Split (3)

train · 937 rows

id stringclasses 179 values	question stringlengths 8.75k 85.9k	answer dict
2001.06354	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: Which method for integration peforms better ensemble or consensus dropout fusion with shared parameters? Context: <<<Title>>> Modality-Balanced Models for Visual Dialogue <<<Abstract>>> The...	{ "references": [ "ensemble model" ], "type": "extractive" }
1910.08210	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: How does propose model model that capture three-way interactions? Context: <<<Title>>> RTFM: Generalising to Novel Environment Dynamics via Reading <<<Abstract>>> Obtaining policies that ca...	{ "references": [ " We first encode text inputs using bidirectional LSTMs, then compute summaries using self-attention and conditional summaries using attention. We concatenate text summaries into text features, which, along with visual features, are processed through consecutive layers. In this case of a textual...
1908.08593	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: How much is performance improved by disabling attention in certain heads? Context: <<<Title>>> Revealing the Dark Secrets of BERT <<<Abstract>>> BERT-based architectures currently give stat...	{ "references": [ "disabling the first layer in the RTE task gives a significant boost, resulting in an absolute performance gain of 3.2%, this operation vary across tasks" ], "type": "extractive" }
1908.08593	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: In which certain heads was attention disabled in experiments? Context: <<<Title>>> Revealing the Dark Secrets of BERT <<<Abstract>>> BERT-based architectures currently give state-of-the-art...	{ "references": [ "single head,disabling a whole layer, that is, all 12 heads in a given layer" ], "type": "extractive" }
1908.08593	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: What handcrafter features-of-interest are used? Context: <<<Title>>> Revealing the Dark Secrets of BERT <<<Abstract>>> BERT-based architectures currently give state-of-the-art performance o...	{ "references": [ "nouns,verbs,pronouns,subjects,objects,negation words,special BERT tokens" ], "type": "extractive" }
1908.08593	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: What subset of GLUE tasks is used? Context: <<<Title>>> Revealing the Dark Secrets of BERT <<<Abstract>>> BERT-based architectures currently give state-of-the-art performance on many NLP ta...	{ "references": [ "MRPC,STS-B,SST-2,QQP,RTE,QNLI,MNLI" ], "type": "extractive" }
1911.02711	Please answer the following question with yes or no based on the given text. You only need to output 'Yes' or 'No' without any additional explanation. Question: Do they predict the sentiment of the review summary? Context: <<<Title>>> Exploring Hierarchical Interaction Between Review and Summary for Better Sentiment ...	{ "references": [ "No" ], "type": "boolean" }
1911.02711	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: Which review dataset do they use? Context: <<<Title>>> Exploring Hierarchical Interaction Between Review and Summary for Better Sentiment Analysis <<<Abstract>>> Sentiment analysis provides...	{ "references": [ "SNAP (Stanford Network Analysis Project)" ], "type": "extractive" }
1910.13890	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: What are the three languages studied in the paper? Context: <<<Title>>> A Latent Morphology Model for Open-Vocabulary Neural Machine Translation <<<Abstract>>> Translation into morphologica...	{ "references": [ "Arabic, Czech and Turkish" ], "type": "extractive" }
1909.01492	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: Which dataset do they use? Context: <<<Title>>> Achieving Verified Robustness to Symbol Substitutions via Interval Bound Propagation <<<Abstract>>> Neural networks are part of many contempo...	{ "references": [ "Stanford Sentiment Treebank (SST) BIBREF15 and AG News BIBREF16" ], "type": "extractive" }
1908.06006	Please answer the following question with yes or no based on the given text. You only need to output 'Yes' or 'No' without any additional explanation. Question: Do they compare to other models appart from HAN? Context: <<<Title>>> Bidirectional Context-Aware Hierarchical Attention Network for Document Understanding <...	{ "references": [ "No" ], "type": "boolean" }
1908.06006	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: What are the datasets used Context: <<<Title>>> Bidirectional Context-Aware Hierarchical Attention Network for Document Understanding <<<Abstract>>> The Hierarchical Attention Network (HAN)...	{ "references": [ "large-scale document classification datasets introduced by BIBREF14" ], "type": "extractive" }
1909.02776	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: What dataset is used for this task? Context: <<<Title>>> Features in Extractive Supervised Single-document Summarization: Case of Persian News <<<Abstract>>> Text summarization has been one...	{ "references": [ "the Pasokh dataset BIBREF42 " ], "type": "extractive" }
1909.02776	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: What features of the document are integrated into vectors of every sentence? Context: <<<Title>>> Features in Extractive Supervised Single-document Summarization: Case of Persian News <<<Ab...	{ "references": [ "Ordinal position,Length of sentence,The Ratio of Nouns,The Ratio of Numerical entities,Cue Words,Cosine position,Relative Length,TF-ISF,POS features,Document sentences,Document words,Topical category,Ratio of Verbs, Ratio of Adjectives, and Ratio of Adverbs" ], "type": "extractive" }
