Papers › DeepStruct: Pretraining of Language Models for Structure Prediction

DeepStruct: Pretraining of Language Models for Structure Prediction

21 May 2022Findings (ACL) 2022 5arXiv:2205.10475archive 2025-07-28

Chenguang Wang, Xiao Liu, Zui Chen, Haoyun Hong, Jie Tang, Dawn Song

We introduce a method for improving the structural understanding abilities of language models. Unlike previous approaches that finetune the models with task-specific augmentation, we pretrain language models on a collection of task-agnostic corpora to generate structures from text. Our structure pretraining enables zero-shot transfer of the learned knowledge that models have about the structure tasks. We study the performance of this approach on 28 datasets, spanning 10 structure prediction tasks including open information extraction, joint entity and relation extraction, named entity recognition, relation classification, semantic role labeling, event extraction, coreference resolution, factual probe, intent detection, and dialogue state tracking. We further enhance the pretraining with the task-specific training sets. We show that a 10B parameter language model transfers non-trivially to most tasks and obtains state-of-the-art performance on 21 of 28 datasets that we evaluate.

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 13 code samples harvested from 1 repository linked to this paper: 11 ran (of which 0 constructed an object rather than computing a result; 10 with no instrument failure: 0 honoured, 1 violated, 9 with no contract checked; 1 where Syntology's instrument failed) · 2 unverified. Where Syntology's instrument failed: 1 where our contract draft was wrong; failures of Syntology's instrument, not of the code.

By repository: official repository: 13 samples from 1 repository, 11 ran (10 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.

Levels written between 2026-09-22 and 2026-09-25 (UTC): the dates Syntology's graph wrote each sample's verification level, not when the code executed.

cgraywang/deepstruct officialmentioned in papermentioned on GitHubpytorchApache-2.0Syntology: 13 samples, 11 ran (1 instrument-failed) · 2 unverified 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

13 samples harvested: 11 ran (of which 0 constructed an object rather than computing a result; 10 with no instrument failure: 0 honoured, 1 violated, 9 with no contract checked; 1 where Syntology's instrument failed) · 2 unverified. “Where Syntology's instrument failed” counts failures 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 · violated contract
1ran · our draft was wrong
9ran
2unverified

Licence: 3 of the 13 samples are pointer only, meaning Syntology does not serve that copy's text. For 3 of them, this paper's copy is under a permissive licence and cleared, but the stored copy of the same code that Syntology serves is not; the licence label's tooltip gives Syntology's recorded reason. This page shows no code text for any sample; each one links to its file in the repository.

Harvested from cgraywang/deepstruct. “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.

A function or class name links to its page when the name occurs in the harvested code of at least 5 papers; the name is a string, not an identity. A census tag appears when Syntology's census ran implementations with that name on one shared input per bucket: it gives how many were compared, from how many papers, and how many distinct outputs they produced, and names this sample's bucket when this sample was one of them.

cgraywang/deepstruct official repository 13 samples, 11 ran (1 instrument-failed) · 2 unverified
FlattenList src/dataset_processing/datasets.py ran fingerprinted Apache-2.0 (permissive) · c1be0c65b775ba03 · report fixture: signature L:list/1/int · returned list · level recorded 2026-09-24
b_cubed in 12 papers src/dataset_processing/coreference_metrics.py ran Apache-2.0 (permissive) here; the stored copy is pointer only · 2d71e2917845b6ac · report fixture: signature clusters:list/1/list, mention_to_gold:dict/0/? · returned tuple · level recorded 2026-09-24
f1 in 57 papers src/dataset_processing/coreference_metrics.py ran Apache-2.0 (permissive) here; the stored copy is pointer only · 5dc251a1bb16c94f · report fixture: signature p_num:int/0/int, p_den:int/0/int, r_num:int/0/int, r_den:int/0/int, beta:int/0/int · returned float · level recorded 2026-09-25
fix in 7 papers src/dataset_processing/datasets.py ran fingerprinted Apache-2.0 (permissive) · 6ad7e61fdd3bfef8 · report fixture: signature string:str/0/str · returned str · level recorded 2026-09-24
get_avg_results src/dataset_processing/evaluate.py ran Apache-2.0 (permissive) · 3e9c0f90d3b4010c · report fixture: signature results:list/1/dict · returned dict · level recorded 2026-09-24
insertSpace in 6 papers src/data_scripts/multi_woz_create_data.py ran · our draft was wrong fingerprinted Apache-2.0 (permissive) · d7503a387ec4bc30 · report fixture: signature token:str/0/str, text:str/0/str · returned str · level recorded 2026-09-22
is_ascii in 11 papers src/data_scripts/multi_woz_create_data.py ran · violated contract Apache-2.0 (permissive) · 1b465b9239cb631c · report fixture: signature s:str/0/str · returned bool · level recorded 2026-09-22
oie_fix src/dataset_processing/output_formats.py ran fingerprinted Apache-2.0 (permissive) · b1f7144387836c66 · report fixture: signature string:str/0/str · returned str · level recorded 2026-09-24
register_dataset in 26 papers src/dataset_processing/datasets.py ran Apache-2.0 (permissive) · 404087a3ff94d6e0 · report fixture: signature dataset_class:callable/0/? · returned type · level recorded 2026-09-24
register_input_format src/dataset_processing/input_formats.py ran Apache-2.0 (permissive) · 4afe64752662fe3b · report fixture: signature format_class:callable/0/? · returned type · level recorded 2026-09-24
register_output_format src/dataset_processing/output_formats.py ran Apache-2.0 (permissive) · ba71a83cd3836e77 · report fixture: signature format_class:callable/0/? · returned type · level recorded 2026-09-24
evaluate_documents in 5 papers src/dataset_processing/coreference_metrics.py unverified Apache-2.0 (permissive) here; the stored copy is pointer only · 44ea1242eeeb38ac · report

