtype=mindspore.float32)
            ...
            >>> inputs.update(
            ...     {
            ...         "visual_embeds": visual_embeds,
            ...         "visual_token_type_ids": visual_token_type_ids,
            ...         "visual_attention_mask": visual_attention_mask,
            ...     }
            ... )
            ...
            >>> labels = mindspore.Tensor(1).unsqueeze(0)  # Batch size 1, Num choices 2
            ...
            >>> outputs = model(**inputs, labels=labels)
            >>> loss = outputs.loss
            >>> scores = outputs.logits
            ```
        """
        return_dict = return_dict if return_dict is not None else self.config.use_return_dict

        outputs = self.visual_bert(
            input_ids,
            attention_mask=attention_mask,
            token_type_ids=token_type_ids,
            position_ids=position_ids,
            head_mask=head_mask,
            inputs_embeds=inputs_embeds,
            visual_embeds=visual_embeds,
            visual_attention_mask=visual_attention_mask,
            visual_token_type_ids=visual_token_type_ids,
            image_text_alignment=image_text_alignment,
            output_attentions=output_attentions,
            output_hidden_states=output_hidden_states,
            return_dict=return_dict,
        )

        # sequence_output = outputs[0]
        pooled_output = outputs[1]
        pooled_output = self.dropout(pooled_output)
        logits = self.cls(pooled_output)
        reshaped_logits = logits

        loss = None
        if labels is not None:
            loss = ops.cross_entropy(reshaped_logits, labels.view(-1).astype(mindspore.int32))

        if not return_dict:
            output = (logits,) + outputs[2:]
            return ((loss,) + output) if loss is not None else output

        return SequenceClassifierOutput(
            loss=loss,
            logits=reshaped_logits,
            hidden_states=outputs.hidden_states,
            attentions=outputs.attentions,
        )

mindnlp.transformers.models.visual_bert.modeling_visual_bert.VisualBertForVisualReasoning.forward(input_ids=None, attention_mask=None, token_type_ids=None, position_ids=None, head_mask=None, inputs_embeds=None, visual_embeds=None, visual_attention_mask=None, visual_token_type_ids=None, image_text_alignment=None, output_attentions=None, output_hidden_states=None, return_dict=None, labels=None)

PARAMETER DESCRIPTION
labels

Labels for computing the sequence classification/regression loss. Indices should be in [0, ..., config.num_labels - 1]. A classification loss is computed (Cross-Entropy) against these labels.

TYPE: `mindspore.Tensor` of shape `(batch_size,)`, *optional* DEFAULT: None

RETURNS DESCRIPTION
Union[Tuple[Tensor], SequenceClassifierOutput]

Union[Tuple[mindspore.Tensor], SequenceClassifierOutput]
Example 
>>> # Assumption: *get_visual_embeddings(image)* gets the visual embeddings of the image in the batch.
>>> from transformers import AutoTokenizer, VisualBertForVisualReasoning
>>> import mindspore
...
>>> tokenizer = AutoTokenizer.from_pretrained("google-bert/bert-base-uncased")
>>> model = VisualBertForVisualReasoning.from_pretrained("uclanlp/visualbert-nlvr2")
...
>>> text = "Who is eating the apple?"
>>> inputs = tokenizer(text, return_tensors="pt")
>>> visual_embeds = get_visual_embeddings(image).unsqueeze(0)
>>> visual_token_type_ids = ops.ones(visual_embeds.shape[:-1], dtype=mindspore.int64)
>>> visual_attention_mask = ops.ones(visual_embeds.shape[:-1], dtype=mindspore.float32)
...
>>> inputs.update(
...     {
...         "visual_embeds": visual_embeds,
...         "visual_token_type_ids": visual_token_type_ids,
...         "visual_attention_mask": visual_attention_mask,
...     }
... )
...
>>> labels = mindspore.Tensor(1).unsqueeze(0)  # Batch size 1, Num choices 2
...
>>> outputs = model(**inputs, labels=labels)
>>> loss = outputs.loss
>>> scores = outputs.logits
Source code in mindnlp/transformers/models/visual_bert/modeling_visual_bert.py 
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
def forward(
    self,
    input_ids: Optional[mindspore.Tensor] = None,
    attention_mask: Optional[mindspore.Tensor] = None,
    token_type_ids: Optional[mindspore.Tensor] = None,
    position_ids: Optional[mindspore.Tensor] = None,
    head_mask: Optional[mindspore.Tensor] = None,
    inputs_embeds: Optional[mindspore.Tensor] = None,
    visual_embeds: Optional[mindspore.Tensor] = None,
    visual_attention_mask: Optional[mindspore.Tensor] = None,
    visual_token_type_ids: Optional[mindspore.Tensor] = None,
    image_text_alignment: Optional[mindspore.Tensor] = None,
    output_attentions: Optional[bool] = None,
    output_hidden_states: Optional[bool] = None,
    return_dict: Optional[bool] = None,
    labels: Optional[mindspore.Tensor] = None,
) -> Union[Tuple[mindspore.Tensor], SequenceClassifierOutput]:
    r"""
    Args:
        labels (`mindspore.Tensor` of shape `(batch_size,)`, *optional*):
            Labels for computing the sequence classification/regression loss. Indices should be in `[0, ...,
            config.num_labels - 1]`. A classification loss is computed (Cross-Entropy) against these labels.

