Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/tomohideshibata/BERT-related-papers
BERT-related papers
https://github.com/tomohideshibata/BERT-related-papers
Last synced: 2 months ago
JSON representation
BERT-related papers
- Host: GitHub
- URL: https://github.com/tomohideshibata/BERT-related-papers
- Owner: tomohideshibata
- Created: 2019-06-12T11:56:44.000Z (about 5 years ago)
- Default Branch: master
- Last Pushed: 2023-08-12T08:28:50.000Z (11 months ago)
- Last Synced: 2024-01-27T14:32:56.296Z (5 months ago)
- Homepage:
- Size: 1.71 MB
- Stars: 2,008
- Watchers: 94
- Forks: 283
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Lists
- awesome-vision-language-pretraining-papers - BERT-related Papers
- awesome-transformer-nlp - tomohideshibata/BERT-related papers
- awesome-stars - tomohideshibata/BERT-related-papers - BERT-related papers (Others)
- awesome-stars - tomohideshibata/BERT-related-papers - BERT-related papers (Others)
- awesome-stars - tomohideshibata/BERT-related-papers - BERT-related papers (Others)
- awesome-stars - tomohideshibata/BERT-related-papers - BERT-related papers (Others)
- awesome-pretrained-models-for-information-retrieval - BERT-related-papers
- awesome-machine-learning-resources - **[List - related-papers?style=social) (Table of Contents)
- awesome-Vision-and-Language-Pre-training - BERT-related Papers
- awesome-xai - [link
- Awesome-Efficient-Transformers - BERT-related Papers
README
# BERT-related Papers
This is a list of BERT-related papers. Any feedback is welcome.
(ChatGPT-related papers are listed at https://github.com/tomohideshibata/ChatGPT-related-papers.)
## Table of Contents
- [Survey paper](#survey-paper)
- [Downstream task](#downstream-task)
- [Generation](#generation)
- [Quality evaluator](#quality-evaluator)
- [Modification (multi-task, masking strategy, etc.)](#modification-multi-task-masking-strategy-etc)
- [Sentence embedding](#sentence-embedding)
- [Transformer variants](#transformer-variants)
- [Probe](#probe)
- [Inside BERT](#inside-bert)
- [Multi-lingual](#multi-lingual)
- [Other than English models](#other-than-english-models)
- [Domain specific](#domain-specific)
- [Multi-modal](#multi-modal)
- [Model compression](#model-compression)
- [Large language model](#large-language-model)
- [Reinforcement learning from human feedback](#reinforcement-learning-from-human-feedback)
- [Misc.](#misc)
## Survey paper
- [Evolution of transfer learning in natural language processing](https://arxiv.org/abs/1910.07370)
- [Pre-trained Models for Natural Language Processing: A Survey](https://arxiv.org/abs/2003.08271)
- [A Survey on Contextual Embeddings](https://arxiv.org/abs/2003.07278)
- [A Survey on Transfer Learning in Natural Language Processing](https://arxiv.org/abs/2007.04239)
- [Which \*BERT? A Survey Organizing Contextualized Encoders](https://arxiv.org/abs/2010.00854) (EMNLP2020)
- [The NLP Cookbook: Modern Recipes for Transformer based Deep Learning Architectures](https://arxiv.org/abs/2104.10640)
- [Pre-Trained Models: Past, Present and Future](https://arxiv.org/abs/2106.07139)
- [A Survey of Transformers](https://arxiv.org/abs/2106.04554)
- [AMMUS : A Survey of Transformer-based Pretrained Models in Natural Language Processing](https://arxiv.org/abs/2108.05542)
- [Paradigm Shift in Natural Language Processing](https://arxiv.org/abs/2109.12575)
- [Recent Advances in Natural Language Processing via Large Pre-Trained Language Models: A Survey](https://arxiv.org/abs/2111.01243)
- [Formal Algorithms for Transformers](https://arxiv.org/abs/2207.09238)
- [A Comprehensive Survey on Pretrained Foundation Models: A History from BERT to ChatGPT](https://arxiv.org/abs/2302.09419)
## Downstream task
### QA, MC, Dialogue
- [Machine Reading Comprehension: The Role of Contextualized Language Models and Beyond](https://arxiv.org/abs/2005.06249)
- [A Survey on Machine Reading Comprehension: Tasks, Evaluation Metrics, and Benchmark Datasets](https://arxiv.org/abs/2006.11880)
- [A BERT Baseline for the Natural Questions](https://arxiv.org/abs/1901.08634)
- [MultiQA: An Empirical Investigation of Generalization and Transfer in Reading Comprehension](https://arxiv.org/abs/1905.13453) (ACL2019)
- [BoolQ: Exploring the Surprising Difficulty of Natural Yes/No Questions](https://arxiv.org/abs/1905.10044) (NAACL2019) [[github](https://github.com/google-research-datasets/boolean-questions)]
- [Natural Perturbation for Robust Question Answering](https://arxiv.org/abs/2004.04849)
- [Unsupervised Domain Adaptation on Reading Comprehension](https://arxiv.org/abs/1911.06137)
- [BERTQA -- Attention on Steroids](https://arxiv.org/abs/1912.10435)
- [Exploring BERT Parameter Efficiency on the Stanford Question Answering Dataset v2.0](https://arxiv.org/abs/2002.10670)
- [Adversarial Augmentation Policy Search for Domain and Cross-Lingual Generalization in Reading Comprehension](https://arxiv.org/abs/2004.06076)
- [Logic-Guided Data Augmentation and Regularization for Consistent Question Answering](https://arxiv.org/abs/2004.10157) (ACL2020)
- [UnifiedQA: Crossing Format Boundaries With a Single QA System](https://arxiv.org/abs/2005.00700)
- [How Can We Know When Language Models Know?](https://arxiv.org/abs/2012.00955)
- [A Multi-Type Multi-Span Network for Reading Comprehension that Requires Discrete Reasoning](https://arxiv.org/abs/1908.05514) (EMNLP2019)
- [A Simple and Effective Model for Answering Multi-span Questions](https://arxiv.org/abs/1909.13375) [[github](https://github.com/eladsegal/tag-based-multi-span-extraction)]
- [Injecting Numerical Reasoning Skills into Language Models](https://arxiv.org/abs/2004.04487) (ACL2020)
- [Towards Question Format Independent Numerical Reasoning: A Set of Prerequisite Tasks](https://arxiv.org/abs/2005.08516)
- [SDNet: Contextualized Attention-based Deep Network for Conversational Question Answering](https://arxiv.org/abs/1812.03593)
- [Multi-hop Question Answering via Reasoning Chains](https://arxiv.org/abs/1910.02610)
- [Select, Answer and Explain: Interpretable Multi-hop Reading Comprehension over Multiple Documents](https://arxiv.org/abs/1911.00484)
- [Multi-step Entity-centric Information Retrieval for Multi-Hop Question Answering](https://arxiv.org/abs/1909.07598) (EMNLP2019 WS)
- [Fine-tuning Multi-hop Question Answering with Hierarchical Graph Network](https://arxiv.org/abs/2004.13821)
- [Unsupervised Alignment-based Iterative Evidence Retrieval for Multi-hop Question Answering](https://www.aclweb.org/anthology/2020.acl-main.414/) (ACL2020)
- [HybridQA: A Dataset of Multi-Hop Question Answering over Tabular and Textual Data](https://arxiv.org/abs/2004.07347)
- [Unsupervised Multi-hop Question Answering by Question Generation](https://arxiv.org/abs/2010.12623) (NAACL2021)
- [End-to-End Open-Domain Question Answering with BERTserini](https://arxiv.org/abs/1902.01718) (NAALC2019)
- [Latent Retrieval for Weakly Supervised Open Domain Question Answering](https://arxiv.org/abs/1906.00300) (ACL2019)
- [Dense Passage Retrieval for Open-Domain Question Answering](https://arxiv.org/abs/2004.04906) (EMNLP2020)
- [Efficient Passage Retrieval with Hashing for Open-domain Question Answering](https://arxiv.org/abs/2106.00882) (ACL2021)
- [End-to-End Training of Neural Retrievers for Open-Domain Question Answering](https://arxiv.org/abs/2101.00408)
- [Domain-matched Pre-training Tasks for Dense Retrieval](https://arxiv.org/abs/2107.13602)
- [Towards Robust Neural Retrieval Models with Synthetic Pre-Training](https://arxiv.org/abs/2104.07800)
- [Simple Entity-Centric Questions Challenge Dense Retrievers](https://arxiv.org/abs/2109.08535) (EMNLP2021) [[github](https://github.com/princeton-nlp/EntityQuestions)]
- [Phrase Retrieval Learns Passage Retrieval, Too](https://arxiv.org/abs/2109.08133) (EMNLP2021) [[github](https://github.com/princeton-nlp/DensePhrases)]
- [Leveraging Passage Retrieval with Generative Models for Open Domain Question Answering](https://arxiv.org/abs/2007.01282)
- [Progressively Pretrained Dense Corpus Index for Open-Domain Question Answering](https://arxiv.org/abs/2005.00038) (EACL2021)
- [Answering Complex Open-Domain Questions with Multi-Hop Dense Retrieval](https://arxiv.org/abs/2009.12756)
- [Multi-Step Reasoning Over Unstructured Text with Beam Dense Retrieval](https://arxiv.org/abs/2104.05883) (NAACL2021) [[github](https://github.com/henryzhao5852/BeamDR)]
- [Retrieve, Read, Rerank, then Iterate: Answering Open-Domain Questions of Varying Reasoning Steps from Text](https://arxiv.org/abs/2010.12527)
- [RocketQA: An Optimized Training Approach to Dense Passage Retrieval for Open-Domain Question Answering](https://arxiv.org/abs/2010.08191)
- [Pre-training Tasks for Embedding-based Large-scale Retrieval](https://arxiv.org/abs/2002.03932) (ICLR2020)
- [Multi-passage BERT: A Globally Normalized BERT Model for Open-domain Question Answering](https://arxiv.org/abs/1908.08167) (EMNLP2019)
- [QED: A Framework and Dataset for Explanations in Question Answering](https://arxiv.org/abs/2009.06354) [[github](https://github.com/google-research-datasets/QED)]
- [Learning to Retrieve Reasoning Paths over Wikipedia Graph for Question Answering](https://arxiv.org/abs/1911.10470) (ICLR2020)
- [Relevance-guided Supervision for OpenQA with ColBERT](https://arxiv.org/abs/2007.00814)
- [RECONSIDER: Re-Ranking using Span-Focused Cross-Attention for Open Domain Question Answering](https://arxiv.org/abs/2010.10757)
- [Joint Passage Ranking for Diverse Multi-Answer Retrieval](https://arxiv.org/abs/2104.08445)
- [SPARTA: Efficient Open-Domain Question Answering via Sparse Transformer Matching Retrieval](https://arxiv.org/abs/2009.13013)
- [Don't Read Too Much into It: Adaptive Computation for Open-Domain Question Answering](https://arxiv.org/abs/2011.05435) (EMNLP2020 WS)
- [Pruning the Index Contents for Memory Efficient Open-Domain QA](https://arxiv.org/abs/2102.10697) [[github](https://github.com/KNOT-FIT-BUT/R2-D2)]
- [Is Retriever Merely an Approximator of Reader?](https://arxiv.org/abs/2010.10999)
- [Neural Retrieval for Question Answering with Cross-Attention Supervised Data Augmentation](https://arxiv.org/abs/2009.13815)
- [RikiNet: Reading Wikipedia Pages for Natural Question Answering](https://arxiv.org/abs/2004.14560) (ACL2020)
- [BERT-kNN: Adding a kNN Search Component to Pretrained Language Models for Better QA](https://arxiv.org/abs/2005.00766)
- [DC-BERT: Decoupling Question and Document for Efficient Contextual Encoding](https://arxiv.org/abs/2002.12591) (SIGIR2020)
- [Learning to Ask Unanswerable Questions for Machine Reading Comprehension](https://arxiv.org/abs/1906.06045) (ACL2019)
- [Unsupervised Question Answering by Cloze Translation](https://arxiv.org/abs/1906.04980) (ACL2019)
- [Reinforcement Learning Based Graph-to-Sequence Model for Natural Question Generation](https://arxiv.org/abs/1908.04942) (ICLR2020)
- [A Recurrent BERT-based Model for Question Generation](https://www.aclweb.org/anthology/D19-5821/) (EMNLP2019 WS)
- [Unsupervised Question Decomposition for Question Answering](https://arxiv.org/abs/2002.09758) [[github](https://github.com/facebookresearch/UnsupervisedDecomposition)]
- [Conversational Question Reformulation via Sequence-to-Sequence Architectures and Pretrained Language Models](https://arxiv.org/abs/2004.01909)
- [Template-Based Question Generation from Retrieved Sentences for Improved Unsupervised Question Answering](https://arxiv.org/abs/2004.11892) (ACL2020)
- [What Are People Asking About COVID-19? A Question Classification Dataset](https://arxiv.org/abs/2005.12522)
- [Learning to Answer by Learning to Ask: Getting the Best of GPT-2 and BERT Worlds](https://arxiv.org/abs/1911.02365)
- [Enhancing Pre-Trained Language Representations with Rich Knowledge for Machine Reading Comprehension](https://www.aclweb.org/anthology/papers/P/P19/P19-1226/) (ACL2019)
- [QA-GNN: Reasoning with Language Models and Knowledge Graphs for Question Answering](https://arxiv.org/abs/2104.06378) (NAACL2021) [[github](https://github.com/michiyasunaga/qagnn)] [[blog](http://ai.stanford.edu/blog/qagnn/)]
- [Incorporating Relation Knowledge into Commonsense Reading Comprehension with Multi-task Learning](https://arxiv.org/abs/1908.04530) (CIKM2019)
- [SG-Net: Syntax-Guided Machine Reading Comprehension](https://arxiv.org/abs/1908.05147)
- [MMM: Multi-stage Multi-task Learning for Multi-choice Reading Comprehension](https://arxiv.org/abs/1910.00458)
- [Cosmos QA: Machine Reading Comprehension with Contextual Commonsense Reasoning](https://arxiv.org/abs/1909.00277) (EMNLP2019)
- [ReClor: A Reading Comprehension Dataset Requiring Logical Reasoning](https://arxiv.org/abs/2002.04326) (ICLR2020)
- [Robust Reading Comprehension with Linguistic Constraints via Posterior Regularization](https://arxiv.org/abs/1911.06948)
- [BAS: An Answer Selection Method Using BERT Language Model](https://arxiv.org/abs/1911.01528)
- [Utilizing Bidirectional Encoder Representations from Transformers for Answer Selection](https://arxiv.org/abs/2011.07208) (AMMCS2019)
- [TANDA: Transfer and Adapt Pre-Trained Transformer Models for Answer Sentence Selection](https://arxiv.org/abs/1911.04118) (AAAI2020)
- [The Cascade Transformer: an Application for Efficient Answer Sentence Selection](https://arxiv.org/abs/2005.02534) (ACL2020)
- [Support-BERT: Predicting Quality of Question-Answer Pairs in MSDN using Deep Bidirectional Transformer](https://arxiv.org/abs/2005.08294)
- [Beat the AI: Investigating Adversarial Human Annotations for Reading Comprehension](https://arxiv.org/abs/2002.00293)
- [Benchmarking Robustness of Machine Reading Comprehension Models](https://arxiv.org/abs/2004.14004)
- [Evaluating NLP Models via Contrast Sets](https://arxiv.org/abs/2004.02709)
- [Undersensitivity in Neural Reading Comprehension](https://arxiv.org/abs/2003.04808)
- [Developing a How-to Tip Machine Comprehension Dataset and its Evaluation in Machine Comprehension by BERT](https://www.aclweb.org/anthology/2020.fever-1.4/) (ACL2020 WS)
- [A Simple but Effective Method to Incorporate Multi-turn Context with BERT for Conversational Machine Comprehension](https://arxiv.org/abs/1905.12848) (ACL2019 WS)
- [FlowDelta: Modeling Flow Information Gain in Reasoning for Conversational Machine Comprehension](https://arxiv.org/abs/1908.05117) (ACL2019 WS)
- [BERT with History Answer Embedding for Conversational Question Answering](https://arxiv.org/abs/1905.05412) (SIGIR2019)
- [GraphFlow: Exploiting Conversation Flow with Graph Neural Networks for Conversational Machine Comprehension](https://arxiv.org/abs/1908.00059) (ICML2019 WS)
- [TAPAS: Weakly Supervised Table Parsing via Pre-training](https://arxiv.org/abs/2004.02349) (ACL2020)
- [TaBERT: Pretraining for Joint Understanding of Textual and Tabular Data](https://arxiv.org/abs/2005.08314) (ACL2020)
- [Understanding tables with intermediate pre-training](https://arxiv.org/abs/2010.00571) (EMNLP2020 Findings)
- [GraPPa: Grammar-Augmented Pre-Training for Table Semantic Parsing](https://arxiv.org/abs/2009.13845) (ICLR2021)
- [Table Search Using a Deep Contextualized Language Model](https://arxiv.org/abs/2005.09207) (SIGIR2020)
- [Open Domain Question Answering over Tables via Dense Retrieval](https://arxiv.org/abs/2103.12011) (NAACL2021)
- [Capturing Row and Column Semantics in Transformer Based Question Answering over Tables](https://arxiv.org/abs/2104.08303) (NAACL2021)
- [MATE: Multi-view Attention for Table Transformer Efficiency](https://arxiv.org/abs/2109.04312) (EMNLP2021)
- [TORQUE: A Reading Comprehension Dataset of Temporal Ordering Questions](https://arxiv.org/abs/2005.00242) (EMNLP2020)
- [Beyond English-only Reading Comprehension: Experiments in Zero-Shot Multilingual Transfer for Bulgarian](https://arxiv.org/abs/1908.01519) (RANLP2019)
- [XQA: A Cross-lingual Open-domain Question Answering Dataset](https://www.aclweb.org/anthology/P19-1227/) (ACL2019)
- [XOR QA: Cross-lingual Open-Retrieval Question Answering](https://arxiv.org/abs/2010.11856) (NAACL2021) [[website](https://nlp.cs.washington.edu/xorqa/)]
- [Cross-Lingual Machine Reading Comprehension](https://arxiv.org/abs/1909.00361) (EMNLP2019)
- [Zero-shot Reading Comprehension by Cross-lingual Transfer Learning with Multi-lingual Language Representation Model](https://arxiv.org/abs/1909.09587)
- [Multilingual Question Answering from Formatted Text applied to Conversational Agents](https://arxiv.org/abs/1910.04659)
- [BiPaR: A Bilingual Parallel Dataset for Multilingual and Cross-lingual Reading Comprehension on Novels](https://arxiv.org/abs/1910.05040) (EMNLP2019)
- [MLQA: Evaluating Cross-lingual Extractive Question Answering](https://arxiv.org/abs/1910.07475)
- [Multilingual Synthetic Question and Answer Generation for Cross-Lingual Reading Comprehension](https://arxiv.org/abs/2010.12008)
- [Synthetic Data Augmentation for Zero-Shot Cross-Lingual Question Answering](https://arxiv.org/abs/2010.12643)
- [Cross-lingual Machine Reading Comprehension with Language Branch Knowledge Distillation](https://arxiv.org/abs/2010.14271) (COLING2020)
- [MKQA: A Linguistically Diverse Benchmark for Multilingual Open Domain Question Answering](https://arxiv.org/abs/2007.15207) [[github](https://github.com/apple/ml-mkqa)]
- [Towards More Equitable Question Answering Systems: How Much More Data Do You Need?](https://arxiv.org/abs/2105.14115) (ACL2021)
- [X-METRA-ADA: Cross-lingual Meta-Transfer Learning Adaptation to Natural Language Understanding and Question Answering](https://arxiv.org/abs/2104.09696) (NAACL2021)
- [Investigating Prior Knowledge for Challenging Chinese Machine Reading Comprehension](https://arxiv.org/abs/1904.09679) (TACL)
- [SberQuAD - Russian Reading Comprehension Dataset: Description and Analysis](https://arxiv.org/abs/1912.09723)
- [DuReaderrobust: A Chinese Dataset Towards Evaluating the Robustness of Machine Reading Comprehension Models](https://arxiv.org/abs/2004.11142)
- [Giving BERT a Calculator: Finding Operations and Arguments with Reading Comprehension](https://arxiv.org/abs/1909.00109) (EMNLP2019)
- [Few-Shot Question Answering by Pretraining Span Selection](https://arxiv.org/abs/2101.00438) (ACL2021)
- [DialoGLUE: A Natural Language Understanding Benchmark for Task-Oriented Dialogue](https://arxiv.org/abs/2009.13570) [[website](https://evalai.cloudcv.org/web/challenges/challenge-page/708/overview)]
- [A Short Survey of Pre-trained Language Models for Conversational AI-A NewAge in NLP](https://arxiv.org/abs/2104.10810)
- [MPC-BERT: A Pre-Trained Language Model for Multi-Party Conversation Understanding](https://arxiv.org/abs/2106.01541) (ACL2021)
- [BERT-DST: Scalable End-to-End Dialogue State Tracking with Bidirectional Encoder Representations from Transformer](https://arxiv.org/abs/1907.03040) (Interspeech2019)
- [Dialog State Tracking: A Neural Reading Comprehension Approach](https://arxiv.org/abs/1908.01946)
- [A Simple but Effective BERT Model for Dialog State Tracking on Resource-Limited Systems](https://arxiv.org/abs/1910.12995) (ICASSP2020)
- [Fine-Tuning BERT for Schema-Guided Zero-Shot Dialogue State Tracking](https://arxiv.org/abs/2002.00181)
- [Goal-Oriented Multi-Task BERT-Based Dialogue State Tracker](https://arxiv.org/abs/2002.02450)
- [Dialogue State Tracking with Pretrained Encoder for Multi-domain Trask-oriented Dialogue Systems](https://arxiv.org/abs/2004.10663)
- [Zero-Shot Transfer Learning with Synthesized Data for Multi-Domain Dialogue State Tracking](https://arxiv.org/abs/2005.00891) (ACL2020)
- [A Fast and Robust BERT-based Dialogue State Tracker for Schema-Guided Dialogue Dataset](https://arxiv.org/abs/2008.12335) (KDD2020 WS)
- [Knowledge-Aware Graph-Enhanced GPT-2 for Dialogue State Tracking](https://arxiv.org/abs/2104.04466)
- [Coreference Augmentation for Multi-Domain Task-Oriented Dialogue State Tracking](https://arxiv.org/abs/2106.08723) (Interspeech2021)
- [ToD-BERT: Pre-trained Natural Language Understanding for Task-Oriented Dialogues](https://arxiv.org/abs/2004.06871) (EMNLP2020)
- [Conversations Are Not Flat: Modeling the Dynamic Information Flow across Dialogue Utterances](https://arxiv.org/abs/2106.02227) (ACL2021)
- [Domain Adaptive Training BERT for Response Selection](https://arxiv.org/abs/1908.04812)
- [Speaker-Aware BERT for Multi-Turn Response Selection in Retrieval-Based Chatbots](https://arxiv.org/abs/2004.03588)
- [Curriculum Learning Strategies for IR: An Empirical Study on Conversation Response Ranking](https://arxiv.org/abs/1912.08555) (ECIR2020)
- [MuTual: A Dataset for Multi-Turn Dialogue Reasoning](https://arxiv.org/abs/2004.04494) (ACL2020)
- [DialBERT: A Hierarchical Pre-Trained Model for Conversation Disentanglement](https://arxiv.org/abs/2004.03760)
- [Generalized Conditioned Dialogue Generation Based on Pre-trained Language Model](https://arxiv.org/abs/2010.11140)
- [BoB: BERT Over BERT for Training Persona-based Dialogue Models from Limited Personalized Data](https://arxiv.org/abs/2106.06169) (ACL2021)
- [Interactive Teaching for Conversational AI](https://arxiv.org/abs/2012.00958) (NeurIPS2020 WS)
- [BERT Goes to Law School: Quantifying the Competitive Advantage of Access to Large Legal Corpora in Contract Understanding](https://arxiv.org/abs/1911.00473)
### Slot filling and Intent Detection
- [A Stack-Propagation Framework with Token-Level Intent Detection for Spoken Language Understanding](https://www.aclweb.org/anthology/D19-1214/) (EMNLP2019)
- [BERT for Joint Intent Classification and Slot Filling](https://arxiv.org/abs/1902.10909)
- [A Co-Interactive Transformer for Joint Slot Filling and Intent Detection](https://arxiv.org/abs/2010.03880) (ICASSP2021)
- [Few-shot Intent Classification and Slot Filling with Retrieved Examples](https://arxiv.org/abs/2104.05763) (NAACL2021)
- [Multi-lingual Intent Detection and Slot Filling in a Joint BERT-based Model](https://arxiv.org/abs/1907.02884)
- [A Comparison of Deep Learning Methods for Language Understanding](https://www.isca-speech.org/archive/Interspeech_2019/abstracts/1262.html) (Interspeech2019)
- [Data Augmentation for Spoken Language Understanding via Pretrained Models](https://arxiv.org/abs/2004.13952)
- [Few-Shot Intent Detection via Contrastive Pre-Training and Fine-Tuning] (EMNLP2021)
- [STIL -- Simultaneous Slot Filling, Translation, Intent Classification, and Language Identification: Initial Results using mBART on MultiATIS++](https://arxiv.org/abs/2010.00760) (AACL-IJCNLP2020) [[github](https://github.com/amazon-research/stil-mbart-multiatis)]
### Analysis
- [Fine-grained Information Status Classification Using Discourse Context-Aware Self-Attention](https://arxiv.org/abs/1908.04755)
- [Neural Aspect and Opinion Term Extraction with Mined Rules as Weak Supervision](https://arxiv.org/abs/1907.03750) (ACL2019)
- [BERT-based Lexical Substitution](https://www.aclweb.org/anthology/P19-1328) (ACL2019)
- [Assessing BERT’s Syntactic Abilities](https://arxiv.org/abs/1901.05287)
- [Investigating Novel Verb Learning in BERT: Selectional Preference Classes and Alternation-Based Syntactic Generalization](https://arxiv.org/abs/2011.02417) (EMNLP2020 WS)
- [Does BERT agree? Evaluating knowledge of structure dependence through agreement relations](https://arxiv.org/abs/1908.09892)
- [Simple BERT Models for Relation Extraction and Semantic Role Labeling](https://arxiv.org/abs/1904.05255)
- [Bridging the Gap in Multilingual Semantic Role Labeling: a Language-Agnostic Approach](https://aclanthology.org/2020.coling-main.120/) (COLING2020)
- [LIMIT-BERT : Linguistic Informed Multi-Task BERT](https://arxiv.org/abs/1910.14296) (EMNLP2020 Findings)
- [Joint Semantic Analysis with Document-Level Cross-Task Coherence Rewards](https://arxiv.org/abs/2010.05567)
- [A Simple BERT-Based Approach for Lexical Simplification](https://arxiv.org/abs/1907.06226)
- [BERT-Based Arabic Social Media Author Profiling](https://arxiv.org/abs/1909.04181)
- [Sentence-Level BERT and Multi-Task Learning of Age and Gender in Social Media](https://arxiv.org/abs/1911.00637)
- [Evaluating the Factual Consistency of Abstractive Text Summarization](https://arxiv.org/abs/1910.12840)