1909.02776	Please answer the following question with yes or no based on the given text. You only need to output 'Yes' or 'No' without any additional explanation. Question: Is new approach tested against state of the art? Context: <<<Title>>> Features in Extractive Supervised Single-document Summarization: Case of Persian News <...	{ "references": [ "No" ], "type": "boolean" }
1909.09018	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: What are all machine learning approaches compared in this work? Context: <<<Title>>> Corporate IT-Support Help-Desk Process Hybrid-Automation Solution with Machine Learning Approach <<<Abst...	{ "references": [ "Feature selection,Random forest,XGBoost,Hierarchical Model" ], "type": "extractive" }
1911.03154	Please answer the following question with yes or no based on the given text. You only need to output 'Yes' or 'No' without any additional explanation. Question: Has there been previous work on SNMT? Context: <<<Title>>> How to Do Simultaneous Translation Better with Consecutive Neural Machine Translation? <<<Abstract...	{ "references": [ "Yes" ], "type": "boolean" }
1911.03154	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: Which languages do they experiment on? Context: <<<Title>>> How to Do Simultaneous Translation Better with Consecutive Neural Machine Translation? <<<Abstract>>> Despite the success of neur...	{ "references": [ "German,English,Chinese" ], "type": "extractive" }
1911.03154	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: What corpora is used? Context: <<<Title>>> How to Do Simultaneous Translation Better with Consecutive Neural Machine Translation? <<<Abstract>>> Despite the success of neural machine transl...	{ "references": [ "IWSLT16,WMT15,NIST" ], "type": "extractive" }
1909.05360	Please answer the following question with yes or no based on the given text. You only need to output 'Yes' or 'No' without any additional explanation. Question: Is this the first paper to propose a joint model for event and temporal relation extraction? Context: <<<Title>>> Joint Event and Temporal Relation Extractio...	{ "references": [ "Yes" ], "type": "boolean" }
1909.05360	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: What datasets were used for this work? Context: <<<Title>>> Joint Event and Temporal Relation Extraction with Shared Representations and Structured Prediction <<<Abstract>>> We propose a jo...	{ "references": [ "TB-Dense, MATRES" ], "type": "extractive" }
2003.12738	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: What baselines other than standard transformers are used in experiments? Context: <<<Title>>> Variational Transformers for Diverse Response Generation <<<Abstract>>> Despite the great promi...	{ "references": [ "attention-based sequence-to-sequence model ,CVAE" ], "type": "extractive" }
2003.12738	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: What three conversational datasets are used for evaluation? Context: <<<Title>>> Variational Transformers for Diverse Response Generation <<<Abstract>>> Despite the great promise of Transfo...	{ "references": [ "MojiTalk ,PersonaChat ,Empathetic-Dialogues" ], "type": "extractive" }
1909.03544	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: What previous approaches did this method outperform? Context: <<<Title>>> Czech Text Processing with Contextual Embeddings: POS Tagging, Lemmatization, Parsing and NER <<<Abstract>>> Contex...	{ "references": [ "Table TABREF44,Table TABREF44,Table TABREF47,Table TABREF47" ], "type": "extractive" }
1909.03544	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: What data is used to build the embeddings? Context: <<<Title>>> Czech Text Processing with Contextual Embeddings: POS Tagging, Lemmatization, Parsing and NER <<<Abstract>>> Contextualized e...	{ "references": [ "large raw Czech corpora available from the LINDAT/CLARIN repository,Czech Wikipedia" ], "type": "extractive" }
1909.12642	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: What is the performance of the model for the German sub-task A? Context: <<<Title>>> HateMonitors: Language Agnostic Abuse Detection in Social Media <<<Abstract>>> Reducing hateful and offe...	{ "references": [ "macro F1 score of 0.62" ], "type": "extractive" }
1909.12642	Please answer the following question with yes or no based on the given text. You only need to output 'Yes' or 'No' without any additional explanation. Question: Is the model tested for language identification? Context: <<<Title>>> HateMonitors: Language Agnostic Abuse Detection in Social Media <<<Abstract>>> Reducing...	{ "references": [ "No" ], "type": "boolean" }
1909.12642	Please answer the following question with yes or no based on the given text. You only need to output 'Yes' or 'No' without any additional explanation. Question: Is the model compared to a baseline model? Context: <<<Title>>> HateMonitors: Language Agnostic Abuse Detection in Social Media <<<Abstract>>> Reducing hatef...	{ "references": [ "No" ], "type": "boolean" }
2003.00639	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: How does framework automatically chooses different curricula at the evolving learning process according to the learning status of the neural dialogue generation model? Context: <<<Title>>> ...	{ "references": [ "The multi-curricula learning scheme is scheduled according to the model's performance on the validation set, where the scheduling mechanism acts as the policy $\\pi $ interacting with the dialogue model to acquire the learning status $s$. The reward of the multi-curricula learning mechanism $m_...
