United States

Many recent studies on large-scale language models have reported successful in-context zero- and few-shot learning ability. However, the in-depth analysis of when in-context learning occurs is still lacking. For example, it is unknown how in-context learning performance changes as the training corpus varies. Here, we investigate the effects of the source and size of the pretraining corpus on in-context learning in HyperCLOVA, a Korean-centric GPT-3 model. From our in-depth investigation, we introduce the following observations: (1) in-context learning performance heavily depends on the corpus domain source, and the size of the pretraining corpus does not necessarily determine the emergence of in-context learning, (2) in-context learning ability can emerge when a language model is trained on a combination of multiple corpora, even when each corpus does not result in in-context learning on its own, (3) pretraining with a corpus related to a downstream task does not always guarantee the competitive in-context learning performance of the downstream task, especially in the few-shot setting, and (4) the relationship between language modeling (measured in perplexity) and in-context learning does not always correlate: e.g., low perplexity does not always imply high in-context few-shot learning performance.

NAACL 2022

On the Effect of Pretraining Corpora on In-context Learning by a Large-scale Language Model

langage model

pretraining corpora

gpt3

in-context learning

large scale

## Welcome to NAACL 2022!
The Annual Meeting of the North American Chapter of the Association for Computational Linguistics (NAACL) is taking place July 10-15, 2022 as a hybrid event, in Seattle, WA and online. We are happy to welcome all of you to conference! 
 
The main conference program features oral presentations, in-person and virtual posters and demo sessions, a plenary session for our best paper presentations and awards, keynote presentations, a plenary panel on the place of linguistics and symbolic structures in NLP, and an Industrial Track panel on Careers in NLP.
 
Posters (including Findings of NAACL 2022) and demos are grouped by areas for both the in-person and the virtual sessions. For the virtual component, the talks will be on Zoom and the posters and the demos will be in GatherTown. The Student Research Workshop and will have an oral session and a poster session.
 
The program also features 6 Tutorials and 26 Workshops.
 
 We wish you a wonderful conference!
 
[The NAACL 2022 Organizing Committee](https://2022.naacl.org/committees/organization/)

2022 Annual Conference of the North American Chapter of the Association for Computational Linguistics

poster

Recent causal probing literature reveals when language models and syntactic probes use similar representations. Such techniques may yield "false negative" causality results: models may use representations of syntax, but probes may have learned to use redundant encodings of the same syntactic information. We demonstrate that models do encode syntactic information redundantly and introduce a new probe design that guides probes to consider all syntactic information present in embeddings. Using these probes, we find evidence for the use of syntax in models where prior methods did not, allowing us to boost model performance by injecting syntactic information into representations.

When Does Syntax Mediate Neural Language Model Performance? Evidence from Dropout Probes

Interpretability methods are developed to understand the working mechanisms of black-box models, which is crucial to their responsible deployment. Fulfilling this goal requires both that the explanations generated by these methods are correct and that people can easily and reliably understand them. While the former has been addressed in prior work, the latter is often overlooked, resulting in informal model understanding derived from a handful of local explanations. In this paper, we introduce explanation summary (ExSum), a mathematical framework for quantifying model understanding, and propose metrics for its quality assessment. On two domains, ExSum highlights various limitations in the current practice, helps develop accurate model understanding, and reveals easily overlooked properties of the model. We also connect understandability to other properties of explanations such as human alignment, robustness, and counterfactual similarity and plausibility.

ExSum: From Local Explanations to Model Understanding

The hidden nature and the limited accessibility of the Dark Web, combined with the lack of public datasets in this domain, make it difficult to study its inherent characteristics such as linguistic properties. Previous works on text classification of Dark Web domain have suggested that the use of deep neural models may be ineffective, potentially due to the linguistic differences between the Dark and Surface Webs. However, not much work has been done to uncover the linguistic characteristics of the Dark Web. This paper introduces CoDA, a publicly available Dark Web dataset consisting of 10000 web documents tailored towards text-based Dark Web analysis. By leveraging CoDA, we conduct a thorough linguistic analysis of the Dark Web and examine the textual differences between the Dark Web and the Surface Web. We also assess the performance of various methods of Dark Web page classification. Finally, we compare CoDA with an existing public Dark Web dataset and evaluate their suitability for various use cases.

Shedding New Light on the Language of the Dark Web

Academic research is an exploratory activity to discover new solutions to problems. By this nature, academic research works perform literature reviews to distinguish their novelties from prior work. In natural language processing, this literature review is usually conducted under the "Related Work" section. The task of related work generation aims to automatically generate the related work section given the rest of the research paper and a list of papers to cite. Prior work on this task has focused on the sentence as the basic unit of generation, neglecting the fact that related work sections consist of variable length text fragments derived from different information sources. As a first step toward a linguistically-motivated related work generation framework, we present a Citation Oriented Related Work Annotation (CORWA) dataset that labels different types of citation text fragments from different information sources. We train a strong baseline model that automatically tags the CORWA labels on massive unlabeled related work section texts. We further suggest a novel framework for human-in-the-loop, iterative, abstractive related work generation.

