Datasets:
Dataset Viewer
The dataset viewer is not available for this subset.
The default config contains 2105 while it should generally contain 3 splits maximum (train/validation/test). If the splits fineweb_fin_199, fineweb_fin_200, fineweb_fin_201, fineweb_fin_202, fineweb_fin_203... are not used to differentiate between training and evaluation, please consider defining configs of this dataset instead. You can find how to define configs instead of splits here: https://huggingface.co/docs/hub/datasets-data-files-configuration
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.
Morfa Raw Data
Raw multilingual text corpus extracted from FineWeb-2 for language model pretraining.
Languages
| Language | Code | Splits |
|---|---|---|
| Finnish | fin |
198 |
| Azerbaijani | aze |
73 |
| Uzbek | uzb |
13 |
| Turkmen | tuk |
1 |
Schema
| Field | Type | Description |
|---|---|---|
text |
string |
Raw document text |
lang |
string |
Language code |
source |
string |
Source dataset identifier |
Usage
from datasets import load_dataset
ds = load_dataset("eokayakca/morfa-raw-data", split="fineweb_aze_001", streaming=True)
for row in ds:
print(row["text"][:100], row["lang"])
Source
All data from HuggingFaceFW/fineweb-2, redistributed under the same license.
- Downloads last month
- 644