RACER IS OP — Odia Web Corpus
Collection
Odia NLP datasets curated by RACER IS OP — web corpora for low-resource Odia language modeling • 6 items • Updated
The dataset viewer is not available for this split.
Job manager crashed while running this job (missing heartbeats).
Error code: JobManagerCrashedError
Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
Odia Web Corpus v2
Expanded and refined version of the Odia web corpus, converted to Parquet format with dedicated train/test/validation splits for reproducible NLP experimentation.
Dataset Details
- Language: Odia (ISO 639-3:
or) - Format: Parquet (columnar, compressed)
- Splits: train (900K), test (50K), validation (50K)
- License: CC-BY-4.0
Data Fields
| Field | Type | Description |
|---|---|---|
text |
string | Cleaned document text |
Usage
from datasets import load_dataset
ds = load_dataset("saidutta69/Odia-Web-Corpus-v2")
print(ds["train"][0]["text"][:200])
Citation
@misc{odia-web-corpus-v2,
author = {Sai Dutta},
title = {Odia Web Corpus v2},
year = {2024},
publisher = {Hugging Face},
url = {https://huggingface.co/datasets/saidutta69/Odia-Web-Corpus-v2}
}
- Downloads last month
- 75