Teaching Small Language Models to Reason
This addresses the challenge of making reasoning capabilities accessible in smaller, more efficient models, which is incremental as it builds on existing chain of thought prompting methods.
The paper tackles the problem of enabling small language models to perform reasoning tasks by using knowledge distillation from a large teacher model, resulting in improved accuracy on datasets like GSM8K from 8.11% to 21.99%.
Chain of thought prompting successfully improves the reasoning capabilities of large language models, achieving state of the art results on a range of datasets. However, these reasoning capabilities only appear to emerge in models with a size of over 100 billion parameters. In this paper, we explore the transfer of such reasoning capabilities to models with less than 100 billion parameters via knowledge distillation. Specifically, we finetune a student model on the chain of thought outputs generated by a larger teacher model. Our experiments show that the proposed method improves task performance across arithmetic, commonsense and symbolic reasoning datasets. For example, the accuracy of T5 XXL on GSM8K improves from 8.11% to 21.99% when finetuned on PaLM-540B generated chains of thought.