Tasks

Coreference ResolutionDialogue State TrackingEvent ExtractionFactual probeIntent DetectionJoint Entity and Relation ExtractionLanguage ModelingLanguage ModellingNamed Entity RecognitionNamed Entity Recognition (NER)Open Information ExtractionPredictionRelation ClassificationRelation ExtractionSemantic Role Labelingcoreference-resolutionnamed-entity-recognition

1 archive task tag without a task page not shown.

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Coreference Resolution CoNLL12 DeepStruct multi-task w/ finetune Average F1 73.1 #1 of 2 Archive leaderboard report
Coreference Resolution CoNLL12 DeepStruct multi-task w/ finetune B3 71.3 #1 of 2 Archive leaderboard report
Coreference Resolution CoNLL12 DeepStruct multi-task w/ finetune CEAFϕ4 73.1 #1 of 2 Archive leaderboard report
Coreference Resolution CoNLL12 DeepStruct multi-task w/ finetune MUC 74.9 #1 of 2 Archive leaderboard report
Coreference Resolution CoNLL12 DeepStruct multi-task Average F1 60.6 #2 of 2 Archive leaderboard report
Coreference Resolution CoNLL12 DeepStruct multi-task B3 57.7 #2 of 2 Archive leaderboard report
Coreference Resolution CoNLL12 DeepStruct multi-task CEAFϕ4 60.2 #2 of 2 Archive leaderboard report
Coreference Resolution CoNLL12 DeepStruct multi-task MUC 63.9 #2 of 2 Archive leaderboard report
Dialogue State Tracking MULTIWOZ 2.1 DeepStruct multi-task w/ finetune Joint Acc 54.2 #1 of 4 Archive leaderboard report
Dialogue State Tracking MULTIWOZ 2.1 DeepStruct multi-task Joint Acc 53.5 #2 of 4 Archive leaderboard report
Event Extraction ACE2005 DeepStruct multi-task Argument Cl 63.9 #1 of 4 Archive leaderboard report
Event Extraction ACE2005 DeepStruct multi-task Argument Id 67.5 #1 of 4 Archive leaderboard report
Event Extraction ACE2005 DeepStruct multi-task Trigger Cl 69.2 #1 of 4 Archive leaderboard report
Event Extraction ACE2005 DeepStruct multi-task Trigger Id 72.7 #1 of 4 Archive leaderboard report
Event Extraction ACE2005 DeepStruct multi-task w/ finetune Argument Cl 56.2 #2 of 4 Archive leaderboard report
Event Extraction ACE2005 DeepStruct multi-task w/ finetune Argument Id 59.4 #2 of 4 Archive leaderboard report
Event Extraction ACE2005 DeepStruct multi-task w/ finetune Trigger Cl 69.8 #2 of 4 Archive leaderboard report
Event Extraction ACE2005 DeepStruct multi-task w/ finetune Trigger Id 73.5 #2 of 4 Archive leaderboard report
Joint Entity and Relation Extraction ACE2005 DeepStruct multi-task Entity F1 90.2 #1 of 3 Archive leaderboard report
Joint Entity and Relation Extraction ACE2005 DeepStruct multi-task Relation F1 58.9 #1 of 3 Archive leaderboard report
Joint Entity and Relation Extraction ACE2005 DeepStruct multi-task w/ finetune Entity F1 90.0 #2 of 3 Archive leaderboard report
Joint Entity and Relation Extraction ACE2005 DeepStruct multi-task w/ finetune Relation F1 66.8 #2 of 3 Archive leaderboard report
Joint Entity and Relation Extraction ACE2005 Deepstruct zero-shot Entity F1 31.8 #3 of 3 Archive leaderboard report
Joint Entity and Relation Extraction ACE2005 Deepstruct zero-shot Relation F1 5.3 #3 of 3 Archive leaderboard report
Joint Entity and Relation Extraction ADE Corpus DeepStruct multi-task w/ finetune Entity F1 91.1 #1 of 3 Archive leaderboard report
Joint Entity and Relation Extraction ADE Corpus DeepStruct multi-task w/ finetune Relation F1 83.8 #1 of 3 Archive leaderboard report
