Skip to content

Instantly share code, notes, and snippets.

@autocorr
Created August 24, 2023 14:49
Show Gist options
  • Save autocorr/896579248e73e70690fb3e054499e014 to your computer and use it in GitHub Desktop.
Save autocorr/896579248e73e70690fb3e054499e014 to your computer and use it in GitHub Desktop.
Example script for running the `variant-nnue-tools` Fairy-Stockfish data generation
#!/usr/bin/env bash
VARIANT=khans
FSF=./fstockfish_bmi
INI=../variants_pychess.ini
NNUE=../nnue/chennis_old.nnue
OUTNAME="$VARIANT"_data_v1.bin
THREADS=20
DEPTH=6
COUNT="$(echo '1 000 000 000' | tr -d ' ')"
RMPVDIFF=200
## For running with an existing neural network:
# setoption name Use NNUE value pure
# setoption name EvalFile value $NNUE
## For running without an existing neural network:
# setoption name Use NNUE value false
## To use an opening book, add "book $BOOK" to the list of arguments.
echo ":: Generating training data for depth=$DEPTH, count=$COUNT"
echo "
uci
setoption name Use NNUE value pure
setoption name EvalFile value $NNUE
setoption name Threads value $THREADS
setoption name Hash value 2048
setoption name VariantPath value $INI
setoption nam UCI_Variant value $VARIANT
isready
generate_training_data depth $DEPTH count $COUNT random_multi_pv 4 random_multi_pv_diff $RMPVDIFF random_move_count 8 random_move_max_ply 20 write_min_ply 5 eval_limit 10000 set_recommended_uci_options data_format bin output_file_name $OUTNAME
quit
" | nice -20 $FSF
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment