JoaoGuiAlves/SFT-JoinPorTurgueseCorpora

---
license: mit
task_categories:
- text-generation
- conversational
language:
- pt
- en
size_categories:
- n<1K
---

# Portuguese SFT Corpora

## Dataset Description

A collection of Portuguese supervised fine-tuning datasets, including translations from OpenAssistant.

### Dataset Structure

This dataset contains translated conversations from the OpenAssistant dataset.

Each entry includes:
- Original English text (`text`)
- Portuguese translation (`text-pt`)
- Conversation metadata (message_id, parent_id, role, etc.)

### Languages

- English (original)
- Portuguese (translation)

### Data Fields

- `message_id`: Unique identifier for the message
- `parent_id`: ID of the parent message (null for root messages)
- `text`: Original English text
- `text-pt`: Portuguese translation
- `role`: Either "prompter" (user) or "assistant" (AI)
- `lang`: Original language code
- Additional metadata fields

### Usage

```python
from datasets import load_dataset

dataset = load_dataset("JoaoGuiAlves/SFT-JoinPorTurgueseCorpora")
```

### Source Data

Translated from [OpenAssistant/oasst1](https://huggingface.co/datasets/OpenAssistant/oasst1)

### License

MIT License