The dataset doc lists that there are 18900 target and 18900 non-target labels on the epoch-level per class. However, in fact, there are 94500 target and 94500 non-target labels (31.5 (trial duration) 60 (presentation rate) 100 (trials) / 2 (target and non-target)).
Additionally, the helper function _add_stim_channel_trial contained an unused parameter ch_name, which should be used on line 171 instead of a hardcoded "stim_trial". This is similar to the correct implementation of _add_stim_channel_epoch.
The dataset doc lists that there are 18900 target and 18900 non-target labels on the epoch-level per class. However, in fact, there are 94500 target and 94500 non-target labels (31.5 (trial duration) 60 (presentation rate) 100 (trials) / 2 (target and non-target)).
Additionally, the helper function
_add_stim_channel_trialcontained an unused parameterch_name, which should be used on line 171 instead of a hardcoded "stim_trial". This is similar to the correct implementation of _add_stim_channel_epoch.Both mistakes are fixed in an upcoming PR.