2003.00639	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: What human judgement metrics are used? Context: <<<Title>>> Learning from Easy to Complex: Adaptive Multi-curricula Learning for Neural Dialogue Generation <<<Abstract>>> Current state-of-t...	{ "references": [ "coherence, logical consistency, fluency and diversity" ], "type": "extractive" }
2003.00639	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: What automatic evaluation metrics are used? Context: <<<Title>>> Learning from Easy to Complex: Adaptive Multi-curricula Learning for Neural Dialogue Generation <<<Abstract>>> Current state...	{ "references": [ "BLEU,embedding-based metrics (Average, Extrema, Greedy and Coherence),, entropy-based metrics (Ent-{1,2}),distinct metrics (Dist-{1,2,3} and Intra-{1,2,3})" ], "type": "extractive" }
2003.00639	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: What state of the art models were used in experiments? Context: <<<Title>>> Learning from Easy to Complex: Adaptive Multi-curricula Learning for Neural Dialogue Generation <<<Abstract>>> Cu...	{ "references": [ "SEQ2SEQ,CVAE,Transformer,HRED,DialogWAE" ], "type": "extractive" }
2003.00639	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: What five dialogue attributes were analyzed? Context: <<<Title>>> Learning from Easy to Complex: Adaptive Multi-curricula Learning for Neural Dialogue Generation <<<Abstract>>> Current stat...	{ "references": [ "Model Confidence,Continuity,Query-relatedness,Repetitiveness,Specificity" ], "type": "extractive" }
2003.00639	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: What three publicly available coropora are used? Context: <<<Title>>> Learning from Easy to Complex: Adaptive Multi-curricula Learning for Neural Dialogue Generation <<<Abstract>>> Current ...	{ "references": [ "PersonaChat BIBREF12,DailyDialog BIBREF13,OpenSubtitles BIBREF7" ], "type": "extractive" }
1909.13668	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: What different properties of the posterior distribution are explored in the paper? Context: <<<Title>>> On the Importance of the Kullback-Leibler Divergence Term in Variational Autoencoders...	{ "references": [ "interdependence between rate and distortion,impact of KL on the sharpness of the approximated posteriors,demonstrate how certain generative behaviours could be imposed on VAEs via a range of maximum channel capacities,some experiments to find if any form of syntactic information is encoded in t...
1909.13668	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: Why does proposed term help to avoid posterior collapse? Context: <<<Title>>> On the Importance of the Kullback-Leibler Divergence Term in Variational Autoencoders for Text Generation <<<Ab...	{ "references": [ "by setting a non-zero positive constraint ($C\\ge 0$) on the KL term ($\|D_{KL}\\big (q_\\phi ({z}\|{x}) \|\| p({z})\\big )-C\|$)" ], "type": "extractive" }
2003.01472	Please answer the following question with yes or no based on the given text. You only need to output 'Yes' or 'No' without any additional explanation. Question: Did they experiment with the tool? Context: <<<Title>>> Seshat: A tool for managing and verifying annotation campaigns of audio data <<<Abstract>>> We introd...	{ "references": [ "Yes" ], "type": "boolean" }
2003.01472	Please answer the following question with yes or no based on the given text. You only need to output 'Yes' or 'No' without any additional explanation. Question: Is this software available to the public? Context: <<<Title>>> Seshat: A tool for managing and verifying annotation campaigns of audio data <<<Abstract>>> We...	{ "references": [ "Yes" ], "type": "boolean" }
2004.01980	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: Which state-of-the-art model is surpassed by 9.68% attraction score? Context: <<<Title>>> Hooks in the Headline: Learning to Generate Headlines with Controlled Styles <<<Abstract>>> Current...	{ "references": [ "pure summarization model NHG" ], "type": "extractive" }
2004.01980	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: How is attraction score measured? Context: <<<Title>>> Hooks in the Headline: Learning to Generate Headlines with Controlled Styles <<<Abstract>>> Current summarization systems only produce...	{ "references": [ "annotators are asked how attractive the headlines are,Likert scale from 1 to 10 (integer values)" ], "type": "extractive" }
2004.01980	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: How is presence of three target styles detected? Context: <<<Title>>> Hooks in the Headline: Learning to Generate Headlines with Controlled Styles <<<Abstract>>> Current summarization syste...	{ "references": [ "human evaluation task about the style strength" ], "type": "extractive" }
