IBLBrainWideBenchTS2#

class ts2.IBLBrainWideBenchTS2(root, split, recording_id, dirname='ibl_brain_wide_bench_2026', transform=None, task=None, mask_token=nan, mask_input=True, context_length=1.0)[source]#

Bases: core.dataset.VariableContextMixin, core.dataset.IBLBrainWideBench2026

Dataset for the IBL BrainWideBench benchmark.

Parameters:

  • root (str) – The root directory of the dataset.

  • split (Literal['train', 'val', 'test']) – The split of the dataset (train, val, test).

  • dirname (str) – The name of the dataset (and the directory containing its data).

  • recording_ids – The recording ids to include in the dataset. If None, all recordings in the dataset related to the regime are included.

  • transform (Optional[Any]) – The transform(s) to apply to the data.

  • task (Optional[Literal['co_smoothing', 'forecasting']]) – The task to include in the dataset.

  • mask_token (float) – Placeholder value for masked entries, for models that need one.

  • mask_input (bool) – Whether the held-out spikes are removed from the model input. Only honored on val: False leaks the held-out activity into the input, for models that cannot take a corrupted one (AEs).

  • context_length (float) – Total sampled window size in seconds (preceding context plus the fixed 1s target). Defaults to 1.0 (no extra context).

dataset_transform(data)[source]#

Defines dataset-level transformations that are applied to all recordings in the Dataset.

This method can be applied on an entire recording or a single slice.

Parameters:

data (Data) – The Data object to apply the transformations to.

Return type:

Data

Returns:

The Data object with the transformations applied.

Example

>>> dataset = IBLBrainWideBench2026(...)
>>> data = dataset.get_recording(...)
>>> data = dataset.dataset_transform(data)
>>> slice = data.slice(0.5, 1.5)
>>> slice = dataset.dataset_transform(slice)
get_sampling_intervals()[source]#

The intervals a sampler may draw windows from: this split’s task-aligned trials.