SOTAVerified

Natural Language Understanding

Natural Language Understanding is an important field of Natural Language Processing which contains various tasks such as text classification, natural language inference and story comprehension. Applications enabled by natural language understanding range from question answering to automated reasoning.

Source: Find a Reasonable Ending for Stories: Does Logic Relation Help the Story Cloze Test?

Papers

Recently AddedMost HypedMost ActiveNeeds VerificationMost Verified

Showing 151–200 of 1978 papers

TitleStatusHype
DialoGLUE: A Natural Language Understanding Benchmark for Task-Oriented DialogueCode1
Joint Prompt Optimization of Stacked LLMs using Variational InferenceCode1
Instructive Decoding: Instruction-Tuned Large Language Models are Self-Refiner from Noisy InstructionsCode1
Debiasing Methods in Natural Language Understanding Make Bias More AccessibleCode1
AdaMix: Mixture-of-Adaptations for Parameter-efficient Model TuningCode1
Decorrelate Irrelevant, Purify Relevant: Overcome Textual Spurious Correlations from a Feature PerspectiveCode1
C-STS: Conditional Semantic Textual SimilarityCode1
CSKG: The CommonSense Knowledge GraphCode1
CTRAN: CNN-Transformer-based Network for Natural Language UnderstandingCode1
Automatic Disfluency Detection from Untranscribed SpeechCode1
Beyond Silent Letters: Amplifying LLMs in Emotion Recognition with Vocal NuancesCode1
A Win-win Deal: Towards Sparse and Robust Pre-trained Language ModelsCode1
data2vec: A General Framework for Self-supervised Learning in Speech, Vision and LanguageCode1
DeLoRA: Decoupling Angles and Strength in Low-rank AdaptationCode1
Efficient Deep Learning: A Survey on Making Deep Learning Models Smaller, Faster, and BetterCode1
Explaining Patterns in Data with Language Models via Interpretable AutopromptingCode1
ConvBERT: Improving BERT with Span-based Dynamic ConvolutionCode1
Conic10K: A Challenging Math Problem Understanding and Reasoning DatasetCode1
A Sequence-to-Sequence Approach to Dialogue State TrackingCode1
A Large-scale Comprehensive Abusiveness Detection Dataset with Multifaceted Labels from RedditCode1
A Simple but Tough-to-Beat Data Augmentation Approach for Natural Language Understanding and GenerationCode1
DroidCall: A Dataset for LLM-powered Android Intent InvocationCode1
ALBETO and DistilBETO: Lightweight Spanish Language ModelsCode1
Dual Learning for Semi-Supervised Natural Language UnderstandingCode1
Convolution-enhanced Evolving Attention NetworksCode1
CONDAQA: A Contrastive Reading Comprehension Dataset for Reasoning about NegationCode1
Efficient Long-Text Understanding with Short-Text ModelsCode1
Efficient Self-supervised Learning with Contextualized Target Representations for Vision, Speech and LanguageCode1
Adapting Pretrained Transformer to Lattices for Spoken Language UnderstandingCode1
Empowering Molecule Discovery for Molecule-Caption Translation with Large Language Models: A ChatGPT PerspectiveCode1
Active Learning by Acquiring Contrastive ExamplesCode1
Endowing Language Models with Multimodal Knowledge Graph RepresentationsCode1
AutoTrans: Automating Transformer Design via Reinforced Architecture SearchCode1
EPIC30M: An Epidemics Corpus Of Over 30 Million Relevant TweetsCode1
ConFiguRe: Exploring Discourse-level Chinese Figures of SpeechCode1
Error Analysis Prompting Enables Human-Like Translation Evaluation in Large Language ModelsCode1
AllenNLP: A Deep Semantic Natural Language Processing PlatformCode1
CPT: A Pre-Trained Unbalanced Transformer for Both Chinese Language Understanding and GenerationCode1
COMET-ATOMIC 2020: On Symbolic and Neural Commonsense Knowledge GraphsCode1
A Surprisingly Robust Trick for Winograd Schema ChallengeCode1
Does It Make Sense? And Why? A Pilot Study for Sense Making and ExplanationCode1
Baby's CoThought: Leveraging Large Language Models for Enhanced Reasoning in Compact ModelsCode1
Belief Revision based Caption Re-ranker with Visual Semantic InformationCode1
Extracting Event Temporal Relations via Hyperbolic GeometryCode1
A Dataset for Statutory Reasoning in Tax Law Entailment and Question AnsweringCode1
A Survey of RWKVCode1
KNOT: Knowledge Distillation using Optimal Transport for Solving NLP TasksCode1
FewCLUE: A Chinese Few-shot Learning Evaluation BenchmarkCode1
Collaborative Multi-Agent Dialogue Model Training Via Reinforcement LearningCode1
Comparison by Conversion: Reverse-Engineering UCCA from Syntax and Lexical SemanticsCode1
Show:102550
← PrevPage 4 of 40Next →
All datasetsPDP60STREUSLELexGLUEDialoGLUE fewshotDialoGLUE fullGLUE

