Papers › SCROLLS: Standardized CompaRison Over Long Language Sequences

SCROLLS: Standardized CompaRison Over Long Language Sequences

10 Jan 2022arXiv:2201.03533archive 2025-07-28

Uri Shaham, Elad Segal, Maor Ivgi, Avia Efrat, Ori Yoran, Adi Haviv, Ankit Gupta, Wenhan Xiong, Mor Geva, Jonathan Berant, Omer Levy

NLP benchmarks have largely focused on short texts, such as sentences and paragraphs, even though long texts comprise a considerable amount of natural language in the wild. We introduce SCROLLS, a suite of tasks that require reasoning over long texts. We examine existing long-text datasets, and handpick ones where the text is naturally long, while prioritizing tasks that involve synthesizing information across the input. SCROLLS contains summarization, question answering, and natural language inference tasks, covering multiple domains, including literature, science, business, and entertainment. Initial baselines, including Longformer Encoder-Decoder, indicate that there is ample room for improvement on SCROLLS. We make all datasets available in a unified text-to-text format and host a live leaderboard to facilitate research on model architecture and pretraining methods.

PaperPDFCodeCode Syntology ran

In Syntology Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

For agents, Syntology's MCP tool lists every function and class Syntology harvested from this paper and whether it ran (how to connect): get_harvested_code_for_paper(arxiv_id="2201.03533")

Code

Syntology Ran 2 of 8 code samples harvested from 1 repository linked to this paper; 6 have no recorded run. Of those that ran: 2 ran · our draft was wrong.

By repository: official repository: 8 samples from 1 repository, 2 ran. The run record, sample by sample. “Ran” means executed on a synthesized input, not that the code is correct or reproduces the paper.

tau-nlp/scrolls officialmentioned in papermentioned on GitHubMIT report
mivg/sled mentioned on GitHubpytorchMIT report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

8 samples harvested; 2 ran; 0 honoured the contract we drafted; 6 have no recorded run. Read from Syntology's graph 2026-09-24; that is when this build read the record, not when the samples ran.

2ran · our draft was wrong
6unverified

Licence: 0 of the 8 samples are pointer only, meaning Syntology does not serve that copy's text. This page shows no code text for any sample; each one links to its file in the repository.

Harvested from tau-nlp/scrolls. “Ran” means the sample executed on a synthesized input. It does not mean the output is correct, and nothing here reproduces the paper's results. “Honoured” and “violated” refer to a contract Syntology drafted from the code itself; “our draft was wrong” and “fixture could not drive it” are failures of Syntology's instrument, not of the code.

Each sample ends with its code_sha256, Syntology's identity for that exact code. An agent fetches the stored sample with Syntology's MCP tool get_code(code_sha256="…") (how to connect); click an identity to copy that call.

Repository labels, per sample. official repository: The archive marks this repository official for the paper. named in the paper: The archive records that the paper mentions this repository; it is not marked official. community (archive-listed): In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper. found in paper text by Syntology: Syntology found this repository in the paper's own text; whether it is the authors' implementation is not asserted. community: Not in the archive's code links for this paper; a community repository Syntology harvested. Samples from a repository marked official are listed first. Licence labels name the repository's licence as recorded at harvest. “Pointer only” means Syntology does not serve that copy's text, for one of four reasons: no licence file was found; the licence was not identified; the licence is recorded as permissive but that copy's record is not marked cleared; or the licence is outside the permissive list Syntology serves text under (MIT, Apache-2.0, BSD and similar). Some licences outside that list permit redistribution, such as WTFPL, and GPL-3.0 under its conditions; they are simply not on the list. Hover a licence label for the reason. File links open the file on GitHub at the default branch, which may have changed since the harvest.

