Analyzing Commonsense Emergence in Few-shot Knowledge Models
This research provides insights into the learning mechanisms of commonsense knowledge in LMs, which is important for researchers developing more efficient and effective knowledge models.
This paper investigates whether commonsense knowledge in language models (LMs) is learned during pretraining or fine-tuning on knowledge graph (KG) tuples. The authors found that LMs rapidly adapt from limited KG examples, suggesting fine-tuning primarily learns an interface to knowledge acquired during pretraining.
Recently, commonsense knowledge models - pretrained language models (LM) fine-tuned on knowledge graph (KG) tuples - showed that considerable amounts of commonsense knowledge can be encoded in the parameters of large language models. However, as parallel studies show that LMs are poor hypothesizers of declarative commonsense relationships on their own, it remains unclear whether this knowledge is learned during pretraining or from fine-tuning on KG examples. To investigate this question, we train commonsense knowledge models in few-shot settings to study the emergence of their commonsense representation abilities. Our results show that commonsense knowledge models can rapidly adapt from limited examples, indicating that KG fine-tuning serves to learn an interface to encoded knowledge learned during pretraining. Importantly, our analysis of absolute, angular, and distributional parameter changes during few-shot fine-tuning provides novel insights into how this interface is learned.