# togethercomputer/RedPajama-Data

The RedPajama-Data repository contains code for preparing large datasets for training large language models.

Repository: https://github.com/togethercomputer/RedPajama-Data
Canonical: https://ross.abutalabs.com/products/redpajama-data
Language: Python
License: Apache-2.0
License Family: permissive
Last push: 2026-06-03T16:00:12+00:00

## Health v2 (maintenance only)
Score: 68/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 85, release rhythm 35, longevity 88
- inputs: {"age_days": 1237, "days_push": 91, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 4980, forks 374 (observed 2026-08-28T04:09:04.178513+00:00)

## What it is
RedPajama-Data provides code and pipelines for building RedPajama-V2, an open dataset with over 30 trillion tokens of web text for training large language models. It processes CommonCrawl snapshots with quality signal computation and deduplication steps.

## Use cases
- build a large pretraining corpus for an LLM
- reproduce the LLaMA training dataset
- deduplicate web crawl text at scale
- compute quality signals for CommonCrawl documents
- filter multilingual web data for model training

## When to choose
- you need a massive open web-scale corpus for LLM pretraining
- you want reproducible pipelines for cleaning and deduplicating CommonCrawl data
- you need quality annotations to filter training documents

## When to avoid
- you need a small curated dataset rather than web-scale crawl data
- you lack the storage and compute to process terabytes of CommonCrawl data
- you only need inference tooling rather than training data preparation

## Facets
- artifact type: dataset
- maturity: active
- function: etl, data-generation, nlp, llm-training
- domain: large-language-models, machine-learning
- platform: python, cloud
- tags: common-crawl, llm-training-data, data-pipeline, deduplication, quality-signals, open-dataset, data-engineering, natural-language-processing, docker, linux

## Member repositories
- togethercomputer/RedPajama-Data (main) score 68

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:09:04.178513+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T18:17:42.262016+00:00, confidence not recorded.
  - readme: https://github.com/togethercomputer/RedPajama-Data (fetched 2026-08-28T04:09:04.178513+00:00, sha 63b166f0506b)
  - registry_pypi: https://pypi.org/pypi/redpajama-data/json (fetched 2026-08-29T08:58:48.699984+00:00, sha a1b4b8c744d1)
- Data as of 2026-08-30T08:39:29.467469+00:00.