DinoResearch/SingaporeGPT-Training

SingaporeGPT Dataset

Overview

SingaporeGPT is a text dataset designed for language model pretraining, fine-tuning, and natural language processing research.

The dataset contains text collected from Singapore-related sources, including news, government publications, and other public information sources. It is designed to provide Singapore-focused language data for training and evaluating machine learning models.

Dataset Information

Dataset name: SingaporeGPT Version: 1.0 License: SingaporeGPT Dataset Terms (SGDT-1.0) Format: Parquet

Files

Example dataset files:

singaporegpttrain-0000.parquet

Data Fields

Each record contains:

Field	Description
text	Extracted text content
url	Original source URL
source	Source identifier

Example:

{
  "text": "Example article text...",
  "url": "https://example.com/article",
  "source": "cna"
}

Sources

The dataset may include content from multiple sources, such as:

• Singapore news websites
• Government publications
• Public information websites

Each source is identified using the source field.

Intended Uses

SingaporeGPT can be used for:

• Language model pretraining experiments
• NLP research
• Text classification
• Information retrieval experiments
• Singapore-focused language understanding tasks

Limitations

This dataset may contain:

• Source-specific writing styles
• Historical information that may become outdated
• Biases present in the original sources
• Text that requires additional filtering before model training

Users should evaluate the dataset before using it in production systems.

Data Processing

The dataset was created using a collection pipeline that:

1. Collects publicly available text sources.
2. Extracts article or document text.
3. Stores provenance information.
4. Saves processed data in Parquet format.

License

This dataset uses:

SingaporeGPT Dataset Terms (SGDT-1.0)

This dataset contains material from multiple sources. Original content remains subject to the rights, licenses, and terms of its respective owners.

SGDT-1.0 does not transfer ownership of third-party content.

Users are responsible for ensuring that their use of the dataset complies with applicable laws, source terms, and licensing requirements.

Citation

If you use SingaporeGPT in research, please cite:

@dataset{singaporegpt,
  title={SingaporeGPT Dataset},
  version={1.0},
  year={2026}
}

Acknowledgements

Thanks to the organizations and publishers whose public information contributes to SingaporeGPT.

Contact

For questions, suggestions, or issues, please open a discussion or issue in this repository.

SingaporeGPT Dataset Terms (SGDT-1.0)

Overview

SingaporeGPT Dataset Terms (SGDT-1.0) is a dataset usage notice for the SingaporeGPT dataset, which contains text data collected from multiple sources for machine learning and language model research.

This dataset may include content from different publishers, organizations, and public sources. Each item retains the rights and terms associated with its original source.

License / Usage Terms

This dataset is provided under SingaporeGPT Dataset Terms (SGDT-1.0).

You may:

• Use the dataset structure and metadata for research and development.
• Analyze the dataset for machine learning and natural language processing experiments.
• Build tools and models using the dataset where you have the necessary rights to use the underlying content.

You may not:

• Claim ownership of third-party content included in this dataset.
• Remove original source attribution, URLs, or provenance information.
• Redistribute source content in ways that violate the original publisher's terms.
• Assume that this dataset grants permission to use copyrighted material beyond applicable rights.

Source Attribution

Each dataset entry includes source information where available:

• text — The extracted text content.
• url — The original webpage URL.
• source — The originating source identifier.

Example:

{
  "text": "Article content...",
  "url": "https://example.com/article",
  "source": "example"
}

Third-Party Content

This dataset may contain material from sources including, but not limited to:

• Government publications
• News organizations
• Public websites
• Other online sources

Rights to third-party materials remain with their respective owners.

SGDT-1.0 does not transfer ownership of third-party content and does not replace the original terms, licenses, or policies of the sources.

Dataset Creator Rights

The creator of this dataset claims rights only over:

• Dataset organization
• Processing methods
• Metadata formatting
• Collection and preparation pipeline

No ownership claim is made over third-party source materials.

Responsible Use

Users should:

• Respect source websites' terms of service.
• Respect copyright and applicable laws.
• Verify permissions before commercial use or redistribution.
• Preserve attribution and provenance information.

Version

Current version:

SGDT-1.0

Contact

For questions about this dataset, please open an issue in the dataset repository.