Update link to run11-best

Rewrite card for v1-v6 and consistent data/ layout

Consistent layout: data/stratified_v1..v6

Update links for SinhaSpeech org

Add stratified_v6 train split (canonicalized spelling/ZWJ/register/typo fixes)

Add stratified_v6 validation split (canonicalized spelling/ZWJ/register/typo fixes)

Add stratified_v6 test split (canonicalized spelling/ZWJ/register/typo fixes)

Add stratified_v5: register-spelling and word-boundary/ZWJ conjunct normalization applied on top of stratified_v4 (407 register pairs, 60 boundary pairs; 13,120 of 154,828 rows changed). Text-only changes, audio bytes unchanged from v4.

Patch 1 row: rule that failed to apply in first batch pass

Patch 8 rows: rules that failed to apply in first batch pass

Update docs: stratified_v4 moved to data/stratified_v4/

Move stratified_v4 under data/

Add stratified_v4 (spacing-normalized) split docs

Upload stratified_v4/train.parquet with huggingface_hub

Upload stratified_v4/validation.parquet with huggingface_hub

Upload stratified_v4/test.parquet with huggingface_hub

Cross-link model repo

Add dataset card

Add stratified_v2 split

Add stratified split

Add stratified_v3 split

initial commit