A straightforward helper that simply averages multiple crops if they are present. Parameters ---------- y_p: np.ndarray The

This is used to provide a custom implementation of taking a filename, and returning a :any:`mne.io.Raw()` instance. If properly const

Provide callable code to create a raw instance from sessions with certain file extensions. This is useful for handling of custom file

Add callbacks to be invoked on successful loading of session and/or thinker. Optionally, these can modify the respective loaded insta

Generates a mapping of sessions and people of the dataset, assuming files are stored in the structure: `toplevel`/(*optional - <versi

This method is used to add trainable modules to the process. Rather than placing objects for training in the __init__ method, they sh

Dumps the dataset to the directory specified by toplevel, with a single file per index. Parameters ---------- toplev

sklearn/keras-like convenience method to simply proceed with training across multiple epochs of the provided dataset Paramet

Create a classifier from a dataset. Parameters ---------- dataset modelargs: dict Options

Creates exclusions to be used by configuratron on next dataset loading. Parameters ---------- save_to_file : str (pa

Accumulates all the sessions from each thinker in the dataset in a nested dictionary. Returns ------- session_dict:

Load the parameters as saved by :py:meth:`save_best()`. Parameters ---------- best: Any

This allows for a distinction between the classification layer(s) and the rest of the network. Using a basic formulation of a network

Maps channels as stored by a :any:`DN3ataset` to the Deep1010 format, will automatically map EOG and extra channels by type. Parameters

Determine the outputs for all loaded data from the dataset Parameters ---------- dataset: DN3ataset, DataLoader

Scan the provided toplevel for all files that may belong to the dataset. Returns ------- files: list

Commits the dataset to numpy-formatted arrays. Useful for saving dataset to disk, or preparing for tools that expect numpy-formatted

Preprocess a particular recording. This is allowed to modify aspects of the recording in-place, but is not strictly advised.

Modifies a batch of tensors. Parameters ---------- x : torch.Tensor, tuple A batch of trial instance ten

Modifies a batch of tensors. Parameters ---------- x : torch.Tensor, tuple The trial tensor, not includin

Parses DN3 configuration files. Checking the DN3 token for listed datasets. Parameters ---------- config_filename :

This provides a workaround for the normal raw recording pipeline so that files are not loaded in any way until they are needed. MNE's

This integrates the loading of multiple datasets, and facilitates over-sampling smaller ones to prevent larger datasets from crowding

With larger datasets, it may be prudent to triage the data based on global statistics. Here spans of time are rejected based on a pro

Base class for that specifies the interface for DN3 datasets.

Interface for bridging mne Raw instances as PyTorch compatible "Dataset". Parameters ---------- raw : mne.io.Raw

Wraps :any:`mne.Epochs` instances so that they conform to the :any:`Recording` API. Parameters ---------- epochs

Collects recordings from multiple people, intended to be of the same task, at different times or conditions. Optionally, can

Uniformly crops the time-dimensions of a batch. Parameters ---------- max_crop_frac: float Th

Crops the time dimension of an entire batch. Parameters ---------- end_crop_frac: float If th

Trial transforms are, for the most part, simply operations that are performed on the loaded tensors when they are fetched via the :me