Support exact 50-line metre task

Add exact 50-sentence benchmark task

Document retained sentences by author

Validate documented metre author shares

Preserve nested metre lines and Hypotactic caesurae

Validate documented metre author counts

Parse Hypotactic verse as sentences before line cropping

Add Stoicheia parsing for complete Hypotactic coverage

Rebuild Sphragis from the complete treebank sentence pool

Synchronize dataset card counts with chunked training

Chunk training splits at fixed task sizes

Polish benchmark overview

Enforce exact verse boundaries and nested CoNLL-U

Represent text chunks as JSON lists

Add SQLite inspection mirror

Remove identifying metadata from metrical lines

Add treebank project bibliography

Fix metrical chunk aggregation

Unify source text across task sizes

manual readme update

Remove CoNLL-U provenance leakage

Validate all CoNLL-U with official loader

Freeze exact evaluation chunks

Add chunk-size dataset variants

Test aggregate genre split ratios

Rebuild work-stratified splits with seed 776

Restore core Hesiodic works

Curate conservative authorship benchmark

Align verse across sentence boundaries

Add Sphregis 0.1 beta dataset

initial commit