PyTorchKR Papers

최신 논문

총 1,245편
태그·기간 등 상세 필터

카테고리 탐색

로그인 후 이 태그 구독

arXiv:2005.14709

Beyond Leaderboards: A survey of methods for revealing weaknesses in Natural Language Inference data and models

Recent years have seen a growing number of publications that analyse Natural Language Inference (NLI) datasets for superficial cues, whether they undermine the complexity of the tasks underlying those datasets and how they impact those models that are optimised and evaluated on this data. This structured survey provides an overview of the evolving research area by categorising reported weaknesses in models and datasets and the methods proposed to reveal and alleviate those weaknesses for the English language. We summarise and discuss the findings and conclude with a set of recommendations for possible future research directions. We hope it will be a useful resource for researchers who propose new datasets, to have a set of tools to assess the suitability and quality of their data to evaluate various phenomena of interest, as well as those who develop novel architectures, to further understand the implications of their improvements with respect to their model's acquired capabilities.

arXiv:2005.14165공식 코드 · Papers with Code

Language Models are Few-Shot Learners

Recent work has demonstrated substantial gains on many NLP tasks and benchmarks by pre-training on a large corpus of text followed by fine-tuning on a specific task. While typically task-agnostic in architecture, this method still requires task-specific fine-tuning datasets of thousands or tens of thousands of examples. By contrast, humans can generally perform a new language task from only a few examples or from simple instructions - something which current NLP systems still largely struggle to do. Here we show that scaling up language models greatly improves task-agnostic, few-shot performance, sometimes even reaching competitiveness with prior state-of-the-art fine-tuning approaches. Specifically, we train GPT-3, an autoregressive language model with 175 billion parameters, 10x more than any previous non-sparse language model, and test its performance in the few-shot setting. For all tasks, GPT-3 is applied without any gradient updates or fine-tuning, with tasks and few-shot demonstrations specified purely via text interaction with the model. GPT-3 achieves strong performance on many NLP datasets, including translation, question-answering, and cloze tasks, as well as several tasks that require on-the-fly reasoning or domain adaptation, such as unscrambling words, using a novel word in a sentence, or performing 3-digit arithmetic. At the same time, we also identify some datasets where GPT-3's few-shot learning still struggles, as well as some datasets where GPT-3 faces methodological issues related to training on large web corpora. Finally, we find that GPT-3 can generate samples of news articles which human evaluators have difficulty distinguishing from articles written by humans. We discuss broader societal impacts of this finding and of GPT-3 in general.

arXiv:2005.12116공식 코드 · Papers with Code

NILE : Natural Language Inference with Faithful Natural Language Explanations

The recent growth in the popularity and success of deep learning models on NLP classification tasks has accompanied the need for generating some form of natural language explanation of the predicted labels. Such generated natural language (NL) explanations are expected to be faithful, i.e., they should correlate well with the model's internal decision making. In this work, we focus on the task of natural language inference (NLI) and address the following question: can we build NLI systems which produce labels with high accuracy, while also generating faithful explanations of its decisions? We propose Natural-language Inference over Label-specific Explanations (NILE), a novel NLI method which utilizes auto-generated label-specific NL explanations to produce labels along with its faithful explanation. We demonstrate NILE's effectiveness over previously reported methods through automated and human evaluation of the produced labels and explanations. Our evaluation of NILE also supports the claim that accurate systems capable of providing testable explanations of their decisions can be designed. We discuss the faithfulness of NILE's explanations in terms of sensitivity of the decisions to the corresponding explanations. We argue that explicit evaluation of faithfulness, in addition to label and explanation accuracy, is an important step in evaluating model's explanations. Further, we demonstrate that task-specific probes are necessary to establish such sensitivity.

