SpanBERT: Improving Pre-training by Representing and Predicting Spans
Open accessThe DOI matches its Crossref record; 53 of the 53 references it cites were found in a registry or an index.
Abstract
We present SpanBERT, a pre-training method that is designed to better represent and predict spans of text. Our approach extends BERT by (1) masking contiguous random spans, rather than random tokens, and (2) training the span boundary representations to predict the entire content of the masked span, without relying on the individual token representations within it. SpanBERT consistently outperforms BERT and our better-tuned baselines, with substantial gains on span selection tasks such as question answering and coreference resolution. In particular, with the same training data and model size as BERT large , our single model obtains 94.6% and 88.7% F1 on SQuAD 1.1 and 2.0 respectively. We also achieve a new state of the art on the OntoNotes coreference resolution task (79.6% F1), strong performance on the TACRED relation extraction benchmark, and even gains on GLUE. 1
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“XLM [46] performs MLM on a concatenation of parallel bilingual sentence pairs, called Translation Language Modeling (TLM).”
From Pre-trained models for natural language processing: A survey (2020) · cites this paper as [46]
Experimental — accuracy not yet measuredIn the citing paper
“Conneau and Lample [46] tried to initialize the entire encoder and decoder by a multilingual pre-trained BERT model and showed a significant improvement could be achieved on unsupervised MT and English-Romanian supervised MT.”
From Pre-trained models for natural language processing: A survey (2020) · cites this paper as [46]
In the citing paper
“In the experiment, this approach can surpass the BERT-style pretraining proposed by Conneau and Lample [46] both on unsupervised MT and English-Romanian supervised MT.”
From Pre-trained models for natural language processing: A survey (2020) · cites this paper as [46]
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A. N., Kaiser, L., & Polosukhin, I. (2017) · arXiv · doi:10.48550/arxiv.1706.03762 · Cited by 120,404
Adam: A Method for Stochastic Optimization
Kingma, D. P., & Ba, J. (2014) · arXiv · doi:10.48550/arxiv.1412.6980 · Cited by 98,946
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
Devlin, J., Chang, M.-W., Lee, K., & Toutanova, K. (2019) · Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers) · doi:10.18653/v1/n19-1423 · Cited by 54,325
RoBERTa: A Robustly Optimized BERT Pretraining Approach
Liu, Y., Ott, M., Goyal, N., Du, J., Joshi, M., Chen, D., Levy, O., Lewis, M., Zettlemoyer, L., & Stoyanov, V. (2019) · arXiv · doi:10.48550/arxiv.1907.11692 · Cited by 16,515
Self-citationDecoupled Weight Decay Regularization
Loshchilov, I., & Hutter, F. (2017) · arXiv · doi:10.48550/arxiv.1711.05101 · Cited by 12,522
Deep Contextualized Word Representations
Peters, M., Neumann, M., Iyyer, M., Gardner, M., Clark, C., Lee, K., & Zettlemoyer, L. (2018) · Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long Papers) · doi:10.18653/v1/n18-1202 · Cited by 11,983
Self-citationRecursive Deep Models for Semantic Compositionality Over a Sentiment Treebank
Socher, R., Perelygin, A., Wu, J., Chuang, J., Manning, C. D., Ng, A., & Potts, C. (2013) · Proceedings of the 2013 Conference on Empirical Methods in Natural Language Processing · doi:10.18653/v1/d13-1170 · Cited by 6,878 · cited 4×
“…Understanding Evaluation (GLUE) benchmark (Wang et al., 2019) consists of 9 sentence-level classification tasks: • Two sentence-level classification tasks including CoLA (Warstadt et al., 2018) for evaluating linguistic acceptability and SST-2 (Socher et al., 2013) for sentiment classification.”
BackgroundInfluentialXLNet: Generalized Autoregressive Pretraining for Language Understanding
Yang, Z., Dai, Z., Yang, Y., Carbonell, J., Salakhutdinov, R., & Le, Q. V. (2019) · arXiv · doi:10.48550/arxiv.1906.08237 · Cited by 6,687 · cited 4×
“While others show the benefits of adding more data (Yang et al., 2019) and increasing model size (Lample and Conneau, 2019), this work demonstrates the importance of designing good pretraining tasks and objectives, which can also have a significant impact.”
BackgroundMethodInfluentialSQuAD: 100,000+ Questions for Machine Comprehension of Text
Rajpurkar, P., Zhang, J., Lopyrev, K., & Liang, P. (2016) · Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing · doi:10.18653/v1/d16-1264 · Cited by 6,208
Universal Language Model Fine-tuning for Text Classification
Howard, J., & Ruder, S. (2018) · Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · doi:10.18653/v1/p18-1031 · Cited by 4,100 · cited 2×
“, 2018) have had immense impact on NLP lately, particularly as methods for initializing a large model before fine-tuning it for a specific task (Howard and Ruder, 2018; Radford et al., 2018; Devlin et al., 2019).”
Method
Research works printed but not linked: 7
- Jimmy Lei Ba, Jamie Ryan Kiros, and Geoffrey E Hinton. 2016. Layer normalization. arXiv preprint arXiv:1607.06450 . · arXiv
- Andrew M Dai and Quoc V Le. 2015. Semisupervised sequence learning. In Advances in Neural Information Processing Systems (NIPS) , pages 3079-3087.
