Unleashing the True Potential of LLMs: A Feedback-Triggered Self-Correction with Long-Term Multipath Decoding

2025-09-09

AI Summary

Large language models (LLMs) frequently generate erroneous content during reasoning, and existing self-correction approaches suffer from unreliable error localization signals and myopic inference caused by autoregressive, token-by-token decoding. To address these limitations, we propose Feedback-Triggered Regeneration with Long-Term Multi-path decoding (FTR-LTM), a novel self-correction framework featuring two core innovations: (1) a feedback-triggered regeneration mechanism that leverages external negative feedback—rather than error-prone internal self-assessment—to halt error propagation; and (2) long-term multi-path decoding, which defers sequence-level evaluation to enable parallel exploration of multiple reasoning paths, thereby overcoming the constraints of conventional single-step prediction. Evaluated on mathematical reasoning and code generation benchmarks, FTR-LTM consistently outperforms state-of-the-art prompt-based self-correction methods, delivering robust and sustained performance gains.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: Generation
• Search and Optimization: Learning to Search

Application Category

• Search and Retrieval-Augmented AI: Large language models for search
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Large Language Models (LLMs) have achieved remarkable performance across diverse tasks, yet their susceptibility to generating incorrect content during inference remains a critical unsolved challenge. While self-correction methods offer potential solutions, their effectiveness is hindered by two inherent limitations: (1) the absence of reliable guidance signals for error localization, and (2) the restricted reasoning depth imposed by conventional next-token decoding paradigms. To address these issues, we propose Feedback-Triggered Regeneration (FTR), a novel framework that synergizes user feedback with enhanced decoding dynamics. Specifically, FTR activates response regeneration only upon receiving negative user feedback, thereby circumventing error propagation from faulty self-assessment while preserving originally correct outputs. Furthermore, we introduce Long-Term Multipath (LTM) decoding, which enables systematic exploration of multiple reasoning trajectories through delayed sequence evaluation, effectively overcoming the myopic decision-making characteristic of standard next-token prediction. Extensive experiments on mathematical reasoning and code generation benchmarks demonstrate that our framework achieves consistent and significant improvements over state-of-the-art prompt-based self-correction methods.

Problem

• LLMs generate incorrect content without reliable error guidance
• Self-correction limited by shallow next-token decoding paradigms
• Need systematic exploration of multiple reasoning trajectories

Innovation

• Feedback-triggered regeneration for error correction
• Long-term multipath decoding for reasoning depth
• User feedback integration to guide regeneration