Unlocking the Black Box of Latent Reasoning: An Interpretability-Guided Approach to Intervention (2026.acl-long)
Copied to clipboard
| Challenge: | Existing methods for large language models (LLMs) lack a coherent representation of reasoning steps. |
| Approach: | They propose a set of latent reasoning interventions that enable latent thinking and decode-time interventions that refine the latent process by imposing the identified geometric and semantic priors. |
| Outcome: | The proposed models unlock latent capabilities and improve reasoning accuracy without any parameter updates. |
Similar Papers
Explainable Chain-of-Thought Reasoning: An Empirical Analysis on State-Aware Reasoning Dynamics (2025.findings-emnlp)
Copied to clipboard
Sheldon Yu, Yuxin Xiong, Junda Wu, Xintong Li, Tong Yu, Xiang Chen, Ritwik Sinha, Jingbo Shang, Julian McAuley
| Challenge: | Recent advances in chain-of-thought prompting have demonstrated the ability of large language models to perform multi-step reasoning. |
| Approach: | They propose a framework to analyze latent dynamics of CoT trajectories for interpretability . they segment generated CoT into discrete reasoning steps and abstract each step into a spectral embedding based on token-level Gram matrices . |
| Outcome: | The proposed framework segments generated CoT steps into discrete reasoning steps, abstracts each step into a spectral embedding based on token-level Gram matrices, and clusters these embeddements into semantically meaningful latent states. |
How Interpretable are Reasoning Explanations from Prompting Large Language Models? (2024.findings-naacl)
Copied to clipboard
| Challenge: | Prompt Engineering has garnered significant attention for enhancing the performance of large language models across a multitude of tasks. |
| Approach: | They propose a simple prompting technique that yields more than 70% improvement in interpretability. |
| Outcome: | The proposed method improves interpretability by 70% across multiple dimensions. |
Assessing Step-by-Step Reasoning against Lexical Negation: A Case Study on Syllogism (2023.emnlp-main)
Copied to clipboard
| Challenge: | Large language models (LLMs) take advantage of step-by-step reasoning instructions . negation is a core linguistic phenomenon that is difficult to process . |
| Approach: | They examine the step-by-step reasoning ability of large language models with a focus on negation . negation is a core linguistic phenomenon that is difficult to process . |
| Outcome: | The proposed models perform better when using chain-of-thought prompting . the results highlight unique limitations in each LLM family . |
Revitalizing Black-Box Interpretability: Actionable Interpretability for LLMs via Proxy Models (2026.acl-long)
Copied to clipboard
| Challenge: | Applying model-agnostic explanations to Large Language Models is hindered by prohibitive computational costs rendering them dormant for real-world applications. |
| Approach: | They propose a budget-friendly proxy framework that leverages efficient models to approximate the decision boundaries of expensive Large Language Models. |
| Outcome: | The proposed framework achieves over 90% fidelity with only 9.5% of the oracle’s cost and is open-source to facilitate future research. |
Resolving the Security-Auditability Dilemma with Auditable Latent Chain-of-Thought Alignment (2026.acl-long)
Copied to clipboard
| Challenge: | Extensive experiments show that ALCA reduces the success rate of adaptive jailbreak attacks by over 40% compared to strong baselines, while preserving performance. |
| Approach: | They propose a framework that decouples internal reasoning from external output and allows the model to reconstruct its latent reasoning into human-readable text for supervision under specific guidance. |
| Outcome: | The proposed framework reduces the success rate of adaptive jailbreak attacks by over 40% compared to baselines while preserving performance. |
Understanding Jailbreak Success: A Study of Latent Space Dynamics in Large Language Models (2026.eacl-long)
Copied to clipboard
| Challenge: | Emerging jailbreaking techniques can still elicit unsafe outputs, presenting an ongoing challenge for model alignment. |
| Approach: | They propose to extract a jailbreak vector from a single class of jailbreaks that works to mitigate jailbreak effectiveness from other, semantically-dissimilar classes. |
| Outcome: | The proposed jailbreak vectors show that they reduce harmfulness in most models, and that they are similar in geometric and effect similarity. |
From Insight to Action: A Novel Framework for Interpretability-Guided Data Selection in Large Language Models (2026.acl-long)
Copied to clipboard
Ling Shi, Xinwei Wu, Xiaohu Zhao, Hao Wang, Heng Liu, Yangyang Liu, Linlong Xu, Longyue Wang, Deyi Xiong, Weihua Luo
| Challenge: | Recent research in mechanistic interpretability has revealed that Large Language models contain disentangled, human-understandable components. |
| Approach: | They propose a framework that first identifies causal task features through frequency recall and interventional filtering, then selects “Feature-Resonant Data” that maximally activates task features for fine-tuning. |
| Outcome: | The proposed framework outperforms existing models on mathematical reasoning, summarization, and translation tasks while using only 50% of the data. |
“Well, Keep Thinking”: Enhancing LLM Reasoning with Adaptive Injection Decoding (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) exhibit strong reasoning abilities, often attributed to few-shot or zero-shot Chain-of-Thought (CoT) prompting. |
| Approach: | They propose a decoding strategy that nudges LLMs to continue reasoning, thereby preventing immature reasoning processes. |
| Outcome: | The proposed method significantly improves LLM reasoning capabilities on diverse reasoning benchmarks. |
Improving Large Language Models Function Calling and Interpretability via Guided-Structured Templates (2025.emnlp-main)
Copied to clipboard
Hy Dang, Tianyi Liu, Zhuofeng Wu, Jingfeng Yang, Haoming Jiang, Tao Yang, Pei Chen, Zhengyang Wang, Helen Wang, Huasheng Li, Bing Yin, Meng Jiang
| Challenge: | Large language models (LLMs) have strong reasoning and tool-use capabilities, yet fail in real-world tool-interactions due to incorrect parameterization, poor tool selection, or misinterpretation of user intent. |
| Approach: | They propose a curriculum-inspired framework that leverages structured reasoning templates to guide LLMs through more deliberate step-by-step instructions for generating function calls. |
| Outcome: | The proposed framework reduces tool-use errors and improves interpretability and transparency of tool-using agents. |
Valid Necessary: Diagnosing Latent Inefficiency in Chain-of-Thought (2026.findings-acl)
Copied to clipboard
| Challenge: | Existing reasoning step evaluators fail to distinguish “valid but inefficient” reasoning steps from necessary reasoning. |
| Approach: | They propose a training-free metric that identifies low-utility steps and a post-hoc compression strategy to quantify their impact on token usage. |
| Outcome: | The proposed metric reduces token consumption by 31–53% while maintaining accuracy at substantially higher compression rates. |