- [Generating Fact Checking Explanations](https://arxiv.org/abs/2004.05773) (ACL2020)
- [NegBERT: A Transfer Learning Approach for Negation Detection and Scope Resolution](https://arxiv.org/abs/1911.04211)
- [xSLUE: A Benchmark and Analysis Platform for Cross-Style Language Understanding and Evaluation](https://arxiv.org/abs/1911.03663)
- [TabFact: A Large-scale Dataset for Table-based Fact Verification](https://arxiv.org/abs/1909.02164) (ICLR2020)
- [Rapid Adaptation of BERT for Information Extraction on Domain-Specific Business Documents](https://arxiv.org/abs/2002.01861)
- [A Focused Study to Compare Arabic Pre-training Models on Newswire IE Tasks](https://arxiv.org/abs/2004.14519)
- [LAMBERT: Layout-Aware (Language) Modeling for information extraction](https://arxiv.org/abs/2002.08087) (ICDAR2021)
- [Keyphrase Extraction from Scholarly Articles as Sequence Labeling using Contextualized Embeddings](https://arxiv.org/abs/1910.08840) (ECIR2020) [[github](https://github.com/midas-research/keyphrase-extraction-as-sequence-labeling-data)]
- [Keyphrase Extraction with Span-based Feature Representations](https://arxiv.org/abs/2002.05407)
- [Keyphrase Prediction With Pre-trained Language Model](https://arxiv.org/abs/2004.10462)
- [Self-Supervised Contextual Keyword and Keyphrase Retrieval with Self-Labelling](https://www.preprints.org/manuscript/201908.0073/v1) [[github](https://github.com/MaartenGr/KeyBERT)]
- [Joint Keyphrase Chunking and Salience Ranking with BERT](https://arxiv.org/abs/2004.13639)
- [Generalizing Natural Language Analysis through Span-relation Representations](https://arxiv.org/abs/1911.03822) (ACL2020) [[github](https://github.com/neulab/cmu-multinlp)]
- [What do you mean, BERT? Assessing BERT as a Distributional Semantics Model](https://arxiv.org/abs/1911.05758)
- [tBERT: Topic Models and BERT Joining Forces for Semantic Similarity Detection](https://www.aclweb.org/anthology/2020.acl-main.630/) (ACL2020)
- [Domain Adaptation with BERT-based Domain Classification and Data Selection](https://www.aclweb.org/anthology/D19-6109/) (EMNLP2019 WS)
- [PERL: Pivot-based Domain Adaptation for Pre-trained Deep Contextualized Embedding Models](https://arxiv.org/abs/2006.09075) (TACL2020)
- [Unsupervised Out-of-Domain Detection via Pre-trained Transformers](https://arxiv.org/abs/2106.00948) (ACL2021) [[github](https://github.com/rivercold/BERT-unsupervised-OOD)]
- [Knowledge Distillation for BERT Unsupervised Domain Adaptation](https://arxiv.org/abs/2010.11478)
- [Sensitive Data Detection and Classification in Spanish Clinical Text: Experiments with BERT](https://arxiv.org/abs/2003.03106) (LREC2020)
- [Does BERT Pretrained on Clinical Notes Reveal Sensitive Data?](https://arxiv.org/abs/2104.07762) (NAACL2021)
- [On the Importance of Word and Sentence Representation Learning in Implicit Discourse Relation Classification](https://arxiv.org/abs/2004.12617) (IJCAI2020)
- [Adapting BERT to Implicit Discourse Relation Classification with a Focus on Discourse Connectives](http://www.lrec-conf.org/proceedings/lrec2020/pdf/2020.lrec-1.144.pdf) (LREC2020)
- [Labeling Explicit Discourse Relations using Pre-trained Language Models](https://arxiv.org/abs/2006.11852) (TSD2020)
- [Causal-BERT : Language models for causality detection between events expressed in text](https://arxiv.org/abs/2012.05453)
- [BERT4SO: Neural Sentence Ordering by Fine-tuning BERT](https://arxiv.org/abs/2103.13584)
- [Document-Level Event Argument Extraction by Conditional Generation](https://arxiv.org/abs/2104.05919) (NAACL2021)
- [Cross-lingual Zero- and Few-shot Hate Speech Detection Utilising Frozen Transformer Language Models and AXEL](https://arxiv.org/abs/2004.13850)
- [Same Side Stance Classification Task: Facilitating Argument Stance Classification by Fine-tuning a BERT Model](https://arxiv.org/abs/2004.11163)
- [Kungfupanda at SemEval-2020 Task 12: BERT-Based Multi-Task Learning for Offensive Language Detection](https://arxiv.org/abs/2004.13432)
- [KEIS@JUST at SemEval-2020 Task 12: Identifying Multilingual Offensive Tweets Using Weighted Ensemble and Fine-Tuned BERT](https://arxiv.org/abs/2005.07820)
- [ALBERT-BiLSTM for Sequential Metaphor Detection](https://www.aclweb.org/anthology/2020.figlang-1.17/) (ACL2020 WS)
- [MelBERT: Metaphor Detection via Contextualized Late Interaction using Metaphorical Identification Theories](https://arxiv.org/abs/2104.13615) (NAACL2021)
- [A BERT-based Dual Embedding Model for Chinese Idiom Prediction](https://arxiv.org/abs/2011.02378) (COLING2020)
- [Should You Fine-Tune BERT for Automated Essay Scoring?](https://www.aclweb.org/anthology/2020.bea-1.15/) (ACL2020 WS)
- [KILT: a Benchmark for Knowledge Intensive Language Tasks](https://arxiv.org/abs/2009.02252) (NAACL2021) [[github](https://github.com/facebookresearch/KILT)]
- [IndoNLU: Benchmark and Resources for Evaluating Indonesian Natural Language Understanding](https://arxiv.org/abs/2009.05387) (AACL-IJCNLP2020)
- [MedFilter: Improving Extraction of Task-relevant Utterances through Integration of Discourse Structure and Ontological Knowledge](https://arxiv.org/abs/2010.02246) (EMNLP2020)
- [ActionBert: Leveraging User Actions for Semantic Understanding of User Interfaces](https://arxiv.org/abs/2012.12350) (AAAI2021)
- [UserBERT: Self-supervised User Representation Learning](https://openreview.net/forum?id=zmgJIjyWSOw)
- [UserBERT: Contrastive User Model Pre-training](https://arxiv.org/abs/2109.01274)
- [Fine-tuning BERT for Low-Resource Natural Language Understanding via Active Learning](https://arxiv.org/abs/2012.02462) (COLING2020)
- [Automatic punctuation restoration with BERT models](https://arxiv.org/abs/2101.07343)
### Word segmentation, parsing, NER
- [BERT Meets Chinese Word Segmentation](https://arxiv.org/abs/1909.09292)
- [Unified Multi-Criteria Chinese Word Segmentation with BERT](https://arxiv.org/abs/2004.05808)
- [RethinkCWS: Is Chinese Word Segmentation a Solved Task?](https://arxiv.org/abs/2011.06858) (EMNLP2020) [[github](https://github.com/neulab/InterpretEval)]
- [Enhancing Chinese Word Segmentation via Pseudo Labels for Practicability](https://aclanthology.org/2021.findings-acl.383/) (ACL2021 Findings)
- [Joint Persian Word Segmentation Correction and Zero-Width Non-Joiner Recognition Using BERT](https://arxiv.org/abs/2010.00287)
- [Toward Fast and Accurate Neural Chinese Word Segmentation with Multi-Criteria Learning](https://arxiv.org/abs/1903.04190)
- [Establishing Strong Baselines for the New Decade: Sequence Tagging, Syntactic and Semantic Parsing with BERT](https://arxiv.org/abs/1908.04943) (FLAIRS-33)
- [Evaluating Contextualized Embeddings on 54 Languages in POS Tagging, Lemmatization and Dependency Parsing](https://arxiv.org/abs/1908.07448)
- [fastHan: A BERT-based Joint Many-Task Toolkit for Chinese NLP](https://arxiv.org/abs/2009.08633)
- [Deep Contextualized Word Embeddings in Transition-Based and Graph-Based Dependency Parsing -- A Tale of Two Parsers Revisited](https://arxiv.org/abs/1908.07397) (EMNLP2019)
- [Is POS Tagging Necessary or Even Helpful for Neural Dependency Parsing?](https://arxiv.org/abs/2003.03204)
- [Parsing as Pretraining](https://arxiv.org/abs/2002.01685) (AAAI2020)
- [Cross-Lingual BERT Transformation for Zero-Shot Dependency Parsing](https://arxiv.org/abs/1909.06775)
- [Recursive Non-Autoregressive Graph-to-Graph Transformer for Dependency Parsing with Iterative Refinement](https://arxiv.org/abs/2003.13118)
- [StructFormer: Joint Unsupervised Induction of Dependency and Constituency Structure from Masked Language Modeling](https://arxiv.org/abs/2012.00857)
- [pyBART: Evidence-based Syntactic Transformations for IE](https://arxiv.org/abs/2005.01306) [[github](https://allenai.github.io/pybart/)]
- [Named Entity Recognition -- Is there a glass ceiling?](https://arxiv.org/abs/1910.02403) (CoNLL2019)
- [A Unified MRC Framework for Named Entity Recognition](https://arxiv.org/abs/1910.11476)
- [Biomedical named entity recognition using BERT in the machine reading comprehension framework](https://arxiv.org/abs/2009.01560)
- [Training Compact Models for Low Resource Entity Tagging using Pre-trained Language Models](https://arxiv.org/abs/1910.06294)
- [Robust Named Entity Recognition with Truecasing Pretraining](https://arxiv.org/abs/1912.07095) (AAAI2020)
- [LTP: A New Active Learning Strategy for Bert-CRF Based Named Entity Recognition](https://arxiv.org/abs/2001.02524)
- [Named Entity Recognition as Dependency Parsing](https://arxiv.org/abs/2005.07150) (ACL2020)
- [Exploring Cross-sentence Contexts for Named Entity Recognition with BERT](https://arxiv.org/abs/2006.01563)
- [CrossNER: Evaluating Cross-Domain Named Entity Recognition](https://arxiv.org/abs/2012.04373) (AAAI2021) [[github](https://github.com/zliucr/CrossNER)]
- [Embeddings of Label Components for Sequence Labeling: A Case Study of Fine-grained Named Entity Recognition](https://arxiv.org/abs/2006.01372) (ACL2020 SRW)
- [BOND: BERT-Assisted Open-Domain Named Entity Recognition with Distant Supervision](https://arxiv.org/abs/2006.15509) (KDD2020) [[github](https://github.com/cliang1453/BOND)]
- [Interpretability Analysis for Named Entity Recognition to Understand System Predictions and How They Can Improve](https://arxiv.org/abs/2004.04564)
- [Single-/Multi-Source Cross-Lingual NER via Teacher-Student Learning on Unlabeled Data in Target Language](https://arxiv.org/abs/2004.12440) (ACL2020)
- [To BERT or Not to BERT: Comparing Task-specific and Task-agnostic Semi-Supervised Approaches for Sequence Tagging](https://arxiv.org/abs/2010.14042) (EMNLP2020)
- [Example-Based Named Entity Recognition](https://arxiv.org/abs/2008.10570)
- [FLERT: Document-Level Features for Named Entity Recognition](https://arxiv.org/abs/2011.06993)
- [Empirical Analysis of Unlabeled Entity Problem in Named Entity Recognition](https://arxiv.org/abs/2012.05426)
- [What's in a Name? Are BERT Named Entity Representations just as Good for any other Name?](https://arxiv.org/abs/2007.06897) (ACL2020 WS)
- [Interpretable Multi-dataset Evaluation for Named Entity Recognition](https://arxiv.org/abs/2011.06854) (EMNLP2020) [[github](https://github.com/neulab/InterpretEval)]
- [Entity Enhanced BERT Pre-training for Chinese NER](https://aclanthology.org/2020.emnlp-main.518/) (EMNLP2020)
- [Lexicon Enhanced Chinese Sequence Labeling Using BERT Adapter](https://arxiv.org/abs/2105.07148) (ACL2021)
- [FLAT: Chinese NER Using Flat-Lattice Transformer](https://aclanthology.org/2020.acl-main.611/) (ACL2020)
- [BioALBERT: A Simple and Effective Pre-trained Language Model for Biomedical Named Entity Recognition](https://arxiv.org/abs/2009.09223)
- [MT-BioNER: Multi-task Learning for Biomedical Named Entity Recognition using Deep Bidirectional Transformers](https://arxiv.org/abs/2001.08904)
- [Knowledge Guided Named Entity Recognition for BioMedical Text](https://arxiv.org/abs/1911.03869)
- [Cross-Lingual Named Entity Recognition Using Parallel Corpus: A New Approach Using XLM-RoBERTa Alignment](https://arxiv.org/abs/2101.11112)
- [Portuguese Named Entity Recognition using BERT-CRF](https://arxiv.org/abs/1909.10649)
- [Towards Lingua Franca Named Entity Recognition with BERT](https://arxiv.org/abs/1912.01389)
- [Larger-Context Tagging: When and Why Does It Work?](https://arxiv.org/abs/2104.04434) (NAACL2021)
### Pronoun/coreference resolution
- [A Brief Survey and Comparative Study of Recent Development of Pronoun Coreference Resolution](https://arxiv.org/abs/2009.12721)
- [Resolving Gendered Ambiguous Pronouns with BERT](https://arxiv.org/abs/1906.01161) (ACL2019 WS)
- [Anonymized BERT: An Augmentation Approach to the Gendered Pronoun Resolution Challenge](https://arxiv.org/abs/1905.01780) (ACL2019 WS)
- [Gendered Pronoun Resolution using BERT and an extractive question answering formulation](https://arxiv.org/abs/1906.03695) (ACL2019 WS)
- [MSnet: A BERT-based Network for Gendered Pronoun Resolution](https://arxiv.org/abs/1908.00308) (ACL2019 WS)
- [Scalable Cross Lingual Pivots to Model Pronoun Gender for Translation](https://arxiv.org/abs/2006.08881)
- [Fill the GAP: Exploiting BERT for Pronoun Resolution](https://www.aclweb.org/anthology/papers/W/W19/W19-3815/) (ACL2019 WS)
- [On GAP Coreference Resolution Shared Task: Insights from the 3rd Place Solution](https://www.aclweb.org/anthology/W19-3816/) (ACL2019 WS)
- [Look Again at the Syntax: Relational Graph Convolutional Network for Gendered Ambiguous Pronoun Resolution](https://arxiv.org/abs/1905.08868) (ACL2019 WS)
- [Unsupervised Pronoun Resolution via Masked Noun-Phrase Prediction](https://arxiv.org/abs/2105.12392) (ACL2021)
- [BERT Masked Language Modeling for Co-reference Resolution](https://www.aclweb.org/anthology/papers/W/W19/W19-3811/) (ACL2019 WS)
- [Coreference Resolution with Entity Equalization](https://www.aclweb.org/anthology/P19-1066/) (ACL2019)
- [BERT for Coreference Resolution: Baselines and Analysis](https://arxiv.org/abs/1908.09091) (EMNLP2019) [[github](https://github.com/mandarjoshi90/coref)]
- [WikiCREM: A Large Unsupervised Corpus for Coreference Resolution](https://arxiv.org/abs/1908.08025) (EMNLP2019)
- [CD2CR: Co-reference Resolution Across Documents and Domains](https://arxiv.org/abs/2101.12637) (EACL2021)
- [Ellipsis Resolution as Question Answering: An Evaluation](https://arxiv.org/abs/1908.11141) (EACL2021)
- [Coreference Resolution as Query-based Span Prediction](https://arxiv.org/abs/1911.01746)
- [Coreferential Reasoning Learning for Language Representation](https://arxiv.org/abs/2004.06870) (EMNLP2020)
- [Revisiting Memory-Efficient Incremental Coreference Resolution](https://arxiv.org/abs/2005.00128)
- [Revealing the Myth of Higher-Order Inference in Coreference Resolution](https://arxiv.org/abs/2009.12013) (EMNLP2020)
- [Coreference Resolution without Span Representations](https://arxiv.org/abs/2101.00434) (ACL2021)
- [Neural Mention Detection](https://arxiv.org/abs/1907.12524) (LREC2020)
- [ZPR2: Joint Zero Pronoun Recovery and Resolution using Multi-Task Learning and BERT](https://www.aclweb.org/anthology/2020.acl-main.482/) (ACL2020)
- [An Empirical Study of Contextual Data Augmentation for Japanese Zero Anaphora Resolution](https://aclanthology.org/2020.coling-main.435/) (COLING2020)
- [BERT-based Cohesion Analysis of Japanese Texts](https://aclanthology.org/2020.coling-main.114/) (COLING2020)
- [Joint Coreference Resolution and Character Linking for Multiparty Conversation](https://arxiv.org/abs/2101.11204)
- [Sequence to Sequence Coreference Resolution](https://www.aclweb.org/anthology/2020.crac-1.5/) (COLING2020 WS)
- [Within-Document Event Coreference with BERT-Based Contextualized Representations](https://arxiv.org/abs/2102.09600)
- [Multi-task Learning Based Neural Bridging Reference Resolution](https://arxiv.org/abs/2003.03666)
- [Bridging Anaphora Resolution as Question Answering](https://arxiv.org/abs/2004.07898) (ACL2020)
- [Fine-grained Information Status Classification Using Discourse Context-Aware BERT](https://arxiv.org/abs/2010.14759) (COLING2020)
### Word sense disambiguation
- [Language Models and Word Sense Disambiguation: An Overview and Analysis](https://arxiv.org/abs/2008.11608)
- [GlossBERT: BERT for Word Sense Disambiguation with Gloss Knowledge](https://arxiv.org/abs/1908.07245) (EMNLP2019)
- [Adapting BERT for Word Sense Disambiguation with Gloss Selection Objective and Example Sentences](https://arxiv.org/abs/2009.11795) (EMNLP2020 Findings)
- [Improved Word Sense Disambiguation Using Pre-Trained Contextualized Word Representations](https://arxiv.org/abs/1910.00194) (EMNLP2019)
- [Using BERT for Word Sense Disambiguation](https://arxiv.org/abs/1909.08358)
- [Language Modelling Makes Sense: Propagating Representations through WordNet for Full-Coverage Word Sense Disambiguation](https://www.aclweb.org/anthology/P19-1569.pdf) (ACL2019)
- [Does BERT Make Any Sense? Interpretable Word Sense Disambiguation with Contextualized Embeddings](https://arxiv.org/abs/1909.10430) (KONVENS2019)
- [An Accurate Model for Predicting the (Graded) Effect of Context in Word Similarity Based on Bert](https://arxiv.org/abs/2005.01006)
- [PolyLM: Learning about Polysemy through Language Modeling](https://arxiv.org/abs/2101.10448) (EACL2021)
- [CluBERT: A Cluster-Based Approach for Learning Sense Distributions in Multiple Languages](https://www.aclweb.org/anthology/2020.acl-main.369/) (ACL2020)
- [Cross-lingual Word Sense Disambiguation using mBERT Embeddings with Syntactic Dependencies](https://arxiv.org/abs/2012.05300)
- [VCDM: Leveraging Variational Bi-encoding and Deep Contextualized Word Representations for Improved Definition Modeling](https://arxiv.org/abs/2010.03124) (EMNLP2020)
### Sentiment analysis
- [Utilizing BERT for Aspect-Based Sentiment Analysis via Constructing Auxiliary Sentence](https://arxiv.org/abs/1903.09588) (NAACL2019)
- [BERT Post-Training for Review Reading Comprehension and Aspect-based Sentiment Analysis](https://arxiv.org/abs/1904.02232) (NAACL2019)
- [Exploiting BERT for End-to-End Aspect-based Sentiment Analysis](https://arxiv.org/abs/1910.00883) (EMNLP2019 WS)
- [Improving BERT Performance for Aspect-Based Sentiment Analysis](https://arxiv.org/abs/2010.11731)
- [Context-Guided BERT for Targeted Aspect-Based Sentiment Analysis](https://arxiv.org/abs/2010.07523)
- [Understanding Pre-trained BERT for Aspect-based Sentiment Analysis](https://arxiv.org/abs/2011.00169) (COLING2020)
- [Does syntax matter? A strong baseline for Aspect-based Sentiment Analysis with RoBERTa](https://arxiv.org/abs/2104.04986) (NAACL2021)
- [Adapt or Get Left Behind: Domain Adaptation through BERT Language Model Finetuning for Aspect-Target Sentiment Classification](https://arxiv.org/abs/1908.11860) (LREC2020)
- [An Investigation of Transfer Learning-Based Sentiment Analysis in Japanese](https://arxiv.org/abs/1905.09642) (ACL2019)
- ["Mask and Infill" : Applying Masked Language Model to Sentiment Transfer](https://arxiv.org/abs/1908.08039)
- [Adversarial Training for Aspect-Based Sentiment Analysis with BERT](https://arxiv.org/abs/2001.11316)
- [Adversarial and Domain-Aware BERT for Cross-Domain Sentiment Analysis](https://www.aclweb.org/anthology/2020.acl-main.370/) (ACL2020)
- [Utilizing BERT Intermediate Layers for Aspect Based Sentiment Analysis and Natural Language Inference](https://arxiv.org/abs/2002.04815)
- [DomBERT: Domain-oriented Language Model for Aspect-based Sentiment Analysis](https://arxiv.org/abs/2004.13816)
- [YASO: A New Benchmark for Targeted Sentiment Analysis](https://arxiv.org/abs/2012.14541)
- [SentiBERT: A Transferable Transformer-Based Architecture for Compositional Sentiment Semantics](https://arxiv.org/abs/2005.04114) (ACL2020)
### Relation extraction
- [Matching the Blanks: Distributional Similarity for Relation Learning](https://arxiv.org/abs/1906.03158) (ACL2019)
- [BERT-Based Multi-Head Selection for Joint Entity-Relation Extraction](https://arxiv.org/abs/1908.05908) (NLPCC2019)
- [Enriching Pre-trained Language Model with Entity Information for Relation Classification](https://arxiv.org/abs/1905.08284)
- [Span-based Joint Entity and Relation Extraction with Transformer Pre-training](https://arxiv.org/abs/1909.07755)
- [Fine-tune Bert for DocRED with Two-step Process](https://arxiv.org/abs/1909.11898)
- [Relation Extraction as Two-way Span-Prediction](https://arxiv.org/abs/2010.04829)
- [Entity, Relation, and Event Extraction with Contextualized Span Representations](https://arxiv.org/abs/1909.03546) (EMNLP2019)
- [Fine-tuning BERT for Joint Entity and Relation Extraction in Chinese Medical Text](https://arxiv.org/abs/1908.07721)
- [Downstream Model Design of Pre-trained Language Model for Relation Extraction Task](https://arxiv.org/abs/2004.03786)
- [Efficient long-distance relation extraction with DG-SpanBERT](https://arxiv.org/abs/2004.03636)
- [Global-to-Local Neural Networks for Document-Level Relation Extraction](https://arxiv.org/abs/2009.10359) (EMNLP2020)
- [DARE: Data Augmented Relation Extraction with GPT-2](https://arxiv.org/abs/2004.13845)
- [Distantly-Supervised Neural Relation Extraction with Side Information using BERT](https://arxiv.org/abs/2004.14443) (IJCNN2020)
- [Improving Distantly-Supervised Relation Extraction through BERT-based Label & Instance Embeddings](https://arxiv.org/abs/2102.01156)
- [An End-to-end Model for Entity-level Relation Extraction using Multi-instance Learning](https://arxiv.org/abs/2102.05980) (EACL2021)
- [ZS-BERT: Towards Zero-Shot Relation Extraction with Attribute Representation Learning](https://arxiv.org/abs/2104.04697) (NAACL2021) [[github](https://github.com/dinobby/ZS-BERT)]
- [AdaPrompt: Adaptive Prompt-based Finetuning for Relation Extraction](https://arxiv.org/abs/2104.07650)
- [Dialogue-Based Relation Extraction](https://arxiv.org/abs/2004.08056) (ACL2020)
- [An Embarrassingly Simple Model for Dialogue Relation Extraction](https://arxiv.org/abs/2012.13873)
- [A Novel Cascade Binary Tagging Framework for Relational Triple Extraction](https://arxiv.org/abs/1909.03227) (ACL2020) [[github](https://github.com/weizhepei/CasRel)]
- [ExpBERT: Representation Engineering with Natural Language Explanations](https://arxiv.org/abs/2005.01932) (ACL2020) [[github](https://github.com/MurtyShikhar/ExpBERT)]
- [AutoRC: Improving BERT Based Relation Classification Models via Architecture Search](https://arxiv.org/abs/2009.10680)
- [Investigation of BERT Model on Biomedical Relation Extraction Based on Revised Fine-tuning Mechanism](https://arxiv.org/abs/2011.00398)
- [Experiments on transfer learning architectures for biomedical relation extraction](https://arxiv.org/abs/2011.12380)
- [Improving BERT Model Using Contrastive Learning for Biomedical Relation Extraction](https://arxiv.org/abs/2104.13913) (BioNLP2021)
- [Cross-Lingual Relation Extraction with Transformers](https://arxiv.org/abs/2010.08652)
- [Improving Scholarly Knowledge Representation: Evaluating BERT-based Models for Scientific Relation Classification](https://arxiv.org/abs/2004.06153)
- [Robustly Pre-trained Neural Model for Direct Temporal Relation Extraction](https://arxiv.org/abs/2004.06216)
- [A BERT-based One-Pass Multi-Task Model for Clinical Temporal Relation Extraction](https://www.aclweb.org/anthology/2020.bionlp-1.7/) (ACL2020 WS)
- [Exploring Contextualized Neural Language Models for Temporal Dependency Parsing](https://arxiv.org/abs/2004.14577)
- [Temporal Reasoning on Implicit Events from Distant Supervision](https://arxiv.org/abs/2010.12753)
- [IMoJIE: Iterative Memory-Based Joint Open Information Extraction](https://arxiv.org/abs/2005.08178) (ACL2020)
- [OpenIE6: Iterative Grid Labeling and Coordination Analysis for Open Information Extraction](https://arxiv.org/abs/2010.03147) (EMNLP2020) [[github](https://github.com/dair-iitd/openie6)]
- [Multi2OIE: Multilingual Open Information Extraction Based on Multi-Head Attention with BERT](https://arxiv.org/abs/2009.08128) (EMNLP2020 Findings)
### Knowledge base
- [KG-BERT: BERT for Knowledge Graph Completion](https://arxiv.org/abs/1909.03193)
- [How Context Affects Language Models' Factual Predictions](https://openreview.net/forum?id=025X0zPfn) (AKBC2020)
- [Inducing Relational Knowledge from BERT](https://arxiv.org/abs/1911.12753) (AAAI2020)
- [Latent Relation Language Models](https://arxiv.org/abs/1908.07690) (AAAI2020)
- [Pretrained Encyclopedia: Weakly Supervised Knowledge-Pretrained Language Model](https://openreview.net/forum?id=BJlzm64tDH) (ICLR2020)
- [Scalable Zero-shot Entity Linking with Dense Entity Retrieval](https://arxiv.org/abs/1911.03814) (EMNLP2020) [[github](https://github.com/facebookresearch/BLINK)]
- [Zero-shot Entity Linking with Efficient Long Range Sequence Modeling](https://arxiv.org/abs/2010.06065) (EMNLP2020 Findings)
- [Investigating Entity Knowledge in BERT with Simple Neural End-To-End Entity Linking](https://www.aclweb.org/anthology/K19-1063/) (CoNLL2019)
- [Improving Entity Linking by Modeling Latent Entity Type Information](https://arxiv.org/abs/2001.01447) (AAAI2020)