Joint Entity and Relation Extraction ADE Corpus DeepStruct multi-task Entity F1 90.5 #2 of 3 Archive leaderboard report
Joint Entity and Relation Extraction ADE Corpus DeepStruct multi-task Relation F1 83.6 #2 of 3 Archive leaderboard report
Joint Entity and Relation Extraction ADE Corpus Deepstruct zero-shot Entity F1 60.7 #3 of 3 Archive leaderboard report
Joint Entity and Relation Extraction ADE Corpus Deepstruct zero-shot Relation F1 10.6 #3 of 3 Archive leaderboard report
Joint Entity and Relation Extraction CoNLL04 DeepStruct multi-task w/ finetune Entity F1 90.7 #1 of 3 Archive leaderboard report
Joint Entity and Relation Extraction CoNLL04 DeepStruct multi-task w/ finetune Relation F1 78.3 #1 of 3 Archive leaderboard report
Joint Entity and Relation Extraction CoNLL04 DeepStruct multi-task Entity F1 88.4 #2 of 3 Archive leaderboard report
Joint Entity and Relation Extraction CoNLL04 DeepStruct multi-task Relation F1 72.8 #2 of 3 Archive leaderboard report
Joint Entity and Relation Extraction CoNLL04 Deepstruct zero-shot Entity F1 48.3 #3 of 3 Archive leaderboard report
Joint Entity and Relation Extraction CoNLL04 Deepstruct zero-shot Relation F1 25.8 #3 of 3 Archive leaderboard report
Joint Entity and Relation Extraction NYT DeepStruct multi-task w/ finetune Entity F1 95.9 #3 of 5 Archive leaderboard report
Joint Entity and Relation Extraction NYT DeepStruct multi-task w/ finetune Relation F1 93.3 #3 of 5 Archive leaderboard report
Joint Entity and Relation Extraction NYT DeepStruct multi-task Entity F1 95.4 #4 of 5 Archive leaderboard report
Joint Entity and Relation Extraction NYT DeepStruct multi-task Relation F1 93.7 #4 of 5 Archive leaderboard report
Joint Entity and Relation Extraction NYT Deepstruct zero-shot Entity F1 60.5 #5 of 5 Archive leaderboard report
Joint Entity and Relation Extraction NYT Deepstruct zero-shot Relation F1 28.6 #5 of 5 Archive leaderboard report
Named Entity Recognition (NER) ACE2005 DeepStruct multi-task w/ finetune F1 86.9 #1 of 2 Archive leaderboard report
Named Entity Recognition (NER) ACE2005 Deepstruct zero-shot F1 28.1 #2 of 2 Archive leaderboard report
Named Entity Recognition (NER) CoNLL03 DeepStruct multi-task F1 93.1 #2 of 5 Archive leaderboard report
Named Entity Recognition (NER) CoNLL03 DeepStruct multi-task w/ finetune F1 93.0 #3 of 5 Archive leaderboard report
Named Entity Recognition (NER) CoNLL03 Deepstruct zero-shot F1 44.4 #4 of 5 Archive leaderboard report
Named Entity Recognition (NER) GENIA DeepStruct multi-task w/ finetune F1 80.8 #1 of 14 Archive leaderboard report
Named Entity Recognition (NER) GENIA DeepStruct multi-task F1 80.2 #3 of 14 Archive leaderboard report
Named Entity Recognition (NER) GENIA Deepstruct zero-shot F1 47.2 #14 of 14 Archive leaderboard report
Named Entity Recognition (NER) OntoNotes DeepStruct multi-task w/ finetune F1 87.8 #1 of 3 Archive leaderboard report
Named Entity Recognition (NER) OntoNotes DeepStruct multi-task F1 87.6 #2 of 3 Archive leaderboard report
Named Entity Recognition (NER) OntoNotes Deepstruct zero-shot F1 2.5 #3 of 3 Archive leaderboard report
Open Information Extraction NYT Deepstruct zero-shot F1 28.9 #1 of 4 Archive leaderboard report
Open Information Extraction NYT DeepStruct multi-task F1 43.6 #2 of 4 Archive leaderboard report
Open Information Extraction NYT DeepStruct multi-task w/ finetune F1 45.0 #3 of 4 Archive leaderboard report
