Despite the widespread adoption of autoregressive language models, explainability evaluation research has predominantly focused on span infilling and masked language models. Evaluating the faithfulness of an explanation method—how accurately it explains the inner workings and decision-making of the model—is challenging because it is difficult to separate the model from its explanation. Most faithfulness evaluation techniques corrupt or remove input tokens deemed important by a particular attribution (feature importance) method and observe the resulting change in the model’s output. However, for autoregressive language models, this approach creates out-of-distribution inputs due to their next-token prediction training objective. In this study, we propose a technique that leverages counterfactual generation to evaluate the faithfulness of attribution methods for autoregressive language models. Our technique generates fluent, in-distribution counterfactuals, making the evaluation protocol more reliable.

This paper explores the efficacy of large language models (LLMs) for Persian. While ChatGPT and consequent LLMs have shown remarkable performance in English, their efficiency for more low-resource languages remains an open question. We present the first comprehensive benchmarking study of LLMs across diverse Persian language tasks. Our primary focus is on GPT-3.5-turbo, but we also include GPT-4 and OpenChat-3.5 to provide a more holistic evaluation. Our assessment encompasses a diverse set of tasks categorized into classic, reasoning, and knowledge-based domains. To enable a thorough comparison, we evaluate LLMs against existing task-specific fine-tuned models. Given the limited availability of Persian datasets for reasoning tasks, we introduce two new benchmarks: one based on elementary school math questions and another derived from the entrance exams for 7th and 10th grades. Our findings reveal that while LLMs, especially GPT-4, excel in tasks requiring reasoning abilities and a broad understanding of general knowledge, they often lag behind smaller pretrained models fine-tuned specifically for particular tasks. Additionally, we observe improved performance when test sets are translated to English before inputting them into GPT-3.5. These results highlight the significant potential for enhancing LLM performance in the Persian language. This is particularly noteworthy due to the unique attributes of Persian, including its distinct alphabet and writing styles. We have made our codes, prompts, and data available here: https://github.com/Ipouyall/Benchmarking_ChatGPT_for_Persian.

Inspired by human cognition, Jiang et al. 2023 create a benchmark for assessing LLMs’ lateral thinking—thinking outside the box. Building upon this benchmark, we investigate how different prompting methods enhance LLMs’ performance on this task to reveal their inherent power for outside-the-box thinking ability. Through participating in SemEval-2024, task 9, Sentence Puzzle sub-task, we explore prompt engineering methods: chain of thoughts (CoT) and direct prompting, enhancing with informative descriptions, and employing contextualizing prompts using a retrieval augmented generation (RAG) pipeline. Our experiments involve three LLMs including GPT-3.5, GPT-4, and Zephyr-7B-beta. We generate a dataset of thinking paths between riddles and options using GPT-4, validated by humans for quality. Findings indicate that compressed informative prompts enhance performance. Dynamic in-context learning enhances model performance significantly. Furthermore, fine-tuning Zephyr on our dataset enhances performance across other commonsense datasets, underscoring the value of innovative thinking.

An emerging solution for explaining Transformer-based models is to use vector-based analysis on how the representations are formed. However, providing a faithful vector-based explanation for a multi-layer model could be challenging in three aspects: (1) Incorporating all components into the analysis, (2) Aggregating the layer dynamics to determine the information flow and mixture throughout the entire model, and (3) Identifying the connection between the vector-based analysis and the model’s predictions. In this paper, we present DecompX to tackle these challenges. DecompX is based on the construction of decomposed token representations and their successive propagation throughout the model without mixing them in between layers. Additionally, our proposal provides multiple advantages over existing solutions for its inclusion of all encoder components (especially nonlinear feed-forward networks) and the classification head. The former allows acquiring precise vectors while the latter transforms the decomposition into meaningful prediction-based values, eliminating the need for norm- or summation-based vector aggregation. According to the standard faithfulness evaluations, DecompX consistently outperforms existing gradient-based and vector-based approaches on various datasets. Our code is available at https://github.com/mohsenfayyaz/DecompX.

