The BridgeTower model was proposed in BridgeTower: Building Bridges Between Encoders in Vision-Language Representative Learning by Xiao Xu, Chenfei Wu, Shachar Rosenman, Vasudev Lal, Wanxiang Che, Nan Duan. The goal of this model is to build a bridge between each uni-modal encoder and the cross-modal encoder to enable comprehensive and detailed interaction at each layer of the cross-modal encoder thus achieving remarkable performance on various downstream tasks with almost negligible additional performance and computational costs.
This paper has been accepted to the AAAI’23 conference.
The abstract from the paper is the following:
Vision-Language (VL) models with the TWO-TOWER architecture have dominated visual-language representation learning in recent years. Current VL models either use lightweight uni-modal encoders and learn to extract, align and fuse both modalities simultaneously in a deep cross-modal encoder, or feed the last-layer uni-modal representations from the deep pre-trained uni-modal encoders into the top cross-modal encoder. Both approaches potentially restrict vision-language representation learning and limit model performance. In this paper, we propose BRIDGETOWER, which introduces multiple bridge layers that build a connection between the top layers of uni-modal encoders and each layer of the crossmodal encoder. This enables effective bottom-up cross-modal alignment and fusion between visual and textual representations of different semantic levels of pre-trained uni-modal encoders in the cross-modal encoder. Pre-trained with only 4M images, BRIDGETOWER achieves state-of-the-art performance on various downstream vision-language tasks. In particular, on the VQAv2 test-std set, BRIDGETOWER achieves an accuracy of 78.73%, outperforming the previous state-of-the-art model METER by 1.09% with the same pre-training data and almost negligible additional parameters and computational costs. Notably, when further scaling the model, BRIDGETOWER achieves an accuracy of 81.15%, surpassing models that are pre-trained on orders-of-magnitude larger datasets.
BridgeTower architecture. Taken from the original paper. This model was contributed by Anahita Bhiwandiwalla, Tiep Le and Shaoyen Tseng. The original code can be found here.
BridgeTower consists of a visual encoder, a textual encoder and cross-modal encoder with multiple lightweight bridge layers. The goal of this approach was to build a bridge between each uni-modal encoder and the cross-modal encoder to enable comprehensive and detailed interaction at each layer of the cross-modal encoder. In principle, one can apply any visual, textual or cross-modal encoder in the proposed architecture.
The BridgeTowerProcessor wraps RobertaTokenizer and BridgeTowerImageProcessor into a single instance to both encode the text and prepare the images respectively.
The following example shows how to run contrastive learning using BridgeTowerProcessor and BridgeTowerForContrastiveLearning.
>>> from transformers import BridgeTowerProcessor, BridgeTowerForContrastiveLearning
>>> import requests
>>> from PIL import Image
>>> url = "http://images.cocodataset.org/val2017/000000039769.jpg"
>>> image = Image.open(requests.get(url, stream=True).raw)
>>> texts = ["An image of two cats chilling on a couch", "A football player scoring a goal"]
>>> processor = BridgeTowerProcessor.from_pretrained("BridgeTower/bridgetower-large-itm-mlm-itc")
>>> model = BridgeTowerForContrastiveLearning.from_pretrained("BridgeTower/bridgetower-large-itm-mlm-itc")
>>> # forward pass
>>> scores = dict()
>>> for text in texts:
... # prepare inputs
... encoding = processor(image, text, return_tensors="pt")
... outputs = model(**encoding)
... scores[text] = outputsThe following example shows how to run image-text retrieval using BridgeTowerProcessor and BridgeTowerForImageAndTextRetrieval.
>>> from transformers import BridgeTowerProcessor, BridgeTowerForImageAndTextRetrieval
>>> import requests
>>> from PIL import Image
>>> url = "http://images.cocodataset.org/val2017/000000039769.jpg"
>>> image = Image.open(requests.get(url, stream=True).raw)
>>> texts = ["An image of two cats chilling on a couch", "A football player scoring a goal"]
>>> processor = BridgeTowerProcessor.from_pretrained("BridgeTower/bridgetower-base-itm-mlm")
>>> model = BridgeTowerForImageAndTextRetrieval.from_pretrained("BridgeTower/bridgetower-base-itm-mlm")
>>> # forward pass
>>> scores = dict()
>>> for text in texts:
... # prepare inputs
... encoding = processor(image, text, return_tensors="pt")
... outputs = model(**encoding)
... scores[text] = outputs.logits[0, 1].item()The following example shows how to run masked language modeling using BridgeTowerProcessor and BridgeTowerForMaskedLM.