2004.01980	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: How is fluency automatically evaluated? Context: <<<Title>>> Hooks in the Headline: Learning to Generate Headlines with Controlled Styles <<<Abstract>>> Current summarization systems only p...	{ "references": [ "fine-tuned the GPT-2 medium model BIBREF51 on our collected headlines and then used it to measure the perplexity (PPL) on the generated outputs" ], "type": "extractive" }
1911.03597	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: What multilingual parallel data is used for training proposed model? Context: <<<Title>>> Zero-Shot Paraphrase Generation with Multilingual Language Models <<<Abstract>>> Leveraging multili...	{ "references": [ "MultiUN BIBREF20,OpenSubtitles BIBREF21" ], "type": "extractive" }
1911.03597	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: How much better are results of proposed model compared to pivoting method? Context: <<<Title>>> Zero-Shot Paraphrase Generation with Multilingual Language Models <<<Abstract>>> Leveraging m...	{ "references": [ "our method outperforms the baseline in both relevance and fluency significantly." ], "type": "extractive" }
2003.08132	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: What representations are presented by this paper? Context: <<<Title>>> Gender Representation in Open Source Speech Resources <<<Abstract>>> With the rise of artificial intelligence (AI) and...	{ "references": [ "the number of speakers of each gender category,their speech duration" ], "type": "extractive" }
2001.02380	Please answer the following question with yes or no based on the given text. You only need to output 'Yes' or 'No' without any additional explanation. Question: Are some models evaluated using this metric, what are the findings? Context: <<<Title>>> A Neural Approach to Discourse Relation Signal Detection <<<Abstract...	{ "references": [ "Yes" ], "type": "boolean" }
2001.02380	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: Where does proposed metric differ from juman judgement? Context: <<<Title>>> A Neural Approach to Discourse Relation Signal Detection <<<Abstract>>> Previous data-driven work investigating ...	{ "references": [ "model points out plausible signals which were passed over by an annotator,it also picks up on a recurring tendency in how-to guides in which the second person pronoun referring to the reader is often the benefactee of some action" ], "type": "extractive" }
2001.02380	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: Where does proposed metric overlap with juman judgement? Context: <<<Title>>> A Neural Approach to Discourse Relation Signal Detection <<<Abstract>>> Previous data-driven work investigating...	{ "references": [ "influence of each word on the score of the correct relation, that impact should and does still correlate with human judgments" ], "type": "extractive" }
2002.00317	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: Which baseline performs best? Context: <<<Title>>> Citation Text Generation <<<Abstract>>> We introduce the task of citation text generation: given a pair of scientific documents, explain t...	{ "references": [ "IR methods perform better than the best neural models" ], "type": "extractive" }
2002.00317	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: Which baselines are explored? Context: <<<Title>>> Citation Text Generation <<<Abstract>>> We introduce the task of citation text generation: given a pair of scientific documents, explain t...	{ "references": [ "GPT2,SciBERT model of BIBREF11" ], "type": "extractive" }
2002.00317	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: What is the size of the corpus? Context: <<<Title>>> Citation Text Generation <<<Abstract>>> We introduce the task of citation text generation: given a pair of scientific documents, explain...	{ "references": [ "8.1 million scientific documents,154K computer science articles,622K citing sentences" ], "type": "extractive" }
2004.04228	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: What models are evaluated with QAGS? Context: <<<Title>>> Asking and Answering Questions to Evaluate the Factual Consistency of Summaries <<<Abstract>>> Practical applications of abstractiv...	{ "references": [ "bert-large-wwm,bert-base,bert-large" ], "type": "extractive" }
2004.04228	Please answer the following question with yes or no based on the given text. You only need to output 'Yes' or 'No' without any additional explanation. Question: Do they use crowdsourcing to collect human judgements? Context: <<<Title>>> Asking and Answering Questions to Evaluate the Factual Consistency of Summaries <...	{ "references": [ "Yes" ], "type": "boolean" }
