First-Step Advantage: Importance of Starting Right in Multi-Step Math Reasoning

Language models can solve complex reasoning tasks better by learning to generate rationales for their predictions. Often these models know how to solve a task, but their auto-regressive decoding nature leads to incorrect results if they start incorrectly. We observe that smaller models, in particular, when corrected, can solve a task that they would have otherwise struggled with. We demonstrate this phenomenon by using a larger model to guide smaller models, which leads to significantly improved performance (up to +24 points on the GSM8K dataset by 7B models). Furthermore, to assist smaller models in initiating the starting step correctly, we propose QuestCoT, where a smaller model first asks itself how to start, before proceeding with a chain of reasoning. On various multistep mathematical reasoning datasets for multiple smaller models, we show that getting the right start can lead to significant performance gains across all models (gains of up to +6 points on GSM8K, +9 on SVAMP, +5 on ASDiv, and +7 on MultiArith). LM Natalia sold 48+72=120 clips in total. Natalia sold clips to 48 of her friends in April.... How many clips did Natalia sell altogether in April and May? INPUT Natalia has 48 clips to sell. April is .. Natalia sold 48/2=24 clips in May. In ..