Open Information Extraction OIE2016 DeepStruct multi-task w/ finetune F1 71.3 #2 of 12 Archive leaderboard report
Open Information Extraction OIE2016 Deepstruct multi-task F1 71.2 #3 of 12 Archive leaderboard report
Open Information Extraction OIE2016 Deepstruct zero-shot F1 28.1 #12 of 12 Archive leaderboard report
Open Information Extraction Penn Treebank Deepstruct zero-shot F1 51 #1 of 4 Archive leaderboard report
Open Information Extraction Penn Treebank DeepStruct multi-task F1 54.5 #2 of 4 Archive leaderboard report
Open Information Extraction Penn Treebank DeepStruct multi-task w/ finetune F1 45,1 #4 of 4 Archive leaderboard report
Open Information Extraction Web Deepstruct zero-shot F1 43.8 #1 of 4 Archive leaderboard report
Open Information Extraction Web DeepStruct multi-task w/ finetune F1 49.1 #2 of 4 Archive leaderboard report
Open Information Extraction Web DeepStruct multi-task F1 50.8 #3 of 4 Archive leaderboard report
Relation Classification FewRel DeepStruct multi-task w/ finetune F1 (10-way 1-shot) 97.8 #1 of 5 Archive leaderboard report
Relation Classification FewRel DeepStruct multi-task w/ finetune F1 (10-way 5-shot) 99.8 #1 of 5 Archive leaderboard report
Relation Classification FewRel DeepStruct multi-task w/ finetune F1 (5-way 1-shot) 98.4 #1 of 5 Archive leaderboard report
Relation Classification FewRel DeepStruct multi-task w/ finetune F1 (5-way 5-shot 100 #1 of 5 Archive leaderboard report
Relation Classification FewRel DeepStruct multi-task F1 (10-way 1-shot) 92.2 #2 of 5 Archive leaderboard report
Relation Classification FewRel DeepStruct multi-task F1 (10-way 5-shot) 94.6 #2 of 5 Archive leaderboard report
Relation Classification FewRel DeepStruct multi-task F1 (5-way 1-shot) 93.6 #2 of 5 Archive leaderboard report
Relation Classification FewRel DeepStruct multi-task F1 (5-way 5-shot 96.4 #2 of 5 Archive leaderboard report
Relation Classification FewRel Deepstruct zero-shot F1 (10-way 1-shot) 67.6 #3 of 5 Archive leaderboard report
Relation Classification FewRel Deepstruct zero-shot F1 (10-way 5-shot) 66.4 #3 of 5 Archive leaderboard report
Relation Classification FewRel Deepstruct zero-shot F1 (5-way 1-shot) 72.4 #3 of 5 Archive leaderboard report
Relation Classification FewRel Deepstruct zero-shot F1 (5-way 5-shot 70.8 #3 of 5 Archive leaderboard report
Relation Classification TACRED Deepstruct zero-shot F1 36.1 #1 of 17 Archive leaderboard report
Relation Classification TACRED DeepStruct multi-task F1 74.9 #15 of 17 Archive leaderboard report
Relation Classification TACRED DeepStruct multi-task w/ finetune F1 76.8 #17 of 17 Archive leaderboard report
Relation Extraction TACRED DeepStruct multi-task w/ finetune F1 76.8 #2 of 40 Archive leaderboard report
Semantic Role Labeling CoNLL05 Brown DeepStruct multi-task w/ finetune F1 92.1 #1 of 2 Archive leaderboard report
Semantic Role Labeling CoNLL05 Brown DeepStruct multi-task F1 92.0 #2 of 2 Archive leaderboard report
Semantic Role Labeling CoNLL05 WSJ DeepStruct multi-task F1 95.5 #1 of 2 Archive leaderboard report
Semantic Role Labeling CoNLL05 WSJ DeepStruct multi-task w/ finetune F1 95.2 #2 of 2 Archive leaderboard report
Semantic Role Labeling CoNLL12 DeepStruct multi-task F1 97.2 #1 of 2 Archive leaderboard report
Semantic Role Labeling CoNLL12 DeepStruct multi-task w/ finetune F1 96.0 #2 of 2 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.

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