Controlled text generation (CTG) seeks to guide large language model (LLM) output, that statistical language generation would conform to desired criteria. The current study presents a novel CTG algorithm that enforces adherence toward specific rhetorical relations in an LLM sentence-completion context by a parser-driven decoding scheme that requires no model fine-tuning. The method is validated both with automatic and human evaluation.

Neural architecture search (NAS) has allowed for the automatic creation of new and effective neural network architectures, offering an alternative to the laborious process of manually designing complex architectures. However, traditional NAS algorithms are slow and require immense amounts of computing power. Recent research has investigated training-free NAS metrics for image classification architectures, drastically speeding up search algorithms. In this paper, we investigate training-free NAS metrics for recurrent neural network (RNN) and BERT-based transformer architectures, targeted towards language modeling tasks. First, we develop a new training-free metric, named hidden covariance, that predicts the trained performance of an RNN architecture and significantly outperforms existing training-free metrics. We experimentally evaluate the effectiveness of the hidden covariance metric on the NAS-Bench-NLP benchmark. Second, we find that the current search space paradigm for transformer architectures is not optimized for training-free neural architecture search. Instead, a simple qualitative analysis can effectively shrink the search space to the best performing architectures. This conclusion is based on our investigation of existing training-free metrics and new metrics developed from recent transformer pruning literature, evaluated on our own benchmark of trained BERT architectures. Ultimately, our analysis shows that the architecture search space and the training-free metric must be developed together in order to achieve effective results. Our source code is available at https://github.com/aaronserianni/training-free-nas.

This paper describes CIC NLP’s submission to the AmericasNLP 2023 Shared Task on machine translation systems for indigenous languages of the Americas. We present the system descriptions for three methods. We used two multilingual models, namely M2M-100 and mBART50, and one bilingual (one-to-one) — Helsinki NLP Spanish-English translation model, and experimented with different transfer learning setups. We experimented with 11 languages from America and report the setups we used as well as the results we achieved. Overall, the mBART setup was able to improve upon the baseline for three out of the eleven languages.

Summary sentences produced by abstractive summarization models may be coherent and comprehensive, but they lack control and rely heavily on reference summaries. The BRIO training paradigm assumes a non-deterministic distribution to reduce the model’s dependence on reference summaries, and improve model performance during inference. This paper presents a straightforward but effective technique to improve abstractive summaries by fine-tuning pre-trained language models, and training them with the BRIO paradigm. We build a text summarization dataset for Vietnamese, called VieSum. We perform experiments with abstractive summarization models trained with the BRIO paradigm on the CNNDM and the VieSum datasets. The results show that the models, trained on basic hardware, outperform all existing abstractive summarization models, especially for Vietnamese.

Multilingual Language Models offer a way to incorporate multiple languages in one model and utilize cross-language transfer learning to improve performance for different Natural Language Processing (NLP) tasks. Despite progress in multilingual models, not all languages are supported as well, particularly in low-resource settings. In this work, we investigate the linguistic representation of different languages in multilingual models. We start by asking the question which languages are supported in popular multilingual models and which languages are left behind. Then, for included languages, we look at models’ learned representations based on language family and dialect and try to understand how models’ learned representations for (1) seen and (2) unseen languages vary across different language groups. In addition, we test and analyze performance on downstream tasks such as text generation and Named Entity Recognition. We observe from our experiments that community-centered models—models that focus on languages of a given family or geographical location and are built by communities who speak them—perform better at distinguishing between languages in the same family for low-resource languages. Our paper contributes to the literature in understanding multilingual models and their shortcomings and offers insights on potential ways to improve them.

Vision-language models can assess visual context in an image and generate descriptive text. While the generated text may be accurate and syntactically correct, it is often overly general. To address this, recent work has used optical character recognition to supplement visual information with text extracted from an image. In this work, we contend that vision-language models can benefit from information that can be extracted from an image, but are not used by current models. We modify previous multimodal frameworks to accept relevant information from any number of auxiliary classifiers. In particular, we focus on person names as an additional set of tokens and create a novel image-caption dataset to facilitate captioning with person names. The dataset, Politicians and Athletes in Captions (PAC), consists of captioned images of well-known people in context. By fine-tuning pretrained models with this dataset, we demonstrate a model that can naturally integrate facial recognition tokens into generated text by training on limited data. For the PAC dataset, we provide a discussion on collection and baseline benchmark scores.

