I read this interesting post and conducted a test myself with NL and EN.
I don't think that this should be run via CafeTran Espresso's ChatGPT feature, but it can be useful to know this technique nevertheless ...
[Top-p=0.3]
[Frequency penalty=0]
Then give it instructions like
Align the following Dutch and English text sand put the alignment into a two-column table. You can merge two sentences into a single cell if necessary, but do not change the source or target segments in any way whatsoever.
If you find a segment in the source without a corresponding target segment, or vice versa, insert that segment into a cell in the correct column, but leave the other corresponding cell empty. Do not merge any such segments, nor try to translate them.
alwayslockyourbike
I read this interesting post and conducted a test myself with NL and EN.
I don't think that this should be run via CafeTran Espresso's ChatGPT feature, but it can be useful to know this technique nevertheless ...