Currently the output looks as follows:

This is not super helpful and does not necessarily lead to good information gained.
Good would be an output along the lines:

"Gold Standard Word"
"1st xxx certainty 20%"
"2nd yyy certainty 10%"
"3rd zzz certainty 8%"
...

Maybe this has been picked up by other people or the original authors. This could very well be in the code2seq repository.