1909.00161	Please answer the following question with yes or no based on the given text. You only need to output 'Yes' or 'No' without any additional explanation. Question: Do they use pretrained models? Context: <<<Title>>> Benchmarking Zero-shot Text Classification: Datasets, Evaluation and Entailment Approach <<<Abstract>>> Z...	{ "references": [ "Yes" ], "type": "boolean" }
1909.00161	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: What are their baseline models? Context: <<<Title>>> Benchmarking Zero-shot Text Classification: Datasets, Evaluation and Entailment Approach <<<Abstract>>> Zero-shot text classification (0...	{ "references": [ "Majority,ESA,Word2Vec ,Binary-BERT" ], "type": "extractive" }
1909.08167	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: How are different domains weighted in WDIRL? Context: <<<Title>>> Weighed Domain-Invariant Representation Learning for Cross-domain Sentiment Analysis <<<Abstract>>> Cross-domain sentiment ...	{ "references": [ "To achieve this purpose, we introduce a trainable class weight $\\mathbf {w}$ to reweigh source domain examples by class when performing DIRL, with $\\mathbf {w}_i > 0$" ], "type": "extractive" }
1909.08167	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: How is DIRL evaluated? Context: <<<Title>>> Weighed Domain-Invariant Representation Learning for Cross-domain Sentiment Analysis <<<Abstract>>> Cross-domain sentiment analysis is currently ...	{ "references": [ "Through the experiments, we empirically studied our analysis on DIRL and the effectiveness of our proposed solution in dealing with the problem it suffered from." ], "type": "extractive" }
1909.04181	Please answer the following question with yes or no based on the given text. You only need to output 'Yes' or 'No' without any additional explanation. Question: Does the paper report F1-scores for the age and language variety tasks? Context: <<<Title>>> BERT-Based Arabic Social Media Author Profiling <<<Abstract>>> W...	{ "references": [ "No" ], "type": "boolean" }
1909.04181	Please answer the following question with yes or no based on the given text. You only need to output 'Yes' or 'No' without any additional explanation. Question: Are the models compared to some baseline models? Context: <<<Title>>> BERT-Based Arabic Social Media Author Profiling <<<Abstract>>> We report our models for...	{ "references": [ "Yes" ], "type": "boolean" }
1909.04181	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: What are the in-house data employed? Context: <<<Title>>> BERT-Based Arabic Social Media Author Profiling <<<Abstract>>> We report our models for detecting age, language variety, and gender...	{ "references": [ "we manually label an in-house dataset of 1,100 users with gender tags,we randomly sample 20,000 tweets for each class from an in-house dataset gold labeled with the same 15 classes as the shared task" ], "type": "extractive" }
1911.06171	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: Which future direction in NLG are discussed? Context: <<<Title>>> Unsupervised Pre-training for Natural Language Generation: A Literature Review <<<Abstract>>> Recently, unsupervised pre-tr...	{ "references": [ "1) How to introduce unsupervised pre-training into NLG tasks with cross-modal context?,2) How to design a generic pre-training algorithm to fit a wide range of NLG tasks?,3) How to reduce the computing resources required for large-scale pre-training?,4) What aspect of knowledge do the pre-train...
1911.06171	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: What experimental phenomena are presented? Context: <<<Title>>> Unsupervised Pre-training for Natural Language Generation: A Literature Review <<<Abstract>>> Recently, unsupervised pre-trai...	{ "references": [ "The advantage of pre-training gradually diminishes with the increase of labeled data,Fixed representations yield better results than fine-tuning in some cases,pre-training the Seq2Seq encoder outperforms pre-training the decoder" ], "type": "extractive" }
1911.06171	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: How strategy-based methods handle obstacles in NLG? Context: <<<Title>>> Unsupervised Pre-training for Natural Language Generation: A Literature Review <<<Abstract>>> Recently, unsupervised...	{ "references": [ "fine-tuning schedules that elaborately design the control of learning rates for optimization,proxy tasks that leverage labeled data to help the pre-trained model better fit the target data distribution,knowledge distillation approaches that ditch the paradigm of initialization with pre-trained ...
