Can Transformers Reason in Fragments of Natural Language?
This addresses the problem of evaluating reasoning capabilities in NLP models for researchers, revealing limitations in current methods.
The paper investigates whether transformer-based language models can detect formally valid inferences in controlled fragments of natural language with increasing complexity, finding they perform well but overfit to superficial patterns rather than learning underlying logical principles.
State-of-the-art deep-learning-based approaches to Natural Language Processing (NLP) are credited with various capabilities that involve reasoning with natural language texts. In this paper we carry out a large-scale empirical study investigating the detection of formally valid inferences in controlled fragments of natural language for which the satisfiability problem becomes increasingly complex. We find that, while transformer-based language models perform surprisingly well in these scenarios, a deeper analysis re-veals that they appear to overfit to superficial patterns in the data rather than acquiring the logical principles governing the reasoning in these fragments.