This folder contains how I turned the data produced by the sampler into .tfrecord format. The script you probably want is to_tfrecord_v2.py. I ran it like this:
num_folds_train=256
num_folds_val=128
mkdir -p logs
parallel -j 4 --will-cite "python to_tfrecord_v2.py -no_1ary -no_2ary -fold={1} -num_folds=${num_folds_val} -fns_list val_fns.txt -out_path gs://YOURPATHHERE/val > logs/val{1}.txt" ::: $(seq 0 $((num_folds_val-1)))
parallel -j 4 --will-cite "python to_tfrecord_v2.py -no_1ary -no_2ary -fold={1} -num_folds=${num_folds_train} -fns_list train_fns.txt -out_path gs://YOURPATHHERE/train > logs/train{1}.txt" ::: $(seq 0 $((num_folds_train-1)))