Please provide your name, affiliation if any, and intended use. Access is requested so the JoeyLLM team can understand who is using the dataset and how it is being used. Commercial use is allowed under the dataset licence. Please cite the dataset if it supports your research, project, publication, model, or public output.

🇦🇺 Australian Web Text — 5B-token Sample 🦘

A 5-billion-token Australian web-text dataset created for the JoeyLLM project. This dataset was sampled from the filtered Australian corpus produced by the JoeyLLM sovereign corpus pipeline. 🌐

The purpose of this dataset is to provide a large-scale Australian text corpus for GPT-style language-model pre-training, continued pre-training, data inspection, and research into regional English language models.

📊 Dataset Summary 📌

Property	Value
Dataset name	Australian Web Text — 5B-token Sample
Project	JoeyLLM
Language	English
Region	Australia
Format	Parquet
Number of output files	1
Target token count	5,000,000,000
Final token count	5,000,000,050
Token count difference	+50 tokens
Main file	`australia_5b_uniform_random.parquet`
Sampling method	Uniform random row sampling
Source	Filtered Australian Common Crawl / FineWeb-style corpus

🎯 Intended Uses

💡 Direct Use

Pre-training and continued pre-training of language models on Australian-domain text. 🤖
Domain-adaptation experiments involving Australian English usage, place names, institutions, and topics. 📝
Research into Common Crawl-derived language-model datasets. 🔬
Inspection and reproducibility of JoeyLLM data pipeline outputs. ⚙️

🚫 Out-of-Scope Use

Extracting personal information or deanonymizing individuals. 🕵️‍♂️
Training models for malicious use, hate speech, harassment, or other harmful applications. 🛑
Treating this dataset as a balanced linguistic corpus or authoritative representation of Australian English. ⚠️

🌍 Source Data

The dataset was created from the filtered Australian corpus stored on the JoeyLLM JupyterHub environment.

The upstream corpus was derived from web-scale data using a JoeyLLM data processing pipeline based on:

Common Crawl / FineWeb-style large-scale web data.
Australian domain-based extraction and country attribution.
Structural filtering and refinement.
Uniform random sampling for the 5B-token release.

The original Australian filtered corpus contained hundreds of billions of tokens. This dataset is a 5B-token uniformly sampled subset of that larger internal corpus.

This dataset should be interpreted as Australia-attributed web text, not necessarily text authored by Australians or officially published in Australia.

🎲 Sampling Method

The final dataset was created using uniform random row sampling across the available Australian parquet files.

The process was:

Scan all filtered Australian parquet files.
Count the total available tokens using the token_count column.
Compute a sampling probability based on the 5B-token target.
Randomly sample rows from all input parquet files using the same probability.
Slightly oversample to avoid falling below the target.
Shuffle the sampled shards.
Randomly trim the sampled data to approximately 5B tokens.
Write the result as a single final parquet file.

This approach avoids simply taking the first rows or the first few parquet files. Each row in the filtered Australian data pool had the same probability of being selected during the sampling stage.

Because rows/documents are not split mid-text, the realised token count slightly overshoots the target: 5,000,000,050 tokens vs 5,000,000,000 target. ⏱️

🧱 Dataset Structure 🗂️

The dataset is stored as a single parquet file:

australia_5b_uniform_random.parquet

Expected core columns include:

Field	Type	Description
`text`	string	Cleaned document body.
`token_count`	int	Token count used for sampling and filtering.
`url`	string	Original source URL, if available.
`id`	string	Stable document identifier, if available.
`dump`	string	Common Crawl dump identifier, if available.
`date`	string	Crawl date, if available.
`language`	string	Language label assigned upstream, expected to be `en`, if present.
`language_score`	float	Language-detector confidence score, if present.
`country`	string	Country attribution, expected to be `Australia`, if present.
`hash`	string	Exact hash value, if exact hashing was enabled.
`simhash`	string	Near-duplicate hash value, if near-duplicate hashing was enabled.

