This paper presents a comprehensive examination of the impact of tokenization strategies and vocabulary sizes on the performance of Arabic language models in downstream natural language processing tasks. Our investigation focused on the effectiveness of four tokenizers across various tasks, including News Classification, Hate Speech Detection, Sentiment Analysis, and Natural Language Inference. Leveraging a diverse set of vocabulary sizes, we scrutinize the intricate interplay between tokenization approaches and model performance. The results reveal that Byte Pair Encoding (BPE) with Farasa outperforms other strategies in multiple tasks, underscoring the significance of morphological analysis in capturing the nuances of the Arabic language. However, challenges arise in sentiment analysis, where dialect specific segmentation issues impact model efficiency. Computational efficiency analysis demonstrates the stability of BPE with Farasa, suggesting its practical viability. Our study uncovers limited impacts of vocabulary size on model performance while keeping the model size unchanged. This is challenging the established beliefs about the relationship between vocabulary, model size, and downstream tasks, emphasizing the need for the study of models' size and their corresponding vocabulary size to generalize across domains and mitigate biases, particularly in dialect based datasets. Paper's recommendations include refining tokenization strategies to address dialect challenges, enhancing model robustness across diverse linguistic contexts, and expanding datasets to encompass the rich dialect based Arabic. This work not only advances our understanding of Arabic language models but also lays the foundation for responsible and ethical developments in natural language processing technologies tailored to the intricacies of the Arabic language.
A syntactic language model (SLM) incrementally generates a sentence with its syntactic tree in a left-to-right manner. We present Generative Pretrained Structured Transformers (GPST), an unsupervised SLM at scale capable of being pre-trained from scratch on raw texts with high parallelism. GPST circumvents the limitations of previous SLMs such as relying on gold trees and sequential training. It consists of two components, a usual SLM supervised by a uni-directional language modeling loss, and an additional composition model, which induces syntactic parse trees and computes constituent representations, supervised by a bi-directional language modeling loss. We propose a representation surrogate to enable joint parallel training of the two models in a hard-EM fashion. We pre-train GPST on OpenWebText, a corpus with $9$ billion tokens, and demonstrate the superiority of GPST over GPT-2 with a comparable size in numerous tasks covering both language understanding and language generation. Meanwhile, GPST also significantly outperforms existing unsupervised SLMs on left-to-right grammar induction, while holding a substantial acceleration on training.
Ensuring factual consistency between the summary and the original document is paramount in summarization tasks. Consequently, considerable effort has been dedicated to detecting inconsistencies. With the advent of Large Language Models (LLMs), recent studies have begun to leverage their advanced language understanding capabilities for inconsistency detection. However, early attempts have shown that LLMs underperform traditional models due to their limited ability to follow instructions and the absence of an effective detection methodology. In this study, we reassess summary inconsistency detection with LLMs, comparing the performances of GPT-3.5 and GPT-4. To advance research in LLM-based inconsistency detection, we propose SIFiD (Summary Inconsistency Detection with Filtered Document) that identify key sentences within documents by either employing natural language inference or measuring semantic similarity between summaries and documents.
Learning better sentence embeddings leads to improved performance for natural language understanding tasks including semantic textual similarity (STS) and natural language inference (NLI). As prior studies leverage large-scale labeled NLI datasets for fine-tuning masked language models to yield sentence embeddings, task performance for languages other than English is often left behind. In this study, we directly compared two data augmentation techniques as potential solutions for monolingual STS: (a) cross-lingual transfer that exploits English resources alone as training data to yield non-English sentence embeddings as zero-shot inference, and (b) machine translation that coverts English data into pseudo non-English training data in advance. In our experiments on monolingual STS in Japanese and Korean, we find that the two data techniques yield performance on par. Rather, we find a superiority of the Wikipedia domain over the NLI domain for these languages, in contrast to prior studies that focused on NLI as training data. Combining our findings, we demonstrate that the cross-lingual transfer of Wikipedia data exhibits improved performance, and that native Wikipedia data can further improve performance for monolingual STS.
Compositional Natural Language Inference has been explored to assess the true abilities of neural models to perform NLI. Yet, current evaluations assume models to have full access to all primitive inferences in advance, in contrast to humans that continuously acquire inference knowledge. In this paper, we introduce the Continual Compositional Generalization in Inference (C2Gen NLI) challenge, where a model continuously acquires knowledge of constituting primitive inference tasks as a basis for compositional inferences. We explore how continual learning affects compositional generalization in NLI, by designing a continual learning setup for compositional NLI inference tasks. Our experiments demonstrate that models fail to compositionally generalize in a continual scenario. To address this problem, we first benchmark various continual learning algorithms and verify their efficacy. We then further analyze C2Gen, focusing on how to order primitives and compositional inference types and examining correlations between subtasks. Our analyses show that by learning subtasks continuously while observing their dependencies and increasing degrees of difficulty, continual learning can enhance composition generalization ability.