1911.06171	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: How architecture-based method handle obstacles in NLG? Context: <<<Title>>> Unsupervised Pre-training for Natural Language Generation: A Literature Review <<<Abstract>>> Recently, unsupervi...	{ "references": [ "task-specific architecture during pre-training (task-specific methods),aim at building a general pre-training architecture to fit all downstream tasks (task-agnostic methods)" ], "type": "extractive" }
2002.06053	Please answer the following question with yes or no based on the given text. You only need to output 'Yes' or 'No' without any additional explanation. Question: Are datasets publicly available? Context: <<<Title>>> Exploring Chemical Space using Natural Language Processing Methodologies for Drug Discovery <<<Abstract...	{ "references": [ "Yes" ], "type": "boolean" }
2002.06053	Please answer the following question with yes or no based on the given text. You only need to output 'Yes' or 'No' without any additional explanation. Question: Is there any concrete example in the paper that shows that this approach had huge impact on drug discovery? Context: <<<Title>>> Exploring Chemical Space usi...	{ "references": [ "Yes" ], "type": "boolean" }
1912.07976	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: How much better is performance of the proposed model compared to the state of the art in these various experiments? Context: <<<Title>>> A Multi-task Learning Model for Chinese-oriented Asp...	{ "references": [ "significantly improves the accuracy and F1 score of aspect polarity classification" ], "type": "extractive" }
1912.07976	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: What was state of the art on SemEval-2014 task4 Restaurant and Laptop dataset? Context: <<<Title>>> A Multi-task Learning Model for Chinese-oriented Aspect Polarity Classification and Aspec...	{ "references": [ "BERT-ADA,BERT-PT, AEN-BERT, SDGCN-BERT" ], "type": "extractive" }
1912.07976	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: What was previous state-of-the-art on four Chinese reviews datasets? Context: <<<Title>>> A Multi-task Learning Model for Chinese-oriented Aspect Polarity Classification and Aspect Term Ext...	{ "references": [ "GANN obtained the state-of-the-art APC performance on the Chinese review datasets" ], "type": "extractive" }
1912.07976	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: In what four Chinese review datasets does LCF-ATEPC achieves state of the art? Context: <<<Title>>> A Multi-task Learning Model for Chinese-oriented Aspect Polarity Classification and Aspec...	{ "references": [ "Car, Phone, Notebook, Camera" ], "type": "extractive" }
1909.09268	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: What is the criteria for a good metric? Context: <<<Title>>> Towards Neural Language Evaluators <<<Abstract>>> We review three limitations of BLEU and ROUGE -- the most popular metrics used...	{ "references": [ "The first one is that it should be highly correlated with human judgement of similarity. The second one is that it should be able to distinguish sentences which are in logical contradiction, logically unrelated or in logical agreement. The third one is that a robust evaluator should also be abl...
1909.09268	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: What are the three limitations? Context: <<<Title>>> Towards Neural Language Evaluators <<<Abstract>>> We review three limitations of BLEU and ROUGE -- the most popular metrics used to asse...	{ "references": [ "High scores to semantically opposite translations/summaries, Low scores to semantically related translations/summaries and High scores to unintelligible translations/summaries." ], "type": "extractive" }
1910.00194	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: Which language(s) are found in the WSD datasets? Context: <<<Title>>> Improved Word Sense Disambiguation Using Pre-Trained Contextualized Word Representations <<<Abstract>>> Contextualized ...	{ "references": [ " WSD is predominantly evaluated on English, we are also interested in evaluating our approach on Chinese" ], "type": "extractive" }
1910.00194	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: What datasets are used for testing? Context: <<<Title>>> Improved Word Sense Disambiguation Using Pre-Trained Contextualized Word Representations <<<Abstract>>> Contextualized word represen...	{ "references": [ "Senseval-2 (SE2), Senseval-3 (SE3), SemEval 2013 task 12 (SE13), and SemEval 2015 task 13 (SE15),OntoNotes Release 5.0" ], "type": "extractive" }
1908.11860	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: By how much does their model outperform the baseline in the cross-domain evaluation? Context: <<<Title>>> Adapt or Get Left Behind: Domain Adaptation through BERT Language Model Finetuning ...	{ "references": [ "$2.2\\%$ absolute accuracy improvement on the laptops test set,$3.6\\%$ accuracy improvement on the restaurants test set" ], "type": "extractive" }
1908.11860	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: What are the performance results? Context: <<<Title>>> Adapt or Get Left Behind: Domain Adaptation through BERT Language Model Finetuning for Aspect-Target Sentiment Classification <<<Abstr...	{ "references": [ "results that for the in-domain training case, our models BERT-ADA Lapt and BERT-ADA Rest achieve performance close to state-of-the-art on the laptops dataset,new state-of-the-art on the restaurants dataset with accuracies of $79.19\\%$ and $87.14\\%$, respectively." ], "type": "extractive" ...
