braindecode.training.iter_trial_prediction_chunks#
- braindecode.training.iter_trial_prediction_chunks(module, dataset, *, windows_per_chunk, return_targets=True, batch_size=1, num_workers=0, loader_kwargs=None)[source]#
Yield predictions in chunks of at most
windows_per_chunkwindows.Takes the arguments of
iter_trial_predictions()and has the same lifecycle. A chunk never spans two trials and overlap removal carries across chunks, so concatenating a trial’s chunks along axis 1 gives itsiter_trial_predictions()output exactly, for the predictions and for time targets. Memory holds one chunk, the loader batches and one window of lookahead, so it does not depend on the trial length.- Parameters:
windows_per_chunk (int) – Maximum number of source windows per chunk.
module – As in
iter_trial_predictions().dataset – As in
iter_trial_predictions().return_targets – As in
iter_trial_predictions().batch_size – As in
iter_trial_predictions().num_workers – As in
iter_trial_predictions().loader_kwargs – As in
iter_trial_predictions().
- Yields:
chunk (TrialPredictionChunk)
Examples
Read the first chunk of a two-window trial and stop.
contextlib.closing()closes the iterator at the end of the block, which gives the model back its training mode.>>> import torch >>> from contextlib import closing >>> from braindecode.training import iter_trial_prediction_chunks >>> model = torch.nn.Conv1d(1, 1, kernel_size=1, bias=False) >>> _ = model.train() >>> windows = [(torch.ones(1, 4), 7, (i, 4 * i, 4 * (i + 1))) for i in range(2)] >>> with closing( ... iter_trial_prediction_chunks( ... model, ... windows, ... windows_per_chunk=1, ... return_targets=False, ... ) ... ) as chunks: ... chunk = next(chunks) ... print( ... chunk.trial_index, ... chunk.prediction_start, ... chunk.prediction.shape, ... chunk.target, ... model.training, ... ) ... 0 0 (1, 4) None False >>> model.training True