- [Global Entity Disambiguation with Pretrained Contextualized Embeddings of Words and Entities](https://arxiv.org/abs/1909.00426)
- [YELM: End-to-End Contextualized Entity Linking](https://arxiv.org/abs/1911.03834)
- [Empirical Evaluation of Pretraining Strategies for Supervised Entity Linking](https://arxiv.org/abs/2005.14253) (AKBC2020)
- [LUKE: Deep Contextualized Entity Representations with Entity-aware Self-attention](https://arxiv.org/abs/2010.01057) (EMNLP2020) [[github](https://github.com/studio-ousia/luke)]
- [Linking Entities to Unseen Knowledge Bases with Arbitrary Schemas](https://arxiv.org/abs/2010.11333)
- [CHOLAN: A Modular Approach for Neural Entity Linking on Wikipedia and Wikidata](https://arxiv.org/abs/2101.09969) (EACL2021)
- [PEL-BERT: A Joint Model for Protocol Entity Linking](https://arxiv.org/abs/2002.00744)
- [End-to-end Biomedical Entity Linking with Span-based Dictionary Matching](https://arxiv.org/abs/2104.10493)
- [Efficient One-Pass End-to-End Entity Linking for Questions](https://arxiv.org/abs/2010.02413) (EMNLP2020) [[github](https://github.com/belindal/BLINK/tree/master/elq)]
- [Cross-Lingual Transfer in Zero-Shot Cross-Language Entity Linking](https://arxiv.org/abs/2010.09828)
- [Entity Linking in 100 Languages](https://arxiv.org/abs/2011.02690) (EMNLP2020) [[github](https://github.com/google-research/google-research/tree/master/dense_representations_for_entity_retrieval/mel)]
- [COMETA: A Corpus for Medical Entity Linking in the Social Media](https://arxiv.org/abs/2010.03295) (EMNLP2020) [[github](https://github.com/cambridgeltl/cometa)]
- [How Can We Know What Language Models Know?](https://arxiv.org/abs/1911.12543) (TACL2020) [[github](https://github.com/jzbjyb/LPAQA)]
- [How to Query Language Models?](https://arxiv.org/abs/2108.01928)
- [Deep Entity Matching with Pre-Trained Language Models](https://arxiv.org/abs/2004.00584)
- [Ultra-Fine Entity Typing with Weak Supervision from a Masked Language Model](https://arxiv.org/abs/2106.04098) (ACL2021)
- [Constructing Taxonomies from Pretrained Language Models](https://arxiv.org/abs/2010.12813) (NAACL2021)
- [Language Models are Open Knowledge Graphs](https://arxiv.org/abs/2010.11967)
- [Can Generative Pre-trained Language Models Serve as Knowledge Bases for Closed-book QA?](https://arxiv.org/abs/2106.01561) (ACL2021)
- [DualTKB: A Dual Learning Bridge between Text and Knowledge Base](https://arxiv.org/abs/2010.14660) (EMNLP2020) [[github](https://github.com/IBM/dualtkb)]
- [Zero-shot Slot Filling with DPR and RAG](https://arxiv.org/abs/2104.08610)
- [How to Avoid Being Eaten by a Grue: Structured Exploration Strategies for Textual Worlds](https://arxiv.org/abs/2006.07409) [[github](https://github.com/rajammanabrolu/Q-BERT)]
- [MLMLM: Link Prediction with Mean Likelihood Masked Language Model](https://arxiv.org/abs/2009.07058)
- [Beyond I.I.D.: Three Levels of Generalization for Question Answering on Knowledge Bases](https://arxiv.org/abs/2011.07743)
### Text classification
- [Deep Learning Based Text Classification: A Comprehensive Review](https://arxiv.org/abs/2004.03705)
- [A Text Classification Survey: From Shallow to Deep Learning](https://arxiv.org/abs/2008.00364)
- [How to Fine-Tune BERT for Text Classification?](https://arxiv.org/abs/1905.05583)
- [X-BERT: eXtreme Multi-label Text Classification with BERT](https://arxiv.org/abs/1905.02331)
- [An Empirical Study on Large-Scale Multi-Label Text Classification Including Few and Zero-Shot Labels](https://arxiv.org/abs/2010.01653) (EMNLP2020)
- [Taming Pretrained Transformers for Extreme Multi-label Text Classification](https://arxiv.org/abs/1905.02331) (KDD2020)
- [Layer-wise Guided Training for BERT: Learning Incrementally Refined Document Representations](https://arxiv.org/abs/2010.05763) (EMNLP2020 WS)
- [DocBERT: BERT for Document Classification](https://arxiv.org/abs/1904.08398)
- [Enriching BERT with Knowledge Graph Embeddings for Document Classification](https://arxiv.org/abs/1909.08402)
- [Classification and Clustering of Arguments with Contextualized Word Embeddings](https://arxiv.org/abs/1906.09821) (ACL2019)
- [BERT for Evidence Retrieval and Claim Verification](https://arxiv.org/abs/1910.02655)
- [Stacked DeBERT: All Attention in Incomplete Data for Text Classification](https://arxiv.org/abs/2001.00137)
- [Cost-Sensitive BERT for Generalisable Sentence Classification with Imbalanced Data](https://arxiv.org/abs/2003.11563)
- [BAE: BERT-based Adversarial Examples for Text Classification](https://arxiv.org/abs/2004.01970) (EMNLP2020)
- [FireBERT: Hardening BERT-based classifiers against adversarial attack](https://arxiv.org/abs/2008.04203) [[github](https://github.com/FireBERT-author/FireBERT)]
- [GAN-BERT: Generative Adversarial Learning for Robust Text Classification with a Bunch of Labeled Examples](https://www.aclweb.org/anthology/2020.acl-main.191/) (ACL2020)
- [Description Based Text Classification with Reinforcement Learning](https://arxiv.org/abs/2002.03067)
- [VGCN-BERT: Augmenting BERT with Graph Embedding for Text Classification](https://arxiv.org/abs/2004.05707)
- [Zero-shot Text Classification via Reinforced Self-training](https://www.aclweb.org/anthology/2020.acl-main.272/) (ACL2020)
- [On Data Augmentation for Extreme Multi-label Classification](https://arxiv.org/abs/2009.10778)
- [Noisy Channel Language Model Prompting for Few-Shot Text Classification](https://arxiv.org/abs/2108.04106)
- [Improving Pretrained Models for Zero-shot Multi-label Text Classification through Reinforced Label Hierarchy Reasoning](https://arxiv.org/abs/2104.01666) (NAACL2021)
- [Towards Evaluating the Robustness of Chinese BERT Classifiers](https://arxiv.org/abs/2004.03742)
- [COVID-Twitter-BERT: A Natural Language Processing Model to Analyse COVID-19 Content on Twitter](https://arxiv.org/abs/2005.07503) [[github](https://github.com/digitalepidemiologylab/covid-twitter-bert)]
- [Large Scale Legal Text Classification Using Transformer Models](https://arxiv.org/abs/2010.12871)
- [BBAEG: Towards BERT-based Biomedical Adversarial Example Generation for Text Classification](https://arxiv.org/abs/2104.01782) (NAACL2021)
- [A Comparison of LSTM and BERT for Small Corpus](https://arxiv.org/abs/2009.05451)
### WSC, WNLI, NLI
- [Exploring Unsupervised Pretraining and Sentence Structure Modelling for Winograd Schema Challenge](https://arxiv.org/abs/1904.09705)
- [A Surprisingly Robust Trick for the Winograd Schema Challenge](https://arxiv.org/abs/1905.06290)
- [WinoGrande: An Adversarial Winograd Schema Challenge at Scale](https://arxiv.org/abs/1907.10641) (AAAI2020)
- [TTTTTackling WinoGrande Schemas](https://arxiv.org/abs/2003.08380)
- [WinoWhy: A Deep Diagnosis of Essential Commonsense Knowledge for Answering Winograd Schema Challenge](https://arxiv.org/abs/2005.05763) (ACL2020)
- [The Sensitivity of Language Models and Humans to Winograd Schema Perturbations](https://arxiv.org/abs/2005.01348) (ACL2020)
- [Precise Task Formalization Matters in Winograd Schema Evaluations](https://arxiv.org/abs/2010.04043) (EMNLP2020)
- [Tackling Domain-Specific Winograd Schemas with Knowledge-Based Reasoning and Machine Learning](https://arxiv.org/abs/2011.12081)
- [A Review of Winograd Schema Challenge Datasets and Approaches](https://arxiv.org/abs/2004.13831)
- [Improving Natural Language Inference with a Pretrained Parser](https://arxiv.org/abs/1909.08217)
- [Are Natural Language Inference Models IMPPRESsive? Learning IMPlicature and PRESupposition](https://arxiv.org/abs/2004.03066)
- [DocNLI: A Large-scale Dataset for Document-level Natural Language Inference](https://arxiv.org/abs/2106.09449) (ACL2021 Findings)
- [Adversarial NLI: A New Benchmark for Natural Language Understanding](https://arxiv.org/abs/1910.14599)
- [Adversarial Analysis of Natural Language Inference Systems](https://arxiv.org/abs/1912.03441) (ICSC2020)
- [ANLIzing the Adversarial Natural Language Inference Dataset](https://arxiv.org/abs/2010.12729)
- [Syntactic Data Augmentation Increases Robustness to Inference Heuristics](https://arxiv.org/abs/2004.11999) (ACL2020)
- [Linguistically-Informed Transformations (LIT): A Method for Automatically Generating Contrast Sets](https://arxiv.org/abs/2010.08580) (EMNLP2020 WS) [[github](https://github.com/leo-liuzy/LIT_auto-gen-contrast-set)]
- [HypoNLI: Exploring the Artificial Patterns of Hypothesis-only Bias in Natural Language Inference](https://arxiv.org/abs/2003.02756) (LREC2020)
- [Use of Machine Translation to Obtain Labeled Datasets for Resource-Constrained Languages](https://arxiv.org/abs/2004.14963) (EMNLP2020) [[github](https://github.com/boun-tabi/NLI-TR)]
- [FarsTail: A Persian Natural Language Inference Dataset](https://arxiv.org/abs/2009.08820)
- [Evaluating BERT for natural language inference: A case study on the CommitmentBank](https://www.aclweb.org/anthology/D19-1630/) (EMNLP2019)
- [Do Neural Models Learn Systematicity of Monotonicity Inference in Natural Language?](https://arxiv.org/abs/2004.14839) (ACL2020)
- [Abductive Commonsense Reasoning](https://arxiv.org/abs/1908.05739) (ICLR2020)
- [Entailment as Few-Shot Learner](https://arxiv.org/abs/2104.14690)
- [Collecting Entailment Data for Pretraining: New Protocols and Negative Results](https://arxiv.org/abs/2004.11997)
- [WANLI: Worker and AI Collaboration for Natural Language Inference Dataset Creation](https://arxiv.org/abs/2201.05955) (EMNLP2022 Findings) [[github](https://github.com/alisawuffles/wanli)]
- [Mining Knowledge for Natural Language Inference from Wikipedia Categories](https://arxiv.org/abs/2010.01239) (EMNLP2020 Findings)
### Commonsense
- [CommonsenseQA: A Question Answering Challenge Targeting Commonsense Knowledge](https://arxiv.org/abs/1811.00937) (NAACL2019)
- [Human Parity on CommonsenseQA: Augmenting Self-Attention with External Attention](https://arxiv.org/abs/2112.03254)
- [HellaSwag: Can a Machine Really Finish Your Sentence?](https://arxiv.org/abs/1905.07830) (ACL2019) [[website](https://rowanzellers.com/hellaswag/)]
- [A Method for Building a Commonsense Inference Dataset Based on Basic Events](https://www.aclweb.org/anthology/2020.emnlp-main.192/) (EMNLP2020) [[website](http://nlp.ist.i.kyoto-u.ac.jp/EN/?KUCI)]
- [Story Ending Prediction by Transferable BERT](https://arxiv.org/abs/1905.07504) (IJCAI2019)
- [Explain Yourself! Leveraging Language Models for Commonsense Reasoning](https://arxiv.org/abs/1906.02361) (ACL2019)
- [Pre-training Is (Almost) All You Need: An Application to Commonsense Reasoning](https://arxiv.org/abs/2004.14074) (ACL2020)
- [Align, Mask and Select: A Simple Method for Incorporating Commonsense Knowledge into Language Representation Models](https://arxiv.org/abs/1908.06725)
- [Informing Unsupervised Pretraining with External Linguistic Knowledge](https://arxiv.org/abs/1909.02339)
- [Commonsense Knowledge + BERT for Level 2 Reading Comprehension Ability Test](https://arxiv.org/abs/1909.03415)
- [BIG MOOD: Relating Transformers to Explicit Commonsense Knowledge](https://arxiv.org/abs/1910.07713)
- [Commonsense Knowledge Mining from Pretrained Models](https://arxiv.org/abs/1909.00505) (EMNLP2019)
- [KagNet: Knowledge-Aware Graph Networks for Commonsense Reasoning](https://arxiv.org/abs/1909.02151) (EMNLP2019)
- [Cracking the Contextual Commonsense Code: Understanding Commonsense Reasoning Aptitude of Deep Contextual Representations](https://www.aclweb.org/anthology/D19-6001/) (EMNLP2019 WS)
- [Do Massively Pretrained Language Models Make Better Storytellers?](https://arxiv.org/abs/1909.10705) (CoNLL2019)
- [PIQA: Reasoning about Physical Commonsense in Natural Language](https://arxiv.org/abs/1911.11641) (AAAI2020)
- [Evaluating Commonsense in Pre-trained Language Models](https://arxiv.org/abs/1911.11931) (AAAI2020)
- [Why Do Masked Neural Language Models Still Need Common Sense Knowledge?](https://arxiv.org/abs/1911.03024)
- [Does BERT Solve Commonsense Task via Commonsense Knowledge?](https://arxiv.org/abs/2008.03945)
- [Unsupervised Commonsense Question Answering with Self-Talk](https://arxiv.org/abs/2004.05483) (EMNLP2020)
- [Knowledge-driven Data Construction for Zero-shot Evaluation in Commonsense Question Answering](https://arxiv.org/abs/2011.03863) (AAAI2021)
- [G-DAUG: Generative Data Augmentation for Commonsense Reasoning](https://arxiv.org/abs/2004.11546)
- [Contrastive Self-Supervised Learning for Commonsense Reasoning](https://arxiv.org/abs/2005.00669) (ACL2020)
- [Differentiable Open-Ended Commonsense Reasoning](https://arxiv.org/abs/2010.14439)
- [Adversarial Training for Commonsense Inference](https://arxiv.org/abs/2005.08156) (ACL2020 WS)
- [Do Fine-tuned Commonsense Language Models Really Generalize?](https://arxiv.org/abs/2011.09159)
- [Do Language Models Perform Generalizable Commonsense Inference?](https://arxiv.org/abs/2106.11533) (ACL2021 Findings)
- [Improving Zero Shot Learning Baselines with Commonsense Knowledge](https://arxiv.org/abs/2012.06236)
- [XCOPA: A Multilingual Dataset for Causal Commonsense Reasoning](https://ducdauge.github.io/files/xcopa.pdf) [[github](https://github.com/cambridgeltl/xcopa)]
- [Do Neural Language Representations Learn Physical Commonsense?](https://arxiv.org/abs/1908.02899) (CogSci2019)
### Extractive summarization
- [HIBERT: Document Level Pre-training of Hierarchical Bidirectional Transformers for Document Summarization](https://arxiv.org/abs/1905.06566) (ACL2019)
- [Deleter: Leveraging BERT to Perform Unsupervised Successive Text Compression](https://arxiv.org/abs/1909.03223)
- [Discourse-Aware Neural Extractive Text Summarization](https://arxiv.org/abs/1910.14142) (ACL2020) [[github](https://github.com/jiacheng-xu/DiscoBERT)]
- [AREDSUM: Adaptive Redundancy-Aware Iterative Sentence Ranking for Extractive Document Summarization](https://arxiv.org/abs/2004.06176)
- [Fact-level Extractive Summarization with Hierarchical Graph Mask on BERT](https://arxiv.org/abs/2011.09739) (COLING2020)
- [Do We Really Need That Many Parameters In Transformer For Extractive Summarization? Discourse Can Help !](https://arxiv.org/abs/2012.02144) (EMNLP2020 WS)
- [Multi-Document Summarization with Determinantal Point Processes and Contextualized Representations](https://arxiv.org/abs/1910.11411) (EMNLP2019 WS)
- [Continual BERT: Continual Learning for Adaptive Extractive Summarization of COVID-19 Literature](https://arxiv.org/abs/2007.03405)
### Grammatical error correction
- [Multi-headed Architecture Based on BERT for Grammatical Errors Correction](https://www.aclweb.org/anthology/papers/W/W19/W19-4426/) (ACL2019 WS)
- [Towards Minimal Supervision BERT-based Grammar Error Correction](https://arxiv.org/abs/2001.03521)
- [Learning to combine Grammatical Error Corrections](https://arxiv.org/abs/1906.03897) (EMNLP2019 WS)
- [LM-Critic: Language Models for Unsupervised Grammatical Error Correction](https://arxiv.org/abs/2109.06822) (EMNLP2021) [[github](https://github.com/michiyasunaga/LM-Critic)]
- [Encoder-Decoder Models Can Benefit from Pre-trained Masked Language Models in Grammatical Error Correction](https://arxiv.org/abs/2005.00987) (ACL2020)
- [Chinese Grammatical Correction Using BERT-based Pre-trained Model](https://arxiv.org/abs/2011.02093) (AACL-IJCNLP2020)
- [Spelling Error Correction with Soft-Masked BERT](https://arxiv.org/abs/2005.07421) (ACL2020)
### IR
- [BEIR: A Heterogenous Benchmark for Zero-shot Evaluation of Information Retrieval Models](https://arxiv.org/abs/2104.08663) [[github](https://github.com/UKPLab/beir)]
- [Pretrained Transformers for Text Ranking: BERT and Beyond](https://arxiv.org/abs/2010.06467)
- [Passage Re-ranking with BERT](https://arxiv.org/abs/1901.04085)
- [Investigating the Successes and Failures of BERT for Passage Re-Ranking](https://arxiv.org/abs/1905.01758)
- [Understanding the Behaviors of BERT in Ranking](https://arxiv.org/abs/1904.07531)
- [Document Expansion by Query Prediction](https://arxiv.org/abs/1904.08375)
- [Improving Document Representations by Generating Pseudo Query Embeddings for Dense Retrieval](https://arxiv.org/abs/2105.03599) (ACL2021)
- [CEDR: Contextualized Embeddings for Document Ranking](https://arxiv.org/abs/1904.07094) (SIGIR2019)
- [Deeper Text Understanding for IR with Contextual Neural Language Modeling](https://arxiv.org/abs/1905.09217) (SIGIR2019)
- [FAQ Retrieval using Query-Question Similarity and BERT-Based Query-Answer Relevance](https://arxiv.org/abs/1905.02851) (SIGIR2019)
- [An Analysis of BERT FAQ Retrieval Models for COVID-19 Infobot](https://openreview.net/forum?id=dGOeF3y_Weh)
- [COUGH: A Challenge Dataset and Models for COVID-19 FAQ Retrieval](https://arxiv.org/abs/2010.12800)
- [Unsupervised FAQ Retrieval with Question Generation and BERT](https://www.aclweb.org/anthology/2020.acl-main.74/) (ACL2020)
- [Multi-Stage Document Ranking with BERT](https://arxiv.org/abs/1910.14424)
- [Learning-to-Rank with BERT in TF-Ranking](https://arxiv.org/abs/2004.08476)
- [Transformer-Based Language Models for Similar Text Retrieval and Ranking](https://arxiv.org/abs/2005.04588)
- [DeText: A Deep Text Ranking Framework with BERT](https://arxiv.org/abs/2008.02460)
- [ColBERT: Efficient and Effective Passage Search via Contextualized Late Interaction over BERT](https://arxiv.org/abs/2004.12832) (SIGIR2020)
- [RepBERT: Contextualized Text Embeddings for First-Stage Retrieval](https://arxiv.org/abs/2006.15498) [[github](https://github.com/jingtaozhan/RepBERT-Index)]
- [Approximate Nearest Neighbor Negative Contrastive Learning for Dense Text Retrieval](https://arxiv.org/abs/2007.00808)
- [Multi-Perspective Semantic Information Retrieval](https://arxiv.org/abs/2009.01938)
- [CharacterBERT and Self-Teaching for Improving the Robustness of Dense Retrievers on Queries with Typos](https://arxiv.org/abs/2204.00716) (SIGIR2022)
- [Expansion via Prediction of Importance with Contextualization](https://arxiv.org/abs/2004.14245) (SIGIR2020)
- [BERT-QE: Contextualized Query Expansion for Document Re-ranking](https://arxiv.org/abs/2009.07258) (EMNLP2020 Findings)
- [Beyond \[CLS\] through Ranking by Generation](https://arxiv.org/abs/2010.03073) (EMNLP2020)
- [Efficient Document Re-Ranking for Transformers by Precomputing Term Representations](https://arxiv.org/abs/2004.14255) (SIGIR2020)
- [Training Curricula for Open Domain Answer Re-Ranking](https://arxiv.org/abs/2004.14269) (SIGIR2020)
- [Efficiently Teaching an Effective Dense Retriever with Balanced Topic Aware Sampling](SIGIR2021)
- [Boosted Dense Retriever](https://arxiv.org/abs/2112.07771)
- [ERNIE-Search: Bridging Cross-Encoder with Dual-Encoder via Self On-the-fly Distillation for Dense Passage Retrieval](https://arxiv.org/abs/2205.09153)
- [Document Ranking with a Pretrained Sequence-to-Sequence Model](https://arxiv.org/abs/2003.06713)
- [A Neural Corpus Indexer for Document Retrieval](https://arxiv.org/abs/2206.02743)
- [COIL: Revisit Exact Lexical Match in Information Retrieval with Contextualized Inverted List](https://arxiv.org/abs/2104.07186) (NAACL2021)
- [Guided Transformer: Leveraging Multiple External Sources for Representation Learning in Conversational Search](https://arxiv.org/abs/2006.07548) (SIGIR2020)
- [Fine-tune BERT for E-commerce Non-Default Search Ranking](https://arxiv.org/abs/2008.09689)
- [IR-BERT: Leveraging BERT for Semantic Search in Background Linking for News Articles](https://arxiv.org/abs/2007.12603)
- [ProphetNet-Ads: A Looking Ahead Strategy for Generative Retrieval Models in Sponsored Search Engine](https://arxiv.org/abs/2010.10789) (NLPCC2020)
- [Rapidly Deploying a Neural Search Engine for the COVID-19 Open Research Dataset: Preliminary Thoughts and Lessons Learned](https://openreview.net/forum?id=PlUA_mgGaPq) (ACL2020 WS)
- [SLEDGE-Z: A Zero-Shot Baseline for COVID-19 Literature Search](https://arxiv.org/abs/2010.05987) (EMNLP2020)
- [Neural Duplicate Question Detection without Labeled Training Data](https://www.aclweb.org/anthology/D19-1171/) (EMNLP2019)
- [Cross-Domain Generalization Through Memorization: A Study of Nearest Neighbors in Neural Duplicate Question Detection](https://arxiv.org/abs/2011.11090)
- [Effective Transfer Learning for Identifying Similar Questions: Matching User Questions to COVID-19 FAQs](https://arxiv.org/abs/2008.13546)
- [Cross-lingual Information Retrieval with BERT](https://arxiv.org/abs/2004.13005)
- [Cross-lingual Retrieval for Iterative Self-Supervised Training](https://arxiv.org/abs/2006.09526) (NeurIPS2020)
- [Graph-based Multilingual Product Retrieval in E-Commerce Search](https://aclanthology.org/2021.naacl-industry.19/) (NAACL2021 Industry)
- [Teaching a New Dog Old Tricks: Resurrecting Multilingual Retrieval Using Zero-shot Learning](https://arxiv.org/abs/1912.13080) (ECIR2020)
- [PROP: Pre-training with Representative Words Prediction for Ad-hoc Retrieval](https://arxiv.org/abs/2010.10137) (WSDM2021)
- [B-PROP: Bootstrapped Pre-training with Representative Words Prediction for Ad-hoc Retrieval](https://arxiv.org/abs/2104.09791) (SIGIR2021)
- [Condenser: a Pre-training Architecture for Dense Retrieval](https://arxiv.org/abs/2104.08253) (EMNLP2021)
- [Augmenting Document Representations for Dense Retrieval with Interpolation and Perturbation](https://arxiv.org/abs/2203.07735) (ACL2022)
- [Mr. TyDi: A Multi-lingual Benchmark for Dense Retrieval](https://arxiv.org/abs/2108.08787) (EMNLP2021 WS) [[github](https://github.com/castorini/mr.tydi)]
## Generation
- [Pretrained Language Models for Text Generation: A Survey](https://arxiv.org/abs/2105.10311) (IJCAI2021 Survey Track)
- [A Survey of Pretrained Language Models Based Text Generation](https://arxiv.org/abs/2201.05273)
- [GLGE: A New General Language Generation Evaluation Benchmark](https://arxiv.org/abs/2011.11928) [[github](https://github.com/microsoft/glge)]
- [BERT has a Mouth, and It Must Speak: BERT as a Markov Random Field Language Model](https://arxiv.org/abs/1902.04094) (NAACL2019 WS)
- [Pretraining-Based Natural Language Generation for Text Summarization](https://arxiv.org/abs/1902.09243)
- [Text Summarization with Pretrained Encoders](https://arxiv.org/abs/1908.08345) (EMNLP2019) [[github (original)](https://github.com/nlpyang/PreSumm)] [[github (huggingface)](https://github.com/huggingface/transformers/tree/master/examples/summarization)]
- [Multi-stage Pretraining for Abstractive Summarization](https://arxiv.org/abs/1909.10599)
- [PEGASUS: Pre-training with Extracted Gap-sentences for Abstractive Summarization](https://arxiv.org/abs/1912.08777)
- [Abstractive Summarization with Combination of Pre-trained Sequence-to-Sequence and Saliency Models](https://arxiv.org/abs/2003.13028)
- [GSum: A General Framework for Guided Neural Abstractive Summarization](https://arxiv.org/abs/2010.08014) (NAACL2021) [[github](https://github.com/neulab/guided_summarization)]
- [STEP: Sequence-to-Sequence Transformer Pre-training for Document Summarization](https://arxiv.org/abs/2004.01853)
- [TLDR: Extreme Summarization of Scientific Documents](https://arxiv.org/abs/2004.15011) [[github](https://github.com/allenai/scitldr)]
- [Product Title Generation for Conversational Systems using BERT](https://arxiv.org/abs/2007.11768)
- [WSL-DS: Weakly Supervised Learning with Distant Supervision for Query Focused Multi-Document Abstractive Summarization](https://arxiv.org/abs/2011.01421) (COLING2020)
- [Constrained Abstractive Summarization: Preserving Factual Consistency with Constrained Generation](https://arxiv.org/abs/2010.12723)
- [Abstractive Query Focused Summarization with Query-Free Resources](https://arxiv.org/abs/2012.14774)
- [Abstractive Summarization of Spoken and Written Instructions with BERT](https://arxiv.org/abs/2008.09676)
- [Language Model as an Annotator: Exploring DialoGPT for Dialogue Summarization](https://arxiv.org/abs/2105.12544) (ACL2021)
- [Coreference-Aware Dialogue Summarization](https://arxiv.org/abs/2106.08556) (SIGDIAL2021)
- [XL-Sum: Large-Scale Multilingual Abstractive Summarization for 44 Languages](https://arxiv.org/abs/2106.13822) (ACL2021 Findings) [[github](https://github.com/csebuetnlp/xl-sum)]
