Free resource · living index

Open datasets for AI grading research, in one place.

There is no single home for the open K-12 essay and writing corpora that AI grading research is built on. Datasets live on Kaggle, Figshare, GitHub, ArXiv, lab sites, and state vendors. K12Eval maintains this list as a public service. Free to use. Credited to original authors. Updated as new corpora drop.

Datasets indexed

60K+

Scored responses covered

EN / PT

Languages represented

How this list is curated: a corpus is included if it contains scored student-written responses, is publicly downloadable (or accessible upon request from the original host), and has been used in published or community AI grading research. K12Eval does not redistribute the data. Each row links you to the original source.

Dataset

Size

Scoring

Language

LEAR Lab Dataset

LEAR Lab

Approximately 24,000 student-written argumentative essays scored 1 to 6. Widely used as a target for automated essay scoring models.

EssaysWritingRubric-scored

DREsS (ArXiv 2402.16733)

1,782 essays from English-as-a-Foreign-Language learners scored by experts on content, organization, and language, plus 6,500+ additional standardized and synthetic samples.

EssaysEFL / ELLRubric-scored

1,782 + 6,500+

Content, organization, language rubrics

English (EFL learners)

ASAP Dataset

Automated Student Assessment Prize (Kaggle / Hewlett)

One of the most widely used datasets in automated essay scoring research. 12,978 essays across eight prompts, covering argumentative, narrative, and source-dependent responses.

EssaysK-12WritingRubric-scored

12,978 essays

Prompt-specific rubrics

English

ASAP (Figshare Pre-Processed)

Figshare

A pre-processed version of the ASAP dataset, hosted on Figshare and extensively reused in published essay-grading research.

EssaysWritingRubric-scored

12,978 essays

Same as ASAP

English

Automated Essay Grading (GitHub)

NishantSushmakar on GitHub

A GitHub repository bundling 12,976 essays with word counts between roughly 150 and 550 words, along with baseline tooling for training and evaluating grading models.

EssaysCode / baselineWriting

12,976 essays

Holistic scores

English

ELL Feedback Prize (Kaggle)

Kaggle / Learning Agency Lab

Roughly 2,700 essays from English Language Learners in grades 8 through 12, scored on six traits (ideas, organization, voice, word choice, sentence fluency, conventions) on a 1 to 5 scale. Well suited to trait-specific scoring and interrater-reliability analyses.

EssaysK-12EFL / ELLTrait-scored

~2,700 essays

6 traits, 1 to 5

English (ELL, grades 8-12)

Learning Agency Lab — Automated Essay Scoring 2.0

The Learning Agency Lab (Kaggle competition)

The largest open-access essay dataset aligned to modern, student-appropriate writing standards. Designed to stress-test contemporary automated scoring systems.

EssaysK-12WritingRubric-scored

Largest open-access AES dataset

Rubrics aligned to current writing standards

English

Essay-BR DatasetBonus

Catalpa-CL

4,570 essays from Brazilian high-school students scored holistically on a 0 to 1000 scale. Useful for cross-linguistic and multilingual scoring research.

EssaysK-12BilingualWriting

4,570 essays

Holistic, 0 to 1000

Portuguese (Brazil)

IELTS Writing Scored Essays (Kaggle)

Kaggle

A collection of IELTS writing samples with band-level scores, useful for studying assessment rubrics used in international English-language proficiency testing.

EssaysWritingRubric-scoredHigh-stakes

Scored writing samples

IELTS band scores

English (IELTS)

PARCC Released Items (New Meridian)

New Meridian

A searchable database of scored and annotated PARCC sample student responses. Valuable as a high-stakes, state-aligned reference for AI-grading evaluation.

K-12High-stakesRubric-scored

Annotated released items

State rubric, annotated sample responses

English

ASAP 2.0 Corpus

scrosseye on GitHub

The newer ASAP 2.0 corpus, a modern successor to the original ASAP dataset, used increasingly in LLM-era AES benchmarks.

EssaysK-12WritingRubric-scored

Refreshed ASAP corpus

Prompt-specific rubrics

English

Each dataset is credited to its original authors; K12Eval only curates and links. If a link breaks or a corpus moves, please let us know.

Suggest a dataset

Beyond the index

This list is the starting point. K12Eval is building what comes next.

The aggregated list above is a free public service. The rest of K12Eval, the production corpus, the IRR-validated subsets, the shared methodology, and the public leaderboard, picks up where these legacy datasets stop.

See the framework →View the leaderboard →Fund the bench →

Datasets /funders [email protected]