arXiv:2005.11223공식 코드 · Papers with Code

L2R2: Leveraging Ranking for Abductive Reasoning

The abductive natural language inference task ($\alpha$NLI) is proposed to evaluate the abductive reasoning ability of a learning system. In the $\alpha$NLI task, two observations are given and the most plausible hypothesis is asked to pick out from the candidates. Existing methods simply formulate it as a classification problem, thus a cross-entropy log-loss objective is used during training. However, discriminating true from false does not measure the plausibility of a hypothesis, for all the hypotheses have a chance to happen, only the probabilities are different. To fill this gap, we switch to a ranking perspective that sorts the hypotheses in order of their plausibilities. With this new perspective, a novel $L2R^2$ approach is proposed under the learning-to-rank framework. Firstly, training samples are reorganized into a ranking form, where two observations and their hypotheses are treated as the query and a set of candidate documents respectively. Then, an ESIM model or pre-trained language model, e.g. BERT or RoBERTa, is obtained as the scoring function. Finally, the loss functions for the ranking task can be either pair-wise or list-wise for training. The experimental results on the ART dataset reach the state-of-the-art in the public leaderboard.

arXiv:2005.10410공식 코드 · Papers with Code

Stance Prediction and Claim Verification: An Arabic Perspective

This work explores the application of textual entailment in news claim verification and stance prediction using a new corpus in Arabic. The publicly available corpus comes in two perspectives: a version consisting of 4,547 true and false claims and a version consisting of 3,786 pairs (claim, evidence). We describe the methodology for creating the corpus and the annotation process. Using the introduced corpus, we also develop two machine learning baselines for two proposed tasks: claim verification and stance prediction. Our best model utilizes pretraining (BERT) and achieves 76.7 F1 on the stance prediction task and 64.3 F1 on the claim verification task. Our preliminary experiments shed some light on the limits of automatic claim verification that relies on claims text only. Results hint that while the linguistic features and world knowledge learned during pretraining are useful for stance prediction, such learned representations from pretraining are insufficient for verifying claims without access to context or evidence.

arXiv:2005.10389공식 코드 · Papers with Code

Pretraining with Contrastive Sentence Objectives Improves Discourse Performance of Language Models

Recent models for unsupervised representation learning of text have employed a number of techniques to improve contextual word representations but have put little focus on discourse-level representations. We propose CONPONO, an inter-sentence objective for pretraining language models that models discourse coherence and the distance between sentences. Given an anchor sentence, our model is trained to predict the text k sentences away using a sampled-softmax objective where the candidates consist of neighboring sentences and sentences randomly sampled from the corpus. On the discourse representation benchmark DiscoEval, our model improves over the previous state-of-the-art by up to 13% and on average 4% absolute across 7 tasks. Our model is the same size as BERT-Base, but outperforms the much larger BERT- Large model and other more recent approaches that incorporate discourse. We also show that CONPONO yields gains of 2%-6% absolute even for tasks that do not explicitly evaluate discourse: textual entailment (RTE), common sense reasoning (COPA) and reading comprehension (ReCoRD).

arXiv:2005.07954공식 코드 · Papers with Code

Logical Inferences with Comparatives and Generalized Quantifiers

Comparative constructions pose a challenge in Natural Language Inference (NLI), which is the task of determining whether a text entails a hypothesis. Comparatives are structurally complex in that they interact with other linguistic phenomena such as quantifiers, numerals, and lexical antonyms. In formal semantics, there is a rich body of work on comparatives and gradable expressions using the notion of degree. However, a logical inference system for comparatives has not been sufficiently developed for use in the NLI task. In this paper, we present a compositional semantics that maps various comparative constructions in English to semantic representations via Combinatory Categorial Grammar (CCG) parsers and combine it with an inference system based on automated theorem proving. We evaluate our system on three NLI datasets that contain complex logical inferences with comparatives, generalized quantifiers, and numerals. We show that the system outperforms previous logic-based systems as well as recent deep learning-based models.

arXiv:2005.08516

Towards Question Format Independent Numerical Reasoning: A Set of Prerequisite Tasks