- [BERT Fine-tuning For Arabic Text Summarization](https://arxiv.org/abs/2004.14135) (ICLR2020 WS)
- [Automatic Text Summarization of COVID-19 Medical Research Articles using BERT and GPT-2](https://arxiv.org/abs/2006.01997)
- [Mixed-Lingual Pre-training for Cross-lingual Summarization](https://arxiv.org/abs/2010.08892) (AACL-IJCNLP2020)
- [PoinT-5: Pointer Network and T-5 based Financial NarrativeSummarisation](https://arxiv.org/abs/2010.04191) (COLING2020 WS)
- [MASS: Masked Sequence to Sequence Pre-training for Language Generation](https://arxiv.org/abs/1905.02450) (ICML2019) [[github](https://github.com/microsoft/MASS)], [[github](https://github.com/microsoft/MASS/tree/master/MASS-fairseq)]
- [JASS: Japanese-specific Sequence to Sequence Pre-training for Neural Machine Translation](https://arxiv.org/abs/2005.03361) (LREC2020)
- [Unified Language Model Pre-training for Natural Language Understanding and Generation](https://arxiv.org/abs/1905.03197) [[github](https://github.com/microsoft/unilm)] (NeurIPS2019)
- [UniLMv2: Pseudo-Masked Language Models for Unified Language Model Pre-Training](https://arxiv.org/abs/2002.12804) [[github](https://github.com/microsoft/unilm)]
- [Dual Inference for Improving Language Understanding and Generation](https://arxiv.org/abs/2010.04246) (EMNLP2020 Findings)
- [All NLP Tasks Are Generation Tasks: A General Pretraining Framework](https://arxiv.org/abs/2103.10360)
- [ProphetNet: Predicting Future N-gram for Sequence-to-Sequence Pre-training](https://arxiv.org/abs/2001.04063) (EMNLP2020 Findings) [[github](https://github.com/microsoft/ProphetNet)]
- [ProphetNet-X: Large-Scale Pre-training Models for English, Chinese, Multi-lingual, Dialog, and Code Generation](https://arxiv.org/abs/2104.08006)
- [Towards Making the Most of BERT in Neural Machine Translation](https://arxiv.org/abs/1908.05672)
- [Improving Neural Machine Translation with Pre-trained Representation](https://arxiv.org/abs/1908.07688)
- [BERT, mBERT, or BiBERT? A Study on Contextualized Embeddings for Neural Machine Translation](https://arxiv.org/abs/2109.04588) (EMNLP2021)
- [On the use of BERT for Neural Machine Translation](https://arxiv.org/abs/1909.12744) (EMNLP2019 WS)
- [Incorporating BERT into Neural Machine Translation](https://openreview.net/forum?id=Hyl7ygStwB) (ICLR2020)
- [Recycling a Pre-trained BERT Encoder for Neural Machine Translation](https://www.aclweb.org/anthology/D19-5603/)
- [Exploring Unsupervised Pretraining Objectives for Machine Translation](https://arxiv.org/abs/2106.05634) (ACL2021 Findings)
- [Reusing a Pretrained Language Model on Languages with Limited Corpora for Unsupervised NMT](https://arxiv.org/abs/2009.07610) (EMNLP2020)
- [Language Models are Good Translators](https://arxiv.org/abs/2106.13627)
- [Leveraging Pre-trained Checkpoints for Sequence Generation Tasks](https://arxiv.org/abs/1907.12461)
- [Mask-Predict: Parallel Decoding of Conditional Masked Language Models](https://arxiv.org/abs/1904.09324) (EMNLP2019)
- [PALM: Pre-training an Autoencoding&Autoregressive Language Model for Context-conditioned Generation](https://arxiv.org/abs/2004.07159) (EMNLP2020)
- [ERNIE-GEN: An Enhanced Multi-Flow Pre-training and Fine-tuning Framework for Natural Language Generation](https://arxiv.org/abs/2001.11314)
- [Non-Autoregressive Text Generation with Pre-trained Language Models](https://arxiv.org/abs/2102.08220) (EACL2021)
- [Cross-Lingual Natural Language Generation via Pre-Training](https://arxiv.org/abs/1909.10481) (AAAI2020) [[github](https://github.com/CZWin32768/XNLG)]
- [PLATO: Pre-trained Dialogue Generation Model with Discrete Latent Variable](https://arxiv.org/abs/1910.07931) (ACL2020)
- [A Tailored Pre-Training Model for Task-Oriented Dialog Generation](https://arxiv.org/abs/2004.13835)
- [Pretrained Language Models for Dialogue Generation with Multiple Input Sources](https://arxiv.org/abs/2010.07576) (EMNLP2020 Findings)
- [Knowledge-Grounded Dialogue Generation with Pre-trained Language Models](https://arxiv.org/abs/2010.08824) (EMNLP2020)
- [Are Pre-trained Language Models Knowledgeable to Ground Open Domain Dialogues?](https://arxiv.org/abs/2011.09708)
- [Open-Domain Dialogue Generation Based on Pre-trained Language Models](https://arxiv.org/abs/2010.12780)
- [LaMDA: Language Models for Dialog Applications](https://arxiv.org/abs/2201.08239)
- [Retrieval-Augmented Transformer-XL for Close-Domain Dialog Generation](https://arxiv.org/abs/2105.09235)
- [Internet-Augmented Dialogue Generation](https://arxiv.org/abs/2107.07566)
- [DialogBERT: Discourse-Aware Response Generation via Learning to Recover and Rank Utterances](https://arxiv.org/abs/2012.01775) (AAAI2021)
- [CG-BERT: Conditional Text Generation with BERT for Generalized Few-shot Intent Detection](https://arxiv.org/abs/2004.01881)
- [QURIOUS: Question Generation Pretraining for Text Generation](https://arxiv.org/abs/2004.11026)
- [Few-Shot NLG with Pre-Trained Language Model](https://arxiv.org/abs/1904.09521) (ACL2020)
- [Text-to-Text Pre-Training for Data-to-Text Tasks](https://arxiv.org/abs/2005.10433)
- [KGPT: Knowledge-Grounded Pre-Training for Data-to-Text Generation](https://arxiv.org/abs/2010.02307) (EMNLP2020)
- [Evaluating Semantic Accuracy of Data-to-Text Generation with Natural Language Inference](https://arxiv.org/abs/2011.10819) (INLG2020)
- [Large Scale Knowledge Graph Based Synthetic Corpus Generation for Knowledge-Enhanced Language Model Pre-training](https://arxiv.org/abs/2010.12688)
- [Structure-Grounded Pretraining for Text-to-SQL](https://arxiv.org/abs/2010.12773)
- [Data Agnostic RoBERTa-based Natural Language to SQL Query Generation](https://arxiv.org/abs/2010.05243)
- [ToTTo: A Controlled Table-To-Text Generation Dataset](https://arxiv.org/abs/2004.14373) (EMNLP2020) [[github](https://github.com/google-research-datasets/ToTTo)]
- [Exploring Fluent Query Reformulations with Text-to-Text Transformers and Reinforcement Learning](https://arxiv.org/abs/2012.10033) (AAAI2021 WS)
- [A Knowledge-Enhanced Pretraining Model for Commonsense Story Generation](https://arxiv.org/abs/2001.05139) (TACL2020) [[github](https://github.com/JianGuanTHU/CommonsenseStoryGen)]
- [MEGATRON-CNTRL: Controllable Story Generation with External Knowledge Using Large-Scale Language Models](https://arxiv.org/abs/2010.00840) (EMNLP2020)
- [Facts2Story: Controlling Text Generation by Key Facts](https://arxiv.org/abs/2012.04332)
- [CommonGen: A Constrained Text Generation Challenge for Generative Commonsense Reasoning](https://arxiv.org/abs/1911.03705) [[github](https://github.com/INK-USC/CommonGen)] [[website](https://inklab.usc.edu/CommonGen/)] (EMNLP2020 Findings)
- [An Enhanced Knowledge Injection Model for Commonsense Generation](https://arxiv.org/abs/2012.00366) (COLING2020)
- [Retrieval Enhanced Model for Commonsense Generation](https://arxiv.org/abs/2105.11174) (ACL2021 Findings)
- [Lexically-constrained Text Generation through Commonsense Knowledge Extraction and Injection](https://arxiv.org/abs/2012.10813) (AAAI2021WS)
- [Pre-training Text-to-Text Transformers for Concept-centric Common Sense](https://arxiv.org/abs/2011.07956)
- [Language Generation with Multi-Hop Reasoning on Commonsense Knowledge Graph](https://arxiv.org/abs/2009.11692) (EMNLP2020)
- [KG-BART: Knowledge Graph-Augmented BART for Generative Commonsense Reasoning](https://arxiv.org/abs/2009.12677)
- [Autoregressive Entity Retrieval](https://arxiv.org/abs/2010.00904) (ICLR2021) [[github](https://github.com/facebookresearch/GENRE)]
- [Multilingual Autoregressive Entity Linking](https://arxiv.org/abs/2103.12528)
- [EIGEN: Event Influence GENeration using Pre-trained Language Models](https://arxiv.org/abs/2010.11764)
- [proScript: Partially Ordered Scripts Generation via Pre-trained Language Models](https://arxiv.org/abs/2104.08251)
- [Goal-Oriented Script Construction](https://arxiv.org/abs/2107.13189) (INLG2021)
- [Contrastive Triple Extraction with Generative Transformer](https://arxiv.org/abs/2009.06207) (AAAI2021)
- [GeDi: Generative Discriminator Guided Sequence Generation](https://arxiv.org/abs/2009.06367)
- [Generating similes effortlessly like a Pro: A Style Transfer Approach for Simile Generation](https://arxiv.org/abs/2009.08942) (EMNLP2020)
- [Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer](https://arxiv.org/abs/1910.10683) (JMLR2020) [[github](https://github.com/google-research/text-to-text-transfer-transformer)]
- [mT5: A massively multilingual pre-trained text-to-text transformer](https://arxiv.org/abs/2010.11934) (NAACL2021) [[github](https://github.com/google-research/multilingual-t5)]
- [nmT5 -- Is parallel data still relevant for pre-training massively multilingual language models?](https://arxiv.org/abs/2106.02171) (ACL2021)
- [mT6: Multilingual Pretrained Text-to-Text Transformer with Translation Pairs](https://arxiv.org/abs/2104.08692)
- [WT5?! Training Text-to-Text Models to Explain their Predictions](https://arxiv.org/abs/2004.14546)
- [NT5?! Training T5 to Perform Numerical Reasoning](https://arxiv.org/abs/2104.07307) [[github](https://github.com/lesterpjy/numeric-t5)]
- [BART: Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and Comprehension](https://arxiv.org/abs/1910.13461) (ACL2020)
- [The GEM Benchmark: Natural Language Generation, its Evaluation and Metrics](https://arxiv.org/abs/2102.01672)
- [GEMv2: Multilingual NLG Benchmarking in a Single Line of Code](https://arxiv.org/abs/2206.11249)
- [Finetuned Language Models Are Zero-Shot Learners](https://arxiv.org/abs/2109.01652) [[blog](https://ai.googleblog.com/2021/10/introducing-flan-more-generalizable.html)]
- [Multitask Prompted Training Enables Zero-Shot Task Generalization](https://arxiv.org/abs/2110.08207)
- [Multilingual Denoising Pre-training for Neural Machine Translation](https://arxiv.org/abs/2001.08210)
- [Best Practices for Data-Efficient Modeling in NLG:How to Train Production-Ready Neural Models with Less Data](https://arxiv.org/abs/2011.03877) (COLING2020)
- [Prefix-Tuning: Optimizing Continuous Prompts for Generation](https://arxiv.org/abs/2101.00190)
- [Unsupervised Pre-training for Natural Language Generation: A Literature Review](https://arxiv.org/abs/1911.06171)
## Quality evaluator
- [BERTScore: Evaluating Text Generation with BERT](https://arxiv.org/abs/1904.09675) (ICLR2020)
- [BERTTune: Fine-Tuning Neural Machine Translation with BERTScore](https://arxiv.org/abs/2106.02208) (ACL2021)
- [Machine Translation Evaluation with BERT Regressor](https://arxiv.org/abs/1907.12679)
- [TransQuest: Translation Quality Estimation with Cross-lingual Transformers](https://arxiv.org/abs/2011.01536) (COLING2020)
- [SumQE: a BERT-based Summary Quality Estimation Model](https://arxiv.org/abs/1909.00578) (EMNLP2019)
- [MoverScore: Text Generation Evaluating with Contextualized Embeddings and Earth Mover Distance](https://arxiv.org/abs/1909.02622) (EMNLP2019) [[github](https://github.com/AIPHES/emnlp19-moverscore)]
- [BERT as a Teacher: Contextual Embeddings for Sequence-Level Reward](https://arxiv.org/abs/2003.02738)
- [Language Model Augmented Relevance Score](https://arxiv.org/abs/2108.08485) (ACL2021)
- [BLEURT: Learning Robust Metrics for Text Generation](https://arxiv.org/abs/2004.04696) (ACL2020)
- [BARTScore: Evaluating Generated Text as Text Generation](https://arxiv.org/abs/2106.11520) [[github](https://github.com/neulab/BARTScore)]
- [Masked Language Model Scoring](https://arxiv.org/abs/1910.14659) (ACL2020)
- [Simple-QE: Better Automatic Quality Estimation for Text Simplification](https://arxiv.org/abs/2012.12382)
## Modification (multi-task, masking strategy, etc.)
- [Multi-Task Deep Neural Networks for Natural Language Understanding](https://arxiv.org/abs/1901.11504) (ACL2019)
- [The Microsoft Toolkit of Multi-Task Deep Neural Networks for Natural Language Understanding](https://arxiv.org/abs/2002.07972)
- [BERT and PALs: Projected Attention Layers for Efficient Adaptation in Multi-Task Learning](https://arxiv.org/abs/1902.02671) (ICML2019)
- [Measuring Massive Multitask Language Understanding](https://arxiv.org/abs/2009.03300) (ICLR2021) [[github](https://github.com/hendrycks/test)]
- [Parameter-efficient Multi-task Fine-tuning for Transformers via Shared Hypernetworks](https://arxiv.org/abs/2106.04489) (ACL2021)
- [Pre-training Text Representations as Meta Learning](https://arxiv.org/abs/2004.05568)
- [Unifying Question Answering and Text Classification via Span Extraction](https://arxiv.org/abs/1904.09286)
- [MATINF: A Jointly Labeled Large-Scale Dataset for Classification, Question Answering and Summarization](https://arxiv.org/abs/2004.12302) (ACL2020)
- [ERNIE: Enhanced Language Representation with Informative Entities](https://arxiv.org/abs/1905.07129) (ACL2019)
- [ERNIE: Enhanced Representation through Knowledge Integration](https://arxiv.org/abs/1904.09223)
- [ERNIE 2.0: A Continual Pre-training Framework for Language Understanding](https://arxiv.org/abs/1907.12412) (AAAI2020)
- [ERNIE 3.0: Large-scale Knowledge Enhanced Pre-training for Language Understanding and Generation](https://arxiv.org/abs/2107.02137)
- [ERNIE-Gram: Pre-Training with Explicitly N-Gram Masked Language Modeling for Natural Language Understanding](https://arxiv.org/abs/2010.12148)
- [XLNet: Generalized Autoregressive Pretraining for Language Understanding](https://arxiv.org/abs/1906.08237) (NeurIPS2019) [[github](https://github.com/zihangdai/xlnet)]
- [MPNet: Masked and Permuted Pre-training for Language Understanding](https://arxiv.org/abs/2004.09297)
- [Pre-Training with Whole Word Masking for Chinese BERT](https://arxiv.org/abs/1906.08101)
- [SpanBERT: Improving Pre-training by Representing and Predicting Spans](https://arxiv.org/abs/1907.10529) (TACL2020) [[github](https://github.com/facebookresearch/SpanBERT)]
- [ConvBERT: Improving BERT with Span-based Dynamic Convolution](https://arxiv.org/abs/2008.02496)
- [Frustratingly Simple Pretraining Alternatives to Masked Language Modeling](https://arxiv.org/abs/2109.01819) (EMNLP2021) [[github](https://github.com/gucci-j/light-transformer-emnlp2021)]
- [TaCL: Improving BERT Pre-training with Token-aware Contrastive Learning](https://arxiv.org/abs/2111.04198) (NAACL2022)
- [ZEN: Pre-training Chinese Text Encoder Enhanced by N-gram Representations](https://aclanthology.org/2020.findings-emnlp.425/) (EMNLP2020 Findings)
- [ZEN 2.0: Continue Training and Adaption for N-gram Enhanced Text Encoders](https://arxiv.org/abs/2105.01279)
- [MVP-BERT: Redesigning Vocabularies for Chinese BERT and Multi-Vocab Pretraining](https://arxiv.org/abs/2011.08539)
- [Adversarial Training for Large Neural Language Models](https://arxiv.org/abs/2004.08994)
- [BERTAC: Enhancing Transformer-based Language Models with Adversarially Pretrained Convolutional Neural Networks](https://aclanthology.org/2021.acl-long.164/) (ACL2021)
- [Train No Evil: Selective Masking for Task-guided Pre-training](https://arxiv.org/abs/2004.09733)
- [Position Masking for Language Models](https://arxiv.org/abs/2006.05676)
- [Masking as an Efficient Alternative to Finetuning for Pretrained Language Models](https://arxiv.org/abs/2004.12406) (EMNLP2020)
- [Variance-reduced Language Pretraining via a Mask Proposal Network](https://arxiv.org/abs/2008.05333)
- [Neural Mask Generator: Learning to Generate Adaptive Word Maskings for Language Model Adaptation](https://arxiv.org/abs/2010.02705) (EMNLP2020)
- [Improving Self-supervised Pre-training via a Fully-Explored Masked Language Model](https://arxiv.org/abs/2010.06040)
- [Contextual Representation Learning beyond Masked Language Modeling](https://arxiv.org/abs/2204.04163) (ACL2022)
- [Curriculum learning for language modeling](https://arxiv.org/abs/2108.02170)
- [Curriculum Learning: A Regularization Method for Efficient and Stable Billion-Scale GPT Model Pre-Training](https://arxiv.org/abs/2108.06084)
- [Focusing More on Conflicts with Mis-Predictions Helps Language Pre-Training](https://arxiv.org/abs/2012.08789)
- [Exploiting Cloze Questions for Few Shot Text Classification and Natural Language Inference](https://arxiv.org/abs/2001.07676) (EACL2021) [[github](https://github.com/timoschick/pet)]
- [It's Not Just Size That Matters: Small Language Models Are Also Few-Shot Learners](https://arxiv.org/abs/2009.07118) (NAACL2021) [[github](https://github.com/timoschick/pet)]
- [Making Pre-trained Language Models Better Few-shot Learners](https://arxiv.org/abs/2012.15723) (ACL2021) [[github](https://github.com/princeton-nlp/LM-BFF)]
- [CrossFit: A Few-shot Learning Challenge for Cross-task Generalization in NLP](https://arxiv.org/abs/2104.08835)
- [Lifelong Learning of Few-shot Learners across NLP Tasks](https://arxiv.org/abs/2104.08808)
- [Don't Stop Pretraining: Adapt Language Models to Domains and Tasks](https://arxiv.org/abs/2004.10964) (ACL2020)
- [Lifelong Pretraining: Continually Adapting Language Models to Emerging Corpora](https://aclanthology.org/2022.naacl-main.351/) (NAACL2022)
- [Towards Continual Knowledge Learning of Language Models](https://arxiv.org/abs/2110.03215) (ICLR2022)
- [An Empirical Investigation Towards Efficient Multi-Domain Language Model Pre-training](https://arxiv.org/abs/2010.00784) [[github](https://github.com/aws-health-ai/multi_domain_lm)]
- [To Pretrain or Not to Pretrain: Examining the Benefits of Pretraining on Resource Rich Tasks](https://arxiv.org/abs/2006.08671) (ACL2020)
- [Revisiting Few-sample BERT Fine-tuning](https://arxiv.org/abs/2006.05987)
- [Blank Language Models](https://arxiv.org/abs/2002.03079)
- [Enabling Language Models to Fill in the Blanks](https://arxiv.org/abs/2005.05339) (ACL2020)
- [Efficient Training of BERT by Progressively Stacking](http://proceedings.mlr.press/v97/gong19a.html) (ICML2019) [[github](https://github.com/gonglinyuan/StackingBERT)]
- [RoBERTa: A Robustly Optimized BERT Pretraining Approach](https://arxiv.org/abs/1907.11692) [[github](https://github.com/pytorch/fairseq/tree/master/examples/roberta)]
- [On Losses for Modern Language Models](https://arxiv.org/abs/2010.01694) (EMNLP2020) [[github](https://github.com/StephAO/olfmlm)]
- [ALBERT: A Lite BERT for Self-supervised Learning of Language Representations](https://arxiv.org/abs/1909.11942) (ICLR2020)
- [Rethinking Embedding Coupling in Pre-trained Language Models](https://openreview.net/forum?id=xpFFI_NtgpW) (ICLR2021)
- [ELECTRA: Pre-training Text Encoders as Discriminators Rather Than Generators](https://openreview.net/forum?id=r1xMH1BtvB) (ICLR2020) [[github](https://github.com/google-research/electra)] [[blog](https://ai.googleblog.com/2020/03/more-efficient-nlp-model-pre-training.html)]
- [Training ELECTRA Augmented with Multi-word Selection](https://arxiv.org/abs/2106.00139) (ACL2021 Findings)
- [Learning to Sample Replacements for ELECTRA Pre-Training](https://arxiv.org/abs/2106.13715) (ACL2021 Findings)
- [SCRIPT: Self-Critic PreTraining of Transformers](https://aclanthology.org/2021.naacl-main.409/) (NAACL2021)
- [Pre-Training Transformers as Energy-Based Cloze Models](https://www.aclweb.org/anthology/2020.emnlp-main.20/) (EMNLP2020) [[github](https://github.com/google-research/electra)]
- [MC-BERT: Efficient Language Pre-Training via a Meta Controller](https://arxiv.org/abs/2006.05744)
- [FreeLB: Enhanced Adversarial Training for Language Understanding](https://openreview.net/forum?id=BygzbyHFvB) (ICLR2020)
- [KERMIT: Generative Insertion-Based Modeling for Sequences](https://arxiv.org/abs/1906.01604)
- [CALM: Continuous Adaptive Learning for Language Modeling](https://arxiv.org/abs/2004.03794)
- [SegaBERT: Pre-training of Segment-aware BERT for Language Understanding](https://arxiv.org/abs/2004.14996)
- [DisSent: Sentence Representation Learning from Explicit Discourse Relations](https://arxiv.org/abs/1710.04334) (ACL2019)
- [Pretraining with Contrastive Sentence Objectives Improves Discourse Performance of Language Models](https://arxiv.org/abs/2005.10389) (ACL2020)
- [CAPT: Contrastive Pre-Training for Learning Denoised Sequence Representations](https://arxiv.org/abs/2010.06351)
- [SLM: Learning a Discourse Language Representation with Sentence Unshuffling](https://arxiv.org/abs/2010.16249) (EMNLP2020)
- [CausalBERT: Injecting Causal Knowledge Into Pre-trained Models with Minimal Supervision](https://arxiv.org/abs/2107.09852)
- [StructBERT: Incorporating Language Structures into Pre-training for Deep Language Understanding](https://arxiv.org/abs/1908.04577) (ICLR2020)
- [Structural Pre-training for Dialogue Comprehension](https://arxiv.org/abs/2105.10956) (ACL2021)
- [Retrofitting Structure-aware Transformer Language Model for End Tasks](https://arxiv.org/abs/2009.07408) (EMNLP2020)
- [Syntax-Enhanced Pre-trained Model](https://arxiv.org/abs/2012.14116)
- [Syntax-Infused Transformer and BERT models for Machine Translation and Natural Language Understanding](https://arxiv.org/abs/1911.06156)
- [Do Syntax Trees Help Pre-trained Transformers Extract Information?](https://arxiv.org/abs/2008.09084)
- [SenseBERT: Driving Some Sense into BERT](https://arxiv.org/abs/1908.05646)
- [Semantics-aware BERT for Language Understanding](https://arxiv.org/abs/1909.02209) (AAAI2020)
- [GiBERT: Introducing Linguistic Knowledge into BERT through a Lightweight Gated Injection Method](https://arxiv.org/abs/2010.12532)
- [K-BERT: Enabling Language Representation with Knowledge Graph](https://arxiv.org/abs/1909.07606)
- [Knowledge Enhanced Contextual Word Representations](https://arxiv.org/abs/1909.04164) (EMNLP2019)
- [Knowledge-Aware Language Model Pretraining](https://arxiv.org/abs/2007.00655)
- [K-Adapter: Infusing Knowledge into Pre-Trained Models with Adapters](https://arxiv.org/abs/2002.01808)
- [JAKET: Joint Pre-training of Knowledge Graph and Language Understanding](https://arxiv.org/abs/2010.00796)
- [E-BERT: Efficient-Yet-Effective Entity Embeddings for BERT](https://arxiv.org/abs/1911.03681) (EMNLP2020)
- [KEPLER: A Unified Model for Knowledge Embedding and Pre-trained Language Representation](https://arxiv.org/abs/1911.06136)
- [Entities as Experts: Sparse Memory Access with Entity Supervision](https://arxiv.org/abs/2004.07202) (EMNLP2020)
- [Exploiting Structured Knowledge in Text via Graph-Guided Representation Learning](https://www.aclweb.org/anthology/2020.emnlp-main.722/) (EMNLP2020)
- [Contextualized Representations Using Textual Encyclopedic Knowledge](https://arxiv.org/abs/2004.12006)
- [CoLAKE: Contextualized Language and Knowledge Embedding](https://arxiv.org/abs/2010.00309) (COLING2020)
- [KI-BERT: Infusing Knowledge Context for Better Language and Domain Understanding](https://arxiv.org/abs/2104.08145)
- [K-XLNet: A General Method for Combining Explicit Knowledge with Language Model Pretraining](https://arxiv.org/abs/2104.10649)
- [Combining pre-trained language models and structured knowledge](https://arxiv.org/abs/2101.12294)
- [Coarse-to-Fine Pre-training for Named Entity Recognition](https://arxiv.org/abs/2010.08210) (EMNLP2020)
- [E.T.: Entity-Transformers. Coreference augmented Neural Language Model for richer mention representations via Entity-Transformer blocks](https://arxiv.org/abs/2011.05431) (COLING2020 WS)
- [REALM: Retrieval-Augmented Language Model Pre-Training](https://arxiv.org/abs/2002.08909) (ICML2020) [[github](https://github.com/google-research/language/tree/master/language/realm)]
- [Simple and Efficient ways to Improve REALM](https://arxiv.org/abs/2104.08710)
- [Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks](https://arxiv.org/abs/2005.11401) (NeurIPS2020)
- [Fine-tune the Entire RAG Architecture (including DPR retriever) for Question-Answering](https://arxiv.org/abs/2106.11517)
- [Joint Retrieval and Generation Training for Grounded Text Generation](https://arxiv.org/abs/2105.06597)
- [Retrieval Augmentation Reduces Hallucination in Conversation](https://arxiv.org/abs/2104.07567)
- [On-The-Fly Information Retrieval Augmentation for Language Models](https://arxiv.org/abs/2007.01528)
- [Current Limitations of Language Models: What You Need is Retrieval](https://arxiv.org/abs/2009.06857)
