Datasets › The Pile

The Pile

Introduced by Leo Gao et al. in The Pile: An 800GB Dataset of Diverse Text for Language Modeling31 Dec 2020 archive 2025-07-28

The Pile is a 825 GiB diverse, open source language modelling data set that consists of 22 smaller, high-quality datasets combined together.

Datasheet: Datasheet for the Pile

Benchmarks archive 2025-07-28

All 1 leaderboard whose dataset resolves to this page shown (sort by any header). "First row" is the archive's own first row at snapshot, in the archive's row order; nothing here re-ranks and metric direction is not asserted.

First row (archive order)PaperCode
Language Modelling The Pile Test-Time Fine-Tuning with SIFT + Llama-3.2 (3B) Bits per byte 0.557 Efficiently Learning at Test-Time: Active Fine-Tuning of LLMs jonhue/activeft 39 Compare

Papers archive 2025-07-28

7 shown of 7 papers with a leaderboard row on this dataset's benchmarks, newest first. The archive's own "papers using this dataset" list was never published, so this is the benchmark-backed subset; the archive's count for this dataset is 467. The Syntology column is from Syntology's graph (read 2026-09-24), stated per sample; it is not part of any archive number.

DateSamples run Syntology
Efficiently Learning at Test-Time: Active Fine-Tuning of LLMs 1 15 10 Oct 2024 ran 2 of 2 samples (0 unverified)
Need a Small Specialized Language Model? Plan Early! 0 4 2 Feb 2024 not harvested
Test-Time Training on Nearest Neighbors for Large Language Models 1 1 29 May 2023 ran 0 of 5 samples (5 unverified)
Hungry Hungry Hippos: Towards Language Modeling with State Space Models 3 2 28 Dec 2022 ran 7 of 15 samples (8 unverified)
GLM-130B: An Open Bilingual Pre-trained Model 9 3 5 Oct 2022 ran 5 of 21 samples (16 unverified)
Knowledge Unlearning for Mitigating Privacy Risks in Language Models 2 6 4 Oct 2022 not harvested
The Pile: An 800GB Dataset of Diverse Text for Language Modeling 22 8 31 Dec 2020 ran 1 of 1 samples (0 unverified)

Dataset loaders archive 2025-07-28

7 loaders as listed in the archive; links are outbound and not re-checked here.

Tasks archive 2025-07-28

License archive 2025-07-28

No licence recorded in the archive. Absence here is not a statement about the dataset's terms.

Modalities archive 2025-07-28

Languages archive 2025-07-28

No language tagged.

Variants archive 2025-07-28

  • The Pile

1 variant name, as the archive lists them.

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections