Big Transfer (BiT)
Overview
The BiT model was proposed in Big Transfer (BiT): General Visual Representation Learning by Alexander Kolesnikov, Lucas Beyer, Xiaohua Zhai, Joan Puigcerver, Jessica Yung, Sylvain Gelly, Neil Houlsby. BiT is a simple recipe for scaling up pre-training of ResNet-like architectures (specifically, ResNetv2). The method results in significant improvements for transfer learning.
The abstract from the paper is the following:
Transfer of pre-trained representations improves sample efficiency and simplifies hyperparameter tuning when training deep neural networks for vision. We revisit the paradigm of pre-training on large supervised datasets and fine-tuning the model on a target task. We scale up pre-training, and propose a simple recipe that we call Big Transfer (BiT). By combining a few carefully selected components, and transferring using a simple heuristic, we achieve strong performance on over 20 datasets. BiT performs well across a surprisingly wide range of data regimes β from 1 example per class to 1M total examples. BiT achieves 87.5% top-1 accuracy on ILSVRC-2012, 99.4% on CIFAR-10, and 76.3% on the 19 task Visual Task Adaptation Benchmark (VTAB). On small datasets, BiT attains 76.8% on ILSVRC-2012 with 10 examples per class, and 97.0% on CIFAR-10 with 10 examples per class. We conduct detailed analysis of the main components that lead to high transfer performance.
Tips:
- BiT models are equivalent to ResNetv2 in terms of architecture, except that: 1) all batch normalization layers are replaced by group normalization, 2) weight standardization is used for convolutional layers. The authors show that the combination of both is useful for training with large batch sizes, and has a significant impact on transfer learning.
This model was contributed by nielsr. The original code can be found here.
Resources
A list of official Hugging Face and community (indicated by π) resources to help you get started with BiT.
- BitForImageClassification is supported by this example script and notebook.
- See also: Image classification task guide
If youβre interested in submitting a resource to be included here, please feel free to open a Pull Request and weβll review it! The resource should ideally demonstrate something new instead of duplicating an existing resource.
BitConfig
class transformers.BitConfig
< source >( num_channels = 3 embedding_size = 64 hidden_sizes = [256, 512, 1024, 2048] depths = [3, 4, 6, 3] layer_type = 'preactivation' hidden_act = 'relu' global_padding = None num_groups = 32 drop_path_rate = 0.0 embedding_dynamic_padding = False output_stride = 32 width_factor = 1 out_features = None out_indices = None **kwargs )
Parameters
- num_channels (
int
, optional, defaults to 3) — The number of input channels. - embedding_size (
int
, optional, defaults to 64) — Dimensionality (hidden size) for the embedding layer. - hidden_sizes (
List[int]
, optional, defaults to[256, 512, 1024, 2048]
) — Dimensionality (hidden size) at each stage. - depths (
List[int]
, optional, defaults to[3, 4, 6, 3]
) — Depth (number of layers) for each stage. - layer_type (
str
, optional, defaults to"preactivation"
) — The layer to use, it can be either"preactivation"
or"bottleneck"
. - hidden_act (
str
, optional, defaults to"relu"
) — The non-linear activation function in each block. If string,"gelu"
,"relu"
,"selu"
and"gelu_new"
are supported. - global_padding (
str
, optional) — Padding strategy to use for the convolutional layers. Can be either"valid"
,"same"
, orNone
. - num_groups (
int
, optional, defaults to 32) — Number of groups used for theBitGroupNormActivation
layers. - drop_path_rate (
float
, optional, defaults to 0.0) — The drop path rate for the stochastic depth. - embedding_dynamic_padding (
bool
, optional, defaults toFalse
) — Whether or not to make use of dynamic padding for the embedding layer. - output_stride (
int
, optional, defaults to 32) — The output stride of the model. - width_factor (
int
, optional, defaults to 1) — The width factor for the model. - out_features (
List[str]
, optional) — If used as backbone, list of features to output. Can be any of"stem"
,"stage1"
,"stage2"
, etc. (depending on how many stages the model has). If unset andout_indices
is set, will default to the corresponding stages. If unset andout_indices
is unset, will default to the last stage. - out_indices (
List[int]
, optional) — If used as backbone, list of indices of features to output. Can be any of 0, 1, 2, etc. (depending on how many stages the model has). If unset andout_features
is set, will default to the corresponding stages. If unset andout_features
is unset, will default to the last stage.
