Learn Beyond The Answer: Training Language Models with Reflection for Mathematical Reasoning
This work addresses the need for better mathematical reasoning in language models, offering a novel approach that complements existing data augmentation techniques, though it appears incremental in nature.
The paper tackles the problem of enhancing language models' mathematical reasoning by introducing reflective augmentation, which embeds problem reflection into training instances to foster deeper understanding, resulting in improved performance in both standard and complex scenarios.
Supervised fine-tuning enhances the problem-solving abilities of language models across various mathematical reasoning tasks. To maximize such benefits, existing research focuses on broadening the training set with various data augmentation techniques, which is effective for standard single-round question-answering settings. Our work introduces a novel technique aimed at cultivating a deeper understanding of the training problems at hand, enhancing performance not only in standard settings but also in more complex scenarios that require reflective thinking. Specifically, we propose reflective augmentation, a method that embeds problem reflection into each training instance. It trains the model to consider alternative perspectives and engage with abstractions and analogies, thereby fostering a thorough comprehension through reflective reasoning. Extensive experiments validate the achievement of our aim, underscoring the unique advantages of our method and its complementary nature relative to existing augmentation techniques.