Converts an amplitude spectrogram to the decibel scale. This computes `20 * log10(spectrogram / reference)`, using basic logarithm properties

Returns a copy of a function f.

Save the modeling files corresponding to a custom model/configuration/tokenizer etc. in a given folder. Optionally adds the proper fields in

Filter a list of files to only keep the ones corresponding to a list of frameworks. Args: files (`List[Union[str, os.PathLike]]`): T

Creates a random float32 tensor

The LM Head layer. This method must be overwritten by all the models that have a lm head. Return: `tf.keras.layers.Layer

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

Function extracts relevant parameters into flatten dict from model params, appends batch normalization statistics if present

Creates a frequency bin conversion matrix used to obtain a mel spectrogram. This is called a *mel filter bank*, and various implementation ex

Converts a power spectrogram to the decibel scale. This computes `10 * log10(spectrogram / reference)`, using basic logarithm properties for

Resets the accumulated gradients on the current replica.

Runs `command` with `subprocess.check_output` and will potentially return the `stdout`. Will also properly capture if an error occured while

Stable wrapper that returns the same output as `tf.nn.softmax`, but that works reliably with XLA on CPU. It is meant as a workaround for the

Add a callback to the current list of [`~transformers.TrainerCallback`]. Args: callback (`type` or [`~transformers.Traine

Small helper to check pseudo-equality of parsed arguments on `ArgumentParser` instances.

Build model inputs from a sequence or a pair of sequence for sequence classification tasks by concatenating and adding special tokens

Get the imports a given module makes. Args: module_fname (`str`): The name of the file of the module where we want to lo

Return the properly gathered arrays and truncate to the number of samples (since the sampler added some extras to get each process a

Set requires_grad=False for each of model.parameters()

Get the `out_features` and `out_indices` so that they are aligned. The logic is as follows: - `out_features` not set, `out_indices`

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

Returns the model's input embeddings layer. Returns: `tf.Variable`: The embeddings layer mapping vocabulary to hidden st

Refines the result of `extract_imports` to remove subfolders and get a proper list of module filenames: if a file as an import `from utils im

Returns the optimizer class and optimizer parameters based on the training arguments. Args: args (`transformers.training

Checks if the dataset implements __len__() and it doesn't raise an error

Convert frequency from mels to hertz. Args: mels (`float` or `np.ndarray`): The frequency, or multiple frequencies, in m

Numpify `tensors` (even if it's a nested list/tuple/dict of tensors).

Normalizes `image` using the mean and standard deviation specified by `mean` and `std`. image = (image - mean) / std Args: imag

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.

Resizes `image` to `(height, width)` specified by `size` using the PIL library. Args: image (`np.ndarray`): The image to

Serializes this instance to a Python dictionary. Returns: `Dict[str, Any]`: Dictionary of all the attributes that make u

Serializes this instance to a JSON string.

Get the index of the word corresponding (i.e. comprising) to an encoded token in a sequence of the batch. Can be called as:

Gather value of `tensors` (tensor or list/tuple of nested tensors) and convert them to numpy before concatenating them to `gathered`

Prepare `inputs` before feeding them to the model, converting them to tensors if they are not already and handling potential state.

Add a key to _index (MinHashLSH) the min_hash is used to query closest matches based on the jaccard_threshold. The new key is either a

Computes the transition scores of sequences given the generation scores (and beam indices, if beam search was used). This is a conven

Create and fill numpy array of size len_of_validation_data * max_length_of_output_tensor Args: start_or_end_logits(:obj:

Downloads a given url in a temporary file. This function is not safe to use in multiple processes. Its only use is for deprecated behavior al

Run evaluation and returns metrics. The calling script will be responsible for providing a method to compute metrics, as they are ta

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

Set requires_grad=False for each of model.parameters()

Return the number of available gpus (regardless of whether torch, tf or jax is used)

Activates gradient checkpointing for the current model. Note that in other frameworks this feature can be referred to as "activation

list(map(f, x))

location will always be cpu for np tensors

location will always be cpu for np tensors

Replace all patterns present in a given text. Args: text (`str`): The text to treat. old_model_patterns (`ModelPatterns`): T

r""" Save a generation configuration object to the directory `save_directory`, so that it can be re-loaded using the [`~GenerationConf

Wrapper function for tokenizing query and supports Args: x (`List[str] or List[List[str]]`): List of stri

Updates attributes of this class with attributes from `update_str`. The expected format is ints, floats and strings as is, and for b

Conv1D layer as defined by Radford et al. for OpenAI GPT (and also used in GPT-2) Basically works like a Linear layer but the weights are