- Adam Fisch, Alon Talmor, Robin Jia, Minjoon Seo, Eunsol Choi, and Danqi Chen. 2019. MRQA 2019 shared task: Evaluating generalization in reading comprehension. In Proceedings of 2nd Machine Reading for Reading Comprehension (MRQA) Workshop at EMNLP .
- Dan Hendrycks and Kevin Gimpel. 2016. Gaussian error linear units (gelus). arXiv preprint arXiv:1606.08415 . · arXiv
- Mandar Joshi, Eunsol Choi, Omer Levy, Daniel Weld, and Luke Zettlemoyer. 2019a. pair2vec: Compositional word-pair embeddings for crosssentence inference. In North American Association for Computational Linguistics (NAACL) , pages 3597-3608.
- Liu Yinhan, 2019, arxiv preprint arXiv:1907.11692 · arXiv
- Mitchell Stern, Noam Shazeer, and Jakob Uszkoreit. 2018. Blockwise parallel decoding for deep autoregressive models. In Advances in Neural Information Processing Systems (NIPS) .
Cited by
Items 401–500 of 1,012
Ling, S., Salazar, J., Liu, Y., & Kirchhoff, K. (2020) · The Speaker and Language Recognition Workshop (Odyssey) · doi:10.21437/odyssey.2020-2 · Cited by 33
“However, we follow SpanBERT [26] in masking spans of inputs instead, to induce higher-level representations by increasing the difficulty of the task (as speech frames are highly correlated over time).”
BackgroundSelf-citationExploring Question-Specific Rewards for Generating Deep Questions
Xie, Y., Pan, L., Wang, D., Kan, M., & Feng, Y. (2020) · International Conference on Computational Linguistics · doi:10.18653/v1/2020.coling-main.228 · Cited by 31
Unified Semantic Typing with Meaningful Label Inference
Huang, J. Y., Li, B., Xu, J., & Chen, M. (2022) · Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (NAACL-HLT) · doi:10.18653/v1/2022.naacl-main.190 · Cited by 40
Jeong, Y., & Kim, E. (2022) · IEEE Access · doi:10.1109/access.2022.3180830 · Cited by 45
“Span-BERT [20] improved MLM to predict spans instead of tokens that are relatively easy to solve.”
Portelli, B., Lenzi, E., Chersoni, E., Serra, G., & Santus, E. (2021) · Conference of the European Chapter of the Association for Computational Linguistics · doi:10.18653/v1/2021.eacl-main.149 · Cited by 29
“Second, whether a model trained to predict coherent spans of text instead of single words can achieve a better performance (Joshi et al., 2019), since our goal is to identify the groups of tokens corresponding to ADEs as precisely as possible.”
BackgroundContextualized Representations Using Textual Encyclopedic KnowledgePreprint
Joshi, M., Lee, K., Luan, Y., & Toutanova, K. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2004.12006 · Cited by 34
“We follow improvements in SpanBERT (Joshi et al., 2020) and mask spans with lengths sampled from a geometric”
BackgroundMethodResultInfluentialSelf-citationZhang, C., Wang, Q., Li, Y., Teng, A., Hu, G., Wuyun, Q., & Zheng, W. H. (2024) · Biomolecules · doi:10.3390/biom14121531 · Cited by 38
Probabilistically Masked Language Model Capable of Autoregressive Generation in Arbitrary Word Order
Liao, Y., Jiang, X., & Liu, Q. (2020) · Proceedings of the conference - Association for Computational Linguistics. Meeting · doi:10.18653/v1/2020.acl-main.24 · Cited by 45
Accurate generation of trigger-action programs with domain-adapted sequence-to-sequence learning
Yusuf, I. N. B., Jiang, L., & Lo, D. (2022) · IEEE International Conference on Program Comprehension (ICPC) · doi:10.1145/3524610.3527922 · Cited by 26
“We choose CodeBERT and RoBERTa as both models have been widely used in various studies [26, 27, 28].”
MethodA Cheaper and Better Diffusion Language Model with Soft-Masked Noise
Chen, J., Zhang, A., Li, M., Smola, A. J., & Yang, D. (2023) · Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/v1/2023.emnlp-main.289 · Cited by 46
Opportunities and Challenges of Foundation Models in Industrial Manufacturing
Ayyat, M., Osman, M., & Nadeem, T. (2025) · IEEE Access · doi:10.1109/access.2025.3589777 · Cited by 31
“Lastly, SpanBERT [69] extended BERT’s masking to spans rather than tokens and trained representations to predict masked span content, achieving further gains in span selection tasks such as question answering and coreference resolution.”
