- ADD indexing (
index
) with token substitutions (Levenshtein over tokens) for alignment
- FIX removed chunks warnings: improve informativeness (
UserWarning: removed chunks: [('y', 2, 3)]
) - ADD tokens to strip from a chunk span for
rebase_chunks
- ADD baselines: random, chance, majority (HMM?)
- ADD partial span match evaluation (extend
compute_spans_stats
)