Attacks on deep learning models are often difficult to identify and therefore are difficult to protect against. This problem is exacerbated by the use of public datasets that typically are not manually inspected before use. In this paper, we offer a solution to this vulnerability by using, during testing, random perturbations such as spelling correction if necessary, substitution by random synonym, or simply drop-ping the word. These perturbations are applied to random words in random sentences to defend NLP models against adversarial attacks. Our Random Perturbations Defense andIncreased Randomness Defense methods are successful in returning attacked models to similar accuracy of models before attacks. The original accuracy of the model used in this work is 80% for sentiment classification. After undergoing attacks, the accuracy drops to an accuracy between 0% and 44%. After applying our defense methods, the accuracy of the model is returned to the original accuracy within statistical significance.

Studies have shown that deep neural networks (DNNs) are vulnerable to adversarial examples – perturbed inputs that cause DNN-based models to produce incorrect results. One robust adversarial attack in the NLP domain is the synonym substitution. In attacks of this variety, the adversary substitutes words with synonyms. Since synonym substitution perturbations aim to satisfy all lexical, grammatical, and semantic constraints, they are difficult to detect with automatic syntax check as well as by humans. In this paper, we propose a structure-free defensive method that is capable of improving the performance of DNN-based models with both clean and adversarial data. Our findings show that replacing the embeddings of the important words in the input samples with the average of their synonyms’ embeddings can significantly improve the generalization of DNN-based classifiers. By doing so, we reduce model sensitivity to particular words in the input samples. Our results indicate that the proposed defense is not only capable of defending against adversarial attacks, but is also capable of improving the performance of DNN-based models when tested on benign data. On average, the proposed defense improved the classification accuracy of the CNN and Bi-LSTM models by 41.30% and 55.66%, respectively, when tested under adversarial attacks. Extended investigation shows that our defensive method can improve the robustness of nonneural models, achieving an average of 17.62% and 22.93% classification accuracy increase on the SVM and XGBoost models, respectively. The proposed defensive method has also shown an average of 26.60% classification accuracy improvement when tested with the infamous BERT model. Our algorithm is generic enough to be applied in any NLP domain and to any model trained on any natural language.

This paper outlines the use of Transformer networks trained to translate math word problems to equivalent arithmetic expressions in infix, prefix, and postfix notations. We compare results produced by a large number of neural configurations and find that most configurations outperform previously reported approaches on three of four datasets with significant increases in accuracy of over 20 percentage points. The best neural approaches boost accuracy by 30% on average when compared to the previous state-of-the-art.

Artificial Neural networks are mathematical models at their core. This truism presents some fundamental difficulty when networks are tasked with Natural Language Processing. A key problem lies in measuring the similarity or distance among vectors in NLP embedding space, since the mathematical concept of distance does not always agree with the linguistic concept. We suggest that the best way to measure linguistic distance among vectors is by employing the Language Model (LM) that created them. We introduce Language Model Distance (LMD) for measuring accuracy of vector transformations based on the Distributional Hypothesis ( LMD Accuracy ). We show the efficacy of this metric by applying it to a simple neural network learning the Procrustes algorithm for bilingual word mapping.

Poetry Generation involves teaching systems to automatically generate text that resembles poetic work. A deep learning system can learn to generate poetry on its own by training on a corpus of poems and modeling the particular style of language. In this paper, we propose taking an approach that fine-tunes GPT-2, a pre-trained language model, to our downstream task of poetry generation. We extend prior work on poetry generation by introducing creative elements. Specifically, we generate poems that express emotion and elicit the same in readers, and poems that use the language of dreams—called dream poetry. We are able to produce poems that correctly elicit the emotions of sadness and joy 87.5 and 85 percent, respectively, of the time. We produce dreamlike poetry by training on a corpus of texts that describe dreams. Poems from this model are shown to capture elements of dream poetry with scores of no less than 3.2 on the Likert scale. We perform crowdsourced human-evaluation for all our poems. We also make use of the Coh-Metrix tool, outlining metrics we use to gauge the quality of text generated.

Recent advances in Natural Language Processing are finding ways to place an emphasis on the hierarchical nature of text instead of representing language as a flat sequence or unordered collection of words or letters. A human reader must capture multiple levels of abstraction and meaning in order to formulate an understanding of a document. In this paper, we address the problem of developing approaches which are capable of working with extremely large and complex literary documents to perform Genre Identification. The task is to assign the literary classification to a full-length book belonging to a corpus of literature, where the works on average are well over 200,000 words long and genre is an abstract thematic concept. We introduce the Gutenberg Dataset for Genre Identification. Additionally, we present a study on how current deep learning models compare to traditional methods for this task. The results are presented as a baseline along with findings on how using an ensemble of chapters can significantly improve results in deep learning methods. The motivation behind the ensemble of chapters method is discussed as the compositionality of subtexts which make up a larger work and contribute to the overall genre.