An Embarrassingly Simple Model for Dialogue Relation Extraction
Xue, F., Sun, A., Zhang, H., Ni, J., & Chng, E. S. (2022) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp43922.2022.9747486 · Cited by 27
Select, Substitute, Search: A New Benchmark for Knowledge-Augmented Visual Question Answering
Jain, A., Kothyari, M., Kumar, V., Jyothi, P., Ramakrishnan, G., & Chakrabarti, S. (2021) · International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR) · doi:10.1145/3404835.3463259 · Cited by 52
Moving on from OntoNotes: Coreference Resolution Model Transfer
Xia, P., & Durme, B. V. (2021) · Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/v1/2021.emnlp-main.425 · Cited by 41
Huang, Y. J., & Kurohashi, S. (2021) · Conference of the European Chapter of the Association for Computational Linguistics · doi:10.18653/v1/2021.eacl-main.265 · Cited by 39
“For coreference resolution, we used the spanBERT-based (Joshi et al., 2020) version of the end-to-end coreference resolver proposed by Lee et al. (2017).”
MethodBoosting Low-Resource Biomedical QA via Entity-Aware Masking Strategies
Pergola, G., Kochkina, E., Gui, L., Liakata, M., & He, Y. (2021) · Conference of the European Chapter of the Association for Computational Linguistics · doi:10.18653/v1/2021.eacl-main.169 · Cited by 50
Song, D., Xu, J., Pang, J., & Huang, H. (2021) · Information Sciences · doi:10.1016/j.ins.2021.05.045 · Cited by 30
Exploiting Document Structures and Cluster Consistencies for Event Coreference Resolution
Tran, H. M., Phung, D., & Nguyen, T. H. (2021) · doi:10.18653/v1/2021.acl-long.374 · Cited by 33
“Event coreference resolution is broadly related to works on entity coreference resolution that aim to resolve nouns phrases/mentions for entities (Raghu-nathan et al., 2010; Ng, 2010; Durrett and Klein, 2013; Lee et al., 2017a; Joshi et al., 2019b,a).”
BackgroundExtensive evaluation of transformer-based architectures for adverse drug events extraction
Scaboro, S., Portelli, B., Chersoni, E., Santus, E., & Serra, G. (2023) · Knowledge-Based Systems · doi:10.1016/j.knosys.2023.110675 · Cited by 30
RefBERT: A Two-Stage Pre-trained Framework for Automatic Rename Refactoring
Liu, H., Wang, Y., Zhao, W., Xu, Y., Wang, J., Li, H., & Ji, R. (2023) · ACM SIGSOFT International Symposium on Software Testing and Analysis (ISSTA) · doi:10.1145/3597926.3598092 · Cited by 36
Improving Gender Fairness of Pre-Trained Language Models without Catastrophic Forgetting
Fatemi, Z., Chen, X., Liu, W., & Xiong, C. (2023) · Proceedings of the conference - Association for Computational Linguistics. Meeting · doi:10.18653/v1/2023.acl-short.108 · Cited by 62
Incivility Detection in Open Source Code Review and Issue Discussions
Ferreira, I., Rafiq, A., & Cheng, J. (2022) · Journal of Systems and Software · doi:10.1016/j.jss.2023.111935 · Cited by 32
“SpanBERT [41] is a pre-trained method to better represent and predict spans of text.”
BackgroundMethodPapadopoulos, D., Papadakis, N. K., & Λίτκε, Α. (2020) · Applied Sciences · doi:10.3390/app10165630 · Cited by 25
“end-to-end coreference resolution model [68] using Span-BERT embeddings [23].”
MethodDependency-based syntax-aware word representations
Zhang, M., Li, Z., Fu, G., & Zhang, M. (2020) · Artificial Intelligence · doi:10.1016/j.artint.2020.103427 · Cited by 26
Transformer-Based Extractive Social Media Question Answering on TweetQA
Butt, S., Ashraf, N., Fahim, H., Sidorov, G., & Gelbukh, A. F. (2021) · Computación y Sistemas · doi:10.13053/cys-25-1-3897 · Cited by 25
Smith, E., Papadopoulos, D., Braschler, M., & Stockinger, K. (2021) · Information Systems · doi:10.1016/j.is.2021.101938 · Cited by 30
Deep learning models for spatial relation extraction in text
Wu, K., Zhang, X., Dang, Y., & Ye, P. (2022) · Geo-spatial Information Science · doi:10.1080/10095020.2022.2076619 · Cited by 26
“…length based on geometric distribution sampling combines continuous tokens into span, and predicting geographic entities based on span can capture the boundaries of geographic entities more accurately and express the potential semantics of geographic entities more realistically (Joshi et al. 2020).”
BackgroundTele-Knowledge Pre-training for Fault Analysis
Chen, Z., Wen, Z., Huang, Y., Chen, M., Geng, Y., Yu, H., Bi, Z., Zhang, Y., Yao, Z., Song, W., et al. (2023) · IEEE International Conference on Data Engineering (ICDE) · doi:10.1109/icde55515.2023.00265 · Cited by 44
“Works like ERNIE [10], spanBERT [42] and StructBERT [43] have extended BERT [9] by incorporating novel token-level and sentence-level pre-training tasks.”