In recent years, there has been significant progress in developing pre-trained language models for NLP. However, these models often struggle when fine-tuned on small datasets. To address this issue, researchers have proposed various adaptation approaches. Prompt-based tuning is arguably the most common way, especially for larger models. Previous research shows that adding contrastive learning to prompt-based fine-tuning is effective as it helps the model generate embeddings that are more distinguishable between classes, and it can also be more sample-efficient as the model learns from positive and negative examples simultaneously. One of the most important components of contrastive learning is data augmentation, but unlike computer vision, effective data augmentation for NLP is still challenging. This paper proposes LM-CPPF, Contrastive Paraphrasing-guided Prompt-based Fine-tuning of Language Models, which leverages prompt-based few-shot paraphrasing using generative language models, especially large language models such as GPT-3 and OPT-175B, for data augmentation. Our experiments on multiple text classification benchmarks show that this augmentation method outperforms other methods, such as easy data augmentation, back translation, and multiple templates.

Neural networks have revolutionized language modeling and excelled in various downstream tasks. However, the extent to which these models achieve compositional generalization comparable to human cognitive abilities remains a topic of debate. While existing approaches in the field have mainly focused on novel architectures and alternative learning paradigms, we introduce a pioneering method harnessing the power of dataset cartography (Swayamdipta et al., 2020). By strategically identifying a subset of compositional generalization data using this approach, we achieve a remarkable improvement in model accuracy, yielding enhancements of up to 10% on CFQ and COGS datasets. Notably, our technique incorporates dataset cartography as a curriculum learning criterion, eliminating the need for hyperparameter tuning while consistently achieving superior performance. Our findings highlight the untapped potential of dataset cartography in unleashing the full capabilities of compositional generalization within Transformer models.

Multilingual pre-training significantly improves many multilingual NLP tasks, including machine translation. Most existing methods are based on some variants of masked language modeling and text-denoising objectives on monolingual data. Multilingual pre-training on monolingual data ignores the availability of parallel data in many language pairs. Also, some other works integrate the available human-generated parallel translation data in their pre-training. This kind of parallel data is definitely helpful, but it is limited even in high-resource language pairs. This paper introduces a novel semi-supervised method, SPDG, that generates high-quality pseudo-parallel data for multilingual pre-training. First, a denoising model is pre-trained on monolingual data to reorder, add, remove, and substitute words, enhancing the pre-training documents’ quality. Then, we generate different pseudo-translations for each pre-training document using dictionaries for word-by-word translation and applying the pre-trained denoising model. The resulting pseudo-parallel data is then used to pre-train our multilingual sequence-to-sequence model, PEACH. Our experiments show that PEACH outperforms existing approaches used in training mT5 and mBART on various translation tasks, including supervised, zero- and few-shot scenarios. Moreover, PEACH’s ability to transfer knowledge between similar languages makes it particularly useful for low-resource languages. Our results demonstrate that with high-quality dictionaries for generating accurate pseudo-parallel, PEACH can be valuable for low-resource languages.

There has been a growing interest in interpreting the underlying dynamics of Transformers. While self-attention patterns were initially deemed as the primary option, recent studies have shown that integrating other components can yield more accurate explanations. This paper introduces a novel token attribution analysis method that incorporates all the components in the encoder block and aggregates this throughout layers. Through extensive quantitative and qualitative experiments, we demonstrate that our method can produce faithful and meaningful global token attributions. Our experiments reveal that incorporating almost every encoder component results in increasingly more accurate analysis in both local (single layer) and global (the whole model) settings. Our global attribution analysis significantly outperforms previous methods on various tasks regarding correlation with gradient-based saliency scores. Our code is freely available at https://github.com/mohsenfayyaz/GlobEnc.

Human languages are full of metaphorical expressions. Metaphors help people understand the world by connecting new concepts and domains to more familiar ones. Large pre-trained language models (PLMs) are therefore assumed to encode metaphorical knowledge useful for NLP systems. In this paper, we investigate this hypothesis for PLMs, by probing metaphoricity information in their encodings, and by measuring the cross-lingual and cross-dataset generalization of this information. We present studies in multiple metaphor detection datasets and in four languages (i.e., English, Spanish, Russian, and Farsi). Our extensive experiments suggest that contextual representations in PLMs do encode metaphorical knowledge, and mostly in their middle layers. The knowledge is transferable between languages and datasets, especially when the annotation is consistent across training and testing sets. Our findings give helpful insights for both cognitive and NLP scientists.