- [Improving language models by retrieving from trillions of tokens](https://arxiv.org/abs/2112.04426) [[blog](https://deepmind.com/research/publications/2021/improving-language-models-by-retrieving-from-trillions-of-tokens)] [[blog](http://jalammar.github.io/illustrated-retrieval-transformer/)]
- [Taking Notes on the Fly Helps BERT Pre-training](https://arxiv.org/abs/2008.01466)
- [Pre-training via Paraphrasing](https://arxiv.org/abs/2006.15020)
- [SKEP: Sentiment Knowledge Enhanced Pre-training for Sentiment Analysis](https://arxiv.org/abs/2005.05635) (ACL2020)
- [Improving Event Duration Prediction via Time-aware Pre-training](https://arxiv.org/abs/2011.02610) (EMNLP2020 Findings)
- [Knowledge-Aware Procedural Text Understanding with Multi-Stage Training](https://arxiv.org/abs/2009.13199)
- [Poly-encoders: Transformer Architectures and Pre-training Strategies for Fast and Accurate Multi-sentence Scoring](https://arxiv.org/abs/1905.01969) (ICLR2020)
- [Rethinking Positional Encoding in Language Pre-training](https://arxiv.org/abs/2006.15595)
- [Improve Transformer Models with Better Relative Position Embeddings](https://arxiv.org/abs/2009.13658) (EMNLP2020 Findings)
- [RoFormer: Enhanced Transformer with Rotary Position Embedding](https://arxiv.org/abs/2104.09864)
- [Position Information in Transformers: An Overview](https://arxiv.org/abs/2102.11090)
- [BoostingBERT:Integrating Multi-Class Boosting into BERT for NLP Tasks](https://arxiv.org/abs/2009.05959)
- [BURT: BERT-inspired Universal Representation from Twin Structure](https://arxiv.org/abs/2004.13947)
- [Universal Text Representation from BERT: An Empirical Study](https://arxiv.org/abs/1910.07973)
- [Symmetric Regularization based BERT for Pair-wise Semantic Reasoning](https://arxiv.org/abs/1909.03405) (SIGIR2020)
- [Beyond 512 Tokens: Siamese Multi-depth Transformer-based Hierarchical Encoder for Document Matching](https://arxiv.org/abs/2004.12297)
- [Hi-Transformer: Hierarchical Interactive Transformer for Efficient and Effective Long Document Modeling](https://arxiv.org/abs/2106.01040) (ACL2021)
- [Transfer Fine-Tuning: A BERT Case Study](https://arxiv.org/abs/1909.00931) (EMNLP2019)
- [Improving Pre-Trained Multilingual Models with Vocabulary Expansion](https://arxiv.org/abs/1909.12440) (CoNLL2019)
- [BERTRAM: Improved Word Embeddings Have Big Impact on Contextualized Model Performance](https://arxiv.org/abs/1910.07181) (ACL2020)
- [A Mixture of h−1 Heads is Better than h Heads](https://arxiv.org/abs/2005.06537) (ACL2020)
- [SesameBERT: Attention for Anywhere](https://arxiv.org/abs/1910.03176)
- [Multi-Head Attention: Collaborate Instead of Concatenate](https://arxiv.org/abs/2006.16362)
- [DeBERTa: Decoding-enhanced BERT with Disentangled Attention](https://arxiv.org/abs/2006.03654) [[github](https://github.com/microsoft/DeBERTa)]
- [Deepening Hidden Representations from Pre-trained Language Models](https://arxiv.org/abs/1911.01940)
- [On the Transformer Growth for Progressive BERT Training](https://arxiv.org/abs/2010.12562)
- [Improving BERT with Self-Supervised Attention](https://arxiv.org/abs/2004.03808)
- [Guiding Attention for Self-Supervised Learning with Transformers](https://arxiv.org/abs/2010.02399) (EMNLP2020 Findings)
- [Improving Disfluency Detection by Self-Training a Self-Attentive Model](https://arxiv.org/abs/2004.05323)
- [Self-training Improves Pre-training for Natural Language Understanding](https://arxiv.org/abs/2010.02194) [[github](https://github.com/facebookresearch/SentAugment)]
- [CERT: Contrastive Self-supervised Learning for Language Understanding](https://arxiv.org/abs/2005.12766)
- [Robust Transfer Learning with Pretrained Language Models through Adapters](https://arxiv.org/abs/2108.02340) (ACL2021)
- [ReadOnce Transformers: Reusable Representations of Text for Transformers](https://arxiv.org/abs/2010.12854) (ACL2021)
- [LV-BERT: Exploiting Layer Variety for BERT](https://arxiv.org/abs/2106.11740) (ACL2021 Findings) [[github](https://github.com/yuweihao/LV-BERT)]
- [Large Product Key Memory for Pretrained Language Models](https://arxiv.org/abs/2010.03881) (EMNLP2020 Findings)
- [Enhancing Pre-trained Language Model with Lexical Simplification](https://arxiv.org/abs/2012.15070)
- [Contextual BERT: Conditioning the Language Model Using a Global State](https://arxiv.org/abs/2010.15778) (COLING2020 WS)
- [SMART: Robust and Efficient Fine-Tuning for Pre-trained Natural Language Models through Principled Regularized Optimization](https://arxiv.org/abs/1911.03437) (ACL2020)
- [Raise a Child in Large Language Model: Towards Effective and Generalizable Fine-tuning](https://arxiv.org/abs/2109.05687) (EMNLP2021) [[github](https://github.com/RunxinXu/ChildTuning)]
- [Token Dropping for Efficient BERT Pretraining](https://arxiv.org/abs/2203.13240) (ACL2022) [[github](https://github.com/tensorflow/models/tree/master/official/projects/token_dropping)]
- [Pay Attention to MLPs](https://arxiv.org/abs/2105.08050)
- [Are Pre-trained Convolutions Better than Pre-trained Transformers?](https://arxiv.org/abs/2105.03322) (ACL2021)
- [Pre-Training a Language Model Without Human Language](https://arxiv.org/abs/2012.11995)
### Tokenization
- [Training Multilingual Pre-trained Language Model with Byte-level Subwords](https://arxiv.org/abs/2101.09012)
- [Byte Pair Encoding is Suboptimal for Language Model Pretraining](https://arxiv.org/abs/2004.03720) (EMNLP2020 Findings)
- [CANINE: Pre-training an Efficient Tokenization-Free Encoder for Language Representation](https://arxiv.org/abs/2103.06874) (TACL2022) [[github](https://github.com/google-research/language/tree/master/language/canine)]
- [ByT5: Towards a token-free future with pre-trained byte-to-byte models](https://arxiv.org/abs/2105.13626) (TACL2022) [[github](https://github.com/google-research/byt5)]
- [Multi-view Subword Regularization](https://arxiv.org/abs/2103.08490) (NAACL2021)
- [Bridging Subword Gaps in Pretrain-Finetune Paradigm for Natural Language Generation](https://arxiv.org/abs/2106.06125) (ACL2021)
- [An Empirical Study of Tokenization Strategies for Various Korean NLP Tasks](https://arxiv.org/abs/2010.02534) (AACL-IJCNLP2020)
- [AMBERT: A Pre-trained Language Model with Multi-Grained Tokenization](https://arxiv.org/abs/2008.11869)
- [LICHEE: Improving Language Model Pre-training with Multi-grained Tokenization](https://arxiv.org/abs/2108.00801) (ACL2021 Findings)
- [Lattice-BERT: Leveraging Multi-Granularity Representations in Chinese Pre-trained Language Models](https://arxiv.org/abs/2104.07204) (NAACL2021)
- [CharBERT: Character-aware Pre-trained Language Model](https://arxiv.org/abs/2011.01513) (COLING2020) [[github](https://github.com/wtma/CharBERT)]
- [CharacterBERT: Reconciling ELMo and BERT for Word-Level Open-Vocabulary Representations From Characters](https://arxiv.org/abs/2010.10392) (COLING2020)
- [Charformer: Fast Character Transformers via Gradient-based Subword Tokenization](https://arxiv.org/abs/2106.12672) [[github](https://github.com/google-research/google-research/tree/master/charformer)]
- [Fast WordPiece Tokenization](https://arxiv.org/abs/2012.15524) (EMNLP2021)
- [MaxMatch-Dropout: Subword Regularization for WordPiece](https://arxiv.org/abs/2209.04126) (COLING2022)
### Prompt
- [Pre-train, Prompt, and Predict: A Systematic Survey of Prompting Methods in Natural Language Processing](https://arxiv.org/abs/2107.13586)
- [AutoPrompt: Eliciting Knowledge from Language Models with Automatically Generated Prompts](https://arxiv.org/abs/2010.15980) (EMNLP2020) [[github](https://github.com/ucinlp/autoprompt)]
- [Calibrate Before Use: Improving Few-Shot Performance of Language Models](https://arxiv.org/abs/2102.09690)
- [Prompt Programming for Large Language Models: Beyond the Few-Shot Paradigm](https://arxiv.org/abs/2102.07350)
- [GPT Understands, Too](https://arxiv.org/abs/2103.10385) [[github](https://github.com/THUDM/P-tuning)]
- [How Many Data Points is a Prompt Worth?](https://arxiv.org/abs/2103.08493) (NAACL2021) [[website](https://huggingface.co/blog/how_many_data_points/)]
- [Learning How to Ask: Querying LMs with Mixtures of Soft Prompts](https://arxiv.org/abs/2104.06599) (NAACL2021)
- [Meta-tuning Language Models to Answer Prompts Better](https://arxiv.org/abs/2104.04670)
- [Fantastically Ordered Prompts and Where to Find Them: Overcoming Few-Shot Prompt Order Sensitivity](https://arxiv.org/abs/2104.08786)
- [The Power of Scale for Parameter-Efficient Prompt Tuning](https://arxiv.org/abs/2104.08691) (EMNLP2021)
- [Cutting Down on Prompts and Parameters: Simple Few-Shot Learning with Language Models](https://arxiv.org/abs/2106.13353)
- [PPT: Pre-trained Prompt Tuning for Few-shot Learning](https://arxiv.org/abs/2109.04332)
- [True Few-Shot Learning with Language Models](https://arxiv.org/abs/2105.11447)
- [Few-shot Sequence Learning with Transformers](https://arxiv.org/abs/2012.09543) (NeurIPS2020 WS)
- [PTR: Prompt Tuning with Rules for Text Classification](https://arxiv.org/abs/2105.11259)
- [Knowledgeable Prompt-tuning: Incorporating Knowledge into Prompt Verbalizer for Text Classification](https://arxiv.org/abs/2108.02035)
- [Discrete and Soft Prompting for Multilingual Models](https://arxiv.org/abs/2109.03630) (EMNLP2021)
- [Reframing Instructional Prompts to GPTk's Language](https://arxiv.org/abs/2109.07830)
- [Multimodal Few-Shot Learning with Frozen Language Models](https://arxiv.org/abs/2106.13884)
- [FLEX: Unifying Evaluation for Few-Shot NLP](https://arxiv.org/abs/2107.07170)
- [Do Prompt-Based Models Really Understand the Meaning of their Prompts?](https://arxiv.org/abs/2109.01247)
- [OpenPrompt: An Open-source Framework for Prompt-learning](https://aclanthology.org/2022.acl-demo.10/) (ACL2022 Demo)
## Sentence embedding
- [Sentence Encoders on STILTs: Supplementary Training on Intermediate Labeled-data Tasks](https://arxiv.org/abs/1811.01088)
- [Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks](https://arxiv.org/abs/1908.10084) (EMNLP2019)
- [Parameter-free Sentence Embedding via Orthogonal Basis](https://arxiv.org/abs/1810.00438) (EMNLP2019)
- [SBERT-WK: A Sentence Embedding Method By Dissecting BERT-based Word Models](https://arxiv.org/abs/2002.06652)
- [On the Sentence Embeddings from Pre-trained Language Models](https://arxiv.org/abs/2011.05864) (EMNLP2020)
- [Semantic Re-tuning with Contrastive Tension](https://openreview.net/forum?id=Ov_sMNau-PF) (ICLR2021)
- [DeCLUTR: Deep Contrastive Learning for Unsupervised Textual Representations](https://arxiv.org/abs/2006.03659) (ACL2021)
- [ConSERT: A Contrastive Framework for Self-Supervised Sentence Representation Transfer](https://arxiv.org/abs/2105.11741) (ACL2021)
- [CLEAR: Contrastive Learning for Sentence Representation](https://arxiv.org/abs/2012.15466)
- [SimCSE: Simple Contrastive Learning of Sentence Embeddings](https://arxiv.org/abs/2104.08821) (EMNLP2021) [[github](https://github.com/princeton-nlp/simcse)]
- [ESimCSE: Enhanced Sample Building Method for Contrastive Learning of Unsupervised Sentence Embedding](https://arxiv.org/abs/2109.04380)
- [Fast, Effective, and Self-Supervised: Transforming Masked Language Models into Universal Lexical and Sentence Encoders](https://arxiv.org/abs/2104.08027) (EMNLP2021) [[github](https://github.com/cambridgeltl/mirror-bert)]
- [TSDAE: Using Transformer-based Sequential Denoising Auto-Encoder for Unsupervised Sentence Embedding Learning](https://arxiv.org/abs/2104.06979) (EMNLP2021 Findings)
- [Trans-Encoder: Unsupervised sentence-pair modelling through self- and mutual-distillations](https://arxiv.org/abs/2109.13059) [[github](https://github.com/amzn/trans-encoder)]
- [Whitening Sentence Representations for Better Semantics and Faster Retrieval](https://arxiv.org/abs/2103.15316) [[github](https://github.com/bojone/BERT-whitening)]
- [Augmented SBERT: Data Augmentation Method for Improving Bi-Encoders for Pairwise Sentence Scoring Tasks](https://arxiv.org/abs/2010.08240) (NAACL2021)
- [DiffCSE: Difference-based Contrastive Learning for Sentence Embeddings](https://arxiv.org/abs/2204.10298) (NAACL2022) [[code](https://github.com/voidism/DiffCSE)]
- [Unsupervised Sentence Representation via Contrastive Learning with Mixing Negatives](https://ojs.aaai.org/index.php/AAAI/article/view/21428) (AAAI2022) [[github](https://github.com/BDBC-KG-NLP/MixCSE_AAAI2022)]
- [Sentence Embeddings by Ensemble Distillation](https://arxiv.org/abs/2104.06719)
- [EASE: Entity-Aware Contrastive Learning of Sentence Embedding](https://arxiv.org/abs/2205.04260) (NAACL2022)
- [Sentence-T5: Scalable Sentence Encoders from Pre-trained Text-to-Text Models](https://arxiv.org/abs/2108.08877)
- [Dual-View Distilled BERT for Sentence Embedding](https://arxiv.org/abs/2104.08675) (SIGIR2021)
- [DefSent: Sentence Embeddings using Definition Sentences](https://arxiv.org/abs/2105.04339) (ACL2021)
- [Paraphrastic Representations at Scale](https://arxiv.org/abs/2104.15114) [[github](https://github.com/jwieting/paraphrastic-representations-at-scale)]
- [Learning Dense Representations of Phrases at Scale](https://arxiv.org/abs/2012.12624) (ACL2021) [[github](https://github.com/princeton-nlp/DensePhrases)]
- [Phrase-BERT: Improved Phrase Embeddings from BERT with an Application to Corpus Exploration](https://arxiv.org/abs/2109.06304) (EMNLP2021)
## Transformer variants
- [Efficient Transformers: A Survey](https://arxiv.org/abs/2009.06732)
- [Adaptive Attention Span in Transformers](https://arxiv.org/abs/1905.07799) (ACL2019)
- [Transformer-XL: Attentive Language Models Beyond a Fixed-Length Context](https://arxiv.org/abs/1901.02860) (ACL2019) [[github](https://github.com/kimiyoung/transformer-xl)]
- [Generating Long Sequences with Sparse Transformers](https://arxiv.org/abs/1904.10509)
- [Do Transformers Need Deep Long-Range Memory](https://arxiv.org/abs/2007.03356) (ACL2020)
- [DA-Transformer: Distance-aware Transformer](https://arxiv.org/abs/2010.06925) (NAACL2021)
- [Adaptively Sparse Transformers](https://arxiv.org/abs/1909.00015) (EMNLP2019)
- [Compressive Transformers for Long-Range Sequence Modelling](https://arxiv.org/abs/1911.05507)
- [The Evolved Transformer](https://arxiv.org/abs/1901.11117) (ICML2019)
- [Reformer: The Efficient Transformer](https://arxiv.org/abs/2001.04451) (ICLR2020) [[github](https://github.com/google/trax/tree/master/trax/models/reformer)]
- [GRET: Global Representation Enhanced Transformer](https://arxiv.org/abs/2002.10101) (AAAI2020)
- [GMAT: Global Memory Augmentation for Transformers](https://arxiv.org/abs/2006.03274)
- [Memory Transformer](https://arxiv.org/abs/2006.11527)
- [Transformer on a Diet](https://arxiv.org/abs/2002.06170) [[github](https://github.com/cgraywang/transformer-on-diet)]
- [A Tensorized Transformer for Language Modeling](https://arxiv.org/abs/1906.09777) (NeurIPS2019)
- [DeFINE: DEep Factorized INput Token Embeddings for Neural Sequence Modeling](https://arxiv.org/abs/1911.12385) (ICLR2020) [[github](https://github.com/sacmehta/delight)]
- [DeLighT: Very Deep and Light-weight Transformer](https://arxiv.org/abs/2008.00623) [[github](https://github.com/sacmehta/delight)]
- [Lite Transformer with Long-Short Range Attention](https://arxiv.org/abs/2004.11886) [[github](https://github.com/mit-han-lab/lite-transformer)] (ICLR2020)
- [Efficient Content-Based Sparse Attention with Routing Transformers](https://openreview.net/forum?id=B1gjs6EtDr)
- [BP-Transformer: Modelling Long-Range Context via Binary Partitioning](https://arxiv.org/abs/1911.04070)
- [Longformer: The Long-Document Transformer](https://arxiv.org/abs/2004.05150) [[github](https://github.com/allenai/longformer)]
- [Big Bird: Transformers for Longer Sequences](https://arxiv.org/abs/2007.14062)
- [Informer: Beyond Efficient Transformer for Long Sequence Time-Series Forecasting](https://arxiv.org/abs/2012.07436) (AAAI2021)
- [Nyströmformer: A Nyström-Based Algorithm for Approximating Self-Attention](https://arxiv.org/abs/2102.03902) (AAAI2021) [[github](https://github.com/mlpen/Nystromformer)]
- [Improving Transformer Models by Reordering their Sublayers](https://arxiv.org/abs/1911.03864) (ACL2020)
- [Highway Transformer: Self-Gating Enhanced Self-Attentive Networks](https://arxiv.org/abs/2004.08178)
- [Mask Attention Networks: Rethinking and Strengthen Transformer](https://arxiv.org/abs/2103.13597) (NAACL2021)
- [Synthesizer: Rethinking Self-Attention in Transformer Models](https://arxiv.org/abs/2005.00743)
- [Query-Key Normalization for Transformers](https://arxiv.org/abs/2010.04245) (EMNLP2020 Findings)
- [Rethinking Attention with Performers](https://arxiv.org/abs/2009.14794) (ICLR2021)
- [FlashAttention: Fast and Memory-Efficient Exact Attention with IO-Awareness](https://arxiv.org/abs/2205.14135)
- [Dynamically Adjusting Transformer Batch Size by Monitoring Gradient Direction Change](https://arxiv.org/abs/2005.02008)
- [HAT: Hardware-Aware Transformers for Efficient Natural Language Processing](https://arxiv.org/abs/2005.14187) (ACL2020) [[github](https://github.com/mit-han-lab/hardware-aware-transformers)]
- [Linformer: Self-Attention with Linear Complexity](https://arxiv.org/abs/2006.04768)
- [What's Hidden in a One-layer Randomly Weighted Transformer?](https://arxiv.org/abs/2109.03939) (EMNLP2021)
- [Transformers are RNNs: Fast Autoregressive Transformers with Linear Attention](https://arxiv.org/abs/2006.16236)
- [Understanding the Difficulty of Training Transformers](https://arxiv.org/abs/2004.08249) (EMNLP2020)
- [Towards Fully 8-bit Integer Inference for the Transformer Model](https://arxiv.org/abs/2009.08034) (IJCAI2020)
- [Extremely Low Bit Transformer Quantization for On-Device Neural Machine Translation](https://arxiv.org/abs/2009.07453)
- [Long Range Arena: A Benchmark for Efficient Transformers](https://arxiv.org/abs/2011.04006)
## Probe
- [A Structural Probe for Finding Syntax in Word Representations](https://aclweb.org/anthology/papers/N/N19/N19-1419/) (NAACL2019)
- [When Bert Forgets How To POS: Amnesic Probing of Linguistic Properties and MLM Predictions](https://arxiv.org/abs/2006.00995)
- [Finding Universal Grammatical Relations in Multilingual BERT](https://arxiv.org/abs/2005.04511) (ACL2020)
- [Probing Multilingual BERT for Genetic and Typological Signals](https://arxiv.org/abs/2011.02070) (COLING2020)
- [Linguistic Knowledge and Transferability of Contextual Representations](https://arxiv.org/abs/1903.08855) (NAACL2019) [[github](https://github.com/nelson-liu/contextual-repr-analysis)]
- [Probing What Different NLP Tasks Teach Machines about Function Word Comprehension](https://arxiv.org/abs/1904.11544) (*SEM2019)
- [BERT Rediscovers the Classical NLP Pipeline](https://arxiv.org/abs/1905.05950) (ACL2019)
- [A Closer Look at How Fine-tuning Changes BERT](https://arxiv.org/abs/2106.14282) (ACL2022)
- [Mediators in Determining what Processing BERT Performs First](https://arxiv.org/abs/2104.06400) (NAACL2021)
- [Probing Neural Network Comprehension of Natural Language Arguments](https://arxiv.org/abs/1907.07355) (ACL2019)
- [Cracking the Contextual Commonsense Code: Understanding Commonsense Reasoning Aptitude of Deep Contextual Representations](https://arxiv.org/abs/1910.01157) (EMNLP2019 WS)
- [What do you mean, BERT? Assessing BERT as a Distributional Semantics Model](https://arxiv.org/abs/1911.05758)
- [Quantity doesn't buy quality syntax with neural language models](https://arxiv.org/abs/1909.00111) (EMNLP2019)
- [Are Pre-trained Language Models Aware of Phrases? Simple but Strong Baselines for Grammar Induction](https://openreview.net/forum?id=H1xPR3NtPB) (ICLR2020)
- [Discourse Probing of Pretrained Language Models](https://arxiv.org/abs/2104.05882) (NAACL2021)
- [oLMpics -- On what Language Model Pre-training Captures](https://arxiv.org/abs/1912.13283)
- [Do Neural Language Models Show Preferences for Syntactic Formalisms?](https://arxiv.org/abs/2004.14096) (ACL2020)
- [Probing for Predicate Argument Structures in Pretrained Language Models](https://aclanthology.org/2022.acl-long.316/) (ACL2022)
- [Perturbed Masking: Parameter-free Probing for Analyzing and Interpreting BERT](https://arxiv.org/abs/2004.14786) (ACL2020)
- [Intermediate-Task Transfer Learning with Pretrained Models for Natural Language Understanding: When and Why Does It Work?](https://arxiv.org/abs/2005.00628) (ACL2020)
- [Probing Linguistic Systematicity](https://arxiv.org/abs/2005.04315) (ACL2020)
- [A Matter of Framing: The Impact of Linguistic Formalism on Probing Results](https://arxiv.org/abs/2004.14999)
- [A Cross-Task Analysis of Text Span Representations](https://www.aclweb.org/anthology/2020.repl4nlp-1.20/) (ACL2020 WS)
- [When Do You Need Billions of Words of Pretraining Data?](https://arxiv.org/abs/2011.04946) [[github](https://github.com/nyu-mll/pretraining-learning-curves)]
- [Picking BERT's Brain: Probing for Linguistic Dependencies in Contextualized Embeddings Using Representational Similarity Analysis](https://arxiv.org/abs/2011.12073)
- [Language Models as Knowledge Bases?](https://arxiv.org/abs/1909.01066) (EMNLP2019) [[github](https://github.com/facebookresearch/LAMA)]
- [BERT is Not a Knowledge Base (Yet): Factual Knowledge vs. Name-Based Reasoning in Unsupervised QA](https://arxiv.org/abs/1911.03681)
- [How Much Knowledge Can You Pack Into the Parameters of a Language Model?](https://arxiv.org/abs/2002.08910) (EMNLP2020)
- [Language Models as Knowledge Bases: On Entity Representations, Storage Capacity, and Paraphrased Queries](https://arxiv.org/abs/2008.09036) (EACL2021)
- [Factual Probing Is [MASK]: Learning vs. Learning to Recall](https://arxiv.org/abs/2104.05240) (NAACL2021) [[github](https://github.com/princeton-nlp/OptiPrompt)]
- [Knowledge Neurons in Pretrained Transformers](https://arxiv.org/abs/2104.08696)
- [DirectProbe: Studying Representations without Classifiers](https://www.aclweb.org/anthology/2021.naacl-main.401/) (NAACL2021)
- [The Language Model Understood the Prompt was Ambiguous: Probing Syntactic Uncertainty Through Generation](https://arxiv.org/abs/2109.07848) (EMNLP2021 WS)
- [X-FACTR: Multilingual Factual Knowledge Retrieval from Pretrained Language Models](https://arxiv.org/abs/2010.06189) (EMNLP2020)
- [Probing BERT in Hyperbolic Spaces](https://arxiv.org/abs/2104.03869) (ICLR2021)
- [Probing Across Time: What Does RoBERTa Know and When?](https://arxiv.org/abs/2104.07885)
- [Do NLP Models Know Numbers? Probing Numeracy in Embeddings](https://arxiv.org/abs/1909.07940) (EMNLP2019)
- [Birds have four legs?! NumerSense: Probing Numerical Commonsense Knowledge of Pre-trained Language Models](https://arxiv.org/abs/2005.00683) [[github](https://github.com/INK-USC/NumerSense)] [[website](https://inklab.usc.edu/NumerSense/)]
- [Negated and Misprimed Probes for Pretrained Language Models: Birds Can Talk, But Cannot Fly](https://www.aclweb.org/anthology/2020.acl-main.698/) (ACL2020)
- [How is BERT surprised? Layerwise detection of linguistic anomalies](https://arxiv.org/abs/2105.07452) (ACL2021)
- [Exploring the Role of BERT Token Representations to Explain Sentence Probing Results](https://arxiv.org/abs/2104.01477)
- [What Does My QA Model Know? Devising Controlled Probes using Expert Knowledge](https://arxiv.org/abs/1912.13337)
- [A Pairwise Probe for Understanding BERT Fine-Tuning on Machine Reading Comprehension](https://arxiv.org/abs/2006.01346)
- [Can BERT Reason? Logically Equivalent Probes for Evaluating the Inference Capabilities of Language Models](https://arxiv.org/abs/2005.00782)
- [Probing Task-Oriented Dialogue Representation from Language Models](https://arxiv.org/abs/2010.13912) (EMNLP2020)
- [Probing for Bridging Inference in Transformer Language Models](https://arxiv.org/abs/2104.09400)
- [BERTering RAMS: What and How Much does BERT Already Know About Event Arguments? -- A Study on the RAMS Dataset](https://arxiv.org/abs/2010.04098) (EMNLP2020 WS)
- [CxGBERT: BERT meets Construction Grammar](https://arxiv.org/abs/2011.04134) (COLING2020) [[github](https://github.com/H-TayyarMadabushi/CxGBERT-BERT-meets-Construction-Grammar)]
- [BERT is to NLP what AlexNet is to CV: Can Pre-Trained Language Models Identify Analogies?](https://arxiv.org/abs/2105.04949) (ACL2021)