    Returns:
        `Union[Tuple[mindspore.Tensor], SequenceClassifierOutput]`

    Example:
        ```python
        >>> # Assumption: *get_visual_embeddings(image)* gets the visual embeddings of the image in the batch.
        >>> from transformers import AutoTokenizer, VisualBertForVisualReasoning
        >>> import mindspore
        ...
        >>> tokenizer = AutoTokenizer.from_pretrained("google-bert/bert-base-uncased")
        >>> model = VisualBertForVisualReasoning.from_pretrained("uclanlp/visualbert-nlvr2")
        ...
        >>> text = "Who is eating the apple?"
        >>> inputs = tokenizer(text, return_tensors="pt")
        >>> visual_embeds = get_visual_embeddings(image).unsqueeze(0)
        >>> visual_token_type_ids = ops.ones(visual_embeds.shape[:-1], dtype=mindspore.int64)
        >>> visual_attention_mask = ops.ones(visual_embeds.shape[:-1], dtype=mindspore.float32)
        ...
        >>> inputs.update(
        ...     {
        ...         "visual_embeds": visual_embeds,
        ...         "visual_token_type_ids": visual_token_type_ids,
        ...         "visual_attention_mask": visual_attention_mask,
        ...     }
        ... )
        ...
        >>> labels = mindspore.Tensor(1).unsqueeze(0)  # Batch size 1, Num choices 2
        ...
        >>> outputs = model(**inputs, labels=labels)
        >>> loss = outputs.loss
        >>> scores = outputs.logits
        ```
    """
    return_dict = return_dict if return_dict is not None else self.config.use_return_dict

    outputs = self.visual_bert(
        input_ids,
        attention_mask=attention_mask,
        token_type_ids=token_type_ids,
        position_ids=position_ids,
        head_mask=head_mask,
        inputs_embeds=inputs_embeds,
        visual_embeds=visual_embeds,
        visual_attention_mask=visual_attention_mask,
        visual_token_type_ids=visual_token_type_ids,
        image_text_alignment=image_text_alignment,
        output_attentions=output_attentions,
        output_hidden_states=output_hidden_states,
        return_dict=return_dict,
    )

    # sequence_output = outputs[0]
    pooled_output = outputs[1]
    pooled_output = self.dropout(pooled_output)
    logits = self.cls(pooled_output)
    reshaped_logits = logits

    loss = None
    if labels is not None:
        loss = ops.cross_entropy(reshaped_logits, labels.view(-1).astype(mindspore.int32))

    if not return_dict:
        output = (logits,) + outputs[2:]
        return ((loss,) + output) if loss is not None else output

    return SequenceClassifierOutput(
        loss=loss,
        logits=reshaped_logits,
        hidden_states=outputs.hidden_states,
        attentions=outputs.attentions,
    )

mindnlp.transformers.models.visual_bert.modeling_visual_bert.VisualBertModel

Bases: VisualBertPreTrainedModel

The model can behave as an encoder (with only self-attention) following the architecture described in Attention is all you need by Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N. Gomez, Lukasz Kaiser and Illia Polosukhin.

