Quality and Limitations
## Quality Notes
- The release is deduplicated and normalized through the preprocessing pipeline.
- The parquet order is arranged to make the dataset viewer open on more representative examples first.
## Known Limitations
- Colloquial text includes slang, transliteration, and non-standard spelling.
- The corpus is not a benchmark suite with fixed splits.
- Metadata is descriptive and should be validated per task.
## Intended Use
Colloquial and code-mixed Nepali text collected from YouTube comments.