## Inside BERT
- [What does BERT learn about the structure of language?](https://hal.inria.fr/hal-02131630/document) (ACL2019)
- [Analyzing Multi-Head Self-Attention: Specialized Heads Do the Heavy Lifting, the Rest Can Be Pruned](https://arxiv.org/abs/1905.09418) (ACL2019) [[github](https://github.com/lena-voita/the-story-of-heads)]
- [Multi-head or Single-head? An Empirical Comparison for Transformer Training](https://arxiv.org/abs/2106.09650)
- [Open Sesame: Getting Inside BERT's Linguistic Knowledge](https://arxiv.org/abs/1906.01698) (ACL2019 WS)
- [Analyzing the Structure of Attention in a Transformer Language Model](https://arxiv.org/abs/1906.04284) (ACL2019 WS)
- [What Does BERT Look At? An Analysis of BERT's Attention](https://arxiv.org/abs/1906.04341) (ACL2019 WS)
- [Do Attention Heads in BERT Track Syntactic Dependencies?](https://arxiv.org/abs/1911.12246)
- [Blackbox meets blackbox: Representational Similarity and Stability Analysis of Neural Language Models and Brains](https://arxiv.org/abs/1906.01539) (ACL2019 WS)
- [Inducing Syntactic Trees from BERT Representations](https://arxiv.org/abs/1906.11511) (ACL2019 WS)
- [A Multiscale Visualization of Attention in the Transformer Model](https://arxiv.org/abs/1906.05714) (ACL2019 Demo)
- [Visualizing and Measuring the Geometry of BERT](https://arxiv.org/abs/1906.02715)
- [How Contextual are Contextualized Word Representations? Comparing the Geometry of BERT, ELMo, and GPT-2 Embeddings](https://arxiv.org/abs/1909.00512) (EMNLP2019)
- [Are Sixteen Heads Really Better than One?](https://arxiv.org/abs/1905.10650) (NeurIPS2019)
- [On the Validity of Self-Attention as Explanation in Transformer Models](https://arxiv.org/abs/1908.04211)
- [Visualizing and Understanding the Effectiveness of BERT](https://arxiv.org/abs/1908.05620) (EMNLP2019)
- [Attention Interpretability Across NLP Tasks](https://arxiv.org/abs/1909.11218)
- [Revealing the Dark Secrets of BERT](https://arxiv.org/abs/1908.08593) (EMNLP2019)
- [Analyzing Redundancy in Pretrained Transformer Models](https://arxiv.org/abs/2004.04010) (EMNLP2020)
- [What's so special about BERT's layers? A closer look at the NLP pipeline in monolingual and multilingual models](https://arxiv.org/abs/2004.06499)
- [Attention Module is Not Only a Weight: Analyzing Transformers with Vector Norms](https://arxiv.org/abs/2004.10102) (ACL2020 SRW)
- [Incorporating Residual and Normalization Layers into Analysis of Masked Language Models](https://arxiv.org/abs/2109.07152) (EMNLP2021)
- [Quantifying Attention Flow in Transformers](https://arxiv.org/abs/2005.00928)
- [Telling BERT's full story: from Local Attention to Global Aggregation](https://arxiv.org/abs/2004.05916) (EACL2021)
- [How Far Does BERT Look At:Distance-based Clustering and Analysis of BERT′s Attention](https://arxiv.org/abs/2011.00943)
- [Contributions of Transformer Attention Heads in Multi- and Cross-lingual Tasks](https://arxiv.org/abs/2108.08375) (ACL2021)
- [What Do Position Embeddings Learn? An Empirical Study of Pre-Trained Language Model Positional Encoding](https://arxiv.org/abs/2010.04903) (EMNLP2020)
- [Investigating BERT's Knowledge of Language: Five Analysis Methods with NPIs](https://arxiv.org/abs/1909.02597) (EMNLP2019)
- [Are Pretrained Language Models Symbolic Reasoners Over Knowledge?](https://arxiv.org/abs/2006.10413) (CoNLL2020)
- [Rethinking the Value of Transformer Components](https://arxiv.org/abs/2011.03803) (COLING2020)
- [Transformer Feed-Forward Layers Are Key-Value Memories](https://arxiv.org/abs/2012.14913)
- [Transformer Feed-Forward Layers Build Predictions by Promoting Concepts in the Vocabulary Space](https://arxiv.org/abs/2203.14680)
- [Investigating Transferability in Pretrained Language Models](https://arxiv.org/abs/2004.14975)
- [What Happens To BERT Embeddings During Fine-tuning?](https://arxiv.org/abs/2004.14448)
- [Analyzing Individual Neurons in Pre-trained Language Models](https://arxiv.org/abs/2010.02695) (EMNLP2020)
- [How fine can fine-tuning be? Learning efficient language models](https://arxiv.org/abs/2004.14129) (AISTATS2020)
- [The Bottom-up Evolution of Representations in the Transformer: A Study with Machine Translation and Language Modeling Objectives](https://arxiv.org/abs/1909.01380) (EMNLP2019)
- [A Primer in BERTology: What we know about how BERT works](https://arxiv.org/abs/2002.12327) (TACL2020)
- [Pretrained Language Model Embryology: The Birth of ALBERT](https://arxiv.org/abs/2010.02480) (EMNLP2020) [[github](https://github.com/d223302/albert-embryology)]
- [Evaluating Saliency Methods for Neural Language Models](https://arxiv.org/abs/2104.05824) (NAACL2021)
- [Investigating Gender Bias in BERT](https://arxiv.org/abs/2009.05021)
- [Measuring and Reducing Gendered Correlations in Pre-trained Models](https://arxiv.org/abs/2010.06032) [[website](https://ai.googleblog.com/2020/10/measuring-gendered-correlations-in-pre.html)]
- [Unmasking Contextual Stereotypes: Measuring and Mitigating BERT's Gender Bias](https://arxiv.org/abs/2010.14534) (COLING2020 WS)
- [Stereotype and Skew: Quantifying Gender Bias in Pre-trained and Fine-tuned Language Models](https://arxiv.org/abs/2101.09688) (EACL2021)
- [CrowS-Pairs: A Challenge Dataset for Measuring Social Biases in Masked Language Models](https://arxiv.org/abs/2010.00133) (EMNLP2020)
- [Unmasking the Mask -- Evaluating Social Biases in Masked Language Models](https://arxiv.org/abs/2104.07496)
- [BERT Knows Punta Cana is not just beautiful, it's gorgeous: Ranking Scalar Adjectives with Contextualised Representations](https://arxiv.org/abs/2010.02686) (EMNLP2020)
- [Does Chinese BERT Encode Word Structure?](https://arxiv.org/abs/2010.07711) (COLING2020) [[github](https://github.com/ylwangy/BERT_zh_Analysis)]
- [How Does BERT Answer Questions? A Layer-Wise Analysis of Transformer Representations](https://arxiv.org/abs/1909.04925) (CIKM2019)
- [Whatcha lookin' at? DeepLIFTing BERT's Attention in Question Answering](https://arxiv.org/abs/1910.06431)
- [What does BERT Learn from Multiple-Choice Reading Comprehension Datasets?](https://arxiv.org/abs/1910.12391)
- [What do Models Learn from Question Answering Datasets?](https://arxiv.org/abs/2004.03490)
- [Towards Interpreting BERT for Reading Comprehension Based QA](https://arxiv.org/abs/2010.08983) (EMNLP2020)
- [Compositional and Lexical Semantics in RoBERTa, BERT and DistilBERT: A Case Study on CoQA](https://arxiv.org/abs/2009.08257) (EMNLP2020)
- [How does BERT’s attention change when you fine-tune? An analysis methodology and a case study in negation scope](https://www.aclweb.org/anthology/2020.acl-main.429/) (ACL2020)
- [Calibration of Pre-trained Transformers](https://arxiv.org/abs/2003.07892)
- [When BERT Plays the Lottery, All Tickets Are Winning](https://arxiv.org/abs/2005.00561) (EMNLP2020)
- [The Lottery Ticket Hypothesis for Pre-trained BERT Networks](https://arxiv.org/abs/2007.12223)
- [What Context Features Can Transformer Language Models Use?](https://arxiv.org/abs/2106.08367) (ACL2021)
- [exBERT: A Visual Analysis Tool to Explore Learned Representations in Transformers Models](https://arxiv.org/abs/1910.05276) [[github](https://github.com/bhoov/exbert)]
- [The Language Interpretability Tool: Extensible, Interactive Visualizations and Analysis for NLP Models](https://arxiv.org/abs/2008.05122) [[github](https://github.com/pair-code/lit)]
- [What Does BERT with Vision Look At?](https://www.aclweb.org/anthology/2020.acl-main.469/) (ACL2020)
- [Behind the Scene: Revealing the Secrets of Pre-trained Vision-and-Language Models]() (ECCV2020)
- [Decoupling the Role of Data, Attention, and Losses in Multimodal Transformers](https://arxiv.org/pdf/2102.00529.pdf) (TACL2021)
- [What Vision-Language Models ‘See’ when they See Scenes](https://arxiv.org/pdf/2109.07301.pdf)
## Multi-lingual
- [A Primer on Pretrained Multilingual Language Models](https://arxiv.org/abs/2107.00676)
- [Multilingual Constituency Parsing with Self-Attention and Pre-Training](https://arxiv.org/abs/1812.11760) (ACL2019)
- [Cross-lingual Language Model Pretraining](https://arxiv.org/abs/1901.07291) (NeurIPS2019) [[github](https://github.com/facebookresearch/XLM)]
- [XLM-E: Cross-lingual Language Model Pre-training via ELECTRA](https://arxiv.org/abs/2106.16138)
- [XLM-K: Improving Cross-Lingual Language Model Pre-Training with Multilingual Knowledge](https://arxiv.org/abs/2109.12573)
- [75 Languages, 1 Model: Parsing Universal Dependencies Universally](https://arxiv.org/abs/1904.02099) (EMNLP2019) [[github](https://github.com/hyperparticle/udify)]
- [Zero-shot Dependency Parsing with Pre-trained Multilingual Sentence Representations](https://arxiv.org/abs/1910.05479) (EMNLP2019 WS)
- [Parsing with Multilingual BERT, a Small Corpus, and a Small Treebank](https://arxiv.org/abs/2009.14124) (EMNLP2020 Findings)
- [Beto, Bentz, Becas: The Surprising Cross-Lingual Effectiveness of BERT](https://arxiv.org/abs/1904.09077) (EMNLP2019)
- [How multilingual is Multilingual BERT?](https://arxiv.org/abs/1906.01502) (ACL2019)
- [How Language-Neutral is Multilingual BERT?](https://arxiv.org/abs/1911.03310)
- [How to Adapt Your Pretrained Multilingual Model to 1600 Languages](https://arxiv.org/abs/2106.02124) (ACL2021)
- [Load What You Need: Smaller Versions of Multilingual BERT](https://arxiv.org/abs/2010.05609) (EMNLP2020) [[github](https://github.com/Geotrend-research/smaller-transformers)]
- [Is Multilingual BERT Fluent in Language Generation?](https://arxiv.org/abs/1910.03806)
- [ZmBART: An Unsupervised Cross-lingual Transfer Framework for Language Generation](https://arxiv.org/abs/2106.01597) (ACL2021 Findings)
- [Unicoder: A Universal Language Encoder by Pre-training with Multiple Cross-lingual Tasks](https://www.aclweb.org/anthology/D19-1252/) (EMNLP2019)
- [BERT is Not an Interlingua and the Bias of Tokenization](https://www.aclweb.org/anthology/D19-6106/) (EMNLP2019 WS)
- [Cross-Lingual Ability of Multilingual BERT: An Empirical Study](https://openreview.net/forum?id=HJeT3yrtDr) (ICLR2020)
- [Multilingual Alignment of Contextual Word Representations](https://arxiv.org/abs/2002.03518) (ICLR2020)
- [Emerging Cross-lingual Structure in Pretrained Language Models](https://arxiv.org/abs/1911.01464) (ACL2020)
- [On the Cross-lingual Transferability of Monolingual Representations](https://arxiv.org/abs/1910.11856)
- [Unsupervised Cross-lingual Representation Learning at Scale](https://arxiv.org/abs/1911.02116) (ACL2020)
- [FILTER: An Enhanced Fusion Method for Cross-lingual Language Understanding](https://arxiv.org/abs/2009.05166)
- [Cross-lingual Alignment Methods for Multilingual BERT: A Comparative Study](https://arxiv.org/abs/2009.14304) (EMNLP2020 Findings)
- [Emerging Cross-lingual Structure in Pretrained Language Models](https://arxiv.org/abs/1911.01464)
- [Can Monolingual Pretrained Models Help Cross-Lingual Classification?](https://arxiv.org/abs/1911.03913)
- [A Study of Cross-Lingual Ability and Language-specific Information in Multilingual BERT](https://arxiv.org/abs/2004.09205)
- [Fully Unsupervised Crosslingual Semantic Textual Similarity Metric Based on BERT for Identifying Parallel Data](https://www.aclweb.org/anthology/K19-1020/) (CoNLL2019)
- [What the \[MASK\]? Making Sense of Language-Specific BERT Models](https://arxiv.org/abs/2003.02912)
- [XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization](https://arxiv.org/abs/2003.11080) (ICML2020)
- [XTREME-R: Towards More Challenging and Nuanced Multilingual Evaluation](https://arxiv.org/abs/2104.07412) (EMNLP2021)
- [XGLUE: A New Benchmark Dataset for Cross-lingual Pre-training, Understanding and Generation](https://arxiv.org/abs/2004.01401)
- [A Systematic Analysis of Morphological Content in BERT Models for Multiple Languages](https://arxiv.org/abs/2004.03032)
- [Extending Multilingual BERT to Low-Resource Languages](https://arxiv.org/abs/2004.13640)
- [Learning Better Universal Representations from Pre-trained Contextualized Language Models](https://arxiv.org/abs/2004.13947)
- [Universal Dependencies according to BERT: both more specific and more general](https://arxiv.org/abs/2004.14620)
- [A Call for More Rigor in Unsupervised Cross-lingual Learning](https://arxiv.org/abs/2004.14958) (ACL2020)
- [Identifying Necessary Elements for BERT's Multilinguality](https://arxiv.org/abs/2005.00396) (EMNLP2020)
- [MAD-X: An Adapter-based Framework for Multi-task Cross-lingual Transfer](https://arxiv.org/abs/2005.00052)
- [From Zero to Hero: On the Limitations of Zero-Shot Cross-Lingual Transfer with Multilingual Transformers](https://arxiv.org/abs/2005.00633)
- [Language Models are Few-shot Multilingual Learners](https://arxiv.org/abs/2109.07684)
- [First Align, then Predict: Understanding the Cross-Lingual Ability of Multilingual BERT](https://arxiv.org/abs/2101.11109) (EACL2021)
- [Multilingual BERT Post-Pretraining Alignment](https://arxiv.org/abs/2010.12547) (NAACL2021)
- [XeroAlign: Zero-Shot Cross-lingual Transformer Alignment](https://arxiv.org/abs/2105.02472) (ACL2021 Findings)
- [Syntax-augmented Multilingual BERT for Cross-lingual Transfer](https://arxiv.org/abs/2106.02134) (ACL2021)
- [Language Representation in Multilingual BERT and its applications to improve Cross-lingual Generalization](https://arxiv.org/abs/2010.10041)
- [VECO: Variable Encoder-decoder Pre-training for Cross-lingual Understanding and Generation](https://openreview.net/forum?id=YjNv-hzM8BE)
- [On the Language Neutrality of Pre-trained Multilingual Representations](https://arxiv.org/abs/2004.05160)
- [Are All Languages Created Equal in Multilingual BERT?](https://arxiv.org/abs/2005.09093) (ACL2020 WS)
- [When Being Unseen from mBERT is just the Beginning: Handling New Languages With Multilingual Language Models](https://arxiv.org/abs/2010.12858)
- [Adapting Monolingual Models: Data can be Scarce when Language Similarity is High](https://arxiv.org/abs/2105.02855) (ACL2021 Findings)
- [Language-agnostic BERT Sentence Embedding](https://arxiv.org/abs/2007.01852)
- [Universal Sentence Representation Learning with Conditional Masked Language Model](https://arxiv.org/abs/2012.14388)
- [WikiBERT models: deep transfer learning for many languages](https://arxiv.org/abs/2006.01538)
- [Inducing Language-Agnostic Multilingual Representations](https://arxiv.org/abs/2008.09112)
- [To What Degree Can Language Borders Be Blurred In BERT-based Multilingual Spoken Language Understanding?](https://arxiv.org/abs/2011.05007) (COLING2020)
- [It's not Greek to mBERT: Inducing Word-Level Translations from Multilingual BERT](https://arxiv.org/abs/2010.08275) (EMNLP2020 WS)
- [XLM-T: A Multilingual Language Model Toolkit for Twitter](https://arxiv.org/abs/2104.12250)
- [A Survey on Recent Approaches for Natural Language Processing in Low-Resource Scenarios](https://arxiv.org/abs/2010.12309)
- [Translation Artifacts in Cross-lingual Transfer Learning](https://arxiv.org/abs/2004.04721) (EMNLP2020)
- [Identifying Cultural Differences through Multi-Lingual Wikipedia](https://arxiv.org/abs/2004.04938)
- [A Supervised Word Alignment Method based on Cross-Language Span Prediction using Multilingual BERT](https://arxiv.org/abs/2004.14516) (EMNLP2020)
- [BERT for Monolingual and Cross-Lingual Reverse Dictionary](https://arxiv.org/abs/2009.14790) (EMNLP2020 Findings)
- [Bilingual Text Extraction as Reading Comprehension](https://arxiv.org/abs/2004.14517)
- [Evaluating Multilingual BERT for Estonian](https://arxiv.org/abs/2010.00454)
- [How Good is Your Tokenizer? On the Monolingual Performance of Multilingual Language Models](https://arxiv.org/abs/2012.15613) (ACL2021) [[github](https://github.com/Adapter-Hub/hgiyt)]
- [Allocating Large Vocabulary Capacity for Cross-lingual Language Model Pre-training](https://arxiv.org/abs/2109.07306) (EMNLP2021)
- [BERTologiCoMix: How does Code-Mixing interact with Multilingual BERT?](https://www.aclweb.org/anthology/2021.adaptnlp-1.12/) (EACL2021 WS)
## Other than English models
- [CamemBERT: a Tasty French Language Model](https://arxiv.org/abs/1911.03894) (ACL2020)
- [On the importance of pre-training data volume for compact language models](https://arxiv.org/abs/2010.03813) (EMNLP2020)
- [FlauBERT: Unsupervised Language Model Pre-training for French](https://arxiv.org/abs/1912.05372) (LREC2020)
- [Multilingual is not enough: BERT for Finnish](https://arxiv.org/abs/1912.07076)
- [BERTje: A Dutch BERT Model](https://arxiv.org/abs/1912.09582)
- [RobBERT: a Dutch RoBERTa-based Language Model](https://arxiv.org/abs/2001.06286)
- [Adaptation of Deep Bidirectional Multilingual Transformers for Russian Language](https://arxiv.org/abs/1905.07213)
- [RussianSuperGLUE: A Russian Language Understanding Evaluation Benchmark](https://arxiv.org/abs/2010.15925) (EMNLP2020)
- [AraBERT: Transformer-based Model for Arabic Language Understanding](https://arxiv.org/abs/2003.00104)
- [ALUE: Arabic Language Understanding Evaluation](https://aclanthology.org/2021.wanlp-1.18/) (EACL2021 WS) [[website](https://www.alue.org/home)]
- [ARBERT & MARBERT: Deep Bidirectional Transformers for Arabic](https://aclanthology.org/2021.acl-long.551/) (ACL2021) [[github](https://github.com/UBC-NLP/marbert)]
- [Pre-Training BERT on Arabic Tweets: Practical Considerations](https://arxiv.org/abs/2102.10684)
- [PhoBERT: Pre-trained language models for Vietnamese](https://arxiv.org/abs/2003.00744)
- [Give your Text Representation Models some Love: the Case for Basque](https://arxiv.org/abs/2004.00033) (LREC2020)
- [ParsBERT: Transformer-based Model for Persian Language Understanding](https://arxiv.org/abs/2005.12515)
- [Leveraging ParsBERT and Pretrained mT5 for Persian Abstractive Text Summarization](https://arxiv.org/abs/2012.11204) (CSICC2021)
- [Pre-training Polish Transformer-based Language Models at Scale](https://arxiv.org/abs/2006.04229)
- [Playing with Words at the National Library of Sweden -- Making a Swedish BERT](https://arxiv.org/abs/2007.01658)
- [KR-BERT: A Small-Scale Korean-Specific Language Model](https://arxiv.org/abs/2008.03979)
- [KoreALBERT: Pretraining a Lite BERT Model for Korean Language Understanding](https://arxiv.org/abs/2101.11363) (ICPR2020)
- [What Changes Can Large-scale Language Models Bring? Intensive Study on HyperCLOVA: Billions-scale Korean Generative Pretrained Transformers](https://arxiv.org/abs/2109.04650) (EMNLP2021)
- [KLUE: Korean Language Understanding Evaluation](https://arxiv.org/abs/2105.09680)
- [WangchanBERTa: Pretraining transformer-based Thai Language Models](https://arxiv.org/abs/2101.09635)
- [FinEst BERT and CroSloEngual BERT: less is more in multilingual models](https://arxiv.org/abs/2006.07890) (TSD2020)
- [GREEK-BERT: The Greeks visiting Sesame Street](https://arxiv.org/abs/2008.12014) (SETN2020)
- [The birth of Romanian BERT](https://arxiv.org/abs/2009.08712) (EMNLP2020 Findings)
- [German's Next Language Model](https://arxiv.org/abs/2010.10906) (COLING2020 Industry Truck)
- [GottBERT: a pure German Language Model](https://arxiv.org/abs/2012.02110)
- [EstBERT: A Pretrained Language-Specific BERT for Estonian](https://arxiv.org/abs/2011.04784)
- [Czert -- Czech BERT-like Model for Language Representation](https://arxiv.org/abs/2103.13031)
- [RobeCzech: Czech RoBERTa, a monolingual contextualized language representation model](https://arxiv.org/abs/2105.11314) (TSD2021)
- [Bertinho: Galician BERT Representations](https://arxiv.org/abs/2103.13799)
- [Pretraining and Fine-Tuning Strategies for Sentiment Analysis of Latvian Tweets](https://arxiv.org/abs/2010.12401)
- [PTT5: Pretraining and validating the T5 model on Brazilian Portuguese data](https://arxiv.org/abs/2008.09144)
- [IndicNLPSuite: Monolingual Corpora, Evaluation Benchmarks and Pre-trained Multilingual Language Models for Indian Languages](https://aclanthology.org/2020.findings-emnlp.445/) (EMNLP2020 Findings)
- [Indic-Transformers: An Analysis of Transformer Language Models for Indian Languages](https://arxiv.org/abs/2011.02323) (NeurIPS2020 WS)
- [IndoLEM and IndoBERT: A Benchmark Dataset and Pre-trained Language Model for Indonesian NLP](https://www.aclweb.org/anthology/2020.coling-main.66/) (COLING2020)
- [IndoBERTweet: A Pretrained Language Model for Indonesian Twitter with Effective Domain-Specific Vocabulary Initialization](https://arxiv.org/abs/2109.04607) (EMNLP2021)
- [IndoNLG: Benchmark and Resources for Evaluating Indonesian Natural Language Generation](https://arxiv.org/abs/2104.08200) (EMNLP2021)
- [AfroMT: Pretraining Strategies and Reproducible Benchmarks for Translation of 8 African Languages](https://arxiv.org/abs/2109.04715) (EMNLP2021)
- [KinyaBERT: a Morphology-aware Kinyarwanda Language Model](https://arxiv.org/abs/2203.08459) (ACL2022)
- [BARThez: a Skilled Pretrained French Sequence-to-Sequence Model](https://arxiv.org/abs/2010.12321)
- [NEZHA: Neural Contextualized Representation for Chinese Language Understanding](https://arxiv.org/abs/1909.00204)
- [Revisiting Pre-Trained Models for Chinese Natural Language Processing](https://arxiv.org/abs/2004.13922) (EMNLP2020 Findings)
- [ChineseBERT: Chinese Pretraining Enhanced by Glyph and Pinyin Information](https://arxiv.org/abs/2106.16038) (ACL2021) [[github](https://github.com/ShannonAI/ChineseBert)]
- [Intrinsic Knowledge Evaluation on Chinese Language Models](https://arxiv.org/abs/2011.14277)
- [CPM: A Large-scale Generative Chinese Pre-trained Language Model](https://arxiv.org/abs/2012.00413) [[github](https://github.com/TsinghuaAI/CPM-Generate)]
- [PanGu-α: Large-scale Autoregressive Pretrained Chinese Language Models with Auto-parallel Computation](https://arxiv.org/abs/2104.12369)
- [CLUECorpus2020: A Large-scale Chinese Corpus for Pre-training Language Model](https://arxiv.org/abs/2003.01355)
- [CLUE: A Chinese Language Understanding Evaluation Benchmark](https://arxiv.org/abs/2004.05986)
- [CUGE: A Chinese Language Understanding and Generation Evaluation Benchmark](https://arxiv.org/abs/2112.13610)
- [FewCLUE: A Chinese Few-shot Learning Evaluation Benchmark](https://arxiv.org/abs/2107.07498)
- [AnchiBERT: A Pre-Trained Model for Ancient ChineseLanguage Understanding and Generation](https://arxiv.org/abs/2009.11473)
- [UER: An Open-Source Toolkit for Pre-training Models](https://arxiv.org/abs/1909.05658) (EMNLP2019 Demo) [[github](https://github.com/dbiir/UER-py)]
## Domain specific
- [AMMU -- A Survey of Transformer-based Biomedical Pretrained Language Models](https://arxiv.org/abs/2105.00827)
- [BioBERT: a pre-trained biomedical language representation model for biomedical text mining](https://arxiv.org/abs/1901.08746)
- [Self-Alignment Pretraining for Biomedical Entity Representations](https://aclanthology.org/2021.naacl-main.334/) (NAACL2021) [[github](https://github.com/cambridgeltl/sapbert)]
- [Learning Domain-Specialised Representations for Cross-Lingual Biomedical Entity Linking](https://aclanthology.org/2021.acl-short.72/) (ACL2021) [[github](https://github.com/cambridgeltl/sapbert)]
- [Transfer Learning in Biomedical Natural Language Processing: An Evaluation of BERT and ELMo on Ten Benchmarking Datasets](https://arxiv.org/abs/1906.05474) (ACL2019 WS)
- [BERT-based Ranking for Biomedical Entity Normalization](https://arxiv.org/abs/1908.03548)
- [PubMedQA: A Dataset for Biomedical Research Question Answering](https://arxiv.org/abs/1909.06146) (EMNLP2019)
- [Pre-trained Language Model for Biomedical Question Answering](https://arxiv.org/abs/1909.08229)
- [How to Pre-Train Your Model? Comparison of Different Pre-Training Models for Biomedical Question Answering](https://arxiv.org/abs/1911.00712)
- [On Adversarial Examples for Biomedical NLP Tasks](https://arxiv.org/abs/2004.11157)
- [An Empirical Study of Multi-Task Learning on BERT for Biomedical Text Mining](https://arxiv.org/abs/2005.02799) (ACL2020 WS)
- [Domain-Specific Language Model Pretraining for Biomedical Natural Language Processing](https://arxiv.org/abs/2007.15779) [[github](https://microsoft.github.io/BLURB/)]
- [Improving Biomedical Pretrained Language Models with Knowledge](https://arxiv.org/abs/2104.10344) (BioNLP2021)
