---
task_categories:
- text2text-generation
language:
- bn
tags:
- bangla
- punctuation-restoration
- nlp
- synthetic-data
- seq2seq
pretty_name: Bangla Punctuation Restoration (Generated)
size_categories:
- 10K<n<100K
---

# ha-pr-bn-aminul-generated

**Dataset Name**: `ha-pr-bn-aminul-generated`
**Language**: Bengali (bn)
**Task**: Punctuation Restoration (Text-to-Text Generation)
**Size**: 11,113 conversations
**License**: CC BY 4.0
**Format**: Human-assistant conversations with punctuation correction

## Dataset Description
This dataset contains **Bangla punctuation restoration data** presented in a **conversational format**. The dataset is designed for training and evaluating models that restore punctuation in Bengali text. Each instance simulates a conversation between a human (who provides unpunctuated Bangla text) and an assistant (who returns the same text with correct punctuation).

## Dataset Summary
- **Language**: Bengali (bn)
- **Task**: Punctuation Restoration (Text-to-Text Generation)
- **Size**: 11,113 conversations
- **Format**: Human → Assistant message pairs


## Dataset Structure

### Data Instances

Each instance is a structured conversation:

```json
{
"input": "শুধু বর্ষা নয় শীত বা গ্রীষ্মেও বেড়ে যাচ্ছে তাপ",
"target": "শুধু বর্ষা নয়, শীত বা গ্রীষ্মেও বেড়ে যাচ্ছে তাপ।"
}
```

## Usage

```python
from datasets import load_dataset

dataset = load_dataset("aminul/ha-pr-bn-aminul-generated")
print(dataset["train"][0])
```

## Dataset Creation

### Source

Texts were **machine-generated** using large language models and represent a mix of:
- News
- Literature
- Wikipedia-style content
- Dialogues

### Preprocessing

1. Generate clean, punctuated Bangla sentences.
2. Remove punctuation to create the input.
3. Store both versions as a conversational pair.
4. Add metadata (source, score).