MethodZhang, S., Wan, D., & Bansal, M. (2023) · Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · doi:10.18653/v1/2023.acl-long.120 · Cited by 10
LingMess: Linguistically Informed Multi Expert Scorers for Coreference Resolution
Otmazgin, S., Cattan, A., & Goldberg, Y. (2023) · Conference of the European Chapter of the Association for Computational Linguistics · doi:10.18653/v1/2023.eacl-main.202 · Cited by 62
Self-Supervised Test-Time Learning for Reading Comprehension
Banerjee, P., Gokhale, T., & Baral, C. R. (2021) · Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (NAACL-HLT) · doi:10.18653/v1/2021.naacl-main.95 · Cited by 42
ELECTRA for Neural Coreference Resolution in Italian
Guarasci, R., Minutolo, A., Damiano, E., Pietro, G. D., Fujita, H., & Esposito, M. (2021) · IEEE Access · doi:10.1109/access.2021.3105278 · Cited by 24
“est language models like ELMo (Embeddings from Language Models) [8], [9] and BERT (Bidirectional Encoder Representations from Transformers) [10]–[12] have been proposed and they have brought significant performance gains over all previous approaches [13].”
ResultSentiment Knowledge Enhanced Self-supervised Learning for Multimodal Sentiment Analysis
Fan, Q., Han, J., He, Y., Zheng, T., & Zheng, G. (2023) · Findings of the Association for Computational Linguistics: ACL (Findings of ACL) · doi:10.18653/v1/2023.findings-acl.821 · Cited by 35
X-News dataset for online news categorization
Yousafzai, S. N., Shahbaz, H., Ali, A., Qamar, A., Nasir, I. M., Tehsin, S., & Damaševičius, R. (2024) · International Journal of Intelligent Computing and Cybernetics · doi:10.1108/ijicc-04-2024-0184 · Cited by 24
A Survey of Few-Shot Learning for Biomedical Time Series
Li, C., Denison, T., & Zhu, T. (2024) · IEEE Reviews in Biomedical Engineering · doi:10.1109/rbme.2024.3492381 · Cited by 40
“Then, multi-source pre-training with unlabeled data is performed using different self-supervised learning strategies: SimCLR [104], Wav2Vec 2.0 [105], MLM [106] and Span-MLM [107].”
InfluentialA survey on cutting-edge relation extraction techniques based on language models
Diaz-Garcia, J. A., & Amador, J. (2025) · Artificial Intelligence Review · doi:10.1007/s10462-025-11280-0 · Cited by 47
Domain-Adaptive Pretraining Methods for Dialogue Understanding
Wu, H., Xu, K., Song, L., Jin, L., Zhang, H., & Song, L. (2021) · doi:10.18653/v1/2021.acl-short.84 · Cited by 28
“Recent advances in pretraining methods (Devlin et al., 2019; Joshi et al., 2020; Yang et al., 2019)”
InfluentialAre Neural Ranking Models Robust?
Wu, C. H., Zhang, R., Guo, J., Fan, Y., & Cheng, X. (2022) · ACM Transactions on Information Systems · doi:10.1145/3534928 · Cited by 51
“, pre-trained model) can achieve promising performance on some IR tasks [28, 43, 56].”
BackgroundSETTI: A S elf-supervised Adv E rsarial Malware De T ection Archi T ecture in an I oT Environment
Golmaryami, M., Taheri, R., Pooranian, Z., Shojafar, M., & Xiao, P. (2022) · ACM Transactions on Multimedia Computing Communications and Applications · doi:10.1145/3536425 · Cited by 28
Hierarchical Graph Convolutional Networks for Structured Long Document Classification
Liu, T., Hu, Y., Wang, B., Sun, Y., Gao, J., & Yin, B. (2022) · IEEE Transactions on Neural Networks and Learning Systems · doi:10.1109/tnnls.2022.3185295 · Cited by 34
“3) Selection Mechanism: It identifies a smaller subset of relevant contexts by training a judge model to feed in the transformer with optional iterations [39], [40].”
BackgroundFine-tuning pretrained transformer encoders for sequence-to-sequence learning
Bao, H., Li, D., Wang, W., Yang, N., Piao, S., & Wei, F. (2023) · International Journal of Machine Learning and Cybernetics · doi:10.1007/s13042-023-01992-6 · Cited by 27
“Among these tasks, MLM has become one of the most successful pretraining targets in natural language processing, and has been widely employed in subsequent research work[6–8, 10, 24, 37].”
Guo, Y. F., Xie, Z., Chen, X., Chen, H., Wang, L., Du, H., Wei, S., Zhao, Y., Li, Q., & Wu, G. (2024) · Neurocomputing · doi:10.1016/j.neucom.2024.127725 · Cited by 28
“However, this approach has been the subject of much debate in recent literature [44, 45, 46, 47].”
BackgroundTo Pretrain or Not to Pretrain: Examining the Benefits of Pretrainng on Resource Rich Tasks
Wang, S., Khabsa, M., & Ma, H. (2020) · Proceedings of the conference - Association for Computational Linguistics. Meeting · doi:10.18653/v1/2020.acl-main.200 · Cited by 26
Hamborg, F. (2020) · Annual Meeting of the Association for Computational Linguistics: Student Research Workshop (ACL SRW) · doi:10.18653/v1/2020.acl-srw.12 · Cited by 41
“Using SpanBERT, a pre-training method focused on spans rather than tokens, the performance is increased to F1=79.6 (Joshi et al., 2019).”
MethodInfluentialDoes Vision-and-Language Pretraining Improve Lexical Grounding?
