# Raw Unihan downloads (regenerated via tools/fetch_unihan.sh)
unihan/
# Raw jieba dict downloads (regenerated via tools/fetch_jieba_dict.sh)
jieba/
# Heteronym candidates (analysis artifact, regenerated via tools/find-heteronym-candidates)
heteronyms_candidates.tsv
# Raw corpus dumps (regenerated via tools/pinyin-fetch-corpus)
corpus/cache/
# Generated weights from corpora (committed snapshot — see ROADMAP item 22)
# (no rule yet — weights/ is committed when it appears)
