Language models (LMs) may appear insensitive to word order changes in natural language understanding (NLU) tasks. In this paper, we propose that linguistic redundancy can explain this phenomenon, whereby word order and other linguistic cues such as case markers provide overlapping and thus redundant information. Our hypothesis is that models exhibit insensitivity to word order when the order provides redundant information, and the degree of insensitivity varies across tasks. We quantify how informative word order is using mutual information (MI) between unscrambled and scrambled sentences. Our results show the effect that the less informative word order is, the more consistent the model's predictions are between unscrambled and scrambled sentences. We also find that the effect varies across tasks: for some tasks, like SST-2, LMs' prediction is almost always consistent with the original one even if the Pointwise-MI (PMI) changes, while for others, like RTE, the consistency is near random when the PMI gets lower, i.e., word order is really important.
When classifying grammatical role, BERT doesn't care about word order... except when it matters
Stanford University(斯坦福大学);University of California, Irvine(加州大学欧文分校);The University of Texas at Austin(得克萨斯大学奥斯汀分校)
因为意义通常可以单独从词汇语义中推断出来,所以词序在自然语言中通常是一个多余的线索。例如,单词chopped、chef和onion更有可能用来表达“The chef chopped the onion”,而不是“The onion chopped the chef”。“最近的工作表明,大语言模型是令人惊讶的词序不变,但至关重要的是,它在很大程度上考虑了自然的原型输入,其中组合意义大多与词汇预期相匹配。为了克服这种混淆,我们探讨语法角色表示在英语BERT和GPT-2,在词汇的期望是不够的情况下,词序知识是必要的正确分类。这种非原型的例子是自然出现的英语句子,有无生命的主体或有生命的物体,或者是我们系统地交换论点来组成像“洋葱切碎了厨师”这样的句子。我们发现,虽然早期层嵌入主要是词汇,词序实际上是至关重要的,在定义后一层表示的单词在语义上的非原型的位置。我们的实验隔离了词序对语境化过程的影响,并强调了模型如何在不常见但关键的情况下使用语境。
Over-representation of phonological features in basic vocabulary doesn't replicate when controlling for spatial and phylogenetic effects
Department of Linguistic and Cultural Evolution, Max-Planck Institute for Evolutionary Anthropology(语言与文化进化系,马克斯·普朗克进化人类学研究所);University of Passau, Chair of Multilingual Computational Linguistics(帕绍大学,多语计算语言学系)
Order Doesn't Matter, But Reasoning Does: Training LLMs with Order-Centric Augmentation
Shanghai Key Laboratory of Data Science, School of Computer Science, Fudan University(上海数据科学关键实验室,计算机科学学院,复旦大学);School of Data Science, Fudan University(数据科学学院,复旦大学);Ant Group(蚂蚁集团)
Dept. Computing Science & Alberta Machine Intelligence Institute (Amii), University of Alberta(计算科学系及阿尔伯塔人工智能研究所(Amii),阿尔伯塔大学);Canada CIFAR AI Chair, Amii(加拿大CIFAR人工智能主席,Amii)