---
license: cc-by-4.0
task_categories:
- translation
language:
- en
- so
tags:
- translation
- parallel
- english
- somali
- en-so
pretty_name: English-Somali Parallel Corpus (Clean 8k)
size_categories:
- 1K<n<10K
---

# English-Somali Parallel Corpus (Clean 8k)

`en-so-parallel-clean-8k` is a clean, formatted English-Somali parallel dataset consisting of 8,912 high-quality translation pairs. This dataset is suitable for training and evaluating machine translation (NMT) models, cross-lingual sentence embedding alignment, and other Somali-English natural language processing (NLP) applications.

The corpus contains everyday conversational phrases, greetings, and common expressions formatted into clean, parallel English (`en`) and Somali (`so`) sentence pairs.

## Dataset Details
- **Total Rows:** 8,912 sentence pairs
- **Total Size:** 661 KB
- **Splits:** `train` (8.91k rows)
- **Columns:**
  - `en`: English sentence (string)
  - `so`: Somali translation (string)

## How to Use
You can easily load this dataset using the Hugging Face `datasets` library:

```python
from datasets import load_dataset

dataset = load_dataset("tufaax/en-so-parallel-clean-8k")
print(dataset["train"][0])