Hugging Face's logo

Datasets:
Zyroxx66
/
somali-pretraining-corpus

Tasks:
Text Generation
Fill-Mask
Modalities:
Text
Formats:
parquet
Languages:
Somali
Size:
10K - 100K
Tags:
somali
af-soomaali
cpt
pretraining
Synthetic
nlp
Libraries:
Datasets
pandas
Polars
License:
Dataset card Data Studio Files
xet
 Community
Dataset Viewer
Auto-converted to Parquet Duplicate
The dataset viewer is not available for this split.
Cannot load the dataset split (in streaming mode) to extract the first rows.
Error code:   StreamingRowsError
Exception:    CastError
Message:      Couldn't cast
id: int64
category: large_string
subcategory: large_string
title: large_string
text: large_string
keywords: large_string
style: large_string
difficulty: large_string
quality_score: double
-- schema metadata --
pandas: '{"index_columns": [], "column_indexes": [], "columns": [{"name":' + 1095
to
{'category': Value('string'), 'subcategory': Value('string'), 'title': Value('string'), 'text': Value('string'), 'keywords': Value('string'), 'style': Value('string'), 'difficulty': Value('string')}
because column names don't match
Traceback:    Traceback (most recent call last):
                File "/src/services/worker/src/worker/utils.py", line 149, in get_rows_or_raise
                  return get_rows(
                      dataset=dataset,
                  ...<4 lines>...
                      column_names=column_names,
                  )
                File "/src/libs/libcommon/src/libcommon/utils.py", line 272, in decorator
                  return func(*args, **kwargs)
                File "/src/services/worker/src/worker/utils.py", line 129, in get_rows
                  rows_plus_one = list(itertools.islice(safe_iter(ds, dataset=dataset), rows_max_number + 1))
                File "/src/services/worker/src/worker/utils.py", line 489, in safe_iter
                  yield from ds.decode(False) if ds.features else ds
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2818, in __iter__
                  for key, example in ex_iterable:
                                      ^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2355, in __iter__
                  for key, pa_table in self._iter_arrow():
                                       ~~~~~~~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2380, in _iter_arrow
                  for key, pa_table in self.ex_iterable._iter_arrow():
                                       ~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 536, in _iter_arrow
                  for key, pa_table in iterator:
                                       ^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 419, in _iter_arrow
                  for key, pa_table in self.generate_tables_fn(**gen_kwags):
                                       ~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/parquet/parquet.py", line 220, in _generate_tables
                  yield Key(file_idx, batch_idx), self._cast_table(pa_table)
                                                  ~~~~~~~~~~~~~~~~^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/parquet/parquet.py", line 156, in _cast_table
                  pa_table = table_cast(pa_table, self.info.features.arrow_schema)
                File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2369, in table_cast
                  return cast_table_to_schema(table, schema)
                File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2297, in cast_table_to_schema
                  raise CastError(
                  ...<3 lines>...
                  )
              datasets.table.CastError: Couldn't cast
              id: int64
              category: large_string
              subcategory: large_string
              title: large_string
              text: large_string
              keywords: large_string
              style: large_string
              difficulty: large_string
              quality_score: double
              -- schema metadata --
              pandas: '{"index_columns": [], "column_indexes": [], "columns": [{"name":' + 1095
              to
              {'category': Value('string'), 'subcategory': Value('string'), 'title': Value('string'), 'text': Value('string'), 'keywords': Value('string'), 'style': Value('string'), 'difficulty': Value('string')}
              because column names don't match

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.

🇸🇴 Somali Open Pre-Training Corpus v2 (SOPC-v2)

Creator: Hamze Jamal (@Zyroxx66)
Language: Somali (Af-Soomaali)
License: CC-BY-4.0
Release Version: 2.0

📌 Overview

The Somali Open Pre-Training Corpus v2 (SOPC-v2) is a massively expanded, high-quality, deduplicated, and domain-balanced dataset designed specifically for Continued Pre-Training (CPT), Domain Adaptation, and Instruction Alignment of Large Language Models (LLMs) such as Qwen 2.5, Llama 3.2, Gemma 2, SmolLM2, and Mistral.