Numerical reasoning is often important to accurately understand the world. Recently, several format-specific datasets have been proposed, such as numerical reasoning in the settings of Natural Language Inference (NLI), Reading Comprehension (RC), and Question Answering (QA). Several format-specific models and architectures in response to those datasets have also been proposed. However, there exists a strong need for a benchmark which can evaluate the abilities of models, in performing question format independent numerical reasoning, as (i) the numerical reasoning capabilities we want to teach are not controlled by question formats, (ii) for numerical reasoning technology to have the best possible application, it must be able to process language and reason in a way that is not exclusive to a single format, task, dataset or domain. In pursuit of this goal, we introduce NUMBERGAME, a multifaceted benchmark to evaluate model performance across numerical reasoning tasks of eight diverse formats. We add four existing question types in our compilation. Two of the new types we add are about questions that require external numerical knowledge, commonsense knowledge and domain knowledge. For building a more practical numerical reasoning system, NUMBERGAME demands four capabilities beyond numerical reasoning: (i) detecting question format directly from data (ii) finding intermediate common format to which every format can be converted (iii) incorporating commonsense knowledge (iv) handling data imbalance across formats. We build several baselines, including a new model based on knowledge hunting using a cheatsheet. However, all baselines perform poorly in contrast to the human baselines, indicating the hardness of our benchmark. Our work takes forward the recent progress in generic system development, demonstrating the scope of these under-explored tasks.

arXiv:2005.06676공식 코드 · Papers with Code

Explaining Black Box Predictions and Unveiling Data Artifacts through Influence Functions

Modern deep learning models for NLP are notoriously opaque. This has motivated the development of methods for interpreting such models, e.g., via gradient-based saliency maps or the visualization of attention weights. Such approaches aim to provide explanations for a particular model prediction by highlighting important words in the corresponding input text. While this might be useful for tasks where decisions are explicitly influenced by individual tokens in the input, we suspect that such highlighting is not suitable for tasks where model decisions should be driven by more complex reasoning. In this work, we investigate the use of influence functions for NLP, providing an alternative approach to interpreting neural text classifiers. Influence functions explain the decisions of a model by identifying influential training examples. Despite the promise of this approach, influence functions have not yet been extensively evaluated in the context of NLP, a gap addressed by this work. We conduct a comparison between influence functions and common word-saliency methods on representative tasks. As suspected, we find that influence functions are particularly useful for natural language inference, a task in which 'saliency maps' may not have clear interpretation. Furthermore, we develop a new quantitative measure based on influence functions that can reveal artifacts in training data.

arXiv:2005.04315공식 코드 · Papers with Code

Probing Linguistic Systematicity

Recently, there has been much interest in the question of whether deep natural language understanding models exhibit systematicity; generalizing such that units like words make consistent contributions to the meaning of the sentences in which they appear. There is accumulating evidence that neural models often generalize non-systematically. We examined the notion of systematicity from a linguistic perspective, defining a set of probes and a set of metrics to measure systematic behaviour. We also identified ways in which network architectures can generalize non-systematically, and discuss why such forms of generalization may be unsatisfying. As a case study, we performed a series of experiments in the setting of natural language inference (NLI), demonstrating that some NLU systems achieve high overall performance despite being non-systematic.

arXiv:2005.04732공식 코드 · Papers with Code

Towards Robustifying NLI Models Against Lexical Dataset Biases

While deep learning models are making fast progress on the task of Natural Language Inference, recent studies have also shown that these models achieve high accuracy by exploiting several dataset biases, and without deep understanding of the language semantics. Using contradiction-word bias and word-overlapping bias as our two bias examples, this paper explores both data-level and model-level debiasing methods to robustify models against lexical dataset biases. First, we debias the dataset through data augmentation and enhancement, but show that the model bias cannot be fully removed via this method. Next, we also compare two ways of directly debiasing the model without knowing what the dataset biases are in advance. The first approach aims to remove the label bias at the embedding level. The second approach employs a bag-of-words sub-model to capture the features that are likely to exploit the bias and prevents the original model from learning these biased features by forcing orthogonality between these two sub-models. We performed evaluations on new balanced datasets extracted from the original MNLI dataset as well as the NLI stress tests, and show that the orthogonality approach is better at debiasing the model while maintaining competitive overall accuracy. Our code and data are available at: https://github.com/owenzx/LexicalDebias-ACL2020

arXiv:2005.02573공식 코드 · Papers with Code

Probing the Natural Language Inference Task with Automated Reasoning Tools

The Natural Language Inference (NLI) task is an important task in modern NLP, as it asks a broad question to which many other tasks may be reducible: Given a pair of sentences, does the first entail the second? Although the state-of-the-art on current benchmark datasets for NLI are deep learning-based, it is worthwhile to use other techniques to examine the logical structure of the NLI task. We do so by testing how well a machine-oriented controlled natural language (Attempto Controlled English) can be used to parse NLI sentences, and how well automated theorem provers can reason over the resulting formulae. To improve performance, we develop a set of syntactic and semantic transformation rules. We report their performance, and discuss implications for NLI and logic-based NLP.

