StepBack resamples errors for efficient test-time reasoning
Original title: StepBack: Step-Level Error-Localized Resampling for Efficient Test-Time Reasoning
StepBack matches self-consistency accuracy while reducing token usage by up to 41.26%.
StepBack is a step-level error-localized resampling method for efficient test-time reasoning. Given draft trajectories, it localizes likely first error positions and resamples the detected error step and its suffix. The approach matches self-consistency accuracy on seven mathematical reasoning and multi-hop QA benchmarks.
- Matches SC accuracy on seven benchmarks
- Reduces token usage by up to 41.26%
- Combines with Adaptive Consistency
From the original article
1 min read at arxiv.orgSelf-consistency (SC) is a widely used decoding strategy for improving chain-of-thought reasoning, but it requires sampling many complete reasoning trajectories with a preset budget. Its efficient variants reduce the cost mainly by stopping full-trajectory sampling early or distributing the number of samples across questions.
By Yang Ouyang · arxiv.org
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.