>>> from transformers import BridgeTowerProcessor, BridgeTowerForMaskedLM
>>> from PIL import Image
>>> import requests
>>> url = "http://images.cocodataset.org/val2017/000000360943.jpg"
>>> image = Image.open(requests.get(url, stream=True).raw).convert("RGB")
>>> text = "a <mask> looking out of the window"
>>> processor = BridgeTowerProcessor.from_pretrained("BridgeTower/bridgetower-base-itm-mlm")
>>> model = BridgeTowerForMaskedLM.from_pretrained("BridgeTower/bridgetower-base-itm-mlm")
>>> # prepare inputs
>>> encoding = processor(image, text, return_tensors="pt")
>>> # forward pass
>>> outputs = model(**encoding)
>>> results = processor.decode(outputs.logits.argmax(dim=-1).squeeze(0).tolist())
>>> print(results)
.a cat looking out of the window.Tips:
( share_cross_modal_transformer_layers = True hidden_act = 'gelu' hidden_size = 768 initializer_factor = 1 layer_norm_eps = 1e-05 share_link_tower_layers = False link_tower_type = 'add' num_attention_heads = 12 num_hidden_layers = 6 tie_word_embeddings = False init_layernorm_from_vision_encoder = False text_config = None vision_config = None **kwargs )
Parameters
bool, optional, defaults to True) —
Whether cross modal transformer layers are shared. str or function, optional, defaults to "gelu") —
The non-linear activation function (function or string) in the encoder and pooler. int, optional, defaults to 768) —
Dimensionality of the encoder layers and the pooler layer. float, optional, defaults to 1) —
A factor for initializing all weight matrices (should be kept to 1, used internally for initialization
testing). float, optional, defaults to 1e-05) —
The epsilon used by the layer normalization layers. bool, optional, defaults to False) —
Whether the bride/link tower layers are shared. str, optional, defaults to "add") —
Type of the bridge/link layer. int, optional, defaults to 12) —
Number of attention heads for each attention layer in the Transformer encoder. int, optional, defaults to 6) —
Number of hidden layers in the Transformer encoder. bool, optional, defaults to False) —
Whether to tie input and output embeddings. bool, optional, defaults to False) —
Whether to init LayerNorm from the vision encoder. dict, optional) —
Dictionary of configuration options used to initialize BridgeTowerTextConfig. dict, optional) —
Dictionary of configuration options used to initialize BridgeTowerVisionConfig. This is the configuration class to store the configuration of a BridgeTowerModel. It is used to instantiate a BridgeTower 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 bridgetower-base BridgeTower/bridgetower-base architecture.
Configuration objects inherit from PretrainedConfig and can be used to control the model outputs. Read the documentation from PretrainedConfig for more information.
Example:
>>> from transformers import BridgeTowerModel, BridgeTowerConfig
>>> # Initializing a BridgeTower BridgeTower/bridgetower-base style configuration
>>> configuration = BridgeTowerConfig()
>>> # Initializing a model from the BridgeTower/bridgetower-base style configuration
>>> model = BridgeTowerModel(configuration)
>>> # Accessing the model configuration
>>> configuration = model.config( text_config: BridgeTowerTextConfig vision_config: BridgeTowerVisionConfig **kwargs )
Instantiate a BridgeTowerConfig (or a derived class) from BridgeTower text model configuration. Returns: BridgeTowerConfig: An instance of a configuration object
( vocab_size = 50265 hidden_size = 768 num_hidden_layers = 12 num_attention_heads = 12 initializer_factor = 1 intermediate_size = 3072 hidden_act = 'gelu' hidden_dropout_prob = 0.1 attention_probs_dropout_prob = 0.1 max_position_embeddings = 514 type_vocab_size = 1 layer_norm_eps = 1e-05 pad_token_id = 1 bos_token_id = 0 eos_token_id = 2 position_embedding_type = 'absolute' use_cache = True **kwargs )
Parameters
int, optional, defaults to 50265) —
Vocabulary size of the text part of the model. Defines the number of different tokens that can be
represented by the inputs_ids passed when calling BridgeTowerModel. int, optional, defaults to 768) —
Dimensionality of the encoder layers and the pooler layer. int, optional, defaults to 12) —
Number of hidden layers in the Transformer encoder. int, optional, defaults to 12) —
Number of attention heads for each attention layer in the Transformer encoder. int, optional, defaults to 3072) —
Dimensionality of the “intermediate” (often named feed-forward) layer in the Transformer encoder. str or Callable, optional, defaults to "gelu") —
The non-linear activation function (function or string) in the encoder and pooler. If string, "gelu",
"relu", "silu" and "gelu_new" are supported. float, optional, defaults to 0.1) —
The dropout probability for all fully connected layers in the embeddings, encoder, and pooler. float, optional, defaults to 0.1) —
The dropout ratio for the attention probabilities. int, optional, defaults to 514) —
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). int, optional, defaults to 2) —
The vocabulary size of the token_type_ids. float, optional, defaults to 1) —
A factor for initializing all weight matrices (should be kept to 1, used internally for initialization
testing). float, optional, defaults to 1e-05) —
The epsilon used by the layer normalization layers. str, optional, defaults to "absolute") —
Type of position embedding. Choose one of "absolute", "relative_key", "relative_key_query". For
positional embeddings use "absolute". For more information on "relative_key", please refer to
Self-Attention with Relative Position Representations (Shaw et al.).