Source code in mindnlp/transformers/models/visual_bert/modeling_visual_bert.py 
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
class VisualBertModel(VisualBertPreTrainedModel):
    """
    The model can behave as an encoder (with only self-attention) following the architecture described in [Attention is
    all you need](https://arxiv.org/abs/1706.03762) by Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit,
    Llion Jones, Aidan N. Gomez, Lukasz Kaiser and Illia Polosukhin.
    """

    def __init__(self, config, add_pooling_layer=True):
        super().__init__(config)
        self.config = config

        self.embeddings = VisualBertEmbeddings(config)
        self.encoder = VisualBertEncoder(config)

        self.pooler = VisualBertPooler(config) if add_pooling_layer else None

        self.bypass_transformer = config.bypass_transformer

        if self.bypass_transformer:
            self.additional_layer = VisualBertLayer(config)

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

    def get_input_embeddings(self):
        return self.embeddings.word_embeddings

    def set_input_embeddings(self, value):
        self.embeddings.word_embeddings = value

    def _prune_heads(self, heads_to_prune):
        """
        Prunes heads of the model. heads_to_prune: dict of {layer_num: list of heads to prune in this layer} See base
        class PreTrainedModel
        """
        for layer, heads in heads_to_prune.items():
            self.encoder.layer[layer].attention.prune_heads(heads)

    def forward(
        self,
        input_ids: Optional[mindspore.Tensor] = None,
        attention_mask: Optional[mindspore.Tensor] = None,
        token_type_ids: Optional[mindspore.Tensor] = None,
        position_ids: Optional[mindspore.Tensor] = None,
        head_mask: Optional[mindspore.Tensor] = None,
        inputs_embeds: Optional[mindspore.Tensor] = None,
        visual_embeds: Optional[mindspore.Tensor] = None,
        visual_attention_mask: Optional[mindspore.Tensor] = None,
        visual_token_type_ids: Optional[mindspore.Tensor] = None,
        image_text_alignment: Optional[mindspore.Tensor] = None,
        output_attentions: Optional[bool] = None,
        output_hidden_states: Optional[bool] = None,
        return_dict: Optional[bool] = None,
    ) -> Union[Tuple[mindspore.Tensor], BaseModelOutputWithPooling]:
        r"""

        Returns:
            `Union[Tuple[mindspore.Tensor], BaseModelOutputWithPooling]`

        Example:
            ```python
            >>> # Assumption: *get_visual_embeddings(image)* gets the visual embeddings of the image.
            >>> from transformers import AutoTokenizer, VisualBertModel
            >>> import mindspore, ops
            ...
            >>> tokenizer = AutoTokenizer.from_pretrained("google-bert/bert-base-uncased")
            >>> model = VisualBertModel.from_pretrained("uclanlp/visualbert-vqa-coco-pre")
            ...
            >>> inputs = tokenizer("The capital of France is Paris.", return_tensors="pt")
            >>> visual_embeds = get_visual_embeddings(image).unsqueeze(0)
            >>> visual_token_type_ids = ops.ones(visual_embeds.shape[:-1], dtype=mindspore.int64)
            >>> visual_attention_mask = ops.ones(visual_embeds.shape[:-1], dtype=mindspore.float32)
            ...
            >>> inputs.update(
            ...     {
            ...         "visual_embeds": visual_embeds,
            ...         "visual_token_type_ids": visual_token_type_ids,
            ...         "visual_attention_mask": visual_attention_mask,
            ...     }
            ... )
            ...
            >>> outputs = model(**inputs)
            ...
            >>> last_hidden_states = outputs.last_hidden_state
            ```
        """