get_metrics_filename tau-nlp/scrolls/evaluator/dataset_evaluator.py official repository ran · our draft was wrong fingerprinted MIT (permissive) · 04757364abe8941b · report
safe_read_csv tau-nlp/scrolls/evaluator/benchmark_evaluator.py official repository ran · our draft was wrong MIT (permissive) · 52aaf6f9e86c2add · report
drop_duplicates_in_input tau-nlp/scrolls/evaluator/dataset_evaluator.py official repository unverified MIT (permissive) · abbee6221df98330 · report
gen tau-nlp/scrolls/baselines/src/utils/namegenerator.py official repository unverified MIT (permissive) · 357bcaff2fb41a9f · report
load_predictions_df tau-nlp/scrolls/evaluator/benchmark_evaluator.py official repository unverified MIT (permissive) · 54fc256b1ee469e2 · report
merge tau-nlp/scrolls/baselines/src/utils/config.py official repository unverified MIT (permissive) · 4c69d00f97bf6280 · report
replace_recursive tau-nlp/scrolls/baselines/src/utils/config.py official repository unverified MIT (permissive) · 3de4ac2599cef7ad · report
unflatten tau-nlp/scrolls/baselines/src/utils/config.py official repository unverified MIT (permissive) · 225dafec388f7a2f · report

Tasks

DecoderLong-range modelingNatural Language InferenceQuestion Answering

Datasets

Introduced by this paper, per the archive.

SCROLLS

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Long-range modeling SCROLLS LED Base Avg. 29.16 #8 of 13 Archive leaderboard report
Long-range modeling SCROLLS LED Base CNLI 71.5 #8 of 13 Archive leaderboard report
Long-range modeling SCROLLS LED Base GovRep 56.2 / 26.6 / 28.8 #8 of 13 Archive leaderboard report
Long-range modeling SCROLLS LED Base Nrtv 18.5 #8 of 13 Archive leaderboard report
Long-range modeling SCROLLS LED Base QALT EM-T/H 25.8 / 25.4 #8 of 13 Archive leaderboard report
Long-range modeling SCROLLS LED Base QMSum 25.1 / 6.7 / 18.8 #8 of 13 Archive leaderboard report
Long-range modeling SCROLLS LED Base Qspr 26.6 #8 of 13 Archive leaderboard report
Long-range modeling SCROLLS LED Base SumScr 24.2 / 4.5 / 15.4 #8 of 13 Archive leaderboard report
Long-range modeling SCROLLS BART Base Avg. 29.01 #9 of 13 Archive leaderboard report
Long-range modeling SCROLLS BART Base CNLI 77.4 #9 of 13 Archive leaderboard report
Long-range modeling SCROLLS BART Base GovRep 47.9 / 18.6 / 22.7 #9 of 13 Archive leaderboard report
Long-range modeling SCROLLS BART Base Nrtv 15.4 #9 of 13 Archive leaderboard report
Long-range modeling SCROLLS BART Base QALT EM-T/H 26.0 / 25.9 #9 of 13 Archive leaderboard report
Long-range modeling SCROLLS BART Base QMSum 30.2 / 8.7 / 20.7 #9 of 13 Archive leaderboard report
Long-range modeling SCROLLS BART Base Qspr 26.3 #9 of 13 Archive leaderboard report
Long-range modeling SCROLLS BART Base SumScr 27.2 / 4.9 / 16.7 #9 of 13 Archive leaderboard report
Long-range modeling SCROLLS Naive Avg. 19.35 #10 of 13 Archive leaderboard report
Long-range modeling SCROLLS Naive CNLI 66 #10 of 13 Archive leaderboard report
Long-range modeling SCROLLS Naive GovRep 45.3 / 17.9 / 20.8 #10 of 13 Archive leaderboard report
Long-range modeling SCROLLS Naive Nrtv 1.5 #10 of 13 Archive leaderboard report
Long-range modeling SCROLLS Naive QALT EM-T/H 25.2 / 26.1 #10 of 13 Archive leaderboard report
Long-range modeling SCROLLS Naive QMSum 14.2 / 2.0 / 9.3 #10 of 13 Archive leaderboard report
Long-range modeling SCROLLS Naive Qspr 3.4 #10 of 13 Archive leaderboard report
Long-range modeling SCROLLS Naive SumScr 19.6 / 1.8 / 11.0 #10 of 13 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Methods

AdamWAttentionAttention DropoutDense ConnectionsDropoutLayer NormalizationLinear LayerLinear Warmup With Linear DecayLongformerMulti-Head AttentionResidual ConnectionSoftmaxWeight DecayWordPiece

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