Yun, T., Sun, C., & Pavlick, E. (2021) · Findings of the Association for Computational Linguistics: EMNLP 2021 · doi:10.18653/v1/2021.findings-emnlp.370 · Cited by 22
A BP Neural Network-Assisted Smart Decision Method for Education Quality
Zhang, D., & Chen, L. (2023) · IEEE Access · doi:10.1109/access.2023.3294804 · Cited by 22
“Therefore, the establishment of ideological and political education evaluation system in colleges must aim at promoting work in this field [11], [12].”
BackgroundEnhancing named entity recognition with external knowledge from large language model
Li, Q., Xie, T., Zhang, J., Ma, K., Su, J., Yang, K., & Wang, H. (2025) · Knowledge-Based Systems · doi:10.1016/j.knosys.2025.113471 · Cited by 23
What’s in a Name? Are BERT Named Entity Representations just as Good for any other Name?
Balasubramanian, S., Jain, N., Jindal, G., Awasthi, A., & Sarawagi, S. (2020) · Workshop on Representation Learning for NLP (RepL4NLP) · doi:10.18653/v1/2020.repl4nlp-1.24 · Cited by 27
Type B Reflexivization as an Unambiguous Testbed for Multilingual Multi-Task Gender Bias
González, A. V., Barrett, M., Hvingelby, R., Webster, K., & Søgaard, A. (2020) · Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/v1/2020.emnlp-main.209 · Cited by 31
Robust Retrieval Augmented Generation for Zero-shot Slot Filling
Glass, M., Rossiello, G., Chowdhury, M. F. M., & Gliozzo, A. (2021) · Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/v1/2021.emnlp-main.148 · Cited by 38
The CODI-CRAC 2021 Shared Task on Anaphora, Bridging, and Discourse Deixis in Dialogue
Khosla, S., Yu, J., Manuvinakurike, R., Ng, V., Poesio, M., Strube, M., & Rosé, C. P. (2021) · doi:10.18653/v1/2021.codi-sharedtask.1 · Cited by 37
“, 2012) has greatly improved in recent years (Wiseman et al., 2015; Lee et al., 2017, 2018; Kantor and Globerson, 2019; Joshi et al., 2020).”
BackgroundMethodNEREL: A Russian Dataset with Nested Named Entities, Relations and Events
Loukachevitch, N., Artemova, E., Huawei Noah’s Ark lab, R., Batura, T. V., Braslavski, P., Denisov, I., Ivanov, V. V., Manandhar, S., Wiseyak, U. S., Pugachev, A., et al. (2021) · International conference Recent advances in natural language processing · doi:10.26615/978-954-452-072-4_100 · Cited by 35
Improving Pretrained Language Model Fine-Tuning With Noise Stability Regularization
Hua, H., Li, X., Dou, D., Xu, C., & Luo, J. (2023) · IEEE Transactions on Neural Networks and Learning Systems · doi:10.1109/tnnls.2023.3330926 · Cited by 37
“L ARGE-SCALE pretrained language models (PLMs) have significantly boosted state-of-the-art performance on nat-ural language processing (NLP) tasks [1], [2], [3], [4], [5], [6].”
From the Detection of Toxic Spans in Online Discussions to the Analysis of Toxic-to-Civil Transfer
Pavlopoulos, J., Laugier, L., Xenos, A., Sorensen, J. S., & Androutsopoulos, I. (2022) · Proceedings of the conference - Association for Computational Linguistics. Meeting · doi:10.18653/v1/2022.acl-long.259 · Cited by 46
“, 2019) and SPANBERT (Joshi et al., 2020) for toxic spans (BERTSEQ, SPAN-BERT-SEQ).”
BackgroundMethodInfluentialCross-document Misinformation Detection based on Event Graph Reasoning
Wu, X., Huang, K.-H., Fung, Y. R., & Ji, H. (2022) · Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (NAACL-HLT) · doi:10.18653/v1/2022.naacl-main.40 · Cited by 47
BertSRC: transformer-based semantic relation classification
Lee, Y., Son, J., & Song, M. (2022) · BMC Medical Informatics and Decision Making · doi:10.1186/s12911-022-01977-5 · Cited by 25
JOIST: A Joint Speech and Text Streaming Model for ASR
Sainath, T. N., Prabhavalkar, R., Bapna, A., Zhang, Y., Huo, Z., Chen, Z., Li, B., Wang, W., & Strohman, T. (2023) · IEEE Spoken Language Technology Workshop (SLT) · doi:10.1109/slt54892.2023.10022774 · Cited by 39
“Following [15, 68], we mask 15% of the up-sampled text IDs with spans of length 5.”
MethodCaliFormer: Leveraging Unlabeled Measurements to Calibrate Sensors with Self-supervised Learning
Wang, H., Liu, Y., Zhao, C., He, J., Ding, W., & Chen, X. (2023) · ACM International Joint Conference on Pervasive and Ubiquitous Computing (UbiComp) · doi:10.1145/3594739.3612917 · Cited by 23
“4, we mask a high proportion of data with the Span Masking mechanism in unlabeled sensor measurement sequences to create a challenging condition in this phase [12].”
