Construct a layernorm module in the T5 style. No bias and no subtraction of mean.

Creates a random float32 tensor

Dict of bias attached to an LM head. The key represents the name of the bias attribute. Return: `tf.Variable`: The weigh

Loads `image` to a PIL Image. Args: image (`str` or `PIL.Image.Image`): The image to convert to the PIL Image format.

Ensure that all layers have a channel count that is divisible by `divisor`. This function is taken from the original TensorFlow repo. It can

Normalizes `image` with `mean` and `std`. Note that this will trigger a conversion of `image` to a NumPy array if it's a PIL Image.

Pads a batch of images to the bottom and right of the image with zeros to the size of largest height and width in the batch and optio

Pads a batch of images to the bottom and right of the image with zeros to the size of largest height and width in the batch and optio

Parses the commit message to detect if a command is there to skip, force all or part of the CI. Args: commit_message (`str`): The co

Framework-agnostic version of `numpy.transpose` that will work on torch/TensorFlow/Jax tensors as well as NumPy arrays.

Verify that the config arguments to be passed to load_backbone are valid

Utility method to check if the model has correctly adapters injected on it.

Computes random mask spans for a given shape. Used to implement [SpecAugment: A Simple Data Augmentation Method for ASR](https://arxiv.org/ab

splits sequence length dim of vectors into `dim_factor_1` and `dim_factor_2` dims

Add `arrays` to the internal storage, Will initialize the storage to the full size at the first arrays passed so that if we're bound

Convert the inner content to tensors. Args: tensor_type (`str` or [`~utils.TensorType`], *optional*): Th

Returns a copy of a function f.

Creates an optimizer with a learning rate schedule using a warmup phase followed by a linear decay. Args: init_lr (`float`):

r""" Example: ```python >>> from transformers import EncoderDecoderModel >>> model = EncoderDecoderModel.from_pretra

r""" Converts a sequence of input tokens into a sequence of mel spectrograms, which are subsequently turned into a speech waveform usi

Returns a dictionary of layers distributed evenly across all devices.

Tries to locate a file in a local folder and repo, downloads and cache it if necessary. Args: path_or_repo (`str` or `os.PathLike`):

Create a schedule with a learning rate that decreases linearly from the initial lr set in the optimizer to 0, after a warmup period during wh

r""" Get the memory footprint of a model. This will return the memory footprint of the current model in bytes. Useful to benchmark the

Retrieve sequence ids from a token list that has no special tokens added. This method is called when adding special tokens using the

CrossEntropyLoss that ignores pad tokens