# openai/gpt-2-output-dataset

Dataset of GPT-2 outputs for research in detection, biases, and more

Repository: https://github.com/openai/gpt-2-output-dataset
Canonical: https://ross.abutalabs.com/products/gpt-2-output-dataset
Language: Python
License: MIT
License Family: permissive
Archived: true
Last push: 2023-12-13T03:03:19+00:00

## Health v2 (maintenance only)
Score: 10/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 0, release rhythm 35, longevity 100
- inputs: {"age_days": 2679, "days_push": 994, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, archived
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 2027, forks 546 (observed 2026-08-28T04:06:06.936798+00:00)

## What it is
A dataset from OpenAI containing 250K WebText documents alongside 250K samples each from four GPT-2 model sizes, both untruncated and Top-K 40 truncated, with train/valid/test splits. It includes download scripts and a detection baseline to support research on detecting model-generated text and studying biases.

## Use cases
- train a classifier to detect GPT-2 generated text
- research biases in language model outputs
- benchmark AI-generated text detection methods
- study how model size affects detectability of samples
- get human vs machine text pairs for NLP experiments
- research adversarial evasion of text detectors

## When to choose
- you need labeled human vs GPT-2 text for detection research
- you want a large-scale benchmark for machine-generated text studies
- you're studying properties of GPT-2 model outputs across sizes

## When to avoid
- you need outputs from modern LLMs like GPT-4 or Llama
- you want a dataset for training language models rather than analyzing them
- you need production-ready detection tooling rather than research data

## Facets
- artifact type: dataset
- maturity: maintenance
- function: machine-learning, nlp, data-generation
- domain: large-language-models, machine-learning
- platform: python, cross-platform
- tags: gpt-2, text-generation, ai-detection, webtext, research-dataset, l, natural-language-processing

## Member repositories
- openai/gpt-2-output-dataset (main) score 10

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:06:06.936798+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 2026-08-30T02:59:42.137601+00:00, confidence not recorded.
  - readme: https://github.com/openai/gpt-2-output-dataset (fetched 2026-08-28T04:06:06.936798+00:00, sha 5d7a9440a4d4)
- Data as of 2026-08-30T08:39:29.467469+00:00.