Self-citationEnriched Pre-trained Transformers for Joint Slot Filling and Intent Detection
Hardalov, M., Koychev, I. K., & Nakov, P. (2023) · International conference Recent advances in natural language processing · doi:10.26615/978-954-452-092-2_054 · Cited by 27
Exposing Shallow Heuristics of Relation Extraction Models with Challenge Data
Rosenman, S., Jacovi, A., & Goldberg, Y. (2020) · Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/v1/2020.emnlp-main.302 · Cited by 34
Coreference Resolution in Research Papers from Multiple Domains
Brack, A., Müller, D. U., Hoppe, A., & Ewerth, R. (2021) · Lecture notes in computer science · doi:10.1007/978-3-030-72113-8_6 · Cited by 25
What GPT Knows About Who is Who
Yang, X., Peynetti, E., Meerman, V., & Tanner, C. (2022) · Workshop on Insights from Negative Results in NLP (Insights) · doi:10.18653/v1/2022.insights-1.10 · Cited by 49
GNN-LM: Language Modeling based on Global Contexts via GNNPreprint
Meng, Y., Zong, S., Li, X., Sun, X., Zhang, T., Wu, F. Q., & Li, J. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2110.08743 · Cited by 35
“BERT (Devlin et al., 2018) proposed the Masked Language Modeling (MLM) pretraining paradigm to train a deep bidirectional Transformer model; RoBERTa (Liu et al., 2019) removed the Next Sentence Prediction (NSP) task in BERT; XLNet (Yang et al., 2019) generalized BERT pretraining to the autoregressive manner; Span-leve…”
BackgroundMethodFrustratingly Simple Pretraining Alternatives to Masked Language Modeling
Yamaguchi, A., Chrysostomou, G., Margatina, K., & Αλέτρας, Ν. (2021) · Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/v1/2021.emnlp-main.249 · Cited by 25
CARES: CAuse Recognition for Emotion in Suicide Notes
Ghosh, S. K., Roy, S., Ekbal, A., & Bhattacharyya, P. (2022) · Lecture notes in computer science · doi:10.1007/978-3-030-99739-7_15 · Cited by 23
“We finetune four pretrained transformer-based models, Multilingual Bidi-rectional Encoder Representations from Transformers (mBERT) Base [5], Span-BERT Base [11], RoBERTa Base [13], and XLM-RoBERTa (XLM-R) [4] models, and for each model, the training is done as follows: 1.”
MethodTiBERT: Tibetan Pre-trained Language Model
Liu, S., Deng, J., Sun, Y., & Zhao, X. (2022) · Conference proceedings/Conference proceedings - IEEE International Conference on Systems, Man, and Cybernetics · doi:10.1109/smc53654.2022.9945074 · Cited by 26
“Studies have shown that masking whole words instead of individual sub-words can improve the performance of the model [24].”
BackgroundCan Pretrained English Language Models Benefit Non-English NLP Systems in Low-Resource Scenarios?
Chi, Z., Huang, H., Liu, L., Bai, Y., Gao, X., & Mao, X.-L. (2023) · IEEE/ACM Transactions on Audio Speech and Language Processing · doi:10.1109/taslp.2023.3267618 · Cited by 25
Machine learning in computational literary studies
Hatzel, H. O., Stiemer, H., Biemann, C., & Gius, E. (2023) · it - Information Technology · doi:10.1515/itit-2023-0041 · Cited by 45
“While recently, in the field of NLP, major advances have been made in terms of scores for coreference resolution systems [34, 94], they do not produce satisfactory results for longer documents in the literature domain (as noted, for example, by Perri et al. [52]) although approaches specifically for…”
BackgroundAlrayzah, A., Alsolami, F., & Saleh, M. (2023) · PeerJ Computer Science · doi:10.7717/peerj-cs.1633 · Cited by 28
Wang, J., Zhang, B., Wang, Y., Zhou, C., Vonsky, M. S., Mitrofanova, L. B., Zou, D., & Li, Q. (2024) · Computerized Medical Imaging and Graphics · doi:10.1016/j.compmedimag.2024.102339 · Cited by 26
Symptom-BERT: Enhancing Cancer Symptom Detection in EHR Clinical Notes
Zeinali, N., Albashayreh, A., Fan, W., & White, S. G. (2024) · Journal of Pain and Symptom Management · doi:10.1016/j.jpainsymman.2024.05.015 · Cited by 27
Bayesian Attention Belief NetworksPreprint
Zhang, S., Fan, X., Chen, B., & Zhou, M. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2106.05251 · Cited by 43
“In an unsupervised manner, this approach learns useful representations that benefit downstream tasks, achieving tremendous success in natural language processing (Devlin et al., 2018; Lan et al., 2019; Liu et al., 2019; Joshi et al., 2020; Radford et al., 2018; Yang et al., 2019), compute vision (Dosovitskiy et al.”
BackgroundOne Size Does Not Fit All: Multi-granularity Patch Generation for Better Automated Program Repair
Lin, B., Wang, S., Wen, M., Chen, L., & Mao, X. (2024) · ACM SIGSOFT International Symposium on Software Testing and Analysis (ISSTA) · doi:10.1145/3650212.3680381 · Cited by 40
“Operating at the individual token level restricts their capacity to grasp broader contextual information, including relationships between groups of words [28].”
