Hello I am looking for a tool that can expand all the games in a lichess database to each and every FEN.
That would work exactly like this python tool https://pypi.org/project/pgn2data/
But it strips the pgn of the %eval tags. Which are cloud computed evaluations for any position. (could be up to multiple SF14 cpu hours per position)
So does anyone have a tool that converts a pgn like this into a list which contains FEN + Evaluation?
dangi12012 wrote: ↑Mon Nov 01, 2021 2:53 pm
Hello I am looking for a tool that can expand all the games in a lichess database to each and every FEN.
That would work exactly like this python tool https://pypi.org/project/pgn2data/
But it strips the pgn of the %eval tags. Which are cloud computed evaluations for any position. (could be up to multiple SF14 cpu hours per position)
So does anyone have a tool that converts a pgn like this into a list which contains FEN + Evaluation?
...
A FEN cannot contain that information (eval) by definition.
What you look for is PGN => EPD (which can have opcodes like e.g. eval) http://www.saremba.de/chessgml/standard ... tm#c16.2.5
Ad hoc I don't remember any tool that does this.
That is perfect my friend. To get literally millions of cpu hours of free evaluations do these steps:
1) Download - https://database.lichess.org/standard/l ... 08.pgn.bz2
2) Extract with multiple Threads - "pbzip2 -v -d -k -m1000 lichess_db_standard_rated_2021-08.pgn"
3) Convert to Position with Eval - "cat lichess_db_standard_rated_2021-08.pgn | pgn-extract --fencomments --commentlines | awk '/eval/ {getline f; print f $0}' > eval_2021-08.txt"