Cover Image

Embedding Perturbation may Better Reflect Intermediate-Step Uncertainty in LLM Reasoning

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Wen, Qihao, Wang, Jiahao, Nan, Yang, He, Pengfei, Tandon, Ravi, Xu, Han
Format:	Preprint
Published:	2026
Subjects:	Machine Learning
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Similar Items

Interpretable Probability Estimation with LLMs via Shapley Reconstruction
by: Nan, Yang, et al.
Published: (2026)

Can Multiple Responses from an LLM Reveal the Sources of Its Uncertainty?
by: Nan, Yang, et al.
Published: (2025)

Trustworthy Actionable Perturbations
by: Friedbaum, Jesse, et al.
Published: (2024)

Fine-Grained Uncertainty Quantification via Collisions
by: Friedbaum, Jesse, et al.
Published: (2024)

Step-wise Rubric Rewards for LLM Reasoning
by: Xie, Weichu, et al.
Published: (2026)

SPLITZ: Certifiable Robustness via Split Lipschitz Randomized Smoothing
by: Zhong, Meiyu, et al.
Published: (2024)

Learning Fair Robustness via Domain Mixup
by: Zhong, Meiyu, et al.
Published: (2024)

Intrinsic Fairness-Accuracy Tradeoffs under Equalized Odds
by: Zhong, Meiyu, et al.
Published: (2024)

Inference Privacy: Properties and Mechanisms
by: Tian, Fengwei, et al.
Published: (2024)

One Step Forward and K Steps Back: Better Reasoning with Denoising Recursion Models
by: Cameron, Chris, et al.
Published: (2026)

CURATE: Scaling-up Differentially Private Causal Graph Discovery
by: Bhattacharjee, Payel, et al.
Published: (2024)

Prompt Fairness: Sub-group Disparities in LLMs
by: Zhong, Meiyu, et al.
Published: (2025)

Speeding up Speculative Decoding via Sequential Approximate Verification
by: Zhong, Meiyu, et al.
Published: (2025)

CPL: Critical Plan Step Learning Boosts LLM Generalization in Reasoning Tasks
by: Wang, Tianlong, et al.
Published: (2024)

Transformer See, Transformer Do: Copying as an Intermediate Step in Learning Analogical Reasoning
by: Hellwig, Philipp, et al.
Published: (2026)

SABER: Small Actions, Big Errors -- Safeguarding Mutating Steps in LLM Agents
by: Cuadron, Alejandro, et al.
Published: (2025)

MARS: Margin and Semantic-Aware Data Augmentation for Reward Modeling
by: Bhattacharjee, Payel, et al.
Published: (2026)

Semantic Step Prediction: Multi-Step Latent Forecasting in LLM Reasoning Trajectories via Step Sampling
by: Yuan, Yidi
Published: (2026)

Transitional Uncertainty with Layered Intermediate Predictions
by: Benkert, Ryan, et al.
Published: (2024)

Offline Reinforcement Learning for LLM Multi-Step Reasoning
by: Wang, Huaijie, et al.
Published: (2024)

Generative Adversarial Reasoner: Enhancing LLM Reasoning with Adversarial Reinforcement Learning
by: Liu, Qihao, et al.
Published: (2025)

Latency-Distortion Tradeoffs in Communicating Classification Results over Noisy Channels
by: Teku, Noel, et al.
Published: (2024)

When LLM Meets Time Series: Can LLMs Perform Multi-Step Time Series Reasoning and Inference
by: Ye, Wen, et al.
Published: (2025)

Accordion-Thinking: Self-Regulated Step Summaries for Efficient and Readable LLM Reasoning
by: Yang, Zhicheng, et al.
Published: (2026)

Better LLM Reasoning via Dual-Play
by: Zhang, Zhengxin, et al.
Published: (2025)

Lethe: Layer- and Time-Adaptive KV Cache Pruning for Reasoning-Intensive LLM Serving
by: Zeng, Hui, et al.
Published: (2025)

Hierarchical Reinforcement Learning with Augmented Step-Level Transitions for LLM Agents
by: Zhen, Shuai, et al.
Published: (2026)

CAMEL: Curvature-Augmented Manifold Embedding and Learning
by: Xu, Nan, et al.
Published: (2023)

Step-Level Sparse Autoencoder for Reasoning Process Interpretation
by: Yang, Xuan, et al.
Published: (2026)

Multi-Faceted Studies on Data Poisoning can Advance LLM Development
by: He, Pengfei, et al.
Published: (2025)

A Single Revision Step Improves Token-Efficient LLM Reasoning
by: Zhang, Yingchuan, et al.
Published: (2026)

LLM Reasoning with Process Rewards for Outcome-Guided Steps
by: Rezaei, Mohammad, et al.
Published: (2026)

Self-playing Adversarial Language Game Enhances LLM Reasoning
by: Cheng, Pengyu, et al.
Published: (2024)

Step-by-Step Reasoning for Math Problems via Twisted Sequential Monte Carlo
by: Feng, Shengyu, et al.
Published: (2024)

BPO: Staying Close to the Behavior LLM Creates Better Online LLM Alignment
by: Xu, Wenda, et al.
Published: (2024)

Explainable Mapper: Charting LLM Embedding Spaces Using Perturbation-Based Explanation and Verification Agents
by: Yan, Xinyuan, et al.
Published: (2025)

What Makes Reasoning Invalid: Echo Reflection Mitigation for Large Language Models
by: He, Chen, et al.
Published: (2025)

Skip the Benchmark: Generating System-Level High-Level Synthesis Data using Generative Machine Learning
by: Liao, Yuchao, et al.
Published: (2024)

Is More Context Always Better? Examining LLM Reasoning Capability for Time Interval Prediction
by: Cao, Yanan, et al.
Published: (2026)

Plausibility Is Not Prediction: Contrastive Evidence for LLM-Based Cellular Perturbation Reasoning
by: Yuan, Xinyu, et al.
Published: (2026)