- Datasheet: unit note with the full length-gate measurement (0/45,043 docs
reach 200 chars; no concatenation) + accept/withdraw decision options; license
at the origin (MDC dataset record + per-row license); MDC Terms clause (viii)
re-identification prohibition carried downstream; scrubber named (pattern
rejection filters, not NERGAL) with a full-shard residual PII scan (clean).
- artifacts/source_findings.md: common_voice_pl finding appended.
Add CC0 Polish Common Voice validated text (v26.0)
This contribution adds a text-only projection of Polish sentences associated
with validated recordings in Mozilla Common Voice Scripted Speech 26.0. It does
not include audio, contributor identifiers, demographic fields or file paths.
- Source rows inspected: `138,798` validated clip records
- Unique normalized sentences retained: `45,043`
- Characters: `2,637,100`
- Whitespace-delimited words: `379,819`
- Tokens: `994,922` (`cl100k_base` proxy)
- License: `CC0-1.0`
- Deterministic data sample: [common_voice_pl.sample.jsonl](https://huggingface.co/datasets/PiotrSty/common-voice-pl-text/blob/v1.0.0/data/common_voice_pl.sample.jsonl)
- Published source snapshot: [PiotrSty/common-voice-pl-text v1.0.0](https://huggingface.co/datasets/PiotrSty/common-voice-pl-text/tree/v1.0.0)
The build rejected `2,557` source rows (`2,538` too short and `19` too long)
and collapsed `91,198` repeated clip rows after within-source normalization.
- Official dataset: `Mozilla Data Collective / Common Voice …
20 days ago
Separate source PRs from releases; add render_sample.py (#42)
- Add eight CC BY-SA 4.0 Polish academic books from ROCK (3c845cd4a0708da984aa5ae0b1f86b1af5c28adf)
Co-authored-by: Piotr Styla <PiotrSty@users.noreply.huggingface.co>
