Step Back to Leap Forward: Self-Backtracking for Boosting Reasoning of Language Models

Open in new window