Building on Version 1 (14,150 documents), v2 expands the corpus to over 34,400+ curated documents (~5.4+ Million Tokens), providing broader thematic coverage, rich idiomatic phrasing, and varied stylistic registers.

📊 Dataset Statistics

Metric Version 1 (SOPC-v1) Version 2 (SOPC-v2)
Total Rows / Documents 14,150 34,498
Estimated Token Count ~2.20M Tokens ~5.40M Tokens
Language Somali (Af-Soomaali) Somali (Af-Soomaali)
License CC-BY-4.0 CC-BY-4.0
Format Parquet / JSON Lines Parquet / JSON Lines / CSV

🎯 Key Features & Improvements in v2

  • 2.4x Scale Expansion: Expanded from 14.1K to 34.5K documents, covering deeper technical, cultural, educational, and modern colloquial contexts.
  • Domain Diversity: Balanced across multiple sectors including Commerce (Ganacsi), Public Health (Caafimaad), Culture & Folk (Dhaqan), News (Warar), Agriculture (Beeraha), Technology (Teknoolojiyada), and Science (Saynis).
  • Multi-Style Representation: Features diverse discourse styles:
    • report (Formal news, announcements, official reports)
    • dialogue (Interviews, multi-turn conversations)
    • faq (Questions and instructional answers)
    • story (Folktales, narrative fiction)
    • article & explanation (Explanatory & informational prose)
  • LLM-Friendly Metadata: Every document includes rich metadata tags (category, subcategory, style, keywords, difficulty) enabling targeted sub-filtering during training.

🗂️ Dataset Schema

Field Name Type Description
category string High-level domain (e.g., Ganacsiga, Caafimaadka, Teknolojiyada)
subcategory string Granular sub-domain (e.g., Dhaqaalaha, Wareysi, Cillad-bixin)
title string Document or section headline
text string Main body content in high-quality Standard Somali
keywords string Relevant keywords separated by commas
style string Writing style (report, dialogue, faq, article, story, explanation)
difficulty string Linguistic difficulty (easy, medium, hard)

🚀 Usage

Loading with Hugging Face datasets 

from datasets import load_dataset

# Load the dataset
dataset = load_dataset("Zyroxx66/somali-pretraining-corpus")

# Inspect a sample
print(dataset['train'][0])

Loading with pandas 

import pandas as pd

df = pd.read_parquet("https://huggingface.co/datasets/Zyroxx66/somali-pretraining-corpus/resolve/main/data/train-00000-of-00001.parquet")
print(f"Total Rows: {len(df)}")
print(df.head())

📝 Example Document 

{
  "category": "Ganacsiga",
  "subcategory": "Dhaqaalaha",
  "title": "Kobaca dhaqaalaha dalka",
  "text": "Warbixin cusub oo ay soo saartay wasaaradda maaliyadda ayaa lagu sheegay in dhaqaalaha dalka uu muujiyay horumar la taaban karo sanadkan...",
  "keywords": "dhaqaale, ganacsi, maaliyadda, kobac",
  "style": "report",
  "difficulty": "medium"
}

📜 License & Citation

This dataset is released under the Creative Commons Attribution 4.0 International (CC-BY-4.0) license. You are free to share, adapt, and build upon this dataset for academic or commercial purposes provided appropriate attribution is given.

@misc{sopcv2_2026,
  author = {Hamze Jamal},
  title = {Somali Open Pre-Training Corpus v2 (SOPC-v2)},
  year = {2026},
  publisher = {Hugging Face},
  journal = {Hugging Face Repository},
  howpublished = {\url{https://huggingface.co/datasets/Zyroxx66/somali-pretraining-corpus}}
}
Downloads last month
72
Number of rows:
34,498
Total file size:
20.5 MB