        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

        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
        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")

        batch_size, seq_length = input_shape

        if visual_embeds is not None:
            visual_input_shape = visual_embeds.shape[:-1]

        if attention_mask is None:
            attention_mask = ops.ones(input_shape)

        if visual_embeds is not None and visual_attention_mask is None:
            visual_attention_mask = ops.ones(visual_input_shape)

        # We can provide a self-attention mask of dimensions [batch_size, from_seq_length, to_seq_length]
        # ourselves in which case we just need to make it broadcastable to all heads.
        if visual_embeds is not None:
            combined_attention_mask = ops.cat((attention_mask, visual_attention_mask), axis=-1)
            extended_attention_mask: mindspore.Tensor = self.get_extended_attention_mask(
                combined_attention_mask, (batch_size, input_shape + visual_input_shape)
            )

        else:
            extended_attention_mask: mindspore.Tensor = self.get_extended_attention_mask(
                attention_mask, (batch_size, input_shape)
            )

        # Prepare head mask if needed
        # 1.0 in head_mask indicate we keep the head
        # attention_probs has shape bsz x n_heads x N x N
        # input head_mask has shape [num_heads] or [num_hidden_layers x num_heads]
        # and head_mask is converted to shape [num_hidden_layers x batch x num_heads x seq_length x seq_length]
        head_mask = self.get_head_mask(head_mask, self.config.num_hidden_layers)

        embedding_output = self.embeddings(
            input_ids=input_ids,
            position_ids=position_ids,
            token_type_ids=token_type_ids,
            inputs_embeds=inputs_embeds,
            visual_embeds=visual_embeds,
            visual_token_type_ids=visual_token_type_ids,
            image_text_alignment=image_text_alignment,
        )

        if self.bypass_transformer and visual_embeds is not None:
            text_length = input_ids.shape[1]
            text_embedding_output = embedding_output[:, :text_length, :]
            visual_embedding_output = embedding_output[:, text_length:, :]

            text_extended_attention_mask = extended_attention_mask[:, :, text_length, :text_length]

            encoded_outputs = self.encoder(
                text_embedding_output,
                attention_mask=text_extended_attention_mask,
                output_attentions=output_attentions,
                output_hidden_states=output_hidden_states,
                return_dict=return_dict,
            )
            sequence_output = encoded_outputs[0]
            concatenated_input = ops.cat((sequence_output, visual_embedding_output), axis=1)
            sequence_output = self.additional_layer(concatenated_input, extended_attention_mask)
            pooled_output = self.pooler(sequence_output) if self.pooler is not None else None

        else:
            encoder_outputs = self.encoder(
                embedding_output,
                attention_mask=extended_attention_mask,
                head_mask=head_mask,
                output_attentions=output_attentions,
                output_hidden_states=output_hidden_states,
                return_dict=return_dict,
            )
            sequence_output = encoder_outputs[0]

            pooled_output = self.pooler(sequence_output) if self.pooler is not None else None

        if not return_dict:
            return (sequence_output, pooled_output) + encoder_outputs[1:]

        return BaseModelOutputWithPooling(
            last_hidden_state=sequence_output,
            pooler_output=pooled_output,
            hidden_states=encoder_outputs.hidden_states,
            attentions=encoder_outputs.attentions,
        )

mindnlp.transformers.models.visual_bert.modeling_visual_bert.VisualBertModel.forward(input_ids=None, attention_mask=None, token_type_ids=None, position_ids=None, head_mask=None, inputs_embeds=None, visual_embeds=None, visual_attention_mask=None, visual_token_type_ids=None, image_text_alignment=None, output_attentions=None, output_hidden_states=None, return_dict=None)

RETURNS DESCRIPTION
Union[Tuple[Tensor], BaseModelOutputWithPooling]