For more information on "relative_key_query", please refer to Method 4 in Improve Transformer Models
with Better Relative Position Embeddings (Huang et al.). bool, optional, defaults to False) —
Whether the model is used as a decoder or not. If False, the model is used as an encoder. bool, optional, defaults to True) —
Whether or not the model should return the last key/values attentions (not used by all models). Only
relevant if config.is_decoder=True. This is the configuration class to store the text configuration of a BridgeTowerModel. The default values here are copied from RoBERTa. Instantiating a configuration with the defaults will yield a similar configuration to that of the bridgetower-base BridegTower/bridgetower-base architecture.
Configuration objects inherit from PretrainedConfig and can be used to control the model outputs. Read the documentation from PretrainedConfig for more information.
( hidden_size = 768 num_hidden_layers = 12 num_channels = 3 patch_size = 16 image_size = 288 initializer_factor = 1 layer_norm_eps = 1e-05 stop_gradient = False share_layernorm = True remove_last_layer = False **kwargs )
Parameters
int, optional, defaults to 768) —
Dimensionality of the encoder layers and the pooler layer. int, optional, defaults to 12) —
Number of hidden layers in visual encoder model. int, optional, defaults to 16) —
The size (resolution) of each patch. int, optional, defaults to 288) —
The size (resolution) of each image. float, optional, defaults to 1) —
A factor for initializing all weight matrices (should be kept to 1, used internally for initialization
testing). float, optional, defaults to 1e-05) —
The epsilon used by the layer normalization layers. bool, optional, defaults to False) —
Whether to stop gradient for training. bool, optional, defaults to True) —
Whether LayerNorm layers are shared. bool, optional, defaults to False) —
Whether to remove the last layer from the vision encoder. This is the configuration class to store the vision configuration of a BridgeTowerModel. Instantiating a configuration with the defaults will yield a similar configuration to that of the bridgetower-base BridgeTower/bridgetower-base architecture.
Configuration objects inherit from PretrainedConfig and can be used to control the model outputs. Read the documentation from PretrainedConfig for more information.
( do_resize: bool = True size: Dict = 288 size_divisor: int = 32 resample: Resampling = <Resampling.BICUBIC: 3> do_rescale: bool = True rescale_factor: Union = 0.00392156862745098 do_normalize: bool = True image_mean: Union = None image_std: Union = None do_center_crop: bool = True do_pad: bool = True **kwargs )
Parameters
bool, optional, defaults to True) —
Whether to resize the image’s (height, width) dimensions to the specified size. Can be overridden by the
do_resize parameter in the preprocess method. Dict[str, int] optional, defaults to 288) —
Resize the shorter side of the input to size["shortest_edge"]. The longer side will be limited to under
int((1333 / 800) * size["shortest_edge"]) while preserving the aspect ratio. Only has an effect if
do_resize is set to True. Can be overridden by the size parameter in the preprocess method. int, optional, defaults to 32) —
The size by which to make sure both the height and width can be divided. Only has an effect if do_resize
is set to True. Can be overridden by the size_divisor parameter in the preprocess method. PILImageResampling, optional, defaults to Resampling.BICUBIC) —
Resampling filter to use if resizing the image. Only has an effect if do_resize is set to True. Can be
overridden by the resample parameter in the preprocess method. bool, optional, defaults to True) —
Whether to rescale the image by the specified scale rescale_factor. Can be overridden by the do_rescale
parameter in the preprocess method. int or float, optional, defaults to 1/255) —
Scale factor to use if rescaling the image. Only has an effect if do_rescale is set to True. Can be
overridden by the rescale_factor parameter in the preprocess method. bool, optional, defaults to True) —
Whether to normalize the image. Can be overridden by the do_normalize parameter in the preprocess
method. Can be overridden by the do_normalize parameter in the preprocess method. float or List[float], optional, defaults to IMAGENET_STANDARD_MEAN) —
Mean to use if normalizing the image. This is a float or list of floats the length of the number of
channels in the image. Can be overridden by the image_mean parameter in the preprocess method. Can be
overridden by the image_mean parameter in the preprocess method. float or List[float], optional, defaults to IMAGENET_STANDARD_STD) —
Standard deviation to use if normalizing the image. This is a float or list of floats the length of the
number of channels in the image. Can be overridden by the image_std parameter in the preprocess method.
Can be overridden by the image_std parameter in the preprocess method. bool, optional, defaults to True) —
Whether to center crop the image. Can be overridden by the do_center_crop parameter in the preprocess
method. bool, optional, defaults to True) —
Whether to pad the image to the (max_height, max_width) of the images in the batch. Can be overridden by
the do_pad parameter in the preprocess method. Constructs a BridgeTower image processor.
