Leap-Of-Thought: Teaching Pre-Trained Models to Systematically Reason Over Implicit Knowledge
This work addresses the challenge of controlled reasoning in open-domain AI systems, offering an incremental step towards models that can improve through user interaction.
The paper tackled the problem of enabling pre-trained language models to systematically reason over both implicit knowledge and explicit statements, demonstrating that models can be trained to perform inference involving taxonomic and world knowledge, chaining, and counting, with results showing generalization beyond the training distribution.
To what extent can a neural network systematically reason over symbolic facts? Evidence suggests that large pre-trained language models (LMs) acquire some reasoning capacity, but this ability is difficult to control. Recently, it has been shown that Transformer-based models succeed in consistent reasoning over explicit symbolic facts, under a "closed-world" assumption. However, in an open-domain setup, it is desirable to tap into the vast reservoir of implicit knowledge already encoded in the parameters of pre-trained LMs. In this work, we provide a first demonstration that LMs can be trained to reliably perform systematic reasoning combining both implicit, pre-trained knowledge and explicit natural language statements. To do this, we describe a procedure for automatically generating datasets that teach a model new reasoning skills, and demonstrate that models learn to effectively perform inference which involves implicit taxonomic and world knowledge, chaining and counting. Finally, we show that "teaching" models to reason generalizes beyond the training distribution: they successfully compose the usage of multiple reasoning skills in single examples. Our work paves a path towards open-domain systems that constantly improve by interacting with users who can instantly correct a model by adding simple natural language statements.