WanJuan-CC: A Safe and High-Quality Open-sourced English Webtext Dataset
This provides a high-quality dataset for researchers and practitioners training language models, but it is incremental as it builds on existing methods for data filtering and processing.
The paper tackles the challenge of constructing large-scale pre-training datasets for language models by creating WanJuan-CC, a safe and high-quality open-sourced English webtext dataset derived from Common Crawl data, resulting in 2.22T tokens of safe data and showing that models trained on it perform better on validation datasets and downstream tasks compared to RefinedWeb.
This paper presents WanJuan-CC, a safe and high-quality open-sourced English webtext dataset derived from Common Crawl data. The study addresses the challenges of constructing large-scale pre-training datasets for language models, which require vast amounts of high-quality data. A comprehensive process was designed to handle Common Crawl data, including extraction, heuristic rule filtering, fuzzy deduplication, content safety filtering, and data quality filtering. From approximately 68 billion original English documents, we obtained 2.22T Tokens of safe data and selected 1.0T Tokens of high-quality data as part of WanJuan-CC. We have open-sourced 100B Tokens from this dataset. The paper also provides statistical information related to data quality, enabling users to select appropriate data according to their needs. To evaluate the quality and utility of the dataset, we trained 1B-parameter and 3B-parameter models using WanJuan-CC and another dataset, RefinedWeb. Results show that WanJuan-CC performs better on validation datasets and downstream tasks.