This is the configuration class to store the configuration of a BitModel. It is used to instantiate an BiT 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 BiT google/bit-50 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 BitConfig, BitModel
>>> # Initializing a BiT bit-50 style configuration
>>> configuration = BitConfig()
>>> # Initializing a model (with random weights) from the bit-50 style configuration
>>> model = BitModel(configuration)
>>> # Accessing the model configuration
>>> configuration = model.config
BitImageProcessor
class transformers.BitImageProcessor
< source >( do_resize: bool = True size: typing.Dict[str, int] = None resample: Resampling = <Resampling.BICUBIC: 3> do_center_crop: bool = True crop_size: typing.Dict[str, int] = None do_rescale: bool = True rescale_factor: typing.Union[int, float] = 0.00392156862745098 do_normalize: bool = True image_mean: typing.Union[float, typing.List[float], NoneType] = None image_std: typing.Union[float, typing.List[float], NoneType] = None do_convert_rgb: bool = True use_square_size: bool = False **kwargs )
Parameters
- do_resize (
bool
, optional, defaults toTrue
) — Whether to resize the image’s (height, width) dimensions to the specifiedsize
. Can be overridden bydo_resize
in thepreprocess
method. - size (
Dict[str, int]
optional, defaults to{"shortest_edge" -- 224}
): Size of the image after resizing. The shortest edge of the image is resized to size[“shortest_edge”], with the longest edge resized to keep the input aspect ratio. Can be overridden bysize
in thepreprocess
method. - resample (
PILImageResampling
, optional, defaults toPILImageResampling.BICUBIC
) — Resampling filter to use if resizing the image. Can be overridden byresample
in thepreprocess
method. - do_center_crop (
bool
, optional, defaults toTrue
) — Whether to center crop the image to the specifiedcrop_size
. Can be overridden bydo_center_crop
in thepreprocess
method. - crop_size (
Dict[str, int]
optional, defaults to 224) — Size of the output image after applyingcenter_crop
. Can be overridden bycrop_size
in thepreprocess
method. - do_rescale (
bool
, optional, defaults toTrue
) — Whether to rescale the image by the specified scalerescale_factor
. Can be overridden bydo_rescale
in thepreprocess
method. - rescale_factor (
int
orfloat
, optional, defaults to1/255
) — Scale factor to use if rescaling the image. Can be overridden byrescale_factor
in thepreprocess
method. do_normalize — Whether to normalize the image. Can be overridden bydo_normalize
in thepreprocess
method. - image_mean (
float
orList[float]
, optional, defaults toOPENAI_CLIP_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 theimage_mean
parameter in thepreprocess
method. - image_std (
float
orList[float]
, optional, defaults toOPENAI_CLIP_MEAN
) — 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 theimage_std
parameter in thepreprocess
method. Can be overridden by theimage_std
parameter in thepreprocess
method. - do_convert_rgb (
bool
, optional, defaults toTrue
) — Whether to convert the image to RGB. - use_square_size (
bool
, optional, defaults toFalse
) — The value to be passed toget_size_dict
asdefault_to_square
when computing the image size. If thesize
argument inget_size_dict
is anint
, it determines whether to default to a square image or not. Note that this attribute is not used in computingcrop_size
via callingget_size_dict
.
Constructs a BiT image processor.