Portelli, B., Scaboro, S., Tonino, R., Chersoni, E., Santus, E., & Serra, G. (2022) · Journal of Medical Internet Research · doi:10.2196/35115 · Cited by 22
“The Symptom Extraction module, in particular, consists of a deep-learning architecture that we created specifically for this task, based on SpanBERT [13], an extension of the bidirectional encoder representations from transformers (BERT) model, which is one of the state-of-the-art models for AE…”
UNIMO-2: End-to-End Unified Vision-Language Grounded Learning
Li, W., Gao, C., Niu, G., Xiao, X., Liu, H., Liu, J., Wu, H., & Wang, H. (2022) · Findings of the Association for Computational Linguistics: ACL (Findings of ACL) · doi:10.18653/v1/2022.findings-acl.251 · Cited by 32
Adapting Coreference Resolution Models through Active Learning
Yuan, M., Xia, P., May, C., Durme, B. V., & Boyd-Graber, J. L. (2022) · Proceedings of the conference - Association for Computational Linguistics. Meeting · doi:10.18653/v1/2022.acl-long.519 · Cited by 29
“, 2020) with SPANBERTLARGE-CASED (Joshi et al., 2020) encoder.”
InfluentialJin, X., & Wang, Y. (2023) · International Workshop on Semantic Evaluation (SemEval) · doi:10.18653/v1/2023.semeval-1.72 · Cited by 24
“BERT-Span (Joshi et al., 2020) is the variation of BERT that uses BERT to train span boundary representations, which are more suitable for entity boundary detection.”
MethodEliciting knowledge from language models with automatically generated continuous prompts
Chen, Y., Yang, G., Wang, D., & Li, D. (2023) · Expert Systems with Applications · doi:10.1016/j.eswa.2023.122327 · Cited by 26
General-purpose pre-trained large cellular models for single-cell transcriptomics
Bian, H., Chen, Y., Luo, E., Wu, X., Hao, M., Wei, L., & Zhang, X. (2024) · National Science Review · doi:10.1093/nsr/nwae340 · Cited by 32
Coreference for Discourse Parsing: A Neural Approach
Guz, G., & Carenini, G. (2020) · Workshop on Computational Approaches to Discourse (CODI) · doi:10.18653/v1/2020.codi-1.17 · Cited by 36
“Similarly to Joshi et al. (2020), we train all of our models with AdamW (Loshchilov and Hutter, 2019) optimizer with learning rate of 1 e − 5 for SpanBERT and 2 e − 4 for model-specific components, with the batch size of 5 and linear decay for 20 epochs.”
MethodResultInfluentialAuxiliary Learning for Relation Extraction
Lyu, S., Cheng, J., Wu, X., Cui, L., Chen, H., & Miao, C. (2020) · IEEE Transactions on Emerging Topics in Computational Intelligence · doi:10.1109/tetci.2020.3040444 · Cited by 24
Frame-based Neural Network for Machine Reading Comprehension
Guo, S., Guan, Y., Tan, H., Li, R., & Li, X. (2021) · Knowledge-Based Systems · doi:10.1016/j.knosys.2021.106889 · Cited by 20
“For instance, SpanBert [43] extended BERT by masking contiguous random spans, rather than random tokens.”
Constrained Multi-Task Learning for Event Coreference Resolution
Lu, J. J., & Ng, V. (2021) · Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (NAACL-HLT) · doi:10.18653/v1/2021.naacl-main.356 · Cited by 24
“We then pass the sequence into a pretrained trans-former encoder used in SpanBERT-large (Joshi et al., 2020) to encode tokens and their contexts.”
MethodUncovering Implicit Gender Bias in Narratives through Commonsense Inference
Huang, T., Brahman, F., Shwartz, V., & Chaturvedi, S. (2021) · Findings of the Association for Computational Linguistics: EMNLP (Findings of EMNLP) · doi:10.18653/v1/2021.findings-emnlp.326 · Cited by 52
Filling the Gaps in Ancient Akkadian Texts: A Masked Language Modelling Approach
Lazar, K., Saret, B., Yehudai, A., Horowitz, W., Wasserman, N., & Stanovsky, G. (2021) · Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/v1/2021.emnlp-main.384 · Cited by 25
“First, they focus on small-scale highly-structured texts, for example, lists (parataxis), such as receipts or census documents (Jursa, 2004).”
BackgroundA systematic review of question answering systems for non-factoid questions
Cortes, E. G., Woloszyn, V., Barone, D. A. C., Möller, S., & Vieira, R. L. R. (2021) · Journal of Intelligent Information Systems · doi:10.1007/s10844-021-00655-8 · Cited by 22
“For example, since 2018, new methods addressing factoid questions have outperformed humans on the Stanford Question Answering Dataset (SQuAD) 1 achieving new state-of-the-art results (Joshi et al., 2020; Young et al., 2018; Shen et al., 2020; Wang et al., 2017).”
