Step-KTO: Optimizing Mathematical Reasoning through Stepwise Binary Feedback
This addresses the need for more interpretable and dependable reasoning in LLMs for mathematical applications, representing an incremental advance over existing methods.
The paper tackles the problem of unreliable reasoning processes in large language models for mathematical tasks by introducing Step-KTO, a training framework that uses stepwise binary feedback, resulting in significant improvements in both final answer accuracy and intermediate reasoning quality on benchmarks like MATH-500.
Large language models (LLMs) have recently demonstrated remarkable success in mathematical reasoning. Despite progress in methods like chain-of-thought prompting and self-consistency sampling, these advances often focus on final correctness without ensuring that the underlying reasoning process is coherent and reliable. This paper introduces Step-KTO, a training framework that combines process-level and outcome-level binary feedback to guide LLMs toward more trustworthy reasoning trajectories. By providing binary evaluations for both the intermediate reasoning steps and the final answer, Step-KTO encourages the model to adhere to logical progressions rather than relying on superficial shortcuts. Our experiments on challenging mathematical benchmarks show that Step-KTO significantly improves both final answer accuracy and the quality of intermediate reasoning steps. For example, on the MATH-500 dataset, Step-KTO achieves a notable improvement in Pass@1 accuracy over strong baselines. These results highlight the promise of integrating stepwise process feedback into LLM training, paving the way toward more interpretable and dependable reasoning capabilities.