2002.09758	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: What off-the-shelf QA model was used to answer sub-questions? Context: <<<Title>>> Unsupervised Question Decomposition for Question Answering <<<Abstract>>> We aim to improve question answe...	{ "references": [ "$\\textsc {BERT}_{\\textsc {BASE}}$ ensemble from BIBREF3" ], "type": "extractive" }
2002.09758	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: How large is the improvement over the baseline? Context: <<<Title>>> Unsupervised Question Decomposition for Question Answering <<<Abstract>>> We aim to improve question answering (QA) by d...	{ "references": [ "3.1 F1 gain on the original dev set,11 F1 gain on the multi-hop dev set,10 F1 gain on the out-of-domain dev set." ], "type": "extractive" }
2002.09758	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: What is the strong baseline that this work outperforms? Context: <<<Title>>> Unsupervised Question Decomposition for Question Answering <<<Abstract>>> We aim to improve question answering (...	{ "references": [ "RoBERTa baseline" ], "type": "extractive" }
1912.08320	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: What are the key issues around whether the gold standard data produced in such an annotation is reliable? Context: <<<Title>>> Garbage In, Garbage Out? Do Machine Learning Application Pape...	{ "references": [ " only 1 in 9 qualitative papers in Human-Computer Interaction reported inter-rater reliability metrics,low-effort responses from crowdworkers" ], "type": "extractive" }
1912.08320	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: How were the machine learning papers from ArXiv sampled? Context: <<<Title>>> Garbage In, Garbage Out? Do Machine Learning Application Papers in Social Computing Report Where Human-Labeled ...	{ "references": [ "sampled all papers published in the Computer Science subcategories of Artificial Intelligence (cs.AI), Machine Learning (cs.LG), Social and Information Networks (cs.SI), Computational Linguistics (cs.CL), Computers and Society (cs.CY), Information Retrieval (cs.IR), and Computer Vision (CS.CV),...
1912.08320	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: What are the core best practices of structured content analysis? Context: <<<Title>>> Garbage In, Garbage Out? Do Machine Learning Application Papers in Social Computing Report Where Human-...	{ "references": [ "“coding scheme” is defined,coders are trained with the coding scheme,Training sometimes results in changes to the coding scheme,calculation of “inter-annotator agreement” or “inter-rater reliability.”,there is a process of “reconciliation” for disagreements" ], "type": "extractive" }
1912.08320	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: In what sense is data annotation similar to structured content analysis? Context: <<<Title>>> Garbage In, Garbage Out? Do Machine Learning Application Papers in Social Computing Report Whe...	{ "references": [ "structured content analysis (also called “closed coding”) is used to turn qualitative or unstructured data of all kinds into structured and/or quantitative data,Projects usually involve teams of “coders” (also called “annotators”, “labelers”, or “reviewers”), with human labor required to “code”...
2002.05058	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: How they add human prefference annotation to fine-tuning process? Context: <<<Title>>> Learning to Compare for Better Training and Evaluation of Open Domain Natural Language Generation Mode...	{ "references": [ "human preference annotation is available,$Q(x_1, x_2) \\in \\lbrace >,<,\\approx \\rbrace $ is the true label for the pair" ], "type": "extractive" }
2002.05058	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: What previous automated evalution approaches authors mention? Context: <<<Title>>> Learning to Compare for Better Training and Evaluation of Open Domain Natural Language Generation Models <...	{ "references": [ "Text Overlap Metrics, including BLEU,Perplexity,Parameterized Metrics" ], "type": "extractive" }
2002.05058	Please answer the following question with yes or no based on the given text. You only need to output 'Yes' or 'No' without any additional explanation. Question: Do the authors suggest that proposed metric replace human evaluation on this task? Context: <<<Title>>> Learning to Compare for Better Training and Evaluatio...	{ "references": [ "No" ], "type": "boolean" }
2002.06675	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: How big are improvements with multilingual ASR training vs single language training? Context: <<<Title>>> Speech Corpus of Ainu Folklore and End-to-end Speech Recognition for Ainu Language ...	{ "references": [ "relative WER improvement of 10%." ], "type": "extractive" }
2002.06675	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: What is the difference between speaker-open and speaker-closed setting? Context: <<<Title>>> Speech Corpus of Ainu Folklore and End-to-end Speech Recognition for Ainu Language <<<Abstract>>...	{ "references": [ "In the speaker-closed condition, two episodes were set aside from each speaker as development and test sets.,In the speaker-open condition, all the data except for the test speaker's were used for training" ], "type": "extractive" }