- [BioMegatron: Larger Biomedical Domain Language Model](https://arxiv.org/abs/2010.06060) (EMNLP2020) [[website](https://developer.nvidia.com/blog/building-state-of-the-art-biomedical-and-clinical-nlp-models-with-bio-megatron/)]
- [Pretrained Language Models for Biomedical and Clinical Tasks: Understanding and Extending the State-of-the-Art](https://www.aclweb.org/anthology/2020.clinicalnlp-1.17/) (EMNLP2020 WS)
- [A pre-training technique to localize medical BERT and enhance BioBERT](https://arxiv.org/abs/2005.07202) [[github](https://github.com/sy-wada/blue_benchmark_with_transformers)]
- [exBERT: Extending Pre-trained Models with Domain-specific Vocabulary Under Constrained Training Resources](https://aclanthology.org/2020.findings-emnlp.129/) [[github](https://github.com/cgmhaicenter/exBERT)] (EMNLP2020 Findings)
- [BERTology Meets Biology: Interpreting Attention in Protein Language Models](https://arxiv.org/abs/2006.15222) (ICLR2021)
- [ClinicalBERT: Modeling Clinical Notes and Predicting Hospital Readmission](https://arxiv.org/abs/1904.05342)
- [Predicting Clinical Diagnosis from Patients Electronic Health Records Using BERT-based Neural Networks](https://arxiv.org/abs/2007.07562) (AIME2020)
- [Publicly Available Clinical BERT Embeddings](https://arxiv.org/abs/1904.03323) (NAACL2019 WS)
- [UmlsBERT: Clinical Domain Knowledge Augmentation of Contextual Embeddings Using the Unified Medical Language System Metathesaurus](https://arxiv.org/abs/2010.10391) (NAACL2021)
- [MT-Clinical BERT: Scaling Clinical Information Extraction with Multitask Learning](https://arxiv.org/abs/2004.10220)
- [A clinical specific BERT developed with huge size of Japanese clinical narrative](https://www.medrxiv.org/content/10.1101/2020.07.07.20148585v1)
- [Clinical Reading Comprehension: A Thorough Analysis of the emrQA Dataset](https://arxiv.org/abs/2005.00574) (ACL2020) [[github](https://github.com/xiangyue9607/CliniRC)]
- [Knowledge-Empowered Representation Learning for Chinese Medical Reading Comprehension: Task, Model and Resources](https://arxiv.org/abs/2008.10327)
- [Classifying Long Clinical Documents with Pre-trained Transformers](https://arxiv.org/abs/2105.06752)
- [Detecting Adverse Drug Reactions from Twitter through Domain-Specific Preprocessing and BERT Ensembling](https://arxiv.org/abs/2005.06634)
- [Progress Notes Classification and Keyword Extraction using Attention-based Deep Learning Models with BERT](https://arxiv.org/abs/1910.05786)
- [BERT-XML: Large Scale Automated ICD Coding Using BERT Pretraining](https://arxiv.org/abs/2006.03685)
- [Prediction of ICD Codes with Clinical BERT Embeddings and Text Augmentation with Label Balancing using MIMIC-III](https://arxiv.org/abs/2008.10492)
- [Infusing Disease Knowledge into BERT for Health Question Answering, Medical Inference and Disease Name Recognition](https://arxiv.org/abs/2010.03746) (EMNLP2020)
- [CheXbert: Combining Automatic Labelers and Expert Annotations for Accurate Radiology Report Labeling Using BERT](https://arxiv.org/abs/2004.09167) (EMNLP2020)
- [Students Need More Attention: BERT-based Attention Model for Small Data with Application to Automatic Patient Message Triage](https://arxiv.org/abs/2006.11991) (MLHC2020)
- [Med-BERT: pre-trained contextualized embeddings on large-scale structured electronic health records for disease prediction](https://arxiv.org/abs/2005.12833) [[github](https://github.com/ZhiGroup/Med-BERT)]
- [SciBERT: Pretrained Contextualized Embeddings for Scientific Text](https://aclanthology.org/D19-1371/) (EMNLP2019) [[github](https://github.com/allenai/scibert)]
- [SPECTER: Document-level Representation Learning using Citation-informed Transformers](https://aclanthology.org/2020.acl-main.207/) (ACL2020) [[github](https://github.com/allenai/specter)]
- [OAG-BERT: Pre-train Heterogeneous Entity-augmented Academic Language Models](https://arxiv.org/abs/2103.02410) [[github](https://github.com/thudm/oag-bert)]
- [PatentBERT: Patent Classification with Fine-Tuning a pre-trained BERT Model](https://arxiv.org/abs/1906.02124)
- [FinBERT: A Pretrained Language Model for Financial Communications](https://arxiv.org/abs/2006.08097)
- [LEGAL-BERT: The Muppets straight out of Law School](https://arxiv.org/abs/2010.02559) (EMNLP2020 Findings)
- [Lawformer: A Pre-trained Language Model for Chinese Legal Long Documents](https://arxiv.org/abs/2105.03887)
- [E-BERT: A Phrase and Product Knowledge Enhanced Language Model for E-commerce](https://arxiv.org/abs/2009.02835)
- [BERT Goes Shopping: Comparing Distributional Models for Product Representations](https://arxiv.org/abs/2012.09807)
- [NewsBERT: Distilling Pre-trained Language Model for Intelligent News Application](https://arxiv.org/abs/2102.04887)
- [Code and Named Entity Recognition in StackOverflow](https://arxiv.org/abs/2005.01634) (ACL2020) [[github](https://github.com/lanwuwei/BERTOverflow)]
- [BERTweet: A pre-trained language model for English Tweets](https://arxiv.org/abs/2005.10200) (EMNLP2020 Demo)
- [TweetBERT: A Pretrained Language Representation Model for Twitter Text Analysis](https://arxiv.org/abs/2010.11091)
- [A Million Tweets Are Worth a Few Points: Tuning Transformers for Customer Service Tasks](https://arxiv.org/abs/2104.07944)
- [Analyzing COVID-19 Tweets with Transformer-based Language Models](https://arxiv.org/abs/2104.10259)
- [Cost-effective Selection of Pretraining Data: A Case Study of Pretraining BERT on Social Media](https://arxiv.org/abs/2010.01150) (EMNLP2020 Findings)
## Multi-modal
- [A Survey on Visual Transformer](https://arxiv.org/abs/2012.12556)
- [Transformers in Vision: A Survey](https://arxiv.org/abs/2101.01169)
- [Vision-Language Pre-training: Basics, Recent Advances, and Future Trends](https://arxiv.org/abs/2210.09263)
- [VideoBERT: A Joint Model for Video and Language Representation Learning](https://arxiv.org/abs/1904.01766) (ICCV2019)
- [ViLBERT: Pretraining Task-Agnostic Visiolinguistic Representations for Vision-and-Language Tasks](https://arxiv.org/abs/1908.02265) (NeurIPS2019)
- [VisualBERT: A Simple and Performant Baseline for Vision and Language](https://arxiv.org/abs/1908.03557)
- [Selfie: Self-supervised Pretraining for Image Embedding](https://arxiv.org/abs/1906.02940)
- [ImageBERT: Cross-modal Pre-training with Large-scale Weak-supervised Image-Text Data](https://arxiv.org/abs/2001.07966)
- [SimVLM: Simple Visual Language Model Pretraining with Weak Supervision](https://arxiv.org/abs/2108.10904) (ICLR2022)
- [Align before Fuse: Vision and Language Representation Learning with Momentum Distillation](https://arxiv.org/abs/2107.07651) (NeurIPS2021) [[github](https://github.com/salesforce/ALBEF)]
- [Contrastive Bidirectional Transformer for Temporal Representation Learning](https://arxiv.org/abs/1906.05743)
- [M-BERT: Injecting Multimodal Information in the BERT Structure](https://arxiv.org/abs/1908.05787)
- [Integrating Multimodal Information in Large Pretrained Transformers](https://arxiv.org/abs/1908.05787)
- [LXMERT: Learning Cross-Modality Encoder Representations from Transformers](https://arxiv.org/abs/1908.07490) (EMNLP2019)
- [Unsupervised Vision-and-Language Pre-training Without Parallel Images and Captions](https://arxiv.org/abs/2010.12831) (NAACL2021)
- [X-LXMERT: Paint, Caption and Answer Questions with Multi-Modal Transformers](https://arxiv.org/abs/2009.11278) (EMNLP2020)
- [Adaptive Transformers for Learning Multimodal Representations](https://arxiv.org/abs/2005.07486) (ACL2020SRW) [[github](https://github.com/prajjwal1/adaptive_transformer)]
- [GEM: A General Evaluation Benchmark for Multimodal Tasks](https://arxiv.org/abs/2106.09889) (ACL2021 Findings) [[github](https://github.com/microsoft/GEM)]
- [Fusion of Detected Objects in Text for Visual Question Answering](https://arxiv.org/abs/1908.05054) (EMNLP2019)
- [VisualMRC: Machine Reading Comprehension on Document Images](https://arxiv.org/abs/2101.11272) (AAAI2021)
- [LambdaNetworks: Modeling long-range Interactions without Attention](https://openreview.net/forum?id=xTJEN-ggl1b) [[github](https://github.com/gsarti/lambda-bert)]
- [BERT representations for Video Question Answering](http://openaccess.thecvf.com/content_WACV_2020/html/Yang_BERT_representations_for_Video_Question_Answering_WACV_2020_paper.html) (WACV2020)
- [Self-supervised pre-training and contrastive representation learning for multiple-choice video QA](https://arxiv.org/abs/2009.08043) (AAAI2021)
- [UNIMO: Towards Unified-Modal Understanding and Generation via Cross-Modal Contrastive Learning](https://arxiv.org/abs/2012.15409) (ACL2021)
- [BLIP: Bootstrapping Language-Image Pre-training for Unified Vision-Language Understanding and Generation](https://arxiv.org/abs/2201.12086) [[github](https://github.com/salesforce/BLIP)]
- [Uni-EDEN: Universal Encoder-Decoder Network by Multi-Granular Vision-Language Pre-training](https://arxiv.org/abs/2201.04026)
- [Contrastive Visual-Linguistic Pretraining](https://arxiv.org/abs/2007.13135)
- [What is More Likely to Happen Next? Video-and-Language Future Event Prediction](https://arxiv.org/abs/2010.07999) (EMNLP2020)
- [VisualGPT: Data-efficient Image Captioning by Balancing Visual Input and Linguistic Knowledge from Pretraining](https://arxiv.org/abs/2102.10407)
- [XGPT: Cross-modal Generative Pre-Training for Image Captioning](https://arxiv.org/abs/2003.01473)
- [Scaling Up Vision-Language Pre-training for Image Captioning](https://arxiv.org/abs/2111.12233)
- [Injecting Semantic Concepts into End-to-End Image Captioning](https://arxiv.org/abs/2112.05230) (CVPR2022)
- [Unified Vision-Language Pre-Training for Image Captioning and VQA](https://arxiv.org/abs/1909.11059) (AAAI2020) [[github](https://github.com/LuoweiZhou/VLP)]
- [TAP: Text-Aware Pre-training for Text-VQA and Text-Caption](https://arxiv.org/abs/2012.04638)
- [An Empirical Study of GPT-3 for Few-Shot Knowledge-Based VQA](https://arxiv.org/abs/2109.05014) (AAAI2022)
- [Transformer is All You Need: Multimodal Multitask Learning with a Unified Transformer](https://arxiv.org/abs/2102.10772)
- [VisualCOMET: Reasoning about the Dynamic Context of a Still Image](https://arxiv.org/abs/2004.10796) (ECCV2020) [[website](http://visualcomet.xyz)]
- [Large-scale Pretraining for Visual Dialog: A Simple State-of-the-Art Baseline](https://arxiv.org/abs/1912.02379)
- [VD-BERT: A Unified Vision and Dialog Transformer with BERT](https://arxiv.org/abs/2004.13278) (EMNLP2020)
- [VL-BERT: Pre-training of Generic Visual-Linguistic Representations](https://arxiv.org/abs/1908.08530) (ICLR2020)
- [Unicoder-VL: A Universal Encoder for Vision and Language by Cross-modal Pre-training](https://arxiv.org/abs/1908.06066)
- [UNITER: Learning UNiversal Image-TExt Representations](https://arxiv.org/abs/1909.11740)
- [ViLT: Vision-and-Language Transformer Without Convolution or Region Supervision](https://arxiv.org/abs/2102.03334)
- [Supervised Multimodal Bitransformers for Classifying Images and Text](https://arxiv.org/abs/1909.02950)
- [InterBERT: Vision-and-Language Interaction for Multi-modal Pretraining](https://arxiv.org/abs/2003.13198)
- [Multimodal Pretraining Unmasked: A Meta-Analysis and a Unified Framework of Vision-and-Language BERTs](https://arxiv.org/abs/2011.15124) (TACL2021)
- [SemVLP: Vision-Language Pre-training by Aligning Semantics at Multiple Levels](https://arxiv.org/abs/2103.07829)
- [LiT : Zero-Shot Transfer with Locked-image Text Tuning](https://arxiv.org/abs/2111.07991) (CVPR2022)
- [WenLan: Bridging Vision and Language by Large-Scale Multi-Modal Pre-Training](https://arxiv.org/abs/2103.06561)
- [Probing Inter-modality: Visual Parsing with Self-Attention for Vision-Language Pre-training](https://arxiv.org/abs/2106.13488) (NeurIPS2021)
- [E2E-VLP: End-to-End Vision-Language Pre-training Enhanced by Visual Learning](https://arxiv.org/abs/2106.01804) (ACL2021)
- [UNIMO-2: End-to-End Unified Vision-Language Grounded Learning](https://arxiv.org/abs/2203.09067) (ACL2022)
- [Grounded Language-Image Pre-training](https://arxiv.org/abs/2112.03857) [[github](https://github.com/microsoft/GLIP)]
- [VLMO: Unified Vision-Language Pre-Training with Mixture-of-Modality-Experts](https://arxiv.org/abs/2111.02358) [[github](https://github.com/microsoft/unilm/tree/master/vlmo)]
- [VinVL: Revisiting Visual Representations in Vision-Language Models](https://arxiv.org/abs/2101.00529)
- [An Empirical Study of Training End-to-End Vision-and-Language Transformers](https://arxiv.org/abs/2111.02387) (CVPR2022) [[github](https://github.com/zdou0830/METER)]
- [Crossing the Format Boundary of Text and Boxes: Towards Unified Vision-Language Modeling](https://arxiv.org/abs/2111.12085)
- [UFO: A UniFied TransfOrmer for Vision-Language Representation Learning](https://arxiv.org/abs/2111.10023)
- [Florence: A New Foundation Model for Computer Vision](https://arxiv.org/abs/2111.11432)
- [Large-Scale Adversarial Training for Vision-and-Language Representation Learning](https://arxiv.org/abs/2006.06195) (NeurIPS2020)
- [Flamingo: a Visual Language Model for Few-Shot Learning](https://arxiv.org/abs/2204.14198)
- [OpenFlamingo: An Open-Source Framework for Training Large Autoregressive Vision-Language Models](https://arxiv.org/abs/2308.01390) [[github](https://github.com/mlfoundations/open_flamingo)]
- [Do DALL-E and Flamingo Understand Each Other?](https://arxiv.org/abs/2212.12249)
- [Conceptual 12M: Pushing Web-Scale Image-Text Pre-Training To Recognize Long-Tail Visual Concepts](https://arxiv.org/abs/2102.08981)
- [Unifying Vision-and-Language Tasks via Text Generation](https://arxiv.org/abs/2102.02779)
- [Scheduled Sampling in Vision-Language Pretraining with Decoupled Encoder-Decoder Network](https://arxiv.org/abs/2101.11562) (AAAI2021)
- [ERNIE-ViL: Knowledge Enhanced Vision-Language Representations Through Scene Graph](https://arxiv.org/abs/2006.16934)
- [KVL-BERT: Knowledge Enhanced Visual-and-Linguistic BERT for Visual Commonsense Reasoning](https://arxiv.org/abs/2012.07000)
- [A Closer Look at the Robustness of Vision-and-Language Pre-trained Models](https://arxiv.org/abs/2012.08673)
- [Self-Supervised learning with cross-modal transformers for emotion recognition](https://arxiv.org/abs/2011.10652) (SLT2020)
- [Vokenization: Improving Language Understanding with Contextualized, Visual-Grounded Supervision](https://arxiv.org/abs/2010.06775) (EMNLP2020)
- [12-in-1: Multi-Task Vision and Language Representation Learning](https://arxiv.org/abs/1912.02315)
- [Multilingual Multimodal Pre-training for Zero-Shot Cross-Lingual Transfer of Vision-Language Models](https://arxiv.org/abs/2103.08849) (NAACL2021)
- [M3P: Learning Universal Representations via Multitask Multilingual Multimodal Pre-training](https://arxiv.org/abs/2006.02635) (CVPR2021)
- [UC2: Universal Cross-lingual Cross-modal Vision-and-Language Pre-training](https://arxiv.org/abs/2104.00332)
- [CM3: A Causal Masked Multimodal Model of the Internet](https://arxiv.org/abs/2201.07520)
- [Retrieval-Augmented Multimodal Language Modeling](https://arxiv.org/abs/2211.12561)
- [Cycle Text-To-Image GAN with BERT](https://arxiv.org/abs/2003.12137)
- [Weak Supervision helps Emergence of Word-Object Alignment and improves Vision-Language Tasks](https://arxiv.org/abs/1912.03063)
- [Oscar: Object-Semantics Aligned Pre-training for Vision-Language Tasks](https://arxiv.org/abs/2004.06165)
- [VIVO: Visual Vocabulary Pre-Training for Novel Object Captioning](https://arxiv.org/abs/2009.13682)
- [DeVLBert: Learning Deconfounded Visio-Linguistic Representations](https://arxiv.org/abs/2008.06884) (ACMMM2020)
- [A Recurrent Vision-and-Language BERT for Navigation](https://arxiv.org/abs/2011.13922)
- [BERT Can See Out of the Box: On the Cross-modal Transferability of Text Representations](https://arxiv.org/abs/2002.10832)
- [Seeing Out of tHe bOx: End-to-End Pre-training for Vision-Language Representation Learning](https://arxiv.org/abs/2104.03135) (CVPR2021)
- [Vision-and-Language or Vision-for-Language? On Cross-Modal Influence in Multimodal Transformers](https://arxiv.org/abs/2109.04448) (EMNLP2021)
- [Pixel-BERT: Aligning Image Pixels with Text by Deep Multi-Modal Transformers](https://arxiv.org/abs/2004.00849)
- [IGLUE: A Benchmark for Transfer Learning across Modalities, Tasks, and Languages](https://arxiv.org/abs/2201.11732)
- [Understanding Advertisements with BERT](https://www.aclweb.org/anthology/2020.acl-main.674/) (ACL2020)
- [BERTERS: Multimodal Representation Learning for Expert Recommendation System with Transformer](https://arxiv.org/abs/2007.07229)
- [FashionBERT: Text and Image Matching with Adaptive Loss for Cross-modal Retrieval](https://arxiv.org/abs/2005.09801) (SIGIR2020)
- [Kaleido-BERT: Vision-Language Pre-training on Fashion Domain](https://arxiv.org/abs/2103.16110) (CVPR2021)
- [LayoutLM: Pre-training of Text and Layout for Document Image Understanding](https://arxiv.org/abs/1912.13318) (KDD2020) [[github](https://github.com/microsoft/unilm/tree/master/layoutlm)]
- [LayoutLMv2: Multi-modal Pre-training for Visually-Rich Document Understanding](https://arxiv.org/abs/2012.14740) (ACL2021)
- [LayoutXLM: Multimodal Pre-training for Multilingual Visually-rich Document Understanding](https://arxiv.org/abs/2104.08836)
- [Unifying Vision, Text, and Layout for Universal Document Processing](https://arxiv.org/abs/2212.02623)
- [LAMPRET: Layout-Aware Multimodal PreTraining for Document Understanding](https://arxiv.org/abs/2104.08405)
- [BROS: A Pre-trained Language Model for Understanding Texts in Document](https://openreview.net/forum?id=punMXQEsPr0)
- [TrOCR: Transformer-based Optical Character Recognition with Pre-trained Models](https://arxiv.org/abs/2109.10282)
- [LayoutReader: Pre-training of Text and Layout for Reading Order Detection](https://arxiv.org/abs/2108.11591) (EMNLP2021)
- [BERT for Large-scale Video Segment Classification with Test-time Augmentation](https://arxiv.org/abs/1912.01127) (ICCV2019WS)
- [Is Space-Time Attention All You Need for Video Understanding?](https://arxiv.org/abs/2102.05095)
- [lamBERT: Language and Action Learning Using Multimodal BERT](https://arxiv.org/abs/2004.07093)
- [Generative Pretraining from Pixels](https://cdn.openai.com/papers/Generative_Pretraining_from_Pixels_V2.pdf) [[github](https://github.com/openai/image-gpt)] [[website](https://openai.com/blog/image-gpt/)]
- [Visual Transformers: Token-based Image Representation and Processing for Computer Vision](https://arxiv.org/abs/2006.03677)
- [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://openreview.net/forum?id=YicbFdNTTy) (ICLR2021)
- [BEiT: BERT Pre-Training of Image Transformers](https://arxiv.org/abs/2106.08254)
- [Zero-Shot Text-to-Image Generation](https://arxiv.org/abs/2102.12092) [[github](https://github.com/openai/DALL-E)] [[website](https://openai.com/blog/dall-e/)]
- [Hierarchical Text-Conditional Image Generation with CLIP Latents](https://arxiv.org/abs/2204.06125) [[website](https://openai.com/dall-e-2/)]
- [Photorealistic Text-to-Image Diffusion Models with Deep Language Understanding](https://arxiv.org/abs/2205.11487) [[website](https://imagen.research.google/)]
- [Scaling Autoregressive Models for Content-Rich Text-to-Image Generation](https://arxiv.org/abs/2206.10789)
- [Learning Transferable Visual Models From Natural Language Supervision](https://cdn.openai.com/papers/Learning_Transferable_Visual_Models_From_Natural_Language_Supervision.pdf) [[github](https://github.com/openai/CLIP)] [[website](https://openai.com/blog/clip/)]
- [How Much Can CLIP Benefit Vision-and-Language Tasks?](https://arxiv.org/abs/2107.06383)
- [EfficientCLIP: Efficient Cross-Modal Pre-training by Ensemble Confident Learning and Language Modeling](https://arxiv.org/abs/2109.04699)
- [e-CLIP: Large-Scale Vision-Language Representation Learning in E-commerce](https://arxiv.org/abs/2207.00208)
- [Chinese CLIP: Contrastive Vision-Language Pretraining in Chinese](https://arxiv.org/abs/2211.01335)
- [Enabling Multimodal Generation on CLIP via Vision-Language Knowledge Distillation](https://arxiv.org/abs/2203.06386) (ACL2022)
- [StyleCLIP: Text-Driven Manipulation of StyleGAN Imagery](https://arxiv.org/abs/2103.17249)
- [Training Vision Transformers for Image Retrieval](https://arxiv.org/abs/2102.05644)
- [LightningDOT: Pre-training Visual-Semantic Embeddings for Real-Time Image-Text Retrieval](https://arxiv.org/abs/2103.08784) (NAACL2021)
- [Colorization Transformer](https://arxiv.org/abs/2102.04432) (ICLR2021) [[github](https://github.com/google-research/google-research/tree/master/coltran)]
- [A Better Use of Audio-Visual Cues: Dense Video Captioning with Bi-modal Transformer](https://arxiv.org/abs/2005.08271) [[website](https://v-iashin.github.io/bmt)]
- [Multimodal Pretraining for Dense Video Captioning](https://arxiv.org/abs/2011.11760) (AACL-IJCNLP2020)
- [Is Space-Time Attention All You Need for Video Understanding?](https://arxiv.org/abs/2102.05095)
- [Less is More: ClipBERT for Video-and-Language Learning via Sparse Sampling](https://arxiv.org/abs/2102.06183) (CVPR2021) [[github](https://github.com/jayleicn/ClipBERT)]
- [VLM: Task-agnostic Video-Language Model Pre-training for Video Understanding](https://arxiv.org/abs/2105.09996) (ACL2021 Findings)
- [VideoCLIP: Contrastive Pre-training for Zero-shot Video-Text Understanding](https://arxiv.org/abs/2109.14084) (EMNLP2021)
- [BERT-hLSTMs: BERT and Hierarchical LSTMs for Visual Storytelling](https://arxiv.org/abs/2012.02128)
- [A Generalist Agent](https://arxiv.org/abs/2205.06175) [[website](https://www.deepmind.com/publications/a-generalist-agent)]
- [SpeechBERT: Cross-Modal Pre-trained Language Model for End-to-end Spoken Question Answering](https://arxiv.org/abs/1910.11559)
- [An Audio-enriched BERT-based Framework for Spoken Multiple-choice Question Answering](https://arxiv.org/abs/2005.12142)
- [vq-wav2vec: Self-Supervised Learning of Discrete Speech Representations](https://arxiv.org/abs/1910.05453)
- [Effectiveness of self-supervised pre-training for speech recognition](https://arxiv.org/abs/1911.03912)
- [wav2vec 2.0: A Framework for Self-Supervised Learning of Speech Representations](https://arxiv.org/abs/2006.11477)
- [Applying wav2vec2.0 to Speech Recognition in various low-resource languages](https://arxiv.org/abs/2012.12121)
- [Pushing the Limits of Semi-Supervised Learning for Automatic Speech Recognition](https://arxiv.org/abs/2010.10504)
- [Speech Recognition by Simply Fine-tuning BERT](https://arxiv.org/abs/2102.00291) (ICASSP2021)
- [Understanding Semantics from Speech Through Pre-training](https://arxiv.org/abs/1909.10924)
- [Speech-XLNet: Unsupervised Acoustic Model Pretraining For Self-Attention Networks](https://arxiv.org/abs/1910.10387)
- [Learning Speech Representations from Raw Audio by Joint Audiovisual Self-Supervision](https://arxiv.org/abs/2007.04134) (ICML2020 WS)
- [Semi-Supervised Spoken Language Understanding via Self-Supervised Speech and Language Model Pretraining](https://arxiv.org/abs/2010.13826)
- [ST-BERT: Cross-modal Language Model Pre-training For End-to-end Spoken Language Understanding](https://arxiv.org/abs/2010.12283)
- [End-to-end spoken language understanding using transformer networks and self-supervised pre-trained features](https://arxiv.org/abs/2011.08238)
- [Speech-language Pre-training for End-to-end Spoken Language Understanding](https://arxiv.org/abs/2102.06283)
- [Jointly Encoding Word Confusion Network and Dialogue Context with BERT for Spoken Language Understanding](https://arxiv.org/abs/2005.11640) (Interspeech2020)
- [AudioCLIP: Extending CLIP to Image, Text and Audio](https://arxiv.org/abs/2106.13043)
- [Audio ALBERT: A Lite BERT for Self-supervised Learning of Audio Representation](https://arxiv.org/abs/2005.08575)
- [Unsupervised Cross-lingual Representation Learning for Speech Recognition](https://arxiv.org/abs/2006.13979)
- [Curriculum Pre-training for End-to-End Speech Translation](https://arxiv.org/abs/2004.10093) (ACL2020)
- [MAM: Masked Acoustic Modeling for End-to-End Speech-to-Text Translation](https://arxiv.org/abs/2010.11445)
