CharDiff-grc-doc_clean / processing_char_bert.py

Commit History

restore_elastic: make dia/boundary masking opt-in (mask_dia_boundary=False by default) instead of always discarding given accents/boundaries
24f5059
verified

Ericu950 commited on

Add restore_elastic (uncertain-lacuna-width restoration); fix decode_restoration for joint dia+boundary masking
ce67eac
verified

Ericu950 commited on

Fix: decode now reads/reinserts the punct plane (comma/middle-dot/colon/period/semicolon were silently dropped on decode)
ac228d5
verified

Ericu950 commited on

Upload processing_char_bert.py with huggingface_hub
d287da5
verified

Ericu950 commited on