In this new era of rapid AI development, especially in language processing, the demand for AI in the legal domain is increasingly critical. In the context where research in other languages such as English, Japanese, and Chinese has been well-established, we introduce the first fundamental research for the Vietnamese language in the legal domain: legal textual entailment recognition through the Vietnamese Language and Speech Processing workshop. In analyzing participants' results, we discuss certain linguistic aspects critical in the legal domain that pose challenges that need to be addressed.
Recent advancements in text summarization, particularly with the advent of Large Language Models (LLMs), have shown remarkable performance. However, a notable challenge persists as a substantial number of automatically-generated summaries exhibit factual inconsistencies, such as hallucinations. In response to this issue, various approaches for the evaluation of consistency for summarization have emerged. Yet, these newly-introduced metrics face several limitations, including lack of interpretability, focus on short document summaries (e.g., news articles), and computational impracticality, especially for LLM-based metrics. To address these shortcomings, we propose Factuality Evaluation of summarization based on Natural language Inference and Claim Extraction (FENICE), a more interpretable and efficient factuality-oriented metric. FENICE leverages an NLI-based alignment between information in the source document and a set of atomic facts, referred to as claims, extracted from the summary. Our metric sets a new state of the art on AGGREFACT, the de-facto benchmark for factuality evaluation. Moreover, we extend our evaluation to a more challenging setting by conducting a human annotation process of long-form summarization. In the hope of fostering research in summarization factuality evaluation, we release the code of our metric and our factuality annotations of long-form summarization at https://github.com/Babelscape/FENICE.
In Natural Language Generation (NLG), contemporary Large Language Models (LLMs) face several challenges, such as generating fluent yet inaccurate outputs and reliance on fluency-centric metrics. This often leads to neural networks exhibiting "hallucinations". The SHROOM challenge focuses on automatically identifying these hallucinations in the generated text. To tackle these issues, we introduce two key components, a data augmentation pipeline incorporating LLM-assisted pseudo-labelling and sentence rephrasing, and a voting ensemble from three models pre-trained on Natural Language Inference (NLI) tasks and fine-tuned on diverse datasets.
We introduce Bonito, an open-source model for conditional task generation that converts unannotated text into task-specific training datasets for instruction tuning. We aim to enable zero-shot task adaptation of large language models on users' specialized, private data. We train Bonito by fine-tuning a pretrained large language model on a new large-scale dataset with 1.65M examples created by remixing existing instruction tuning datasets into meta-templates. The meta-templates for a dataset produce training examples where the input is the unannotated text and the task attribute and the output consists of the instruction and the response. We use Bonito to generate synthetic tasks for seven datasets from specialized domains with unannotated text across three task types -- yes-no question answering, extractive question answering, and natural language inference -- and adapt language models. We show that Bonito significantly improves the average performance of pretrained and instruction tuned models over the de facto self supervised baseline. For example, adapting Mistral-Instruct-v2 and instruction tuned variants of Mistral and Llama2 with Bonito improves the strong zero-shot performance by 22.1 F1 points whereas the next word prediction objective undoes some of the benefits of instruction tuning and reduces the average performance by 0.8 F1 points. We conduct additional experiments with Bonito to understand the effects of the domain, the size of the training set, and the choice of alternative synthetic task generators. Overall, we show that learning with synthetic instruction tuning datasets is an effective way to adapt language models to new domains. The model, dataset, and code are available at https://github.com/BatsResearch/bonito.
The superior performance of supervised classification methods in the information extraction (IE) area heavily relies on a large amount of gold standard data. Recent zero-shot classification methods converted the task to other NLP tasks (e.g., textual entailment) and used off-the-shelf models of these NLP tasks to directly perform inference on the test data without using a large amount of IE annotation data. A potentially valuable by-product of these methods is the large-scale silver standard data, i.e., pseudo-labeled data by the off-the-shelf models of other NLP tasks. However, there is no further investigation into the use of these data. In this paper, we propose a new framework, Clean-LaVe, which aims to utilize silver standard data to enhance the zero-shot performance. Clean-LaVe includes four phases: (1) Obtaining silver data; (2) Identifying relatively clean data from silver data; (3) Finetuning the off-the-shelf model using clean data; (4) Inference on the test data. The experimental results show that Clean-LaVe can outperform the baseline by 5% and 6% on TACRED and Wiki80 dataset in the zero-shot relation classification task, and by 3%-7% on Smile (Korean and Polish) in the zero-shot cross-lingual relation classification task, and by 8% on ACE05-E+ in the zero-shot event argument classification task. The code is share in https://github.com/wjw136/Clean_LaVe.git.
