Upload docs/QUALITY_AND_LIMITATIONS.md with huggingface_hub
Browse files
docs/QUALITY_AND_LIMITATIONS.md
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# Quality and Limitations
|
| 2 |
+
|
| 3 |
+
## Quality Notes
|
| 4 |
+
- The release is deduplicated and normalized through the preprocessing pipeline.
|
| 5 |
+
- The parquet order is arranged to make the dataset viewer open on more representative examples first.
|
| 6 |
+
|
| 7 |
+
## Known Limitations
|
| 8 |
+
- Colloquial text includes slang, transliteration, and non-standard spelling.
|
| 9 |
+
- The corpus is not a benchmark suite with fixed splits.
|
| 10 |
+
- Metadata is descriptive and should be validated per task.
|
| 11 |
+
|
| 12 |
+
## Intended Use
|
| 13 |
+
Full combined Nepali text corpus covering colloquial, formal, and encyclopedia-style text.
|