preprocess
< source >( images: typing.Union[ForwardRef('PIL.Image.Image'), numpy.ndarray, ForwardRef('torch.Tensor'), typing.List[ForwardRef('PIL.Image.Image')], typing.List[numpy.ndarray], typing.List[ForwardRef('torch.Tensor')]] do_resize: bool = None size: typing.Dict[str, int] = None resample: Resampling = None do_center_crop: bool = None crop_size: int = None do_rescale: bool = None rescale_factor: float = None do_normalize: bool = None image_mean: typing.Union[float, typing.List[float], NoneType] = None image_std: typing.Union[float, typing.List[float], NoneType] = None do_convert_rgb: bool = None return_tensors: typing.Union[str, transformers.utils.generic.TensorType, NoneType] = None data_format: typing.Optional[transformers.image_utils.ChannelDimension] = <ChannelDimension.FIRST: 'channels_first'> input_data_format: typing.Union[str, transformers.image_utils.ChannelDimension, NoneType] = None **kwargs )
Parameters
- images (
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, setdo_rescale=False
. - do_resize (
bool
, optional, defaults toself.do_resize
) — Whether to resize the image. - size (
Dict[str, int]
, optional, defaults toself.size
) — Size of the image after resizing. Shortest edge of the image is resized to size[“shortest_edge”], with the longest edge resized to keep the input aspect ratio. - resample (
int
, optional, defaults toself.resample
) — Resampling filter to use if resizing the image. This can be one of the enumPILImageResampling
. Only has an effect ifdo_resize
is set toTrue
. - do_center_crop (
bool
, optional, defaults toself.do_center_crop
) — Whether to center crop the image. - crop_size (
Dict[str, int]
, optional, defaults toself.crop_size
) — Size of the center crop. Only has an effect ifdo_center_crop
is set toTrue
. - do_rescale (
bool
, optional, defaults toself.do_rescale
) — Whether to rescale the image. - rescale_factor (
float
, optional, defaults toself.rescale_factor
) — Rescale factor to rescale the image by ifdo_rescale
is set toTrue
. - do_normalize (
bool
, optional, defaults toself.do_normalize
) — Whether to normalize the image. - image_mean (
float
orList[float]
, optional, defaults toself.image_mean
) — Image mean to use for normalization. Only has an effect ifdo_normalize
is set toTrue
. - image_std (
float
orList[float]
, optional, defaults toself.image_std
) — Image standard deviation to use for normalization. Only has an effect ifdo_normalize
is set toTrue
. - do_convert_rgb (
bool
, optional, defaults toself.do_convert_rgb
) — Whether to convert the image to RGB. - return_tensors (
str
orTensorType
, optional) — The type of tensors to return. Can be one of:- Unset: Return a list of
np.ndarray
. TensorType.TENSORFLOW
or'tf'
: Return a batch of typetf.Tensor
.TensorType.PYTORCH
or'pt'
: Return a batch of typetorch.Tensor
.TensorType.NUMPY
or'np'
: Return a batch of typenp.ndarray
.TensorType.JAX
or'jax'
: Return a batch of typejax.numpy.ndarray
.
- Unset: Return a list of
- data_format (
ChannelDimension
orstr
, optional, defaults toChannelDimension.FIRST
) — The channel dimension format for the output image. Can be one of:"channels_first"
orChannelDimension.FIRST
: image in (num_channels, height, width) format."channels_last"
orChannelDimension.LAST
: image in (height, width, num_channels) format.- Unset: Use the channel dimension format of the input image.
- input_data_format (
ChannelDimension
orstr
, 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"
orChannelDimension.FIRST
: image in (num_channels, height, width) format."channels_last"
orChannelDimension.LAST
: image in (height, width, num_channels) format."none"
orChannelDimension.NONE
: image in (height, width) format.
Preprocess an image or batch of images.
BitModel
class transformers.BitModel
< source >( config )
Parameters
- config (BitConfig) — Model configuration class with all the parameters of the model. Initializing with a config file does not load the weights associated with the model, only the configuration. Check out the from_pretrained() method to load the model weights.
The bare BiT model outputting raw features without any specific head on top. This model is a PyTorch 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.
forward
< source >( pixel_values: Tensor output_hidden_states: typing.Optional[bool] = None return_dict: typing.Optional[bool] = None ) β transformers.modeling_outputs.BaseModelOutputWithPoolingAndNoAttention
or tuple(torch.FloatTensor)
Parameters
- pixel_values (
torch.FloatTensor
of shape(batch_size, num_channels, height, width)
) — Pixel values. Pixel values can be obtained using AutoImageProcessor. See BitImageProcessor.call() for details. - output_hidden_states (
bool
, optional) — Whether or not to return the hidden states of all layers. Seehidden_states
under returned tensors for more detail. - return_dict (
bool
, optional) — Whether or not to return a ModelOutput instead of a plain tuple.
Returns
transformers.modeling_outputs.BaseModelOutputWithPoolingAndNoAttention
or tuple(torch.FloatTensor)
A transformers.modeling_outputs.BaseModelOutputWithPoolingAndNoAttention
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 (BitConfig) and inputs.