Union[Tuple[mindspore.Tensor], BaseModelOutputWithPooling]
Example 
>>> # Assumption: *get_visual_embeddings(image)* gets the visual embeddings of the image.
>>> from transformers import AutoTokenizer, VisualBertModel
>>> import mindspore, ops
...
>>> tokenizer = AutoTokenizer.from_pretrained("google-bert/bert-base-uncased")
>>> model = VisualBertModel.from_pretrained("uclanlp/visualbert-vqa-coco-pre")
...
>>> inputs = tokenizer("The capital of France is Paris.", return_tensors="pt")
>>> visual_embeds = get_visual_embeddings(image).unsqueeze(0)
>>> visual_token_type_ids = ops.ones(visual_embeds.shape[:-1], dtype=mindspore.int64)
>>> visual_attention_mask = ops.ones(visual_embeds.shape[:-1], dtype=mindspore.float32)
...
>>> inputs.update(
...     {
...         "visual_embeds": visual_embeds,
...         "visual_token_type_ids": visual_token_type_ids,
...         "visual_attention_mask": visual_attention_mask,
...     }
... )
...
>>> outputs = model(**inputs)
...
>>> last_hidden_states = outputs.last_hidden_state
Source code in mindnlp/transformers/models/visual_bert/modeling_visual_bert.py 
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
def forward(
    self,
    input_ids: Optional[mindspore.Tensor] = None,
    attention_mask: Optional[mindspore.Tensor] = None,
    token_type_ids: Optional[mindspore.Tensor] = None,
    position_ids: Optional[mindspore.Tensor] = None,
    head_mask: Optional[mindspore.Tensor] = None,
    inputs_embeds: Optional[mindspore.Tensor] = None,
    visual_embeds: Optional[mindspore.Tensor] = None,
    visual_attention_mask: Optional[mindspore.Tensor] = None,
    visual_token_type_ids: Optional[mindspore.Tensor] = None,
    image_text_alignment: Optional[mindspore.Tensor] = None,
    output_attentions: Optional[bool] = None,
    output_hidden_states: Optional[bool] = None,
    return_dict: Optional[bool] = None,
) -> Union[Tuple[mindspore.Tensor], BaseModelOutputWithPooling]:
    r"""

    Returns:
        `Union[Tuple[mindspore.Tensor], BaseModelOutputWithPooling]`

    Example:
        ```python
        >>> # Assumption: *get_visual_embeddings(image)* gets the visual embeddings of the image.
        >>> from transformers import AutoTokenizer, VisualBertModel
        >>> import mindspore, ops
        ...
        >>> tokenizer = AutoTokenizer.from_pretrained("google-bert/bert-base-uncased")
        >>> model = VisualBertModel.from_pretrained("uclanlp/visualbert-vqa-coco-pre")
        ...
        >>> inputs = tokenizer("The capital of France is Paris.", return_tensors="pt")
        >>> visual_embeds = get_visual_embeddings(image).unsqueeze(0)
        >>> visual_token_type_ids = ops.ones(visual_embeds.shape[:-1], dtype=mindspore.int64)
        >>> visual_attention_mask = ops.ones(visual_embeds.shape[:-1], dtype=mindspore.float32)
        ...
        >>> inputs.update(
        ...     {
        ...         "visual_embeds": visual_embeds,
        ...         "visual_token_type_ids": visual_token_type_ids,
        ...         "visual_attention_mask": visual_attention_mask,
        ...     }
        ... )
        ...
        >>> outputs = model(**inputs)
        ...
        >>> last_hidden_states = outputs.last_hidden_state
        ```
    """

    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

    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
    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")

    batch_size, seq_length = input_shape

    if visual_embeds is not None:
        visual_input_shape = visual_embeds.shape[:-1]

    if attention_mask is None:
        attention_mask = ops.ones(input_shape)

    if visual_embeds is not None and visual_attention_mask is None:
        visual_attention_mask = ops.ones(visual_input_shape)

    # We can provide a self-attention mask of dimensions [batch_size, from_seq_length, to_seq_length]
    # ourselves in which case we just need to make it broadcastable to all heads.
    if visual_embeds is not None:
        combined_attention_mask = ops.cat((attention_mask, visual_attention_mask), axis=-1)
        extended_attention_mask: mindspore.Tensor = self.get_extended_attention_mask(
            combined_attention_mask, (batch_size, input_shape + visual_input_shape)
        )

    else:
        extended_attention_mask: mindspore.Tensor = self.get_extended_attention_mask(
            attention_mask, (batch_size, input_shape)
        )