🏗️ Dataset Creation

🔍 Curation Rationale

Regional linguistic nuances — including Australian spelling, slang, place names, institutions, public services, media references, and local web conventions — are often diluted in global web-scale datasets. JoeyLLM provides targeted Australian web-text samples to support research into foundation models with stronger Australian and regional coverage. 🏙️

🧹 Cleaning Pipeline

The upstream dataset was processed using a FineWeb-style pipeline including:

Language filtering: documents retained only when classified as English with sufficient language confidence. ✨
Quality filtering: heuristics to reduce low-quality pages, boilerplate, repetitive text, navigation text, and obvious extraction artefacts. 🧽
Australian attribution: targeted selection using signals such as .au domains, URL features, crawl metadata, and Australian content signals. 📍
Structural filtering: removal or reduction of obvious web artefacts and low-value extracted content.
Deduplication / hashing: duplicate tracking and filtering were applied upstream where available. ✂️

🛡️ Personal and Sensitive Information

This dataset is derived from public web crawls and may contain names, contact details, opinions, offensive content, copyrighted text, or other sensitive material. No dedicated PII masking was performed. ⚠️

Users should apply additional filtering, redaction, and safety review before using this dataset in production systems or public-facing models.

🚀 Loading the Dataset 💻

from datasets import load_dataset

# Load the full 5B-token sample
ds = load_dataset("JoeyLLM/Australian-dataset-5b", split="train")

print(ds)
print(ds[0]["text"][:500])

For streaming, which is recommended for rapid inspection:

from datasets import load_dataset

ds = load_dataset("JoeyLLM/Australian-dataset-5b", split="train", streaming=True)

for ex in ds.take(3):
    print(ex.get("url"), ex["token_count"])

If loading from a local parquet file:

from datasets import load_dataset

ds = load_dataset(
    "parquet",
    data_files="australia_5b_uniform_random.parquet",
    split="train",
)

print(ds)

⚠️ Limitations and Known Issues

This dataset is derived from public web crawl data and inherits the usual limitations of Common Crawl-derived corpora.

Known limitations include:

Heuristic country attribution. Australia attribution is based on automated signals and may contain false positives or non-Australian content.
Web-text noise. Boilerplate, navigation text, advertisements, duplicate fragments, low-quality pages, and formatting artefacts may remain.
Residual duplication. Deduplication and hashing reduce duplicate content, but near-duplicates may remain.
Potential personal information. Public web data may contain personal names, contact details, or other sensitive material.
Copyright and source terms. The underlying text originates from public web pages and may remain subject to the rights and terms of the original publishers.
Not balanced by domain or genre. The dataset reflects the distribution of selected web crawl data rather than a deliberately balanced linguistic corpus.
Single-file layout. The dataset is provided as one parquet file, which may be less convenient for some distributed training workflows than a sharded dataset.

📜 License ⚖️

The dataset card, metadata, selection, and processing outputs are released under CC BY 4.0.

Commercial use is allowed under the dataset licence, subject to attribution and the other terms of CC BY 4.0. The underlying text is derived from publicly crawled web pages via Common Crawl and may remain subject to the rights, licences, and terms of the original publishers. Users are responsible for ensuring that their downstream use complies with applicable law and source terms.

📚 Citation ✒️

A citation entry for the JoeyLLM project paper will be added once available. Until then, please cite this dataset card by URL:

@misc{joeyllm_australian_5b,
  title        = {Australian Web Text -- 5B-token Sample},
  author       = {JoeyLLM Team},
  year         = {2026},
  howpublished = {https://huggingface.co/datasets/JoeyLLM/Australian-dataset-5b}
}

🙏 Acknowledgements 🤝

Built using Common Crawl data and a FineWeb-style processing pipeline. While dataset selection, cleaning, sampling, and publication were carried out by the JoeyLLM team, this project would not have been possible without the invaluable tools, feedback, and ongoing support of the broader open-source AI community. We extend our deepest gratitude to all open-source contributors and researchers whose collaborative efforts continue to drive this field forward. 🌟

📬 Dataset Card Contact ✉️

For inquiries regarding research access, collaboration, or dataset questions, please contact:

Matthew Altenburg AI Scientist & Lead Researcher, JoeyLLM matthew.altenburg@anu.edu.au Backup: mattaltenburg@gmail.com


Double-check the actual repo name. If it is not `JoeyLLM/Australian-dataset-5b`, update the two `load_dataset(...)` lines and the citation URL.

Downloads last month: 28

Total file size:

10.5 GB