A possible explanation for the impressive performance of masked language model (MLM) pre-training is that such models have learned to represent the syntactic structures prevalent in classical NLP pipelines. In this paper, we propose a different explanation: MLMs succeed on downstream tasks almost entirely due to their ability to model higher-order word co-occurrence statistics. To demonstrate this, we pre-train MLMs on sentences with randomly shuffled word order, and show that these models still achieve high accuracy after fine-tuning on many downstream tasks -- including on tasks specifically designed to be challenging for models that ignore word order. Our models perform surprisingly well according to some parametric syntactic probes, indicating possible deficiencies in how we test representations for syntactic information. Overall, our results show that purely distributional information largely explains the success of pre-training, and underscore the importance of curating challenging evaluation datasets that require deeper linguistic knowledge.
Less but Better: Generalization Enhancement of Ordinal Embedding via Distributional Margin
Institute of Information Engineering, Chinese Academy of Sciences（中国科学院信息工程研究所）；School of Cyber Security, University of Chinese Academy of Sciences（中国科学院大学网络空间安全学院）；Institute of Computing Technology, Chinese Academy of Sciences（中国科学院计算技术研究所）
Humans acquire language continually with much more limited access to data samples at a time, as compared to contemporary NLP systems. To study this human-like language acquisition ability, we present VisCOLL, a visually grounded language learning task, which simulates the continual acquisition of compositional phrases from streaming visual scenes. In the task, models are trained on a paired image-caption stream which has shifting object distribution; while being constantly evaluated by a visually-grounded masked language prediction task on held-out test sets. VisCOLL compounds the challenges o…
This paper studies the problem of injecting factual knowledge into large pre-trained language models. We train adapter modules on parts of the ConceptNet knowledge graph using the masked language modeling objective and evaluate the success of the method by a series of probing experiments on the LAMA probe. Mean P@K curves for different configurations indicate that the technique is effective, increasing the performance on subsets of the LAMA probe for large values of k by adding as little as 2.1% additional parameters to the original models.
Bocconi University（博科尼大学）；Coveo；Coveo AI Labs（Coveo人工智能实验室）
Word embeddings (e.g., word2vec) have been applied successfully to eCommerce products through~\textit{prod2vec}. Inspired by the recent performance improvements on several NLP tasks brought by contextualized embeddings, we propose to transfer BERT-like architectures to eCommerce: our model -- ~\textit{Prod2BERT} -- is trained to generate representations of products through masked session modeling. Through extensive experiments over multiple shops, different tasks, and a range of design choices, we systematically compare the accuracy of~\textit{Prod2BERT} and~\textit{prod2vec} embeddings: while…
Language models have proven to be very useful when adapted to specific domains. Nonetheless, little research has been done on the adaptation of domain-specific BERT models in the French language. In this paper, we focus on creating a language model adapted to French legal text with the goal of helping law professionals. We conclude that some specific tasks do not benefit from generic language models pre-trained on large amounts of data. We explore the use of smaller architectures in domain-specific sub-languages and their benefits for French legal text. We prove that domain-specific pre-traine…
University of Washington（华盛顿大学）；Amazon（亚马逊）；Allen Institute for Artificial Intelligence（艾伦人工智能研究所）
Contrastively trained language-image models such as CLIP, ALIGN, and BASIC have demonstrated unprecedented robustness to multiple challenging natural distribution shifts. Since these language-image models differ from previous training approaches in several ways, an important question is what causes the large robustness gains. We answer this question via a systematic experimental investigation. Concretely, we study five different possible causes for the robustness gains: (i) the training set size, (ii) the training distribution, (iii) language supervision at training time, (iv) language supervi…
Lexical semantic typology has identified important cross-linguistic generalizations about the variation and commonalities in polysemy patterns---how languages package up meanings into words. Recent computational research has enabled investigation of lexical semantics at a much larger scale, but little work has explored lexical typology across semantic domains, nor the factors that influence cross-linguistic similarities. We present a novel computational framework that quantifies semantic affinity, the cross-linguistic similarity of lexical semantics for a concept. Our approach defines a common…
Cross-modal alignment is essential for vision-language pre-training (VLP) models to learn the correct corresponding information across different modalities. For this purpose, inspired by the success of masked language modeling (MLM) tasks in the NLP pre-training area, numerous masked modeling tasks have been proposed for VLP to further promote cross-modal interactions. The core idea of previous masked modeling tasks is to focus on reconstructing the masked tokens based on visible context for learning local-to-local alignment. However, most of them pay little attention to the global semantic fe…
A fundamental characteristic of natural language is the high rate at which speakers produce novel expressions. Because of this novelty, a heavy-tail of rare events accounts for a significant amount of the total probability mass of distributions in language (Baayen, 2001). Standard language modeling metrics such as perplexity quantify the performance of language models (LM) in aggregate. As a result, we have relatively little understanding of whether neural LMs accurately estimate the probability of sequences in this heavy-tail of rare events. To address this gap, we develop a controlled evalua…
Multi-Granularity Tibetan Textual Adversarial Attack Method Based on Masked Language Model
School of Information Science and Technology, Tibet University（西藏大学信息科学技术学院）；Collaborative Innovation Center for Tibet Informatization by MOE and Tibet Autonomous Region（教育部与西藏自治区西藏信息化协同创新中心）；Engineering Research Center of Tibetan Information Technology, Ministry of Education（教育部藏信息技术工程研究中心）
Learnability-Informed Fine-Tuning of Diffusion Language Models
Department of Computer Science and Engineering, Texas A\&M University, College Station, TX, USA（计算机科学与工程系，德克萨斯A&M大学，College Station, TX, USA）；Department of Electrical and Computer Engineering, Texas A\&M University, College Station, TX, USA（电气与计算机工程系，德克萨斯A&M大学，College Station, TX, USA）
Con-ReCall: Detecting Pre-training Data in LLMs via Contrastive Decoding
National University of Singapore（新加坡国立大学）；University of California, Merced（加州大学默塞德分校）；University of California, Los Angeles（加州大学洛杉矶分校）；University of Queensland（昆士兰大学）
In this work, we explore joint energy-based model (EBM) training during the finetuning of pretrained text encoders (e.g., Roberta) for natural language understanding (NLU) tasks. Our experiments show that EBM training can help the model reach a better calibration that is competitive to strong baselines, with little or no loss in accuracy. We discuss three variants of energy functions (namely scalar, hidden, and sharp-hidden) that can be defined on top of a text encoder, and compare them in experiments. Due to the discreteness of text data, we adopt noise contrastive estimation (NCE) to train t…