    # Prepare head mask if needed
    # 1.0 in head_mask indicate we keep the head
    # attention_probs has shape bsz x n_heads x N x N
    # input head_mask has shape [num_heads] or [num_hidden_layers x num_heads]
    # and head_mask is converted to shape [num_hidden_layers x batch x num_heads x seq_length x seq_length]
    head_mask = self.get_head_mask(head_mask, self.config.num_hidden_layers)

    embedding_output = self.embeddings(
        input_ids=input_ids,
        position_ids=position_ids,
        token_type_ids=token_type_ids,
        inputs_embeds=inputs_embeds,
        visual_embeds=visual_embeds,
        visual_token_type_ids=visual_token_type_ids,
        image_text_alignment=image_text_alignment,
    )

    if self.bypass_transformer and visual_embeds is not None:
        text_length = input_ids.shape[1]
        text_embedding_output = embedding_output[:, :text_length, :]
        visual_embedding_output = embedding_output[:, text_length:, :]

        text_extended_attention_mask = extended_attention_mask[:, :, text_length, :text_length]

        encoded_outputs = self.encoder(
            text_embedding_output,
            attention_mask=text_extended_attention_mask,
            output_attentions=output_attentions,
            output_hidden_states=output_hidden_states,
            return_dict=return_dict,
        )
        sequence_output = encoded_outputs[0]
        concatenated_input = ops.cat((sequence_output, visual_embedding_output), axis=1)
        sequence_output = self.additional_layer(concatenated_input, extended_attention_mask)
        pooled_output = self.pooler(sequence_output) if self.pooler is not None else None

    else:
        encoder_outputs = self.encoder(
            embedding_output,
            attention_mask=extended_attention_mask,
            head_mask=head_mask,
            output_attentions=output_attentions,
            output_hidden_states=output_hidden_states,
            return_dict=return_dict,
        )
        sequence_output = encoder_outputs[0]

        pooled_output = self.pooler(sequence_output) if self.pooler is not None else None

    if not return_dict:
        return (sequence_output, pooled_output) + encoder_outputs[1:]

    return BaseModelOutputWithPooling(
        last_hidden_state=sequence_output,
        pooler_output=pooled_output,
        hidden_states=encoder_outputs.hidden_states,
        attentions=encoder_outputs.attentions,
    )

mindnlp.transformers.models.visual_bert.modeling_visual_bert.VisualBertPreTrainedModel

Bases: PreTrainedModel

An abstract class to handle weights initialization and a simple interface for downloading and loading pretrained models.

Source code in mindnlp/transformers/models/visual_bert/modeling_visual_bert.py 
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
class VisualBertPreTrainedModel(PreTrainedModel):
    """
    An abstract class to handle weights initialization and a simple interface for downloading and loading pretrained
    models.
    """

    config_class = VisualBertConfig
    base_model_prefix = "visual_bert"
    supports_gradient_checkpointing = True

    def _init_weights(self, cell):
        """Initialize the weights"""
        if isinstance(cell, (nn.Linear, nn.Embedding)):
            # Slightly different from the TF version which uses truncated_normal for initialization
            # cf https://github.com/pytorch/pytorch/pull/5617
            cell.weight.set_data(initializer(Normal(self.config.initializer_range),
                                                    cell.weight.shape, cell.weight.dtype))
        elif isinstance(cell, nn.LayerNorm):
            cell.weight.set_data(initializer('ones', cell.weight.shape, cell.weight.dtype))
            cell.bias.set_data(initializer('zeros', cell.bias.shape, cell.bias.dtype))
        if isinstance(cell, nn.Linear) and cell.bias is not None:
            cell.bias.set_data(initializer('zeros', cell.bias.shape, cell.bias.dtype))