CORWA: A Citation-Oriented Related Work Annotation Dataset

Modern unsupervised machine translation (MT) systems reach reasonable translation quality under clean and controlled data conditions. As the performance gap between supervised and unsupervised MT narrows, it is interesting to ask whether the different training methods result in systematically different output beyond what is visible via quality metrics like adequacy or BLEU. We compare translations from supervised and unsupervised MT systems of similar quality, finding that unsupervised output is more fluent and more structurally different in comparison to human translation than is supervised MT. We then demonstrate a way to combine the benefits of both methods into a single system which results in improved adequacy and fluency as rated by human evaluators. Our results open the door to interesting discussions about how supervised and unsupervised MT might be different yet mutually-beneficial.

On Systematic Style Differences between Unsupervised and Supervised MT and an Application for High-Resource Machine Translation

Vocabulary selection, or lexical shortlisting, is a well-known technique to improve latency of Neural Machine Translation models by constraining the set of allowed output words during inference. The chosen set is typically determined by separately trained alignment model parameters, independent of the source-sentence context at inference time. While vocabulary selection appears competitive with respect to automatic quality metrics in prior work, we show that it can fail to select the right set of output words, particularly for semantically non-compositional linguistic phenomena such as idiomatic expressions, leading to reduced translation quality as perceived by humans. Trading off latency for quality by increasing the size of the allowed set is often not an option in real-world scenarios. We propose a model of vocabulary selection, integrated into the neural translation model, that predicts the set of allowed output words from contextualized encoder representations. This restores translation quality of an unconstrained system, as measured by human evaluations on WMT newstest2020 and idiomatic expressions, at an inference latency competitive with alignment-based selection using aggressive thresholds, thereby removing the dependency on separately trained alignment models.

The Devil is in the Details: On the Pitfalls of Vocabulary Selection in Neural Machine Translation

Social media rumours, a form of misinformation, can mislead the public and cause significant economic and social disruption. Motivated by the observation that the user network — which captures who engages with a story — and the comment network — which captures how they react to it—provide complementary signals for rumour detection. In this paper, we propose DUCK (rumour detection with user and comment networks) for rumour detection on social media. We study how to leverage transformers and graph attention networks to jointly model the contents and the structure of social media conversations, as well as the network of users who engage in these conversations. Over four widely used benchmark rumour datasets in English and Chinese, we show that DUCK produces superior performance
for detecting rumours, creating a new state-of-the-art.

DUCK: Rumour Detection on Social Media by Modelling User and  Comment Propagation Networks

Little attention has been paid on EArly Rumor Detection (EARD), and EARD performance was evaluated inappropriately on a few datasets where the actual early-stage information is largely missing. To reverse such situation, we construct BEARD, a new Benchmark dataset for EARD, based on claims from fact-checking websites by trying to gather as many early relevant posts as possible. We also pro- pose HEARD, a novel model based on neural Hawkes process for EARD, which can guide a generic rumor detection model to make timely, accurate and stable predictions. Experiments show that HEARD achieves effective EARD performance on two commonly used general rumor detection datasets and our BEARD dataset.

Early Rumor Detection Using Neural Hawkes Process with a New Benchmark Dataset

Task-oriented parsing (TOP) aims to convert natural language into machine-readable representations of specific tasks, such as setting an alarm. A popular approach to TOP is to apply seq2seq models to generate linearized parse trees. A more recent line of work argues that pretrained seq2seq2 models are better at generating outputs that are themselves natural language, so they replace linearized parse trees with canonical natural-language paraphrases that can then be easily translated into parse trees, resulting in so-called naturalized parsers. In this work we continue to explore naturalized semantic parsing by presenting a general reduction of TOP to abstractive question answering that overcomes some limitations of canonical paraphrasing. Experimental results show that our QA-based technique outperforms state-of-the-art methods in full-data settings while achieving dramatic improvements in few-shot settings.

Compositional Task-Oriented Parsing as Abstractive Question Answering

We examine the extent to which, in principle, different syntactic and semantic graph representations can complement and improve neural language modeling. Specifically, by conditioning on a subgraph encapsulating the locally relevant sentence history, can a model make better next-word predictions than a pretrained sequential language model alone? With an ensemble setup consisting of GPT-2 and ground-truth graphs from one of 7 different formalisms, we find that the graph information indeed improves perplexity and other metrics. Moreover, this architecture provides a new way to compare different frameworks of linguistic representation. In our oracle graph setup, training and evaluating on English WSJ, semantic constituency structures prove most useful to language modeling performance—outpacing syntactic constituency structures as well as syntactic and semantic dependency structures.

On the Effect of Pretraining Corpora on In-context Learning by a Large-scale Language Model

Downloads

Next from NAACL 2022

When Does Syntax Mediate Neural Language Model Performance? Evidence from Dropout Probes

Similar lecture

Improving In-Context Few-Shot Learning via Self-Supervised Training