It has been shown that NLI models are usually biased with respect to the word-overlap between the premise and the hypothesis, as they take this feature as a primary cue for predicting the entailment label. In this paper, we focus on an overlooked aspect of the overlap bias in the NLI models: the reverse word-overlap bias. Our experimental results demonstrate that current NLI systems are also highly biased towards the non-entailment label on instances with low overlap and that existing debiasing methods, which are reportedly successful on challenge datasets, are generally ineffective in addressing this category of bias.Through a set of analyses, we investigate the reasons for the emergence of the overlap bias and the role of minority examples in mitigating this bias.For the former, we find that the word overlap bias does not stem from pre-training, and in the latter, we observe that in contrast to the accepted assumption, eliminating minority examples does not affect the generalizability of debiasing methods with respect to the overlap bias.

Community Question Answering (CQA) forums provide answers to many real-life questions. These forums are trendy among machine learning researchers due to their large size. Automatic answer selection, answer ranking, question retrieval, expert finding, and fact-checking are example learning tasks performed using CQA data. This paper presents PerCQA, the first Persian dataset for CQA. This dataset contains the questions and answers crawled from the most well-known Persian forum. After data acquisition, we provide rigorous annotation guidelines in an iterative process and then the annotation of question-answer pairs in SemEvalCQA format. PerCQA contains 989 questions and 21,915 annotated answers. We make PerCQA publicly available to encourage more research in Persian CQA. We also build strong benchmarks for the task of answer selection in PerCQA by using mono- and multi-lingual pre-trained language models.

This paper presents the AliEdalat team’s methodology and results in SemEval-2022 Task 4: Patronizing and Condescending Language (PCL) Detection. This task aims to detect the presence of PCL and PCL categories in text in order to prevent further discrimination against vulnerable communities. We use an ensemble of three basic models to detect the presence of PCL: fine-tuned bigbird, fine-tuned mpnet, and BERT+BiGRU. The ensemble model performs worse than the baseline due to overfitting and achieves an F1-score of 0.3031. We offer another solution to resolve the submitted model’s problem. We consider the different categories of PCL separately. To detect each category of PCL, we act like a PCL detector. Instead of BERT+BiGRU, we use fine-tuned roberta in the models. In PCL category detection, our model outperforms the baseline model and achieves an F1-score of 0.2531. We also present new models for detecting two categories of PCL that outperform the submitted models.

Despite the progress made in recent years in addressing natural language understanding (NLU) challenges, the majority of this progress remains to be concentrated on resource-rich languages like English. This work focuses on Persian language, one of the widely spoken languages in the world, and yet there are few NLU datasets available for this language. The availability of high-quality evaluation datasets is a necessity for reliable assessment of the progress on different NLU tasks and domains. We introduce ParsiNLU, the first benchmark in Persian language that includes a range of language understanding tasks—reading comprehension, textual entailment, and so on. These datasets are collected in a multitude of ways, often involving manual annotations by native speakers. This results in over 14.5k new instances across 6 distinct NLU tasks. Additionally, we present the first results on state-of-the-art monolingual and multilingual pre-trained language models on this benchmark and compare them with human performance, which provides valuable insights into our ability to tackle natural language understanding challenges in Persian. We hope ParsiNLU fosters further research and advances in Persian language understanding.1

Neural NLP models tend to rely on spurious correlations between labels and input features to perform their tasks. Minority examples, i.e., examples that contradict the spurious correlations present in the majority of data points, have been shown to increase the out-of-distribution generalization of pre-trained language models. In this paper, we first propose using example forgetting to find minority examples without prior knowledge of the spurious correlations present in the dataset. Forgettable examples are instances either learned and then forgotten during training or never learned. We show empirically how these examples are related to minorities in our training sets. Then, we introduce a new approach to robustify models by fine-tuning our models twice, first on the full training data and second on the minorities only. We obtain substantial improvements in out-of-distribution generalization when applying our approach to the MNLI, QQP and FEVER datasets.

Pretrained language models achieve state-of-the-art results on many NLP tasks, but there are still many open questions about how and why they work so well. We investigate the contextualization of words in BERT. We quantify the amount of contextualization, i.e., how well words are interpreted in context, by studying the extent to which semantic classes of a word can be inferred from its contextualized embedding. Quantifying contextualization helps in understanding and utilizing pretrained language models. We show that the top layer representations support highly accurate inference of semantic classes; that the strongest contextualization effects occur in the lower layers; that local context is mostly sufficient for contextualizing words; and that top layer representations are more task-specific after finetuning while lower layer representations are more transferable. Finetuning uncovers task-related features, but pretrained knowledge about contextualization is still well preserved.

