Open zkokaja opened 1 year ago
Does this suffice: https://github.com/hassonlab/247-encoding/blob/e0b7468824bc950f15dd8e47f9b7c4bdb3615109/scripts/tfsenc_read_datum.py#L117
or we need one for the onset/offset as well?
no harm in dropping them as well, just in case.
similar to how we do for embeddings
df["is_nan"] = df["embeddings"].apply(lambda x: np.isnan(x).all())