Reasoning is a fundamental component of language understanding. Recent prompting techniques, such as chain of thought, have consistently improved LLMs’ performance on various reasoning tasks. Nevertheless, there is still little understanding of what triggers reasoning abilities in LLMs in the inference stage. In this paper, we investigate the effect of the input representation on the reasoning abilities of LLMs. We hypothesize that representing natural language tasks as code can enhance specific reasoning abilities such as entity tracking or logical reasoning. To study this, we propose code prompting, a methodology we operationalize as a chain of prompts that transforms a natural language problem into code and directly prompts the LLM using the generated code without resorting to external code execution. We find that code prompting exhibits a high-performance boost for multiple LLMs (up to 22.52 percentage points on GPT 3.5, 7.75 on Mixtral, and 16.78 on Mistral) across multiple conditional reasoning datasets. We then conduct comprehensive experiments to understand how the code representation triggers reasoning abilities and which capabilities are elicited in the underlying models. Our analysis on GPT 3.5 reveals that the code formatting of the input problem is essential for performance improvement. Furthermore, the code representation improves sample efficiency of in-context learning and facilitates state tracking of entities.

In recent years, large language models (LLMs) have shown remarkable capabilities at scale, particularly at generating text conditioned on a prompt. In our work, we investigate the use of LLMs to augment training data of smaller language models (SLMs) with automatically generated counterfactual (CF) instances – i.e. minimally altered inputs – in order to improve out-of-domain (OOD) performance of SLMs in the extractive question answering (QA) setup. We show that, across various LLM generators, such data augmentation consistently enhances OOD performance and improves model calibration for both confidence-based and rationale-augmented calibrator models. Furthermore, these performance improvements correlate with higher diversity of CF instances in terms of their surface form and semantic content. Finally, we show that CF augmented models which are easier to calibrate also exhibit much lower entropy when assigning importance, indicating that rationale-augmented calibrators prefer concise explanations.

A popular approach to unveiling the black box of neural NLP models is to leverage saliency methods, which assign scalar importance scores to each input component. A common practice for evaluating whether an interpretability method is faithful has been to use evaluation-by-agreement – if multiple methods agree on an explanation, its credibility increases. However, recent work has found that saliency methods exhibit weak rank correlations even when applied to the same model instance and advocated for alternative diagnostic methods. In our work, we demonstrate that rank correlation is not a good fit for evaluating agreement and argue that Pearson-r is a better-suited alternative. We further show that regularization techniques that increase faithfulness of attention explanations also increase agreement between saliency methods. By connecting our findings to instance categories based on training dynamics, we show that the agreement of saliency method explanations is very low for easy-to-learn instances. Finally, we connect the improvement in agreement across instance categories to local representation space statistics of instances, paving the way for work on analyzing which intrinsic model properties improve their predisposition to interpretability methods.

Twitter has slowly but surely established itself as a forum for disseminating, analysing and promoting NLP research. The trend of researchers promoting work not yet peer-reviewed (preprints) by posting concise summaries presented itself as an opportunity to collect and combine multiple modalities of data. In scope of this paper, we (1) construct a dataset of Twitter threads in which researchers promote NLP preprints and (2) evaluate whether it is possible to predict the popularity of a thread based on the content of the Twitter thread, paper content and user metadata. We experimentally show that it is possible to predict popularity of threads promoting research based on their content, and that predictive performance depends on modelling textual input, indicating that the dataset could present value for related areas of NLP research such as citation recommendation and abstractive summarization.

The attention mechanism has quickly become ubiquitous in NLP. In addition to improving performance of models, attention has been widely used as a glimpse into the inner workings of NLP models. The latter aspect has in the recent years become a common topic of discussion, most notably in recent work of Jain and Wallace; Wiegreffe and Pinter. With the shortcomings of using attention weights as a tool of transparency revealed, the attention mechanism has been stuck in a limbo without concrete proof when and whether it can be used as an explanation. In this paper, we provide an explanation as to why attention has seen rightful critique when used with recurrent networks in sequence classification tasks. We propose a remedy to these issues in the form of a word level objective and our findings give credibility for attention to provide faithful interpretations of recurrent models.

Natural language processing has greatly benefited from the introduction of the attention mechanism. However, standard attention models are of limited interpretability for tasks that involve a series of inference steps. We describe an iterative recursive attention model, which constructs incremental representations of input data through reusing results of previously computed queries. We train our model on sentiment classification datasets and demonstrate its capacity to identify and combine different aspects of the input in an easily interpretable manner, while obtaining performance close to the state of the art.

This paper describes our system for fine-grained sentiment scoring of news headlines submitted to SemEval 2017 task 5–subtask 2. Our system uses a feature-light method that consists of a Support Vector Regression (SVR) with various kernels and word vectors as features. Our best-performing submission scored 3rd on the task out of 29 teams and 4th out of 45 submissions with a cosine score of 0.733.

In this paper, we describe our preliminary study on annotating event mention as a part of our research on high-precision news event extraction models. To this end, we propose a two-layer annotation scheme, designed to separately capture the functional and conceptual aspects of event mentions. We hypothesize that the precision of models can be improved by modeling and extracting separately the different aspects of news events, and then combining the extracted information by leveraging the complementarities of the models. In addition, we carry out a preliminary annotation using the proposed scheme and analyze the annotation quality in terms of inter-annotator agreement.

We present a preliminary study on predicting news values from headline text and emotions. We perform a multivariate analysis on a dataset manually annotated with news values and emotions, discovering interesting correlations among them. We then train two competitive machine learning models – an SVM and a CNN – to predict news values from headline text and emotions as features. We find that, while both models yield a satisfactory performance, some news values are more difficult to detect than others, while some profit more from including emotion information.