Word embeddings typically represent different meanings of a word in a single conflated vector. Empirical analysis of embeddings of ambiguous words is currently limited by the small size of manually annotated resources and by the fact that word senses are treated as unrelated individual concepts. We present a large dataset based on manual Wikipedia annotations and word senses, where word senses from different words are related by semantic classes. This is the basis for novel diagnostic tests for an embedding’s content: we probe word embeddings for semantic classes and analyze the embedding space by classifying embeddings into semantic classes. Our main findings are: (i) Information about a sense is generally represented well in a single-vector embedding – if the sense is frequent. (ii) A classifier can accurately predict whether a word is single-sense or multi-sense, based only on its embedding. (iii) Although rare senses are not well represented in single-vector embeddings, this does not have negative impact on an NLP application whose performance depends on frequent senses.

Large scale knowledge graphs (KGs) such as Freebase are generally incomplete. Reasoning over multi-hop (mh) KG paths is thus an important capability that is needed for question answering or other NLP tasks that require knowledge about the world. mh-KG reasoning includes diverse scenarios, e.g., given a head entity and a relation path, predict the tail entity; or given two entities connected by some relation paths, predict the unknown relation between them. We present ROPs, recurrent one-hop predictors, that predict entities at each step of mh-KB paths by using recurrent neural networks and vector representations of entities and relations, with two benefits: (i) modeling mh-paths of arbitrary lengths while updating the entity and relation representations by the training signal at each step; (ii) handling different types of mh-KG reasoning in a unified framework. Our models show state-of-the-art for two important multi-hop KG reasoning tasks: Knowledge Base Completion and Path Query Answering.

Embedding models typically associate each word with a single real-valued vector, representing its different properties. Evaluation methods, therefore, need to analyze the accuracy and completeness of these properties in embeddings. This requires fine-grained analysis of embedding subspaces. Multi-label classification is an appropriate way to do so. We propose a new evaluation method for word embeddings based on multi-label classification given a word embedding. The task we use is fine-grained name typing: given a large corpus, find all types that a name can refer to based on the name embedding. Given the scale of entities in knowledge bases, we can build datasets for this task that are complementary to the current embedding evaluation datasets in: they are very large, contain fine-grained classes, and allow the direct evaluation of embeddings without confounding factors like sentence context.

Accurate and complete knowledge bases (KBs) are paramount in NLP. We employ mul-itiview learning for increasing the accuracy and coverage of entity type information in KBs. We rely on two metaviews: language and representation. For language, we consider high-resource and low-resource languages from Wikipedia. For representation, we consider representations based on the context distribution of the entity (i.e., on its embedding), on the entity’s name (i.e., on its surface form) and on its description in Wikipedia. The two metaviews language and representation can be freely combined: each pair of language and representation (e.g., German embedding, English description, Spanish name) is a distinct view. Our experiments on entity typing with fine-grained classes demonstrate the effectiveness of multiview learning. We release MVET, a large multiview — and, in particular, multilingual — entity typing dataset we created. Mono- and multilingual fine-grained entity typing systems can be evaluated on this dataset.

Entities are essential elements of natural language. In this paper, we present methods for learning multi-level representations of entities on three complementary levels: character (character patterns in entity names extracted, e.g., by neural networks), word (embeddings of words in entity names) and entity (entity embeddings). We investigate state-of-the-art learning methods on each level and find large differences, e.g., for deep learning models, traditional ngram features and the subword model of fasttext (Bojanowski et al., 2016) on the character level; for word2vec (Mikolov et al., 2013) on the word level; and for the order-aware model wang2vec (Ling et al., 2015a) on the entity level. We confirm experimentally that each level of representation contributes complementary information and a joint representation of all three levels improves the existing embedding based baseline for fine-grained entity typing by a large margin. Additionally, we show that adding information from entity descriptions further improves multi-level representations of entities.

In this paper, we address two different types of noise in information extraction models: noise from distant supervision and noise from pipeline input features. Our target tasks are entity typing and relation extraction. For the first noise type, we introduce multi-instance multi-label learning algorithms using neural network models, and apply them to fine-grained entity typing for the first time. Our model outperforms the state-of-the-art supervised approach which uses global embeddings of entities. For the second noise type, we propose ways to improve the integration of noisy entity type predictions into relation extraction. Our experiments show that probabilistic predictions are more robust than discrete predictions and that joint training of the two tasks performs best.