← All companies & organizations

Research orgs / Organization profile

EleutherAI

What EleutherAI does

EleutherAI is an open research organization that develops and releases open-weight large language models, datasets, and model evaluation tooling, with an emphasis on mechanistic interpretability and model evaluation. The group’s best-known open artifacts include the GPT-Neo model family, GPT-J, GPT-NeoX-20B, The Pile pretraining dataset, the Pythia training/checkpoint suite for interpretability and training-dynamics work, and the lm-evaluation-harness framework used to run standardized evaluations across language models.

News

Company record

Source map · 2 recurring channels · 71 references
EleutherAIOfficial Site · monitor↗MTS Frontier / Control RoomCurated Feed · monitor↗Official organization sourceOfficial Reference · reference↗sec.govRegulatory · reference↗2.zoppoz.workers.devReporting · reference↗abgt.assembly.ca.govReporting · reference↗aclanthology.orgReporting · reference↗aethir.comReporting · reference↗aiwiki.aiReporting · reference↗alison.legislature.state.al.usReporting · reference↗app.grantmaking.aiReporting · reference↗appropriations.house.govReporting · reference↗arc.renaissancephilanthropy.orgReporting · reference↗arxiv.orgReporting · reference↗assets-eu.researchsquare.comReporting · reference↗assets.ctfassets.netReporting · reference↗canva.comReporting · reference↗causeiq.comReporting · reference↗cbinsights.comReporting · reference↗cbo.govReporting · reference↗cerre.euReporting · reference↗citeseerx.ist.psu.eduReporting · reference↗coefficientgiving.orgReporting · reference↗consumerrights.wikiReporting · reference↗coreweave.comReporting · reference↗crunchbase.comReporting · reference↗csv-storage.forexpros.comReporting · reference↗dash.harvard.eduReporting · reference↗davidwidder.meReporting · reference↗downloads.regulations.govReporting · reference↗en.wikipedia.orgReporting · reference↗es.wikipedia.orgReporting · reference↗forum.effectivealtruism.orgReporting · reference↗funderscope.comReporting · reference↗gatesfoundation.orgReporting · reference↗gitextract.comReporting · reference↗github.comReporting · reference↗github.itlym.cnReporting · reference↗goodventures.orgReporting · reference↗grants.openai.comReporting · reference↗huggingface.coReporting · reference↗impactaid.ed.govReporting · reference↗impala.digitalReporting · reference↗investors.axecompute.comReporting · reference↗judiciary.senate.govReporting · reference↗kindora.coReporting · reference↗lci.ca.govReporting · reference↗linkedin.comReporting · reference↗longtermwiki.comReporting · reference↗loora.comReporting · reference↗mozillafoundation.orgReporting · reference↗nextomoro.comReporting · reference↗omidyar.comReporting · reference↗open.mozilla.orgReporting · reference↗openai.comReporting · reference↗openreview.netReporting · reference↗papers.coolReporting · reference↗pdfs.assets.alphaxiv.orgReporting · reference↗philanthropy.orgReporting · reference↗phillipring.comReporting · reference↗proceedings.mlr.pressReporting · reference↗projects.propublica.orgReporting · reference↗reddit.comReporting · reference↗renaissancephilanthropy.orgReporting · reference↗rockefellerfoundation.orgReporting · reference↗roundpaper.comReporting · reference↗s205.q4cdn.comReporting · reference↗stability.aiReporting · reference↗stpp.fordschool.umich.eduReporting · reference↗tabpi.orgReporting · reference↗techcrunch.comReporting · reference↗together.aiReporting · reference↗x.aiReporting · reference↗

Still resolving: Newsroom · Official X

What it builds

GPT-Neo (and related GPT-J / GPT-NeoX releases)

EleutherAI open-sources large decoder-only language model code and releases open-weight model checkpoints under permissive licenses, including the GPT-Neo and GPT-J model families and GPT-NeoX-20B.

source ↗
The Pile

The Pile is an open, large-scale curated text dataset for language-model pretraining, released by EleutherAI.

source ↗
Pythia

Pythia is a suite of interpretability- and training-dynamics-focused model checkpoints and accompanying analysis code released by EleutherAI.

source ↗
lm-evaluation-harness

lm-evaluation-harness is EleutherAI’s open framework for running standardized few-shot and zero-shot evaluations of language models across many tasks.

source ↗

Milestones & partnerships