( images: Union do_resize: Optional = None size: Optional = None size_divisor: Optional = None resample: Resampling = None do_rescale: Optional = None rescale_factor: Optional = None do_normalize: Optional = None image_mean: Union = None image_std: Union = None do_pad: Optional = None do_center_crop: Optional = None return_tensors: Union = None data_format: ChannelDimension = <ChannelDimension.FIRST: 'channels_first'> input_data_format: Union = None **kwargs )
Parameters
ImageInput) —
Image to preprocess. Expects a single or batch of images with pixel values ranging from 0 to 255. If
passing in images with pixel values between 0 and 1, set do_rescale=False. bool, optional, defaults to self.do_resize) —
Whether to resize the image. Dict[str, int], optional, defaults to self.size) —
Controls the size of the image after resize. The shortest edge of the image is resized to
size["shortest_edge"] whilst preserving the aspect ratio. If the longest edge of this resized image
is > int(size["shortest_edge"] * (1333 / 800)), then the image is resized again to make the longest
edge equal to int(size["shortest_edge"] * (1333 / 800)). int, optional, defaults to self.size_divisor) —
The image is resized to a size that is a multiple of this value. PILImageResampling, optional, defaults to self.resample) —
Resampling filter to use if resizing the image. Only has an effect if do_resize is set to True. bool, optional, defaults to self.do_rescale) —
Whether to rescale the image values between [0 - 1]. float, optional, defaults to self.rescale_factor) —
Rescale factor to rescale the image by if do_rescale is set to True. bool, optional, defaults to self.do_normalize) —
Whether to normalize the image. float or List[float], optional, defaults to self.image_mean) —
Image mean to normalize the image by if do_normalize is set to True. float or List[float], optional, defaults to self.image_std) —
Image standard deviation to normalize the image by if do_normalize is set to True. bool, optional, defaults to self.do_pad) —
Whether to pad the image to the (max_height, max_width) in the batch. If True, a pixel mask is also
created and returned. bool, optional, defaults to self.do_center_crop) —
Whether to center crop the image. If the input size is smaller than crop_size along any edge, the
image is padded with 0’s and then center cropped. str or TensorType, optional) —
The type of tensors to return. Can be one of:np.ndarray.TensorType.TENSORFLOW or 'tf': Return a batch of type tf.Tensor.TensorType.PYTORCH or 'pt': Return a batch of type torch.Tensor.TensorType.NUMPY or 'np': Return a batch of type np.ndarray.TensorType.JAX or 'jax': Return a batch of type jax.numpy.ndarray.ChannelDimension or str, optional, defaults to ChannelDimension.FIRST) —
The channel dimension format for the output image. Can be one of:"channels_first" or ChannelDimension.FIRST: image in (num_channels, height, width) format."channels_last" or ChannelDimension.LAST: image in (height, width, num_channels) format.ChannelDimension or str, optional) —
The channel dimension format for the input image. If unset, the channel dimension format is inferred
from the input image. Can be one of:"channels_first" or ChannelDimension.FIRST: image in (num_channels, height, width) format."channels_last" or ChannelDimension.LAST: image in (height, width, num_channels) format."none" or ChannelDimension.NONE: image in (height, width) format.Preprocess an image or batch of images.
( image_processor tokenizer )
Parameters
BridgeTowerImageProcessor) —
An instance of BridgeTowerImageProcessor. The image processor is a required input. RobertaTokenizerFast) —
An instance of [‘RobertaTokenizerFast`]. The tokenizer is a required input. Constructs a BridgeTower processor which wraps a Roberta tokenizer and BridgeTower image processor into a single processor.
BridgeTowerProcessor offers all the functionalities of BridgeTowerImageProcessor and
RobertaTokenizerFast. See the docstring of call() and
decode() for more information.
( images text: Union = None add_special_tokens: bool = True padding: Union = False truncation: Union = None max_length: Optional = None stride: int = 0 pad_to_multiple_of: Optional = None return_token_type_ids: Optional = None return_attention_mask: Optional = None return_overflowing_tokens: bool = False return_special_tokens_mask: bool = False return_offsets_mapping: bool = False return_length: bool = False verbose: bool = True return_tensors: Union = None **kwargs )
This method uses BridgeTowerImageProcessor.call() method to prepare image(s) for the model, and RobertaTokenizerFast.call() to prepare text for the model.
Please refer to the docstring of the above two methods for more information.
( config )
Parameters
The bare BridgeTower Model transformer outputting BridgeTowerModelOutput object without any specific head on top.
This model is a PyTorch torch.nn.Module <https://pytorch.org/docs/stable/nn.html#torch.nn.Module>_ subclass. Use
it as a regular PyTorch Module and refer to the PyTorch documentation for all matter related to general usage and
behavior.
( input_ids: Optional = None attention_mask: Optional = None token_type_ids: Optional = None pixel_values: Optional = None pixel_mask: Optional = None head_mask: Optional = None inputs_embeds: Optional = None image_embeds: Optional = None image_token_type_idx: Optional = None output_attentions: Optional = None output_hidden_states: Optional = None return_dict: Optional = None labels: Optional = None ) → transformers.models.bridgetower.modeling_bridgetower.BridgeTowerModelOutput or tuple(torch.FloatTensor)
Parameters
torch.LongTensor of shape ({0})) —
Indices of input sequence tokens in the vocabulary. Indices can be obtained using AutoTokenizer. See
PreTrainedTokenizer.encode() and PreTrainedTokenizer.call() for details. What are input
IDs? torch.FloatTensor of shape ({0}), optional) —
Mask to avoid performing attention on padding token indices. Mask values selected in [0, 1]:
torch.LongTensor of shape ({0}), optional) —
Segment token indices to indicate first and second portions of the inputs. Indices are selected in [0, 1]:
torch.FloatTensor of shape (batch_size, num_channels, height, width)) —
Pixel values. Pixel values can be obtained using BridgeTowerImageProcessor. See
BridgeTowerImageProcessor.call() for details. torch.LongTensor of shape (batch_size, height, width), optional) —
Mask to avoid performing attention on padding pixel values. Mask values selected in [0, 1]:
What are attention masks? <../glossary.html#attention-mask>__torch.FloatTensor of shape (num_heads,) or (num_layers, num_heads), optional) —
Mask to nullify selected heads of the self-attention modules. Mask values selected in [0, 1]:
torch.FloatTensor of shape ({0}, hidden_size), optional) —
Optionally, instead of passing input_ids you can choose to directly pass an embedded representation. This
is useful if you want more control over how to convert input_ids indices into associated vectors than the
model’s internal embedding lookup matrix. torch.FloatTensor of shape (batch_size, num_patches, hidden_size), optional) —
Optionally, instead of passing pixel_values, you can choose to directly pass an embedded representation.
This is useful if you want more control over how to convert pixel_values into patch embeddings. int, optional) —
bool, optional) —
Whether or not to return the attentions tensors of all attention layers. See attentions under returned
tensors for more detail. bool, optional) —
Whether or not to return the hidden states of all layers. See hidden_states under returned tensors for
more detail. bool, optional) —
Whether or not to return a ModelOutput instead of a plain tuple. bool, optional) —
If set to True, hidden states are returned as a list containing the hidden states of text, image, and
cross-modal components respectively. i.e. (hidden_states_text, hidden_states_image, hidden_states_cross_modal) where each element is a list of the hidden states of the corresponding
modality. hidden_states_txt/img are a list of tensors corresponding to unimodal hidden states and
hidden_states_cross_modal is a list of tuples containing cross_modal_text_hidden_states and
cross_modal_image_hidden_states of each brdige layer. torch.LongTensor of shape (batch_size,), optional) —
Labels are currently not supported. Returns
transformers.models.bridgetower.modeling_bridgetower.BridgeTowerModelOutput or tuple(torch.FloatTensor)
A transformers.models.bridgetower.modeling_bridgetower.BridgeTowerModelOutput or a tuple of
torch.FloatTensor (if return_dict=False is passed or when config.return_dict=False) comprising various
elements depending on the configuration (BridgeTowerConfig) and inputs.
text_features (torch.FloatTensor of shape (batch_size, text_sequence_length, hidden_size)) — Sequence of hidden-states at the text output of the last layer of the model.
image_features (torch.FloatTensor of shape (batch_size, image_sequence_length, hidden_size)) — Sequence of hidden-states at the image output of the last layer of the model.
pooler_output (torch.FloatTensor of shape (batch_size, hidden_size x 2)) — Concatenation of last layer hidden-state of the first token of the text and image sequence (classification
token), respectively, after further processing through layers used for auxiliary pretraining tasks.
hidden_states (tuple(torch.FloatTensor), optional, returned when output_hidden_states=True is passed or when config.output_hidden_states=True) — Tuple of torch.FloatTensor (one for the output of the embeddings, if the model has an embedding layer, +
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 optional initial embedding outputs.
attentions (tuple(torch.FloatTensor), optional, returned when output_attentions=True is passed or when config.output_attentions=True) — Tuple of torch.FloatTensor (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.
The BridgeTowerModel forward method, overrides the __call__ special method.
Although the recipe for forward pass needs to be defined within this function, one should call the Module
instance afterwards instead of this since the former takes care of running the pre and post processing steps while
the latter silently ignores them.
Examples:
>>> from transformers import BridgeTowerProcessor, BridgeTowerModel
>>> from PIL import Image
>>> import requests
>>> # prepare image and text
>>> url = "http://images.cocodataset.org/val2017/000000039769.jpg"
>>> image = Image.open(requests.get(url, stream=True).raw)
>>> text = "hello world"
>>> processor = BridgeTowerProcessor.from_pretrained("BridgeTower/bridgetower-base")
>>> model = BridgeTowerModel.from_pretrained("BridgeTower/bridgetower-base")
>>> inputs = processor(image, text, return_tensors="pt")
>>> outputs = model(**inputs)
>>> outputs.keys()
odict_keys(['text_features', 'image_features', 'pooler_output'])( config )
Parameters
BridgeTower Model with a image-text contrastive head on top computing image-text contrastive loss.
This model is a PyTorch torch.nn.Module <https://pytorch.org/docs/stable/nn.html#torch.nn.Module>_ subclass. Use
it as a regular PyTorch Module and refer to the PyTorch documentation for all matter related to general usage and
behavior.