We study existing approaches to leverage off-the-shelf Natural Language Inference (NLI) models for the evaluation of summary faithfulness and argue that these are sub-optimal due to the granularity level considered for premises and hypotheses. That is, the smaller content unit considered as hypothesis is a sentence and premises are made up of a fixed number of document sentences. We propose a novel approach, namely InFusE, that uses a variable premise size and simplifies summary sentences into shorter hypotheses. Departing from previous studies which focus on single short document summarisation, we analyse NLI based faithfulness evaluation for diverse summarisation tasks. We introduce DiverSumm, a new benchmark comprising long form summarisation (long documents and summaries) and diverse summarisation tasks (e.g., meeting and multi-document summarisation). In experiments, InFusE obtains superior performance across the different summarisation tasks. Our code and data are available at https://github.com/HJZnlp/infuse.
Program-of-Thought (PoT) replaces natural language-based Chain-of-Thought (CoT) as the most popular method in Large Language Models (LLMs) mathematical reasoning tasks by utilizing external tool calls to circumvent computational errors. However, our evaluation of the GPT-4 and Llama series reveals that using PoT introduces more reasoning errors, such as incorrect formulas or flawed logic, compared to CoT. To address this issue, we propose Human-Think Language (HTL), which leverages a suite of strategies that help integrate PoT and CoT, encompassing: (1) a new generation paradigm that uses full CoT reasoning to control code generation. (2) Focus Attention, that directs model attention to the CoT reasoning during PoT to generate more logical code. (3) reinforcement learning that utilizes the accuracy of both CoT and PoT responses as rewards to prevent repetitive reasoning steps in LLMs when solving difficult math problems. Our method achieves an average improvement of 6.5% on the Llama-Base model and 4.3% on the Mistral-Base model across 8 mathematical calculation datasets. It also shows significant effectiveness on five out-of-domain datasets by controlling the model's information flow, exhibiting strong transferability. Additionally, HTL shows the most significant improvement in non-mathematical natural language inference task, contributing to a unified reasoning task framework
Large language models (LLMs) exhibit powerful general intelligence across diverse scenarios, including their integration into chatbots. However, a vital challenge of LLM-based chatbots is that they may produce hallucinated content in responses, which significantly limits their applicability. Various efforts have been made to alleviate hallucination, such as retrieval augmented generation and reinforcement learning with human feedback, but most of them require additional training and data annotation. In this paper, we propose a novel post-hoc Citation-Enhanced Generation (CEG) approach combined with retrieval argumentation. Unlike previous studies that focus on preventing hallucinations during generation, our method addresses this issue in a post-hoc way. It incorporates a retrieval module to search for supporting documents relevant to the generated content, and employs a natural language inference-based citation generation module. Once the statements in the generated content lack of reference, our model can regenerate responses until all statements are supported by citations. Note that our method is a training-free plug-and-play plugin that is capable of various LLMs. Experiments on various hallucination-related datasets show our framework outperforms state-of-the-art methods in both hallucination detection and response regeneration on three benchmarks. Our codes and dataset will be publicly available.
Online hate detection suffers from biases incurred in data sampling, annotation, and model pre-training. Therefore, measuring the averaged performance over all examples in held-out test data is inadequate. Instead, we must identify specific model weaknesses and be informed when it is more likely to fail. A recent proposal in this direction is HateCheck, a suite for testing fine-grained model functionalities on synthesized data generated using templates of the kind "You are just a [slur] to me." However, despite enabling more detailed diagnostic insights, the HateCheck test cases are often generic and have simplistic sentence structures that do not match the real-world data. To address this limitation, we propose GPT-HateCheck, a framework to generate more diverse and realistic functional tests from scratch by instructing large language models (LLMs). We employ an additional natural language inference (NLI) model to verify the generations. Crowd-sourced annotation demonstrates that the generated test cases are of high quality. Using the new functional tests, we can uncover model weaknesses that would be overlooked using the original HateCheck dataset.