- [Multilingual Speech Translation with Efficient Finetuning of Pretrained Models](https://arxiv.org/abs/2010.12829) (ACL2021)
- [Multilingual Byte2Speech Text-To-Speech Models Are Few-shot Spoken Language Learners](https://arxiv.org/abs/2103.03541)
- [Towards Transfer Learning for End-to-End Speech Synthesis from Deep Pre-Trained Language Models](https://arxiv.org/abs/1906.07307)
- [To BERT or Not To BERT: Comparing Speech and Language-based Approaches for Alzheimer's Disease Detection](https://arxiv.org/abs/2008.01551) (Interspeech2020)
- [BERT for Joint Multichannel Speech Dereverberation with Spatial-aware Tasks](https://arxiv.org/abs/2010.10892)
## Model compression
- [Compression of Deep Learning Models for Text: A Survey](https://arxiv.org/abs/2008.05221)
- [Distilling Task-Specific Knowledge from BERT into Simple Neural Networks](https://arxiv.org/abs/1903.12136)
- [Patient Knowledge Distillation for BERT Model Compression](https://arxiv.org/abs/1908.09355) (EMNLP2019)
- [Small and Practical BERT Models for Sequence Labeling](https://arxiv.org/abs/1909.00100) (EMNLP2019)
- [TinyBERT: Distilling BERT for Natural Language Understanding](https://arxiv.org/abs/1909.10351) [[github](https://github.com/huawei-noah/Pretrained-Language-Model/tree/master/TinyBERT)]
- [DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lighter](https://arxiv.org/abs/1910.01108) (NeurIPS2019 WS) [[github](https://github.com/huggingface/transformers/tree/master/examples/distillation)]
- [Contrastive Distillation on Intermediate Representations for Language Model Compression](https://arxiv.org/abs/2009.14167) (EMNLP2020)
- [Knowledge Distillation from Internal Representations](https://arxiv.org/abs/1910.03723) (AAAI2020)
- [Reinforced Multi-Teacher Selection for Knowledge Distillation](https://arxiv.org/abs/2012.06048) (AAAI2021)
- [ALP-KD: Attention-Based Layer Projection for Knowledge Distillation](https://arxiv.org/abs/2012.14022) (AAAI2021)
- [Dynamic Knowledge Distillation for Pre-trained Language Models](https://arxiv.org/abs/2109.11295) (EMNLP2021)
- [Distilling Linguistic Context for Language Model Compression](https://arxiv.org/abs/2109.08359) (EMNLP2021)
- [Improving Task-Agnostic BERT Distillation with Layer Mapping Search](https://arxiv.org/abs/2012.06153)
- [PoWER-BERT: Accelerating BERT inference for Classification Tasks](https://arxiv.org/abs/2001.08950)
- [WaLDORf: Wasteless Language-model Distillation On Reading-comprehension](https://arxiv.org/abs/1912.06638)
- [Extremely Small BERT Models from Mixed-Vocabulary Training](https://arxiv.org/abs/1909.11687) (EACL2021)
- [BERT-of-Theseus: Compressing BERT by Progressive Module Replacing](https://arxiv.org/abs/2002.02925) (EMNLP2020)
- [Compressing BERT: Studying the Effects of Weight Pruning on Transfer Learning](https://arxiv.org/abs/2002.08307) (ACL2020 SRW)
- [MiniLM: Deep Self-Attention Distillation for Task-Agnostic Compression of Pre-Trained Transformers](https://arxiv.org/abs/2002.10957)
- [Extract then Distill: Efficient and Effective Task-Agnostic BERT Distillation](https://arxiv.org/abs/2104.11928)
- [Compressing Large-Scale Transformer-Based Models: A Case Study on BERT](https://arxiv.org/abs/2002.11985)
- [Train Large, Then Compress: Rethinking Model Size for Efficient Training and Inference of Transformers](https://arxiv.org/abs/2002.11794)
- [Well-Read Students Learn Better: On the Importance of Pre-training Compact Models](https://arxiv.org/abs/1908.08962)
- [MobileBERT: a Compact Task-Agnostic BERT for Resource-Limited Devices](https://arxiv.org/abs/2004.02984) (ACL2020)
- [Distilling Knowledge from Pre-trained Language Models via Text Smoothing](https://arxiv.org/abs/2005.03848)
- [DynaBERT: Dynamic BERT with Adaptive Width and Depth](https://arxiv.org/abs/2004.04037)
- [Reducing Transformer Depth on Demand with Structured Dropout](https://arxiv.org/abs/1909.11556)
- [DeeBERT: Dynamic Early Exiting for Accelerating BERT Inference](https://www.aclweb.org/anthology/2020.acl-main.204/) (ACL2020)
- [BERT Loses Patience: Fast and Robust Inference with Early Exit](https://arxiv.org/abs/2006.04152) [[github](https://github.com/JetRunner/PABEE)] [[github](https://github.com/huggingface/transformers/tree/master/examples/bert-loses-patience)]
- [Accelerating BERT Inference for Sequence Labeling via Early-Exit](https://arxiv.org/abs/2105.13878) (ACL2021)
- [Elbert: Fast Albert with Confidence-Window Based Early Exit](https://arxiv.org/abs/2107.00175)
- [RomeBERT: Robust Training of Multi-Exit BERT](https://arxiv.org/abs/2101.09755)
- [TR-BERT: Dynamic Token Reduction for Accelerating BERT Inference](https://arxiv.org/abs/2105.11618) (NAACL2021)
- [FastBERT: a Self-distilling BERT with Adaptive Inference Time](https://www.aclweb.org/anthology/2020.acl-main.537/) (ACL2020)
- [Distilling Large Language Models into Tiny and Effective Students using pQRNN](https://arxiv.org/abs/2101.08890)
- [Towards Non-task-specific Distillation of BERT via Sentence Representation Approximation](https://arxiv.org/abs/2004.03097)
- [LadaBERT: Lightweight Adaptation of BERT through Hybrid Model Compression](https://arxiv.org/abs/2004.04124) (COLING2020)
- [Poor Man's BERT: Smaller and Faster Transformer Models](https://arxiv.org/abs/2004.03844)
- [schuBERT: Optimizing Elements of BERT](https://arxiv.org/abs/2005.06628) (ACL2020)
- [BERT-EMD: Many-to-Many Layer Mapping for BERT Compression with Earth Mover's Distance](https://arxiv.org/abs/2010.06133) (EMNLP2020) [[github](https://github.com/lxk00/BERT-EMD)]
- [One Teacher is Enough? Pre-trained Language Model Distillation from Multiple Teachers](https://arxiv.org/abs/2106.01023) (ACL2021 Findings)
- [From Dense to Sparse: Contrastive Pruning for Better Pre-trained Language Model Compression](https://arxiv.org/abs/2112.07198) (AAAI2022)
- [TinyMBERT: Multi-Stage Distillation Framework for Massive Multi-lingual NER](https://arxiv.org/abs/2004.05686) (ACL2020)
- [XtremeDistil: Multi-stage Distillation for Massive Multilingual Models](https://www.aclweb.org/anthology/2020.acl-main.202/) (ACL2020)
- [Robustly Optimized and Distilled Training for Natural Language Understanding](https://arxiv.org/abs/2103.08809)
- [Structured Pruning of Large Language Models](https://arxiv.org/abs/1910.04732)
- [Movement Pruning: Adaptive Sparsity by Fine-Tuning](https://arxiv.org/abs/2005.07683) [[github](https://github.com/huggingface/transformers/tree/master/examples/movement-pruning)]
- [Efficient Transformer-based Large Scale Language Representations using Hardware-friendly Block Structured Pruning](https://arxiv.org/abs/2009.08065) (EMNLP2020 Findings)
- [Pruning Redundant Mappings in Transformer Models via Spectral-Normalized Identity Prior](https://www.aclweb.org/anthology/2020.findings-emnlp.64/) (EMNLP2020 Findings)
- [Parameter-Efficient Transfer Learning with Diff Pruning](https://arxiv.org/abs/2012.07463)
- [FastFormers: Highly Efficient Transformer Models for Natural Language Understanding](https://arxiv.org/abs/2010.13382) (EMNLP2020 WS) [[github](https://github.com/microsoft/fastformers)]
- [AutoTinyBERT: Automatic Hyper-parameter Optimization for Efficient Pre-trained Language Models](https://arxiv.org/abs/2107.13686) (ACL2021) [[github](https://github.com/huawei-noah/Pretrained-Language-Model/tree/master/AutoTinyBERT)]
- [Adapt-and-Distill: Developing Small, Fast and Effective Pretrained Language Models for Domains](https://arxiv.org/abs/2106.13474) (ACL2021 Findings)
- [Distilling BERT into Simple Neural Networks with Unlabeled Transfer Data](https://arxiv.org/abs/1910.01769)
- [AdaBERT: Task-Adaptive BERT Compression with Differentiable Neural Architecture Search](https://arxiv.org/abs/2001.04246)
- [SqueezeBERT: What can computer vision teach NLP about efficient neural networks?](https://arxiv.org/abs/2006.11316)
- [Optimizing Transformers with Approximate Computing for Faster, Smaller and more Accurate NLP Models](https://arxiv.org/abs/2010.03688)
- [An Approximation Algorithm for Optimal Subarchitecture Extraction](https://arxiv.org/abs/2010.08512) [[github](https://github.com/alexa/bort/)]
- [Structured Pruning of a BERT-based Question Answering Model](https://arxiv.org/abs/1910.06360)
- [DeFormer: Decomposing Pre-trained Transformers for Faster Question Answering](https://arxiv.org/abs/2005.00697) (ACL2020)
- [Distilling Knowledge Learned in BERT for Text Generation](https://arxiv.org/abs/1911.03829) (ACL2020)
- [Distilling the Knowledge of BERT for Sequence-to-Sequence ASR](https://arxiv.org/abs/2008.03822) (Interspeech2020)
- [Pre-trained Summarization Distillation](https://arxiv.org/abs/2010.13002)
- [Understanding BERT Rankers Under Distillation](https://arxiv.org/abs/2007.11088) (ICTIR2020)
- [Simplified TinyBERT: Knowledge Distillation for Document Retrieval](https://arxiv.org/abs/2009.07531)
- [Exploring the Limits of Simple Learners in Knowledge Distillation for Document Classification with DocBERT](https://www.aclweb.org/anthology/2020.repl4nlp-1.10/) (ACL2020 WS)
- [TextBrewer: An Open-Source Knowledge Distillation Toolkit for Natural Language Processing](https://arxiv.org/abs/2002.12620) (ACL2020 Demo)
- [TopicBERT for Energy Efficient Document Classification](https://arxiv.org/abs/2010.16407) (EMNLP2020 Findings)
- [MiniVLM: A Smaller and Faster Vision-Language Model](https://arxiv.org/abs/2012.06946)
- [Compressing Visual-linguistic Model via Knowledge Distillation](https://arxiv.org/abs/2104.02096)
- [Playing Lottery Tickets with Vision and Language](https://arxiv.org/abs/2104.11832)
- [Q-BERT: Hessian Based Ultra Low Precision Quantization of BERT](https://arxiv.org/abs/1909.05840)
- [Q8BERT: Quantized 8Bit BERT](https://arxiv.org/abs/1910.06188) (NeurIPS2019 WS)
- [Training with Quantization Noise for Extreme Model Compression](https://arxiv.org/abs/2004.07320) (ICLR2021)
- [Hardware Acceleration of Fully Quantized BERT for Efficient Natural Language Processing](https://arxiv.org/abs/2103.02800)
- [BinaryBERT: Pushing the Limit of BERT Quantization](https://arxiv.org/abs/2012.15701) (ACL2021)
- [I-BERT: Integer-only BERT Quantization](https://arxiv.org/abs/2101.01321)
- [ROSITA: Refined BERT cOmpreSsion with InTegrAted techniques](https://arxiv.org/abs/2103.11367) (AAAI2021)
- [TernaryBERT: Distillation-aware Ultra-low Bit BERT](https://arxiv.org/abs/2009.12812) (EMNLP2020)
- [EdgeBERT: Optimizing On-Chip Inference for Multi-Task NLP](https://arxiv.org/abs/2011.14203)
- [Optimizing Inference Performance of Transformers on CPUs](https://arxiv.org/abs/2102.06621)
## Large language model
- [Language Models are Unsupervised Multitask Learners](https://d4mucfpksywv.cloudfront.net/better-language-models/language-models.pdf) [[github](https://github.com/openai/gpt-2)]
- [Language Models are Few-Shot Learners](https://arxiv.org/abs/2005.14165) (NeurIPS2020) [[github](https://github.com/openai/gpt-3)]
- [Language Models as Few-Shot Learner for Task-Oriented Dialogue Systems](https://arxiv.org/abs/2008.06239)
- [OPT: Open Pre-trained Transformer Language Models](https://arxiv.org/abs/2205.01068) [[website](https://ai.facebook.com/blog/democratizing-access-to-large-scale-language-models-with-opt-175b/)]
- [GPT-NeoX-20B: An Open-Source Autoregressive Language Model](https://arxiv.org/abs/2204.06745)
- [Scaling Language Models: Methods, Analysis & Insights from Training Gopher](https://arxiv.org/abs/2112.11446)
- [Switch Transformers: Scaling to Trillion Parameter Models with Simple and Efficient Sparsity](https://arxiv.org/abs/2101.03961)
- [GLaM: Efficient Scaling of Language Models with Mixture-of-Experts](https://arxiv.org/abs/2112.06905) [[blog](https://ai.googleblog.com/2021/12/more-efficient-in-context-learning-with.html)]
- [Training Compute-Optimal Large Language Models](https://arxiv.org/abs/2203.15556)
- [PaLM: Scaling Language Modeling with Pathways](https://arxiv.org/abs/2204.02311) [[blog](https://ai.googleblog.com/2022/04/pathways-language-model-palm-scaling-to.html)]
- [LLaMA: Open and Efficient Foundation Language Models](https://arxiv.org/abs/2302.13971)
- [Pythia: A Suite for Analyzing Large Language Models Across Training and Scaling](https://arxiv.org/abs/2304.01373) [[github](https://github.com/EleutherAI/pythia)]
- [PolyLM: An Open Source Polyglot Large Language Model](https://arxiv.org/abs/2307.06018)
- [Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism](https://arxiv.org/abs/1909.08053)
- [Efficient Large-Scale Language Model Training on GPU Clusters Using Megatron-LM](https://arxiv.org/abs/2104.04473)
- [Using DeepSpeed and Megatron to Train Megatron-Turing NLG 530B, A Large-Scale Generative Language Model](https://arxiv.org/abs/2201.11990) [[blog](https://www.microsoft.com/en-us/research/blog/using-deepspeed-and-megatron-to-train-megatron-turing-nlg-530b-the-worlds-largest-and-most-powerful-generative-language-model/)]
- [DeepSpeed Inference: Enabling Efficient Inference of Transformer Models at Unprecedented Scale](https://arxiv.org/abs/2207.00032) [[github](https://github.com/microsoft/DeepSpeed)]
- [ZeRO: Memory Optimizations Toward Training Trillion Parameter Models](https://arxiv.org/abs/1910.02054)
- [ZeRO++: Extremely Efficient Collective Communication for Giant Model Training](https://arxiv.org/abs/2306.10209) [[blog](https://www.microsoft.com/en-us/research/blog/deepspeed-zero-a-leap-in-speed-for-llm-and-chat-model-training-with-4x-less-communication/)]
- [ZeRO-Infinity: Breaking the GPU Memory Wall for Extreme Scale Deep Learning](https://arxiv.org/abs/2104.07857)
## Reinforcement learning from human feedback
- [Fine-Tuning Language Models from Human Preferences](https://arxiv.org/abs/1909.08593) [[github](https://github.com/openai/lm-human-preferences)] [[blog](https://openai.com/blog/fine-tuning-gpt-2/)]
- [Training language models to follow instructions with human feedback](https://arxiv.org/abs/2203.02155) [[github](https://github.com/openai/following-instructions-human-feedback)] [[blog](https://openai.com/blog/instruction-following/)]
- [WebGPT: Browser-assisted question-answering with human feedback](https://arxiv.org/abs/2112.09332) [[blog](https://openai.com/blog/webgpt/)]
- [Improving alignment of dialogue agents via targeted human judgements](https://arxiv.org/abs/2209.14375)
- [Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedback](https://arxiv.org/abs/2204.05862)
- [Training Language Models with Language Feedback](https://arxiv.org/abs/2204.14146) (ACL2022 WS)
- [Self-Instruct: Aligning Language Model with Self Generated Instructions](https://arxiv.org/abs/2212.10560) [[github](https://github.com/yizhongw/self-instruct)]
- [Is ChatGPT a General-Purpose Natural Language Processing Task Solver?](https://arxiv.org/abs/2302.06476)
- [ChatGPT: A Meta-Analysis after 2.5 Months](https://arxiv.org/abs/2302.13795)
## Misc.
- [Extracting Training Data from Large Language Models](https://arxiv.org/abs/2012.07805)
- [Generative Language Modeling for Automated Theorem Proving](https://arxiv.org/abs/2009.03393)
- [Do you have the right scissors? Tailoring Pre-trained Language Models via Monte-Carlo Methods](https://www.aclweb.org/anthology/2020.acl-main.314/) (ACL2020)
- [jiant: A Software Toolkit for Research on General-Purpose Text Understanding Models](https://arxiv.org/abs/2003.02249) [[github](https://github.com/nyu-mll/jiant/)]
- [Cloze-driven Pretraining of Self-attention Networks](https://arxiv.org/abs/1903.07785)
- [Learning and Evaluating General Linguistic Intelligence](https://arxiv.org/abs/1901.11373)
- [To Tune or Not to Tune? Adapting Pretrained Representations to Diverse Tasks](https://arxiv.org/abs/1903.05987) (ACL2019 WS)
- [Learning to Speak and Act in a Fantasy Text Adventure Game](https://www.aclweb.org/anthology/D19-1062/) (EMNLP2019)
- [A Two-Stage Masked LM Method for Term Set Expansion](https://arxiv.org/abs/2005.01063) (ACL2020)
- [Cold-start Active Learning through Self-supervised Language Modeling](https://arxiv.org/abs/2010.09535) (EMNLP2020)
- [Conditional BERT Contextual Augmentation](https://arxiv.org/abs/1812.06705)
- [Data Augmentation using Pre-trained Transformer Models](https://arxiv.org/abs/2003.02245) (AACL-IJCNLP2020) [[github](https://github.com/varinf/TransformersDataAugmentation)]
- [Mixup-Transfomer: Dynamic Data Augmentation for NLP Tasks](https://arxiv.org/abs/2010.02394) (COLING2020)
- [GPT3Mix: Leveraging Large-scale Language Models for Text Augmentation](https://arxiv.org/abs/2104.08826)
- [Unsupervised Text Style Transfer with Padded Masked Language Models](https://arxiv.org/abs/2010.01054) (EMNLP2020)
- [Assessing Discourse Relations in Language Generation from Pre-trained Language Models](https://arxiv.org/abs/2004.12506)
- [Large Batch Optimization for Deep Learning: Training BERT in 76 minutes](https://arxiv.org/abs/1904.00962) (ICLR2020)
- [Accelerated Large Batch Optimization of BERT Pretraining in 54 minutes](https://arxiv.org/abs/2006.13484)
- [IsoBN: Fine-Tuning BERT with Isotropic Batch Normalization](https://arxiv.org/abs/2005.02178) (AAAI2021)
- [Multi-node Bert-pretraining: Cost-efficient Approach](https://arxiv.org/abs/2008.00177)
- [How to Train BERT with an Academic Budget](https://arxiv.org/abs/2104.07705)
- [Amazon SageMaker Model Parallelism: A General and Flexible Framework for Large Model Training](https://arxiv.org/abs/2111.05972)
- [PatrickStar: Parallel Training of Pre-trained Models via Chunk-based Memory Management](https://arxiv.org/abs/2108.05818) [[github](https://github.com/Tencent/PatrickStar)]
- [1-bit Adam: Communication Efficient Large-Scale Training with Adam's Convergence Speed](https://arxiv.org/abs/2102.02888)
- [TeraPipe: Token-Level Pipeline Parallelism for Training Large-Scale Language Models](https://arxiv.org/abs/2102.07988)
- [Efficient Large-Scale Language Model Training on GPU Clusters](https://arxiv.org/abs/2104.04473)
- [Scaling Laws for Neural Language Models](https://arxiv.org/abs/2001.08361)
- [Scaling Laws for Autoregressive Generative Modeling](https://arxiv.org/abs/2010.14701)
- [Scale Efficiently: Insights from Pre-training and Fine-tuning Transformers](https://arxiv.org/abs/2109.10686)
- [The Pile: An 800GB Dataset of Diverse Text for Language Modeling](https://arxiv.org/abs/2101.00027) [[website](https://pile.eleuther.ai/)]
- [Deduplicating Training Data Makes Language Models Better](https://arxiv.org/abs/2107.06499)
- [Mixout: Effective Regularization to Finetune Large-scale Pretrained Language Models](https://openreview.net/forum?id=HkgaETNtDB) (ICLR2020)
- [A Mutual Information Maximization Perspective of Language Representation Learning](https://openreview.net/forum?id=Syx79eBKwr) (ICLR2020)
- [Is BERT Really Robust? Natural Language Attack on Text Classification and Entailment](https://arxiv.org/abs/1907.11932) (AAAI2020)
- [Weight Poisoning Attacks on Pre-trained Models](https://arxiv.org/abs/2004.06660) (ACL2020)
- [BERT-ATTACK: Adversarial Attack Against BERT Using BERT](https://arxiv.org/abs/2004.09984) (EMNLP2020)
- [BERT-Defense: A Probabilistic Model Based on BERT to Combat Cognitively Inspired Orthographic Adversarial Attacks](https://arxiv.org/abs/2106.01452) (ACL2021 Findings)
- [Model Extraction and Adversarial Transferability, Your BERT is Vulnerable!](https://arxiv.org/abs/2103.10013) (NAACL2021)
- [Adv-BERT: BERT is not robust on misspellings! Generating nature adversarial samples on BERT](https://arxiv.org/abs/2003.04985)
- [Robust Encodings: A Framework for Combating Adversarial Typos](https://www.aclweb.org/anthology/2020.acl-main.245/) (ACL2020)
- [On the Robustness of Language Encoders against Grammatical Errors](https://arxiv.org/abs/2005.05683) (ACL2020)
- [Evaluating the Robustness of Neural Language Models to Input Perturbations](https://arxiv.org/abs/2108.12237) (EMNLP2021)
- [Pretrained Transformers Improve Out-of-Distribution Robustness](https://arxiv.org/abs/2004.06100) (ACL2020) [[github](https://github.com/camelop/NLP-Robustness)]
- ["You are grounded!": Latent Name Artifacts in Pre-trained Language Models](https://arxiv.org/abs/2004.03012) (EMNLP2020)
- [The Right Tool for the Job: Matching Model and Instance Complexities](https://arxiv.org/abs/2004.07453) (ACL2020) [[github](https://github.com/allenai/sledgehammer)]
- [Unsupervised Domain Clusters in Pretrained Language Models](https://arxiv.org/abs/2004.02105) (ACL2020)
- [Thieves on Sesame Street! Model Extraction of BERT-based APIs](https://arxiv.org/abs/1910.12366) (ICLR2020)
- [Graph-Bert: Only Attention is Needed for Learning Graph Representations](https://arxiv.org/abs/2001.05140)
- [Graph-Aware Transformer: Is Attention All Graphs Need?](https://arxiv.org/abs/2006.05213)
- [CodeBERT: A Pre-Trained Model for Programming and Natural Languages](https://arxiv.org/abs/2002.08155) (EMNLP2020 Findings)
- [Unsupervised Translation of Programming Languages](https://arxiv.org/abs/2006.03511)
- [Unified Pre-training for Program Understanding and Generation](https://arxiv.org/abs/2103.06333) (NAACL2021)
- [MathBERT: A Pre-Trained Model for Mathematical Formula Understanding](https://arxiv.org/abs/2105.00377)
- [Investigating Math Word Problems using Pretrained Multilingual Language Models](https://arxiv.org/abs/2105.08928)
- [Measuring and Improving BERT's Mathematical Abilities by Predicting the Order of Reasoning](https://arxiv.org/abs/2106.03921) (ACL2021)
- [Pre-train or Annotate? Domain Adaptation with a Constrained Budget](https://arxiv.org/abs/2109.04711) (EMNLP2021)
- [Item-based Collaborative Filtering with BERT](https://www.aclweb.org/anthology/2020.ecnlp-1.8/) (ACL2020 WS)
- [RecoBERT: A Catalog Language Model for Text-Based Recommendations](https://arxiv.org/abs/2009.13292)
- [Fine-Tuning Pretrained Language Models: Weight Initializations, Data Orders, and Early Stopping](https://arxiv.org/abs/2002.06305)
- [Extending Machine Language Models toward Human-Level Language Understanding](https://arxiv.org/abs/1912.05877)
- [Climbing towards NLU: On Meaning, Form, and Understanding in the Age of Data](https://openreview.net/forum?id=GKTvAcb12b) (ACL2020)
- [Are Larger Pretrained Language Models Uniformly Better? Comparing Performance at the Instance Level](https://arxiv.org/abs/2105.06020) (ACL2021 Findings) [[github](https://github.com/ruiqi-zhong/acl2021-instance-level)]
- [Glyce: Glyph-vectors for Chinese Character Representations](https://arxiv.org/abs/1901.10125)
- [Back to the Future -- Sequential Alignment of Text Representations](https://arxiv.org/abs/1909.03464)
- [Improving Cuneiform Language Identification with BERT](https://www.aclweb.org/anthology/papers/W/W19/W19-1402/) (NAACL2019 WS)
- [Generating Derivational Morphology with BERT](https://arxiv.org/abs/2005.00672)
- [BERT has a Moral Compass: Improvements of ethical and moral values of machines](https://arxiv.org/abs/1912.05238)
- [MusicBERT: Symbolic Music Understanding with Large-Scale Pre-Training](https://arxiv.org/abs/2106.05630) (ACL2021 Findings)
- [SMILES-BERT: Large Scale Unsupervised Pre-Training for Molecular Property Prediction](https://dl.acm.org/citation.cfm?id=3342186) (ACM-BCB2019)
- [ChemBERTa: Large-Scale Self-Supervised Pretraining for Molecular Property Prediction](https://arxiv.org/abs/2010.09885)
- [BERT Learns (and Teaches) Chemistry](https://arxiv.org/abs/2007.16012)
- [Prediction of RNA-protein interactions using a nucleotide language model](https://www.biorxiv.org/content/10.1101/2021.04.27.441365v1)
- [Sketch-BERT: Learning Sketch Bidirectional Encoder Representation from Transformers by Self-supervised Learning of Sketch Gestalt](https://arxiv.org/abs/2005.09159) (CVPR2020)
- [The Chess Transformer: Mastering Play using Generative Language Models](https://arxiv.org/abs/2008.04057)
- [The Go Transformer: Natural Language Modeling for Game Play](https://arxiv.org/abs/2007.03500)
- [On the comparability of Pre-trained Language Models](https://arxiv.org/abs/2001.00781)
- [Transformers: State-of-the-art Natural Language Processing](https://arxiv.org/abs/1910.03771)
- [The Cost of Training NLP Models: A Concise Overview](https://arxiv.org/abs/2004.08900)