( input_ids: Optional = None attention_mask: Optional = None token_type_ids: Optional = None pixel_values: Optional = None pixel_mask: Optional = None head_mask: Optional = None inputs_embeds: Optional = None image_embeds: Optional = None output_attentions: Optional = None output_hidden_states: Optional = True return_dict: Optional = None return_loss: Optional = None ) → transformers.models.bridgetower.modeling_bridgetower.BridgeTowerContrastiveOutput or tuple(torch.FloatTensor)
Parameters
torch.LongTensor of shape ({0})) —
Indices of input sequence tokens in the vocabulary. Indices can be obtained using AutoTokenizer. See
PreTrainedTokenizer.encode() and PreTrainedTokenizer.call() for details. What are input
IDs? torch.FloatTensor of shape ({0}), optional) —
Mask to avoid performing attention on padding token indices. Mask values selected in [0, 1]:
torch.LongTensor of shape ({0}), optional) —
Segment token indices to indicate first and second portions of the inputs. Indices are selected in [0, 1]:
torch.FloatTensor of shape (batch_size, num_channels, height, width)) —
Pixel values. Pixel values can be obtained using BridgeTowerImageProcessor. See
BridgeTowerImageProcessor.call() for details. torch.LongTensor of shape (batch_size, height, width), optional) —
Mask to avoid performing attention on padding pixel values. Mask values selected in [0, 1]:
What are attention masks? <../glossary.html#attention-mask>__torch.FloatTensor of shape (num_heads,) or (num_layers, num_heads), optional) —
Mask to nullify selected heads of the self-attention modules. Mask values selected in [0, 1]:
torch.FloatTensor of shape ({0}, hidden_size), optional) —
Optionally, instead of passing input_ids you can choose to directly pass an embedded representation. This
is useful if you want more control over how to convert input_ids indices into associated vectors than the
model’s internal embedding lookup matrix. torch.FloatTensor of shape (batch_size, num_patches, hidden_size), optional) —
Optionally, instead of passing pixel_values, you can choose to directly pass an embedded representation.
This is useful if you want more control over how to convert pixel_values into patch embeddings. int, optional) —
bool, optional) —
Whether or not to return the attentions tensors of all attention layers. See attentions under returned
tensors for more detail. bool, optional) —
Whether or not to return the hidden states of all layers. See hidden_states under returned tensors for
more detail. bool, optional) —
Whether or not to return a ModelOutput instead of a plain tuple. bool, optional) —
Whether or not to return the contrastive loss. Returns
transformers.models.bridgetower.modeling_bridgetower.BridgeTowerContrastiveOutput or tuple(torch.FloatTensor)
A transformers.models.bridgetower.modeling_bridgetower.BridgeTowerContrastiveOutput or a tuple of
torch.FloatTensor (if return_dict=False is passed or when config.return_dict=False) comprising various
elements depending on the configuration (BridgeTowerConfig) and inputs.
torch.FloatTensor of shape (1,), optional, returned when return_loss is True — Image-text contrastive loss.torch.FloatTensor of shape (batch_size, sequence_length, config.vocab_size)) — Prediction scores of the language modeling head (scores for each vocabulary token before SoftMax).torch.FloatTensor), optional, returned when model is initialized with with_projection=True) — The text embeddings obtained by applying the projection layer to the pooler_output.torch.FloatTensor), optional, returned when model is initialized with with_projection=True) — The image embeddings obtained by applying the projection layer to the pooler_output.torch.FloatTensor), optional, returned when model is initialized with with_projection=True) — The text-image cross-modal embeddings obtained by applying the projection layer to the pooler_output.tuple(torch.FloatTensor), optional, returned when output_hidden_states=True is passed or when config.output_hidden_states=True) — Tuple of torch.FloatTensor (one for the output of the embeddings, if the model has an embedding layer, +
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 optional initial embedding outputs.tuple(torch.FloatTensor), optional, returned when output_attentions=True is passed or when config.output_attentions=True) — Tuple of torch.FloatTensor (one for each layer) of shape (batch_size, num_heads, sequence_length, sequence_length).The BridgeTowerForContrastiveLearning forward method, overrides the __call__ special method.
Although the recipe for forward pass needs to be defined within this function, one should call the Module
instance afterwards instead of this since the former takes care of running the pre and post processing steps while
the latter silently ignores them.
Examples:
>>> from transformers import BridgeTowerProcessor, BridgeTowerForContrastiveLearning
>>> import requests
>>> from PIL import Image
>>> import torch
>>> image_urls = [
... "https://farm4.staticflickr.com/3395/3428278415_81c3e27f15_z.jpg",
... "http://images.cocodataset.org/val2017/000000039769.jpg",
... ]
>>> texts = ["two dogs in a car", "two cats sleeping on a couch"]
>>> images = [Image.open(requests.get(url, stream=True).raw) for url in image_urls]
>>> processor = BridgeTowerProcessor.from_pretrained("BridgeTower/bridgetower-large-itm-mlm-itc")
>>> model = BridgeTowerForContrastiveLearning.from_pretrained("BridgeTower/bridgetower-large-itm-mlm-itc")
>>> inputs = processor(images, texts, padding=True, return_tensors="pt")
>>> loss = model(**inputs, return_loss=True).loss
>>> inputs = processor(images, texts[::-1], padding=True, return_tensors="pt")
>>> loss_swapped = model(**inputs, return_loss=True).loss
>>> print("Loss", round(loss.item(), 4))
Loss 0.0019
>>> print("Loss with swapped images", round(loss_swapped.item(), 4))
Loss with swapped images 2.126( config )
Parameters
BridgeTower Model with a language modeling head on top as done during pretraining.