Decoder-based large language models (LLMs) have shown high performance on many tasks in natural language processing. This is also true for sentence embedding learning, where a decoder-based model, PromptEOL, has achieved the best performance on semantic textual similarity (STS) tasks. However, PromptEOL requires a manually annotated natural language inference (NLI) dataset for fine-tuning. We aim to improve sentence embeddings without using large manually annotated datasets by automatically generating an NLI dataset with an LLM and using it for fine-tuning of PromptEOL. To achieve this, we explore methods of data generation suitable for sentence embedding learning in this study. Specifically, we will focus on automatic dataset generation through few-shot learning and explore the appropriate methods to leverage few-shot examples. Experimental results on the STS tasks demonstrate that our approach outperforms existing models in settings without large manually annotated datasets.
Recent language models enable new opportunities for structured reasoning with text, such as the construction of intuitive, proof-like textual entailment trees without relying on brittle formal logic. However, progress in this direction has been hampered by a long-standing lack of a clear protocol for determining what valid compositional entailment is. This absence causes noisy datasets and limited performance gains by modern neuro-symbolic engines. To address these problems, we formulate a consistent and theoretically grounded approach to annotating decompositional entailment and evaluate its impact on LLM-based textual inference. We find that our new dataset, RDTE (Recognizing Decompositional Textual Entailment), has a substantially higher internal consistency (+9%) than prior decompositional entailment datasets. We also find that training an RDTE-oriented entailment classifier via knowledge distillation and employing it in an entailment tree reasoning engine significantly improves both accuracy and proof quality, illustrating the practical benefit of this advance for textual inference.
Factual inconsistencies pose a significant hurdle for the faithful summarization by generative models. While a major direction to enhance inconsistency detection is to derive stronger Natural Language Inference (NLI) models, we propose an orthogonal aspect that underscores the importance of incorporating task-specific taxonomy into the inference. To this end, we consolidate key error types of inconsistent facts in summaries, and incorporate them to facilitate both the zero-shot and supervised paradigms of LLMs. Extensive experiments on ten datasets of five distinct domains suggest that, zero-shot LLM inference could benefit from the explicit solution space depicted by the error type taxonomy, and achieves state-of-the-art performance overall, surpassing specialized non-LLM baselines, as well as recent LLM baselines. We further distill models that fuse the taxonomy into parameters through our designed prompt completions and supervised training strategies, efficiently substituting state-of-the-art zero-shot inference with much larger LLMs.
Natural Language Inference (NLI) remains an important benchmark task for LLMs. NLI datasets are a springboard for transfer learning to other semantic tasks, and NLI models are standard tools for identifying the faithfulness of model-generated text. There are several large scale NLI datasets today, and models have improved greatly by hill-climbing on these collections. Yet their realistic performance on out-of-distribution/domain data is less well-understood. We explore the opportunity for synthetic high-quality datasets to adapt NLI models for zero-shot use in downstream applications across new and unseen text domains. We demonstrate a new approach for generating NLI data in diverse domains and lengths, so far not covered by existing training sets. The resulting examples have meaningful premises, the hypotheses are formed in creative ways rather than simple edits to a few premise tokens, and the labels have high accuracy. We show that models trained on this data ($685$K synthetic examples) have the best generalization to completely new downstream test settings. On the TRUE benchmark, a T5-small model trained with our data improves around $7\%$ on average compared to training on the best alternative dataset. The improvements are more pronounced for smaller models, while still meaningful on a T5 XXL model. We also demonstrate gains on test sets when in-domain training data is augmented with our domain-general synthetic data.
Credal sets are sets of probability distributions that are considered as candidates for an imprecisely known ground-truth distribution. In machine learning, they have recently attracted attention as an appealing formalism for uncertainty representation, in particular due to their ability to represent both the aleatoric and epistemic uncertainty in a prediction. However, the design of methods for learning credal set predictors remains a challenging problem. In this paper, we make use of conformal prediction for this purpose. More specifically, we propose a method for predicting credal sets in the classification task, given training data labeled by probability distributions. Since our method inherits the coverage guarantees of conformal prediction, our conformal credal sets are guaranteed to be valid with high probability (without any assumptions on model or distribution). We demonstrate the applicability of our method to natural language inference, a highly ambiguous natural language task where it is common to obtain multiple annotations per example.
Despite great performance on many tasks, language models (LMs) still struggle with reasoning, sometimes providing responses that cannot possibly be true because they stem from logical incoherence. We call such responses \textit{strong hallucinations} and prove that they follow from an LM's computation of its internal representations for logical operators and outputs from those representations. Focusing on negation, we provide a novel solution in which negation is treated not as another element of a latent representation, but as \textit{an operation over an LM's latent representations that constrains how they may evolve}. We show that our approach improves model performance in cloze prompting and natural language inference tasks with negation without requiring training on sparse negative data.