Do Prompt-Based Models Really Understand the Meaning of their Prompts?

Recently, a boom of papers have shown extraordinary progress in few-shot learning with various prompt-based models. Such success can give the impression that prompts help models to learn faster in the same way that humans learn faster when provided with task instructions expressed in natural language. In this study, we experiment with over 30 prompts manually written for natural language inference (NLI). We find that models learn just as fast with many prompts that are intentionally irrelevant or even pathologically misleading as they do with instructively good prompts. Additionally, we find that model performance is more dependent on the choice of the LM target words (a.k.a. the verbalizer that converts LM vocabulary prediction to class labels) than on the text of the prompt itself. In sum, we find little evidence that suggests existing prompt-based models truly understand the meaning of their given prompts.

Which Linguist Invented the Lightbulb? Presupposition Verification for Question-Answering

Many Question-Answering (QA) datasets contain unanswerable questions, but their treatment in QA systems remains primitive. Our analysis of the Natural Questions (Kwiatkowski et al. 2019) dataset reveals that a substantial portion of unanswerable questions ($sim$21%) can be explained based on the presence of unverifiable presuppositions. We discuss the shortcomings of current models in handling such questions, and describe how an improved system could handle them. Through a user preference study, we demonstrate that the oracle behavior of our proposed system that provides responses based on presupposition failure is preferred over the oracle behavior of existing QA systems. Then we discuss how our proposed system could be implemented, presenting a novel framework that breaks down the problem into three steps: presupposition generation, presupposition verification and explanation generation. We report our progress in tackling each subproblem, and present a preliminary approach to integrating these steps into an existing QA system. We find that adding presuppositions and their verifiability to an existing model yields modest gains in downstream performance and unanswerability detection. The biggest bottleneck is the verification component, which needs to be substantially improved for the integrated system to approach ideal behavior -- even transfer from the best entailment models currently falls short.

A Visuospatial Dataset for Naturalistic Verb Learning

We introduce a new dataset for training and evaluating grounded language models. Our data is collected within a virtual reality environment and is designed to emulate the quality of language data to which a pre-verbal child is likely to have access: That is, naturalistic, spontaneous speech paired with richly grounded visuospatial context. We use the collected data to compare several distributional semantics models for verb learning. We evaluate neural models based on 2D (pixel) features as well as feature-engineered models based on 3D (symbolic, spatial) features, and show that neither modeling approach achieves satisfactory performance. Our results are consistent with evidence from child language acquisition that emphasizes the difficulty of learning verbs from naive distributional data. We discuss avenues for future work on cognitively-inspired grounded language learning, and release our corpus with the intent of facilitating research on the topic.

Self-play for Data Efficient Language Acquisition

When communicating, people behave consistently across conversational roles: People understand the words they say and are able to produce the words they hear. To date, artificial agents developed for language tasks have lacked such symmetry, meaning agents trained to produce language are unable to understand it and vice-versa. In this work, we exploit the symmetric nature of communication in order to improve both the efficiency and quality of language acquisition in learning agents. Specifically, we consider the setting in which an agent must learn to both understand and generate words in an existing language, but with the assumption that access to interaction with oracle speakers of the language is very limited. We show that using self-play as a substitute for direct supervision enables the agent to transfer its knowledge across roles (e.g. training as a listener but testing as a speaker) and make better inferences about the ground truth lexicon using only a handful of interactions with the oracle.

What Happens To BERT Embeddings During Fine-tuning?

While there has been much recent work studying how linguistic information is encoded in pre-trained sentence representations, comparatively little is understood about how these models change when adapted to solve downstream tasks. Using a suite of analysis techniques (probing classifiers, Representational Similarity Analysis, and model ablations), we investigate how fine-tuning affects the representations of the BERT model. We find that while fine-tuning necessarily makes significant changes, it does not lead to catastrophic forgetting of linguistic phenomena. We instead find that fine-tuning primarily affects the top layers of BERT, but with noteworthy variation across tasks. In particular, dependency parsing reconfigures most of the model, whereas SQuAD and MNLI appear to involve much shallower processing. Finally, we also find that fine-tuning has a weaker effect on representations of out-of-domain sentences, suggesting room for improvement in model generalization.