arXiv:2005.00630공식 코드 · Papers with Code

KLEJ: Comprehensive Benchmark for Polish Language Understanding

In recent years, a series of Transformer-based models unlocked major improvements in general natural language understanding (NLU) tasks. Such a fast pace of research would not be possible without general NLU benchmarks, which allow for a fair comparison of the proposed methods. However, such benchmarks are available only for a handful of languages. To alleviate this issue, we introduce a comprehensive multi-task benchmark for the Polish language understanding, accompanied by an online leaderboard. It consists of a diverse set of tasks, adopted from existing datasets for named entity recognition, question-answering, textual entailment, and others. We also introduce a new sentiment analysis task for the e-commerce domain, named Allegro Reviews (AR). To ensure a common evaluation scheme and promote models that generalize to different NLU tasks, the benchmark includes datasets from varying domains and applications. Additionally, we release HerBERT, a Transformer-based model trained specifically for the Polish language, which has the best average performance and obtains the best results for three out of nine tasks. Finally, we provide an extensive evaluation, including several standard baselines and recently proposed, multilingual Transformer-based models.

arXiv:2005.00882공식 코드 · Papers with Code

Improving Truthfulness of Headline Generation

Most studies on abstractive summarization report ROUGE scores between system and reference summaries. However, we have a concern about the truthfulness of generated summaries: whether all facts of a generated summary are mentioned in the source text. This paper explores improving the truthfulness in headline generation on two popular datasets. Analyzing headlines generated by the state-of-the-art encoder-decoder model, we show that the model sometimes generates untruthful headlines. We conjecture that one of the reasons lies in untruthful supervision data used for training the model. In order to quantify the truthfulness of article-headline pairs, we consider the textual entailment of whether an article entails its headline. After confirming quite a few untruthful instances in the datasets, this study hypothesizes that removing untruthful instances from the supervision data may remedy the problem of the untruthful behaviors of the model. Building a binary classifier that predicts an entailment relation between an article and its headline, we filter out untruthful instances from the supervision data. Experimental results demonstrate that the headline generation model trained on filtered supervision data shows no clear difference in ROUGE scores but remarkable improvements in automatic and manual evaluations of the generated headlines.

arXiv:2005.00719

Probing the Probing Paradigm: Does Probing Accuracy Entail Task Relevance?

Although neural models have achieved impressive results on several NLP benchmarks, little is understood about the mechanisms they use to perform language tasks. Thus, much recent attention has been devoted to analyzing the sentence representations learned by neural encoders, through the lens of `probing' tasks. However, to what extent was the information encoded in sentence representations, as discovered through a probe, actually used by the model to perform its task? In this work, we examine this probing paradigm through a case study in Natural Language Inference, showing that models can learn to encode linguistic properties even if they are not needed for the task on which the model was trained. We further identify that pretrained word embeddings play a considerable role in encoding these properties rather than the training task itself, highlighting the importance of careful controls when designing probing experiments. Finally, through a set of controlled synthetic tasks, we demonstrate models can encode these properties considerably above chance-level even when distributed in the data as random noise, calling into question the interpretation of absolute claims on probing tasks.

arXiv:2005.00661공식 코드 · Papers with Code

On Faithfulness and Factuality in Abstractive Summarization

It is well known that the standard likelihood training and approximate decoding objectives in neural text generation models lead to less human-like responses for open-ended tasks such as language modeling and story generation. In this paper we have analyzed limitations of these models for abstractive document summarization and found that these models are highly prone to hallucinate content that is unfaithful to the input document. We conducted a large scale human evaluation of several neural abstractive summarization systems to better understand the types of hallucinations they produce. Our human annotators found substantial amounts of hallucinated content in all model generated summaries. However, our analysis does show that pretrained models are better summarizers not only in terms of raw metrics, i.e., ROUGE, but also in generating faithful and factual summaries as evaluated by humans. Furthermore, we show that textual entailment measures better correlate with faithfulness than standard metrics, potentially leading the way to automatic evaluation metrics as well as training and decoding criteria.

arXiv:2005.00979공식 코드 · Papers with Code

How Does Selective Mechanism Improve Self-Attention Networks?

