Papers › LongT5: Efficient Text-To-Text Transformer for Long Sequences

LongT5: Efficient Text-To-Text Transformer for Long Sequences

15 Dec 2021Findings (NAACL) 2022 7arXiv:2112.07916archive 2025-07-28

Mandy Guo, Joshua Ainslie, David Uthus, Santiago Ontanon, Jianmo Ni, Yun-Hsuan Sung, Yinfei Yang

Recent work has shown that either (1) increasing the input length or (2) increasing model size can improve the performance of Transformer-based neural models. In this paper, we present a new model, called LongT5, with which we explore the effects of scaling both the input length and model size at the same time. Specifically, we integrated attention ideas from long-input transformers (ETC), and adopted pre-training strategies from summarization pre-training (PEGASUS) into the scalable T5 architecture. The result is a new attention mechanism we call Transient Global (TGlobal), which mimics ETC's local/global attention mechanism, but without requiring additional side-inputs. We are able to achieve state-of-the-art results on several summarization tasks and outperform the original T5 models on question answering tasks.

PaperPDFConference PDFCodeCode Syntology ran

In Syntology View this paper on Syntology, its page in Syntology's graph. That page adds which of its functions are also present in other papers' code Syntology holds, with dates, and its dated run record as JSON and a README badge. It also lists the repositories linked to the paper, the abstract and the calls for agents.

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

For agents, on Syntology's MCP service (how to connect):

Code

Syntology 1 code sample harvested from 1 repository linked to this paper: 1 ran (of which 0 constructed an object rather than computing a result; 0 with no instrument failure: 0 honoured, 0 violated, 0 with no contract checked; 1 where Syntology's instrument failed) · 0 unverified. Where Syntology's instrument failed: 1 where our contract draft was wrong; failures of Syntology's instrument, not of the code. Every run on this page is a failure of Syntology's instrument, not of the code.

By repository: official repository: 1 sample from 1 repository, 1 ran (0 with no instrument failure; 1 where Syntology's instrument failed). The run record, sample by sample. “Ran” means executed on a synthesized input, not that the code is correct or reproduces the paper.

No sample is linked to this paper from 1 of its repositories that Syntology harvested for other papers:

Levels written on 2026-09-22 (UTC): the dates Syntology's graph wrote each sample's verification level, not when the code executed.

google-research/longt5 officialmentioned in papertfSyntology: 1 sample, 1 ran (1 instrument-failed) · 0 unverified report
utsjiyaoli/qa-attack mentioned on GitHubpytorchSyntology: not harvested report
MindSpore-scientific-2/code-14 mindsporeSyntology: not harvested report
pwc-1/Paper-9 mindsporeSyntology: no sample linked to this paper (harvested for another paper). Syntology's graph links these samples to RandAugment: Practical automated data augmentation with a reduced search space (arXiv:1909.13719; that paper's own run record: 58 ran (of which 1 constructed an object rather than computing a result; 7 with no instrument failure: 0 honoured, 1 violated, 6 with no contract checked; 51 where Syntology's instrument failed) · 7 unverified); nothing checks that they implement this paper's method. 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.

A paper named beside a repository with no sample linked to this paper is shown with that paper's own run record, not this paper's: “ran” means executed on a synthesized input, not that the code is correct, and “where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code. Papers are listed in arXiv-id order, at most three per repository.

Code Syntology ran Syntology

1 sample harvested: 1 ran (of which 0 constructed an object rather than computing a result; 0 with no instrument failure: 0 honoured, 0 violated, 0 with no contract checked; 1 where Syntology's instrument failed) · 0 unverified. “Where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code. Every run on this page is a failure of Syntology's instrument, not of the code. Syntology's record for this page has not changed since , the first build that kept a record date for it; when this build read Syntology's graph is in the build record. That date is not when the samples ran.

1ran · our draft was wrong

Licence: 1 of the 1 sample is 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 google-research/longt5. 1 sample whose link to this paper records no repository is listed under one of this paper's own repositories, because identical code was harvested from it for another paper. “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. A line beginning “fixture:” under a sample that ran gives the arguments Syntology's fixture called the sample with, each as name:type/rank/element kind (? where none was recorded), and the type of what the fixture's run returned, exactly as the harness recorded them; the arguments and the returned type were recorded by two separate runs of the same fixture. No output value is shown, and a returned type says nothing about whether the output was right. “Level recorded” gives the UTC date when Syntology's graph wrote the sample's verification level, not when the code executed; where the level was written before Syntology's graph kept that date, the row says “level recorded, date not recorded”. “Honoured” and “violated” refer to a contract Syntology drafted from the code itself. “Our draft was wrong” and “the contract check could not run” are failures of Syntology's instrument, not of the code: the sample ran on Syntology's synthesized input, and then the drafted contract did not describe it, or the check itself raised an error (RAISES), lacked a dependency (DEP_MISSING) or ran out of time (TIMEOUT); hover the label for which. A metamorphic tier, where shown, is evidence of the output's shape, not that the sample implements the method.