MethodPretrained AI Models: Performativity, Mobility, and Change
Varshney, L. R., Keskar, N. S., & Socher, R. (2019) · arXiv · doi:10.48550/arxiv.1909.03290 · Cited by 20
“As a notable example, consider SpanBERT (not a fine-tuning of BERT, but a new pretrained model inspired by BERT) designed to be better at inductive transfer to new language tasks [35]; its developers are primarily from Facebook but also have participants from academia.”
BackgroundLightPAFF: A Two-Stage Distillation Framework for Pre-training and Fine-tuningPreprint
Song, K., Sun, H., Tan, X., Qin, T., Lu, J., Liu, H., & Liu, T. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2004.12817 · Cited by 47
“Recently, the pre-training and fine-tuning frameworks (Devlin et al., 2018; Radford et al., 2018; 2019; Yang et al., 2019b; Liu et al., 2019; Lample & Conneau, 2019; Song et al., 2019) have achieved great success in NLP by learning universal language representations on large-scale language corpus…”
BackgroundJoint Keyphrase Chunking and Salience Ranking with BERT.Preprint
Sun, S. Y., Xiong, C., Liu, Z., Liu, Z., & Bao, J. (2020) · arXiv (Cornell University) · Cited by 20
CIDER: Commonsense Inference for Dialogue Explanation and Reasoning
Ghosal, D., Hong, P., Shen, S., Majumder, N., Mihalcea, R., & Poria, S. (2021) · Proceedings of the 22nd Annual Meeting of the Special Interest Group on Discourse and Dialogue · doi:10.18653/v1/2021.sigdial-1.33 · Cited by 19
Fan, J., & Cohn, T. (2021) · Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (NAACL-HLT) · doi:10.18653/v1/2021.naacl-main.125 · Cited by 25
“The representation of token nodes and predicate nodes is the contextualized embeddings from the SpanBERT encoder, denoted as h w and h p respectively.”
MethodInfluentialReasonBERT: Pre-trained to Reason with Distant Supervision
Deng, X., Su, Y., Lees, A., Wu, Y., Yu, C., & Sun, H. (2021) · Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/v1/2021.emnlp-main.494 · Cited by 35
“We also include the masked language modeling (MLM) objective in pre-training to leverage other tokens in the input that are not entities.”
BackgroundMethodResultInfluentialAutomated hate speech detection and span extraction in underground hacking and extremist forums
Zhou, L., Caines, A., Pete, I., & Hutchings, A. (2022) · Natural Language Engineering · doi:10.1017/s1351324922000262 · Cited by 28
“Similar to BERT + token, it is based on sequence labelling except it fine-tunes SpanBERT (Joshi et al. 2019).”
SPOT: Knowledge-Enhanced Language Representations for Information Extraction
Li, J., Katsis, Y., Baldwin, T., Kim, H., Bartko, A. P., McAuley, J., & Hsu, C. (2022) · ACM International Conference on Information and Knowledge Management (CIKM) · doi:10.1145/3511808.3557459 · Cited by 21
A deep neural network model for coreference resolution in geological domain
Wan, B., Dong, S., Chu, D., Li, H., Liu, Y., Fu, J., Fang, F., Li, S., & Zhou, D. (2023) · Information Processing & Management · doi:10.1016/j.ipm.2023.103268 · Cited by 21
Self-citationWang, Y., Sun, Y., Fu, Y., Zhu, D., & Tian, Z. (2024) · IEEE Transactions on Instrumentation and Measurement · doi:10.1109/tim.2024.3374300 · Cited by 26
Dance2MIDI: Dance-driven multi-instrument music generation
Han, B., Li, Y., Shen, Y., Ren, Y., & Han, F. (2024) · Computational Visual Media · doi:10.1007/s41095-024-0417-1 · Cited by 22
“To address this issue, TBPTT [28] employs a Transformer-XL [27] generator in conjunction with a pre-trained Span-BERT [29] discriminator for long symbolic music generation, which enhances training stability.”
Metadata
- Authors
- Joshi, M., Chen, D., Liu, Y., Weld, D. S., Zettlemoyer, L., & Levy, O.
- Affiliations
- University of Washington (US)
- Princeton University (US)
- Meta (United States) (US)
- Venue
- Transactions of the Association for Computational Linguistics
- Publisher
- MIT Press
- Published
- 2020-12
- Volume
- 8
- Pages
- 64-77
- ISSN
- 2307-387X (ISSN-L)
- OpenAlex
- W2962753370
- Cited by
- 1,827
- References
- 51
- Citations by year
- 2026: 84 · 2025: 169 · 2024: 214 · 2023: 373 · 2022: 342 · 2021: 393 · 2020: 224
- FWCI
- 130.82
- Type
- journal article
- Language
- en
- Open access
- diamond
- doi.org
- Licence
- cc-by
- Locations
- Transactions of the Association for Computational Linguistics (publishedVersion, cc-by)
- arXiv (Cornell University) (submittedVersion)
- DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
- arXiv (Cornell University)
- Funders
- University of Washington
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Topic Modeling
- Physical Sciences › Computer Science › Artificial Intelligence › Natural Language Processing Techniques
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Multimodal Machine Learning Applications
- Keywords
- SpanBERT, BERT pre-training, span prediction, question answering, coreference resolution, relation extraction, text representation learning
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.