Create and initialize the parameters shared by all compacter modules

This function creates the shared random matrices A and B that are used across all Vera layers. These matrices are frozen and initialized acco

Copied from rabeehk/compacter seq2seq/hypercomplex/kronecker.py Kronecker product of matrices a and b with leading batch dimensions. Batch d

Loads a pre-trained pytorch adapter module from the local file system or a remote location. Args: adapter_name_or_path (

Loads a AdapterFusion module from the given directory. Args: save_directory (str): The directory from where to load the

Loads a prediction head module from the given directory. Args: save_directory (str): The directory from where to load th

Loads an adapter setup from the local file system or a remote location. Args: adapter_setup_name_or_path (str): can be either:

Averages the given states along the batch size dimension by the given weights. This is e.g. used by the Average composition block. IMPORTANT:

Parses an adapter configuration string into a list of tuples. Each tuple constists of an adapter config identifier and dictionary.

Performs computations after the forward pass of the adapter block(s). This e.g. includes applying the residual connection and layer n

Post-processes the predictions of a question-answering model to convert them to answers that are substrings of the original contexts. This is

Redirects loading from the archived Hub repository to HuggingFace Model Hub. Args: specifier (str): A string specifying the adapter

The callable returned by this method is used to optionally rename the module weights after loading. Args: old_name (str)

Repeats the given state along the batch size dimension for the given number of times. This is e.g. used by the Parallel composition block. IMP

Resets weights of a LoRA module merged using `model.merge_adapter(name)`.

Resizes position embeddings of the model if :obj:`new_num_position_embeddings != config.max_position_embeddings`. Arguments:

Restores the model output from a JSON Lines (.jsonl) file as a list of lists for the specified model. Args: config (str): Name of the adap

Saves an adapter and its configuration file to a directory, so that it can be reloaded using the `load()` method. Args:

Saves a AdapterFusion module into the given directory. Args: save_directory (str): The directory to save the weights in.

Saves all AdapterFusion layers of this model together with their configuration to subfolders of the given location. Args:

Saves all prediction heads of this model to subfolders of the given location. Args: save_directory (str): Path to the base direct

Save model output to a JSON Lines (.jsonl) file as a dictionary. Each line represents one model, where the key is the model name (specified by adapter

Tie the weights between the input embeddings and the output embeddings. If the :obj:`torchscript` flag is set in the configuration,

Converts the config class to a Python dict.

Generate a local filename from a url. Convert `url` into a hashed filename in a reproducible way. If `etag` is specified, append its hash to

Performs simple validations of a list of configurations to check whether they can be combined to a common setup. Args:

Makes required changes to a model config class to allow usage with adapters. Args: config (PretrainedConfig): The config to be wrapp