1909.08041	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: How do they train the retrieval modules? Context: <<<Title>>> Revealing the Importance of Semantic Retrieval for Machine Reading at Scale <<<Abstract>>> Machine Reading at Scale (MRS) is a ...	{ "references": [ "We treated the neural semantic retrieval at both the paragraph and sentence level as binary classification problems with models' parameters updated by minimizing binary cross entropy loss." ], "type": "extractive" }
1909.08041	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: How do they model the neural retrieval modules? Context: <<<Title>>> Revealing the Importance of Semantic Retrieval for Machine Reading at Scale <<<Abstract>>> Machine Reading at Scale (MRS...	{ "references": [ "BERT-Base BIBREF2 to provide the state-of-the-art contextualized modeling" ], "type": "extractive" }
1909.08041	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: Retrieval at what level performs better, sentence level or paragraph level? Context: <<<Title>>> Revealing the Importance of Semantic Retrieval for Machine Reading at Scale <<<Abstract>>> M...	{ "references": [ "This seems to indicate that the downstream QA module relies more on the upstream paragraph-level retrieval whereas the verification module relies more on the upstream sentence-level retrieval." ], "type": "extractive" }
1909.09270	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: Which languages are evaluated? Context: <<<Title>>> Named Entity Recognition with Partially Annotated Training Data <<<Abstract>>> Supervised machine learning assumes the availability of fu...	{ "references": [ "Bengali,English, German, Spanish, Dutch,Amharic,Arabic,Hindi,Somali " ], "type": "extractive" }
2003.09586	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: How much is decoding speed increased by increasing encoder and decreasing decoder depth? Context: <<<Title>>> Analyzing Word Translation of Transformer Layers <<<Abstract>>> The Transformer...	{ "references": [ "the Transformer with 10 encoder layers and 2 decoder layers is $2.32$ times as fast as the 6-layer Transformer" ], "type": "extractive" }
1911.03270	Please answer the following question with yes or no based on the given text. You only need to output 'Yes' or 'No' without any additional explanation. Question: Does the paper report the performance on the task of a Neural Machine Translation model? Context: <<<Title>>> Char-RNN and Active Learning for Hashtag Segmen...	{ "references": [ "No" ], "type": "boolean" }
1911.03270	Please answer the following question with yes or no based on the given text. You only need to output 'Yes' or 'No' without any additional explanation. Question: Is the RNN model evaluated against any baseline? Context: <<<Title>>> Char-RNN and Active Learning for Hashtag Segmentation <<<Abstract>>> We explore the abi...	{ "references": [ "Yes" ], "type": "boolean" }
1911.03270	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: Which languages are used in the paper? Context: <<<Title>>> Char-RNN and Active Learning for Hashtag Segmentation <<<Abstract>>> We explore the abilities of character recurrent neural netwo...	{ "references": [ "English,Russian" ], "type": "extractive" }
2004.03762	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: What metrics are used for evaluation? Context: <<<Title>>> Generating Narrative Text in a Switching Dynamical System <<<Abstract>>> Early work on narrative modeling used explicit plans and ...	{ "references": [ "ROUGE BIBREF29 and METEOR BIBREF30" ], "type": "extractive" }
2004.03762	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: What baselines are used? Context: <<<Title>>> Generating Narrative Text in a Switching Dynamical System <<<Abstract>>> Early work on narrative modeling used explicit plans and goals to gene...	{ "references": [ "a two layer recurrent neural language model with GRU cells of hidden size 512,a two layer neural sequence to sequence model equipped with bi-linear attention function with GRU cells of hidden size 512,a linear dynamical system,semi-supervised SLDS models with varying amount of labelled sentimen...
1909.07593	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: Which model is used to capture the implicit structure? Context: <<<Title>>> Learning Explicit and Implicit Structures for Targeted Sentiment Analysis <<<Abstract>>> Targeted sentiment analy...	{ "references": [ "Bi-directional LSTM,self-attention " ], "type": "extractive" }
1909.07593	Please extract a concise answer without any additional explanation for the following question based on the given text. Question: How is the robustness of the model evaluated? Context: <<<Title>>> Learning Explicit and Implicit Structures for Targeted Sentiment Analysis <<<Abstract>>> Targeted sentiment analysis is th...	{ "references": [ "10-fold cross validation" ], "type": "extractive" }

Subsets and Splits

No community queries yet

The top public SQL queries from the community will appear here once available.