Each sample ends with its code_sha256. A sample's code_sha256 is Syntology's identity for that exact code. It is printed in full (16 hexadecimal characters for every sample in this build), and get_code(code_sha256="…") takes it exactly as printed. An agent calls it on Syntology's MCP service (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 are grouped by repository: a repository marked official first, then by how directly the repository is tied to the paper, then in the archive's order; never by how many ran. 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 six 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; the licence is outside the permissive list Syntology serves text under (MIT, Apache-2.0, BSD and similar); this paper's copy was not located, so its licence is not recorded; or this paper's copy is cleared, but the stored copy of the same code that Syntology serves is not. 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.

google-research/longt5 official repository 1 sample, 1 ran (1 instrument-failed) · 0 unverified
parse_json_line longt5/data/nq_preprocess.py identical code Syntology harvested from this repository for another paper; this paper's link records no repository, path or licence of its own ran · our draft was wrong licence of this copy not recorded · 3df820a3fa7f5f80 · report fixture: signature line:str/0/str · returned dict · level recorded 2026-09-22

Tasks

Abstractive Text SummarizationLong-range modelingMulti-Document SummarizationQuestion AnsweringText Summarization

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Abstractive Text Summarization CNN / Daily Mail LongT5 ROUGE-1 43.94 #21 of 53 Archive leaderboard report
Abstractive Text Summarization CNN / Daily Mail LongT5 ROUGE-2 21.40 #21 of 53 Archive leaderboard report
Abstractive Text Summarization CNN / Daily Mail LongT5 ROUGE-L 41.28 #21 of 53 Archive leaderboard report
Long-range modeling SCROLLS LongT5 XL Avg. 42.53 #2 of 13 Archive leaderboard report
Long-range modeling SCROLLS LongT5 XL CNLI 88.2 #2 of 13 Archive leaderboard report
Long-range modeling SCROLLS LongT5 XL GovRep 61.1 / 32.3 / 33.7 #2 of 13 Archive leaderboard report
Long-range modeling SCROLLS LongT5 XL Nrtv 29.3 #2 of 13 Archive leaderboard report
Long-range modeling SCROLLS LongT5 XL QALT EM-T/H 46.0 / 42.1 #2 of 13 Archive leaderboard report
Long-range modeling SCROLLS LongT5 XL QMSum 34.9 / 11.8 / 23.5 #2 of 13 Archive leaderboard report
Long-range modeling SCROLLS LongT5 XL Qspr 53.1 #2 of 13 Archive leaderboard report
Long-range modeling SCROLLS LongT5 XL SumScr 35.8 / 9.6 / 21.1 #2 of 13 Archive leaderboard report
Long-range modeling SCROLLS LongT5 Large Avg. 41.03 #3 of 13 Archive leaderboard report
Long-range modeling SCROLLS LongT5 Large CNLI 87.3 #3 of 13 Archive leaderboard report
Long-range modeling SCROLLS LongT5 Large GovRep 61.3/32.2/33.8 #3 of 13 Archive leaderboard report
Long-range modeling SCROLLS LongT5 Large Nrtv 27.2 #3 of 13 Archive leaderboard report
Long-range modeling SCROLLS LongT5 Large QALT EM-T/H 40.6 / 38.6 #3 of 13 Archive leaderboard report
Long-range modeling SCROLLS LongT5 Large QMSum 35.1 / 12.0 / 23.3 #3 of 13 Archive leaderboard report
Long-range modeling SCROLLS LongT5 Large Qspr 52.3 #3 of 13 Archive leaderboard report
Long-range modeling SCROLLS LongT5 Large SumScr 60.3 / 31.1 / 32.8 #3 of 13 Archive leaderboard report
Long-range modeling SCROLLS LongT5 Base Avg. 38.6 #5 of 13 Archive leaderboard report
Long-range modeling SCROLLS LongT5 Base CNLI 85.6 #5 of 13 Archive leaderboard report
Long-range modeling SCROLLS LongT5 Base GovRep 57.7 / 30.0 / 31.4 #5 of 13 Archive leaderboard report
Long-range modeling SCROLLS LongT5 Base Nrtv 23.0 #5 of 13 Archive leaderboard report
Long-range modeling SCROLLS LongT5 Base QALT EM-T/H 37.9 / 36.6 #5 of 13 Archive leaderboard report
Long-range modeling SCROLLS LongT5 Base QMSum 33.9 / 11.0 / 22.8 #5 of 13 Archive leaderboard report
Long-range modeling SCROLLS LongT5 Base Qspr 46.6 #5 of 13 Archive leaderboard report
Long-range modeling SCROLLS LongT5 Base SumScr 34.8 / 9.6 / 21.1 #5 of 13 Archive leaderboard report
Multi-Document Summarization Multi-News LongT5 ROUGE-1 48.17 #2 of 6 Archive leaderboard report
Multi-Document Summarization Multi-News LongT5 ROUGE-2 19.43 #2 of 6 Archive leaderboard report
Multi-Document Summarization Multi-News LongT5 ROUGE-SU4 24.94 #2 of 6 Archive leaderboard report
Text Summarization Arxiv HEP-TH citation graph LongT5 ROUGE-1 48.35 #9 of 28 Archive leaderboard report
Text Summarization Arxiv HEP-TH citation graph LongT5 ROUGE-2 21.92 #9 of 28 Archive leaderboard report
Text Summarization Arxiv HEP-TH citation graph LongT5 ROUGE-L 44.27 #9 of 28 Archive leaderboard report
Text Summarization BigPatent LongT5 ROUGE-1 76.87 #1 of 2 Archive leaderboard report
Text Summarization BigPatent LongT5 ROUGE-2 66.06 #1 of 2 Archive leaderboard report
Text Summarization BigPatent LongT5 ROUGE-L 70.76 #1 of 2 Archive leaderboard report
Text Summarization Pubmed LongT5 ROUGE-1 50.23 #4 of 29 Archive leaderboard report
Text Summarization Pubmed LongT5 ROUGE-2 24.76 #4 of 29 Archive leaderboard report
Text Summarization Pubmed LongT5 ROUGE-L 46.67 #4 of 29 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

AdafactorAttentionAttention DropoutBPEDense ConnectionsDropoutGated Linear UnitInverse Square Root ScheduleLayer NormalizationLinear LayerMulti-Head AttentionResidual ConnectionSentencePieceSoftmaxT5

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