analiza aiqbench
APCD introduces a novel inference-time framework that enhances factual reliability in large language models by dynamically expanding decoding paths at points of high uncertainty. This method improves accuracy across general and medical benchmarks without requiring model retraining, addressing critical limitations in single-trajectory generation approaches. The research advances the field of reliable LLM deployment by balancing diversity and coherence through adaptive multi-path contrastive mechanisms.
APCD: Adaptive Path-Contrastive Decoding for Reliable Large Language Model Generation
arXiv:2605.09492v3 Announce Type: replace Abstract: Reliable text generation is critical for deploying large language models (LLMs) in real-world applications, particularly in high-stakes domains such as medicine. To improve factual reliability, various inference-time methods have been proposed, including logit-level methods that modify token probability distributions and representation-level methods that manipulate intermediate model representations. However, most existing approaches operate on a single decoding trajectory, limiting their ability to explore alternative reasoning paths and making them susceptible to error accumulation. To address this limitation, we propose Adaptive Path-Contrastive Decoding (APCD), an adaptive multi-path contrastive decoding framework that improves factual reliability without model retraining or fine-tuning. APCD comprises two key components: Entropy-Driven Path Expansion, which adaptively expands the decoding process only at high-uncertainty decision points, and Divergence-Aware Path Contrast, which dynamically regulates contrastive interactions among parallel decoding paths based on their distributional divergence to balance diversity and coherence. We evaluate APCD on four LLM backbones across eight benchmarks spanning both general-domain and medical question answering tasks. Experimental results demonstrate that APCD consistently outperforms strong inference-time baselines in factual accuracy while maintaining competitive inference efficiency. These results demonstrate the robustness and generalizability of APCD across diverse models and tasks, highlighting its effectiveness as a practical multi-path decoding framework for reliable LLM deployment, particularly in high-stakes domains such as medicine. Code is available at https://github.com/zty-king/APCD.
Abstract:Reliable text generation is critical for deploying large language models (LLMs) in real-world applications, particularly in high-stakes domains such as medicine. To improve factual reliability, various inference-time methods have been proposed, including logit-level methods that modify token probability distributions and representation-level methods that manipulate intermediate model representations. However, most existing approaches operate on a single decoding trajectory, limiting their ability to explore alternative reasoning paths and making them susceptible to error accumulation. To address this limitation, we propose Adaptive Path-Contrastive Decoding (APCD), an adaptive multi-path contrastive decoding framework that improves factual reliability without model retraining or fine-tuning. APCD comprises two key components: Entropy-Driven Path Expansion, which adaptively expands the decoding process only at high-uncertainty decision points, and Divergence-Aware Path Contrast, which dynamically regulates contrastive interactions among parallel decoding paths based on their distributional divergence to balance diversity and coherence. We evaluate APCD on four LLM backbones across eight benchmarks spanning both general-domain and medical question answering tasks. Experimental results demonstrate that APCD consistently outperforms strong inference-time baselines in factual accuracy while maintaining competitive inference efficiency. These results demonstrate the robustness and generalizability of APCD across diverse models and tasks, highlighting its effectiveness as a practical multi-path decoding framework for reliable LLM deployment, particularly in high-stakes domains such as medicine. Code is available at this https URL.
This is an extended journal version submitted to ESWA. It builds upon a previously withdrawn conference manuscript (ACL format). The core research work remains unchanged, with substantial extensions including additional ablation experiments and deeper analysis to meet journal requirements
Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
arXiv:2605.09492 [cs.CL]
(or arXiv:2605.09492v3 [cs.CL] for this version)
Text integral sub licența deschisă a publicației · 270 cuvinte
Subiecte conexe
Adaptive Parallel Reasoning: The Next Paradigm in Efficient Inference Scaling
BAIR Blog · 8 mai 2026
TAD: Token-Adaptive Contrastive Decoding with Confidence-Guided Gating for Hallucination Mitigation in Large Audio-Language Models
arXiv · 9 sept. 2026
[AINews] High Taste Testers Ep. 1
Latent Space · 17 oct. 2026
Enabling Speculative Speculative Decoding on MI300X
AMD · 29 mai 2026