Benchmark Results

#ModelMetricClaimedVerifiedStatus
1HNNAccuracy90—Unverified
2BERT-large 340MAccuracy78.3—Unverified
3UDSSM-II (ensemble)Accuracy78.3—Unverified
4UDSSM-I (ensemble)Accuracy76.7—Unverified
5DSSMAccuracy75—Unverified
6UDSSM-IIAccuracy75—Unverified
7BERT-base 110M + MASAccuracy68.3—Unverified
8USSM + Supervised Deepnet + 3 Knowledge BasesAccuracy66.7—Unverified
9Word-level CNN+LSTM (full scoring)Accuracy60—Unverified
10Subword-level Transformer LMAccuracy58.3—Unverified
#ModelMetricClaimedVerifiedStatus
1BERT (pred POS/lemmas)Tags (Full) Acc82.5—Unverified
2BERT (none)Tags (Full) Acc82—Unverified
3BERT (gold POS/lemmas)Tags (Full) Acc81—Unverified
4GloVe (gold POS/lemmas)Tags (Full) Acc79.3—Unverified
5RoBERTa + LinearFull F1 (Preps)78.2—Unverified
6GloVe (none)Tags (Full) Acc77.5—Unverified
7GloVe (pred POS/lemmas)Tags (Full) Acc77.1—Unverified
8SVM (feature-rich, gold syntax)Role F1 (Preps)62.2—Unverified
9BiLSTM + MLP (gold syntax)Role F1 (Preps)62.2—Unverified
10SVM (feature-rich, auto syntax)Role F1 (Preps)58.2—Unverified
#ModelMetricClaimedVerifiedStatus
1CaseLaw-BERTCaseHOLD75.6—Unverified
2Legal-BERTCaseHOLD75.1—Unverified
3DeBERTaCaseHOLD72.1—Unverified
4LongformerCaseHOLD72—Unverified
5RoBERTaCaseHOLD71.7—Unverified
6BERTCaseHOLD70.7—Unverified
7BigBirdCaseHOLD70.4—Unverified
#ModelMetricClaimedVerifiedStatus
1ConvBERT-DGAverage74.6—Unverified
2ConvBERT-DG + Pre + MultiAverage73.8—Unverified
3mslmAverage73.49—Unverified
4ConvBERT + Pre + MultiAverage68.22—Unverified
5BanLanGenAverage39.16—Unverified
#ModelMetricClaimedVerifiedStatus
1ConvBERT + Pre + MultiAverage86.89—Unverified
2mslmAverage85.83—Unverified
3ConvBERT-DG + Pre + MultiAverage85.34—Unverified
#ModelMetricClaimedVerifiedStatus
1MT-DNN-SMARTAverage89.9—Unverified
2BERT-LARGEAverage82.1—Unverified