This model is a PyTorch torch.nn.Module <https://pytorch.org/docs/stable/nn.html#torch.nn.Module>_ subclass. Use
it as a regular PyTorch Module and refer to the PyTorch documentation for all matter related to general usage and
behavior.
( input_ids: Optional = None attention_mask: Optional = None token_type_ids: Optional = None pixel_values: Optional = None pixel_mask: Optional = None head_mask: Optional = None inputs_embeds: Optional = None image_embeds: Optional = None output_attentions: Optional = None output_hidden_states: Optional = None return_dict: Optional = None labels: Optional = None ) → transformers.modeling_outputs.MaskedLMOutput or tuple(torch.FloatTensor)
Parameters
torch.LongTensor of shape (batch_size, sequence_length)) —
Indices of input sequence tokens in the vocabulary. Indices can be obtained using AutoTokenizer. See
PreTrainedTokenizer.encode() and PreTrainedTokenizer.call() for details. What are input
IDs? torch.FloatTensor of shape (batch_size, sequence_length), optional) —
Mask to avoid performing attention on padding token indices. Mask values selected in [0, 1]:
torch.LongTensor of shape (batch_size, sequence_length), optional) —
Segment token indices to indicate first and second portions of the inputs. Indices are selected in [0, 1]:
torch.FloatTensor of shape (batch_size, num_channels, height, width)) —
Pixel values. Pixel values can be obtained using BridgeTowerImageProcessor. See
BridgeTowerImageProcessor.call() for details. torch.LongTensor of shape (batch_size, height, width), optional) —
Mask to avoid performing attention on padding pixel values. Mask values selected in [0, 1]:
What are attention masks? <../glossary.html#attention-mask>__torch.FloatTensor of shape (num_heads,) or (num_layers, num_heads), optional) —
Mask to nullify selected heads of the self-attention modules. Mask values selected in [0, 1]:
torch.FloatTensor of shape (batch_size, sequence_length, hidden_size), optional) —
Optionally, instead of passing input_ids you can choose to directly pass an embedded representation. This
is useful if you want more control over how to convert input_ids indices into associated vectors than the
model’s internal embedding lookup matrix. torch.FloatTensor of shape (batch_size, num_patches, hidden_size), optional) —
Optionally, instead of passing pixel_values, you can choose to directly pass an embedded representation.
This is useful if you want more control over how to convert pixel_values into patch embeddings. int, optional) —
bool, optional) —
Whether or not to return the attentions tensors of all attention layers. See attentions under returned
tensors for more detail. bool, optional) —
Whether or not to return the hidden states of all layers. See hidden_states under returned tensors for
more detail. bool, optional) —
Whether or not to return a ModelOutput instead of a plain tuple. torch.LongTensor of shape (batch_size, 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] Returns
transformers.modeling_outputs.MaskedLMOutput or tuple(torch.FloatTensor)
A transformers.modeling_outputs.MaskedLMOutput or a tuple of
torch.FloatTensor (if return_dict=False is passed or when config.return_dict=False) comprising various
elements depending on the configuration (BridgeTowerConfig) and inputs.
loss (torch.FloatTensor of shape (1,), optional, returned when labels is provided) — Masked language modeling (MLM) loss.
logits (torch.FloatTensor of shape (batch_size, sequence_length, config.vocab_size)) — Prediction scores of the language modeling head (scores for each vocabulary token before SoftMax).
hidden_states (tuple(torch.FloatTensor), optional, returned when output_hidden_states=True is passed or when config.output_hidden_states=True) — Tuple of torch.FloatTensor (one for the output of the embeddings, if the model has an embedding layer, +
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 optional initial embedding outputs.
attentions (tuple(torch.FloatTensor), optional, returned when output_attentions=True is passed or when config.output_attentions=True) — Tuple of torch.FloatTensor (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.
The BridgeTowerForMaskedLM forward method, overrides the __call__ special method.
Although the recipe for forward pass needs to be defined within this function, one should call the Module
instance afterwards instead of this since the former takes care of running the pre and post processing steps while
the latter silently ignores them.
Examples:
>>> from transformers import BridgeTowerProcessor, BridgeTowerForMaskedLM
>>> from PIL import Image
>>> import requests
>>> url = "http://images.cocodataset.org/val2017/000000360943.jpg"
>>> image = Image.open(requests.get(url, stream=True).raw).convert("RGB")
>>> text = "a <mask> looking out of the window"
>>> processor = BridgeTowerProcessor.from_pretrained("BridgeTower/bridgetower-base-itm-mlm")
>>> model = BridgeTowerForMaskedLM.from_pretrained("BridgeTower/bridgetower-base-itm-mlm")
>>> # prepare inputs
>>> encoding = processor(image, text, return_tensors="pt")
>>> # forward pass
>>> outputs = model(**encoding)
>>> results = processor.decode(outputs.logits.argmax(dim=-1).squeeze(0).tolist())
>>> print(results)
.a cat looking out of the window.( config )
Parameters
BridgeTower Model transformer with a classifier head on top (a linear layer on top of the final hidden state of the [CLS] token) for image-to-text matching.
