RedPajama v2

Aggregates 100+ billion deduplicated documents (30 trillion tokens) from 84 CommonCrawl dumps across 5 languages (English, German, French, Spanish, Italian). Each document is pre-annotated with 40+ quality signals including perplexity scores, deduplication hashes, content classifiers, and toxicity ratings computed via a standardized pipeline. The architecture processes raw CommonCrawl HTML through text extraction, deduplication, and multi-dimensional quality scoring, enabling downstream users to apply custom filtering strategies without reprocessing the raw data.

Implements deduplication across 100+ billion documents using hash-based matching to identify and remove duplicate content from CommonCrawl. The pipeline computes deduplication hashes for each document and filters the raw 100+ trillion token corpus down to 30 trillion deduplicated tokens. This approach preserves document boundaries (unlike token-level deduplication) and produces deterministic, reproducible results across reprocessing runs.

Provides the entire 30 trillion token corpus, processing scripts, and quality annotations as free, open-source resources with no licensing restrictions. Users can download, modify, redistribute, and use the data for any purpose including commercial applications. This open approach enables broad research access and community-driven improvements without vendor lock-in.

Computes perplexity scores for each document using a reference language model, enabling quantitative assessment of text quality and language model fitness. The perplexity metric measures how well a pre-trained model predicts the document; lower perplexity indicates higher-quality, more coherent text. These pre-computed scores allow users to filter documents by quality threshold without running inference themselves, and to study the relationship between perplexity and downstream model performance.

Annotates each document with content classifiers and toxicity ratings, enabling category-based filtering and safety-aware data curation. The pipeline applies pre-trained classifiers to categorize document content (e.g., news, forums, documentation) and compute toxicity scores. These annotations are pre-computed and stored with each document, allowing users to filter by content type or toxicity threshold without running inference themselves.

Publishes end-to-end processing scripts on GitHub that convert raw CommonCrawl HTML to deduplicated, annotated documents. The pipeline is fully open-source, enabling users to understand, verify, and reproduce the data processing methodology. Scripts handle HTML-to-text conversion, deduplication, quality signal computation, and filtering, allowing researchers to reprocess data with custom parameters or apply the same methodology to new CommonCrawl dumps.

Enables users to apply custom filtering strategies by combining 40+ pre-computed quality signals (perplexity, toxicity, content classifiers, deduplication hashes, etc.). Rather than providing pre-filtered 'ready-to-train' datasets, RedPajama v2 provides the raw signals and lets users define their own filtering logic. This architecture supports comparative studies of curation strategies and enables organizations to apply domain-specific or value-aligned filtering without reprocessing the base dataset.

Provides 30 trillion tokens across 5 languages (English, German, French, Spanish, Italian) with consistent quality signal annotations applied uniformly across all languages. The architecture processes each language through the same deduplication, quality scoring, and classification pipeline, enabling comparative studies of language-specific data characteristics and training multilingual models on a standardized base dataset. Language-specific processing details are not documented, but the consistent annotation methodology enables cross-language analysis.

Aggregates data from 84 CommonCrawl dumps (100+ trillion raw tokens) into a single, deduplicated, consistently-annotated dataset. The architecture handles the complexity of processing massive-scale web data including deduplication across dumps, consistent quality signal computation, and language-specific filtering. This enables users to work with a unified, large-scale web corpus without managing multiple CommonCrawl dumps or implementing their own aggregation pipeline.

Publishes processing scripts on GitHub enabling users to understand, validate, and extend the data processing pipeline. Scripts cover HTML-to-text conversion, deduplication, quality signal computation, and filtering. This transparency enables reproducible research, allows users to apply custom modifications, and supports community contributions. Users can inspect the exact methodology used for corpus creation and adapt it for their own data sources.

Distributes the 30 trillion token corpus via HuggingFace Datasets, enabling users to download, stream, or access subsets without managing raw files directly. HuggingFace integration provides standardized data loading APIs compatible with PyTorch, TensorFlow, and other ML frameworks. Users can load documents with quality annotations, apply filters, and create training dataloaders with minimal code.