-
last_hidden_state (
torch.FloatTensor
of shape(batch_size, num_channels, height, width)
) β Sequence of hidden-states at the output of the last layer of the model. -
pooler_output (
torch.FloatTensor
of shape(batch_size, hidden_size)
) β Last layer hidden-state after a pooling operation on the spatial dimensions. -
hidden_states (
tuple(torch.FloatTensor)
, optional, returned whenoutput_hidden_states=True
is passed or whenconfig.output_hidden_states=True
) β Tuple oftorch.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, num_channels, height, width)
.Hidden-states of the model at the output of each layer plus the optional initial embedding outputs.
The BitModel 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.
Example:
>>> from transformers import AutoImageProcessor, BitModel
>>> import torch
>>> from datasets import load_dataset
>>> dataset = load_dataset("huggingface/cats-image")
>>> image = dataset["test"]["image"][0]
>>> image_processor = AutoImageProcessor.from_pretrained("google/bit-50")
>>> model = BitModel.from_pretrained("google/bit-50")
>>> inputs = image_processor(image, return_tensors="pt")
>>> with torch.no_grad():
... outputs = model(**inputs)
>>> last_hidden_states = outputs.last_hidden_state
>>> list(last_hidden_states.shape)
[1, 2048, 7, 7]
BitForImageClassification
class transformers.BitForImageClassification
< source >( config )
Parameters
- config (BitConfig) — Model configuration class with all the parameters of the model. Initializing with a config file does not load the weights associated with the model, only the configuration. Check out the from_pretrained() method to load the model weights.
BiT Model with an image classification head on top (a linear layer on top of the pooled features), e.g. for ImageNet.
This model is a PyTorch 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.
forward
< source >( pixel_values: typing.Optional[torch.FloatTensor] = None labels: typing.Optional[torch.LongTensor] = None output_hidden_states: typing.Optional[bool] = None return_dict: typing.Optional[bool] = None ) β transformers.modeling_outputs.ImageClassifierOutputWithNoAttention or tuple(torch.FloatTensor)
Parameters
- pixel_values (
torch.FloatTensor
of shape(batch_size, num_channels, height, width)
) — Pixel values. Pixel values can be obtained using AutoImageProcessor. See BitImageProcessor.call() for details. - output_hidden_states (
bool
, optional) — Whether or not to return the hidden states of all layers. Seehidden_states
under returned tensors for more detail. - return_dict (
bool
, optional) — Whether or not to return a ModelOutput instead of a plain tuple. - labels (
torch.LongTensor
of shape(batch_size,)
, optional) — Labels for computing the image classification/regression loss. Indices should be in[0, ..., config.num_labels - 1]
. Ifconfig.num_labels > 1
a classification loss is computed (Cross-Entropy).
Returns
transformers.modeling_outputs.ImageClassifierOutputWithNoAttention or tuple(torch.FloatTensor)
A transformers.modeling_outputs.ImageClassifierOutputWithNoAttention 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 (BitConfig) and inputs.
- loss (
torch.FloatTensor
of shape(1,)
, optional, returned whenlabels
is provided) β Classification (or regression if config.num_labels==1) loss. - logits (
torch.FloatTensor
of shape(batch_size, config.num_labels)
) β Classification (or regression if config.num_labels==1) scores (before SoftMax). - hidden_states (
tuple(torch.FloatTensor)
, optional, returned whenoutput_hidden_states=True
is passed or whenconfig.output_hidden_states=True
) β Tuple oftorch.FloatTensor
(one for the output of the embeddings, if the model has an embedding layer, + one for the output of each stage) of shape(batch_size, num_channels, height, width)
. Hidden-states (also called feature maps) of the model at the output of each stage.
The BitForImageClassification 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.
Example:
>>> from transformers import AutoImageProcessor, BitForImageClassification
>>> import torch
>>> from datasets import load_dataset
>>> dataset = load_dataset("huggingface/cats-image")
>>> image = dataset["test"]["image"][0]
>>> image_processor = AutoImageProcessor.from_pretrained("google/bit-50")
>>> model = BitForImageClassification.from_pretrained("google/bit-50")
>>> inputs = image_processor(image, return_tensors="pt")
>>> with torch.no_grad():
... logits = model(**inputs).logits
>>> # model predicts one of the 1000 ImageNet classes
>>> predicted_label = logits.argmax(-1).item()
>>> print(model.config.id2label[predicted_label])
tiger cat