This model is a PyTorch torch.nn.Module <https://pytorch.org/docs/stable/nn.html#torch.nn.Module>_ subclass. Use
it as a regular PyTorch Module and refer to the PyTorch documentation for all matter related to general usage and
behavior.
( input_ids: Optional = None attention_mask: Optional = None token_type_ids: Optional = None pixel_values: Optional = None pixel_mask: Optional = None head_mask: Optional = None inputs_embeds: Optional = None image_embeds: Optional = None output_attentions: Optional = None output_hidden_states: Optional = None return_dict: Optional = None labels: Optional = None ) → transformers.modeling_outputs.SequenceClassifierOutput or tuple(torch.FloatTensor)
Parameters
torch.LongTensor of shape ({0})) —
Indices of input sequence tokens in the vocabulary. Indices can be obtained using AutoTokenizer. See
PreTrainedTokenizer.encode() and PreTrainedTokenizer.call() for details. What are input
IDs? torch.FloatTensor of shape ({0}), optional) —
Mask to avoid performing attention on padding token indices. Mask values selected in [0, 1]:
torch.LongTensor of shape ({0}), optional) —
Segment token indices to indicate first and second portions of the inputs. Indices are selected in [0, 1]:
torch.FloatTensor of shape (batch_size, num_channels, height, width)) —
Pixel values. Pixel values can be obtained using BridgeTowerImageProcessor. See
BridgeTowerImageProcessor.call() for details. torch.LongTensor of shape (batch_size, height, width), optional) —
Mask to avoid performing attention on padding pixel values. Mask values selected in [0, 1]:
What are attention masks? <../glossary.html#attention-mask>__torch.FloatTensor of shape (num_heads,) or (num_layers, num_heads), optional) —
Mask to nullify selected heads of the self-attention modules. Mask values selected in [0, 1]:
torch.FloatTensor of shape ({0}, hidden_size), optional) —
Optionally, instead of passing input_ids you can choose to directly pass an embedded representation. This
is useful if you want more control over how to convert input_ids indices into associated vectors than the
model’s internal embedding lookup matrix. torch.FloatTensor of shape (batch_size, num_patches, hidden_size), optional) —
Optionally, instead of passing pixel_values, you can choose to directly pass an embedded representation.
This is useful if you want more control over how to convert pixel_values into patch embeddings. int, optional) —
bool, optional) —
Whether or not to return the attentions tensors of all attention layers. See attentions under returned
tensors for more detail. bool, optional) —
Whether or not to return the hidden states of all layers. See hidden_states under returned tensors for
more detail. bool, optional) —
Whether or not to return a ModelOutput instead of a plain tuple. torch.LongTensor of shape (batch_size, 1), optional) —
Labels for computing the image-text matching loss. 0 means the pairs don’t match and 1 means they match.
The pairs with 0 will be skipped for calculation. Returns
transformers.modeling_outputs.SequenceClassifierOutput or tuple(torch.FloatTensor)
A transformers.modeling_outputs.SequenceClassifierOutput or a tuple of
torch.FloatTensor (if return_dict=False is passed or when config.return_dict=False) comprising various
elements depending on the configuration (BridgeTowerConfig) and inputs.
loss (torch.FloatTensor of shape (1,), optional, returned when labels is provided) — Classification (or regression if config.num_labels==1) loss.
logits (torch.FloatTensor of shape (batch_size, config.num_labels)) — Classification (or regression if config.num_labels==1) scores (before SoftMax).
hidden_states (tuple(torch.FloatTensor), optional, returned when output_hidden_states=True is passed or when config.output_hidden_states=True) — Tuple of torch.FloatTensor (one for the output of the embeddings, if the model has an embedding layer, +
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 optional initial embedding outputs.
attentions (tuple(torch.FloatTensor), optional, returned when output_attentions=True is passed or when config.output_attentions=True) — Tuple of torch.FloatTensor (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.
The BridgeTowerForImageAndTextRetrieval forward method, overrides the __call__ special method.
Although the recipe for forward pass needs to be defined within this function, one should call the Module
instance afterwards instead of this since the former takes care of running the pre and post processing steps while
the latter silently ignores them.
Examples:
>>> from transformers import BridgeTowerProcessor, BridgeTowerForImageAndTextRetrieval
>>> import requests
>>> from PIL import Image
>>> url = "http://images.cocodataset.org/val2017/000000039769.jpg"
>>> image = Image.open(requests.get(url, stream=True).raw)
>>> texts = ["An image of two cats chilling on a couch", "A football player scoring a goal"]
>>> processor = BridgeTowerProcessor.from_pretrained("BridgeTower/bridgetower-base-itm-mlm")
>>> model = BridgeTowerForImageAndTextRetrieval.from_pretrained("BridgeTower/bridgetower-base-itm-mlm")
>>> # forward pass
>>> scores = dict()
>>> for text in texts:
... # prepare inputs
... encoding = processor(image, text, return_tensors="pt")
... outputs = model(**encoding)
... scores[text] = outputs.logits[0, 1].item()