# This example collects its training record into `data/` and writes what it generates
# into `output/`. Both are re-creatable by re-running the script, so neither is committed.
# The repository-wide .gitignore already excludes them at any depth; this file says so
# here, where someone reading the folder will see it.
data/
output/

# Raw per-realisation dumps and figures are regenerable from the scripts, and large.
# FINDINGS.md and results/summary_*.csv are committed: they are the evidence FIX-1 cites.
results/raw/
results/*.log
figures/

# The repo-root .gitignore excludes *.csv wholesale, which would silently swallow the
# summary tables this folder exists to produce. Re-include them explicitly.
!results/summary_*.csv