Self-attention networks (SANs) with selective mechanism has produced substantial improvements in various NLP tasks by concentrating on a subset of input words. However, the underlying reasons for their strong performance have not been well explained. In this paper, we bridge the gap by assessing the strengths of selective SANs (SSANs), which are implemented with a flexible and universal Gumbel-Softmax. Experimental results on several representative NLP tasks, including natural language inference, semantic role labelling, and machine translation, show that SSANs consistently outperform the standard SANs. Through well-designed probing experiments, we empirically validate that the improvement of SSANs can be attributed in part to mitigating two commonly-cited weaknesses of SANs: word order encoding and structure modeling. Specifically, the selective mechanism improves SANs by paying more attention to content words that contribute to the meaning of the sentence. The code and data are released at https://github.com/xwgeng/SSAN.

arXiv:2004.14519공식 코드 · Papers with Code

An Empirical Study of Pre-trained Transformers for Arabic Information Extraction

Multilingual pre-trained Transformers, such as mBERT (Devlin et al., 2019) and XLM-RoBERTa (Conneau et al., 2020a), have been shown to enable the effective cross-lingual zero-shot transfer. However, their performance on Arabic information extraction (IE) tasks is not very well studied. In this paper, we pre-train a customized bilingual BERT, dubbed GigaBERT, that is designed specifically for Arabic NLP and English-to-Arabic zero-shot transfer learning. We study GigaBERT's effectiveness on zero-short transfer across four IE tasks: named entity recognition, part-of-speech tagging, argument role labeling, and relation extraction. Our best model significantly outperforms mBERT, XLM-RoBERTa, and AraBERT (Antoun et al., 2020) in both the supervised and zero-shot transfer settings. We have made our pre-trained models publicly available at https://github.com/lanwuwei/GigaBERT.

arXiv:2004.14623공식 코드 · Papers with Code

Neural Natural Language Inference Models Partially Embed Theories of Lexical Entailment and Negation

We address whether neural models for Natural Language Inference (NLI) can learn the compositional interactions between lexical entailment and negation, using four methods: the behavioral evaluation methods of (1) challenge test sets and (2) systematic generalization tasks, and the structural evaluation methods of (3) probes and (4) interventions. To facilitate this holistic evaluation, we present Monotonicity NLI (MoNLI), a new naturalistic dataset focused on lexical entailment and negation. In our behavioral evaluations, we find that models trained on general-purpose NLI datasets fail systematically on MoNLI examples containing negation, but that MoNLI fine-tuning addresses this failure. In our structural evaluations, we look for evidence that our top-performing BERT-based model has learned to implement the monotonicity algorithm behind MoNLI. Probes yield evidence consistent with this conclusion, and our intervention experiments bolster this, showing that the causal dynamics of the model mirror the causal dynamics of this algorithm on subsets of MoNLI. This suggests that the BERT model at least partially embeds a theory of lexical entailment and negation at an algorithmic level.

arXiv:2004.14963공식 코드 · Papers with Code

Data and Representation for Turkish Natural Language Inference

Large annotated datasets in NLP are overwhelmingly in English. This is an obstacle to progress in other languages. Unfortunately, obtaining new annotated resources for each task in each language would be prohibitively expensive. At the same time, commercial machine translation systems are now robust. Can we leverage these systems to translate English-language datasets automatically? In this paper, we offer a positive response for natural language inference (NLI) in Turkish. We translated two large English NLI datasets into Turkish and had a team of experts validate their translation quality and fidelity to the original labels. Using these datasets, we address core issues of representation for Turkish NLI. We find that in-language embeddings are essential and that morphological parsing can be avoided where the training set is large. Finally, we show that models trained on our machine-translated datasets are successful on human-translated evaluation sets. We share all code, models, and data publicly.