Imagen de Portada

Improving Latent Reasoning in LLMs via Soft Concept Mixing

Fuente: arXiv

Guardado en:

Detalles Bibliográficos
Autores principales:	Wang, Kang, Duan, Xiangyu, Du, Tianyi
Formato:	Preprint
Publicado:	2025
Materias:	Computation and Language
Acceso en línea:	Acceder al recurso
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

Ejemplares similares

Soft Thinking: Unlocking the Reasoning Potential of LLMs in Continuous Concept Space
por: Zhang, Zhen, et al.
Publicado: (2025)

RuPLaR : Efficient Latent Compression of LLM Reasoning Chains with Rule-Based Priors From Multi-Step to One-Step
por: Luo, Xiaocheng, et al.
Publicado: (2026)

On the Intrinsic Self-Correction Capability of LLMs: Uncertainty and Latent Concept
por: Liu, Guangliang, et al.
Publicado: (2024)

SoftCoT: Soft Chain-of-Thought for Efficient Reasoning with LLMs
por: Xu, Yige, et al.
Publicado: (2025)

Token Assorted: Mixing Latent and Text Tokens for Improved Language Model Reasoning
por: Su, DiJia, et al.
Publicado: (2025)

Answer-Centric or Reasoning-Driven? Uncovering the Latent Memory Anchor in LLMs
por: Wu, Yang, et al.
Publicado: (2025)

Geometric Latent Reasoning Induces Shorter Generations in LLMs
por: Kumar, Shashi, et al.
Publicado: (2026)

Ko-MuSR: A Multistep Soft Reasoning Benchmark for LLMs Capable of Understanding Korean
por: Park, Chanwoo, et al.
Publicado: (2025)

ReGenesis: LLMs can Grow into Reasoning Generalists via Self-Improvement
por: Peng, Xiangyu, et al.
Publicado: (2024)

Latent Thinking Optimization: Your Latent Reasoning Language Model Secretly Encodes Reward Signals in Its Latent Thoughts
por: Du, Hanwen, et al.
Publicado: (2025)

Can LLMs Reason with Rules? Logic Scaffolding for Stress-Testing and Improving LLMs
por: Wang, Siyuan, et al.
Publicado: (2024)

Hybrid Latent Reasoning via Reinforcement Learning
por: Yue, Zhenrui, et al.
Publicado: (2025)

Chain of Preference Optimization: Improving Chain-of-Thought Reasoning in LLMs
por: Zhang, Xuan, et al.
Publicado: (2024)

LLM Latent Reasoning as Chain of Superposition
por: Deng, Jingcheng, et al.
Publicado: (2025)

LaDiR: Latent Diffusion Enhances LLMs for Text Reasoning
por: Kang, Haoqiang, et al.
Publicado: (2025)

Latent-GRPO: Group Relative Policy Optimization for Latent Reasoning
por: Deng, Jingcheng, et al.
Publicado: (2026)

SwiReasoning: Switch-Thinking in Latent and Explicit for Pareto-Superior Reasoning LLMs
por: Shi, Dachuan, et al.
Publicado: (2025)

Impacts of Continued Legal Pre-Training and IFT on LLMs' Latent Representations of Human-Defined Legal Concepts
por: Ho, Shaun
Publicado: (2024)

Scaling Latent Reasoning via Looped Language Models
por: Zhu, Rui-Jie, et al.
Publicado: (2025)

UniVLR: Unifying Text and Vision in Visual Latent Reasoning for Multimodal LLMs
por: Jiang, Houcheng, et al.
Publicado: (2026)

LaSR: Context-Aware Speech Recognition via Latent Reasoning
por: Liu, Heyang, et al.
Publicado: (2026)

Rectify Evaluation Preference: Improving LLMs' Critique on Math Reasoning via Perplexity-aware Reinforcement Learning
por: Tian, Changyuan, et al.
Publicado: (2025)

EnigmaToM: Improve LLMs' Theory-of-Mind Reasoning Capabilities with Neural Knowledge Base of Entity States
por: Xu, Hainiu, et al.
Publicado: (2025)

Wait, We Don't Need to "Wait"! Removing Thinking Tokens Improves Reasoning Efficiency
por: Wang, Chenlong, et al.
Publicado: (2025)

Concept-Reversed Winograd Schema Challenge: Evaluating and Improving Robust Reasoning in Large Language Models via Abstraction
por: Han, Kaiqiao, et al.
Publicado: (2024)

LLMs are Single-threaded Reasoners: Demystifying the Working Mechanism of Soft Thinking
por: Wu, Junhong, et al.
Publicado: (2025)

Constructive Circuit Amplification: Improving Math Reasoning in LLMs via Targeted Sub-Network Updates
por: Prakash, Nikhil, et al.
Publicado: (2025)

Format-Adapter: Improving Reasoning Capability of LLMs by Adapting Suitable Format
por: Wang, Dingzirui, et al.
Publicado: (2025)

Benchmarking and Understanding Compositional Relational Reasoning of LLMs
por: Ni, Ruikang, et al.
Publicado: (2024)

LLMs Improving LLMs: Agentic Discovery for Test-Time Scaling
por: Zheng, Tong, et al.
Publicado: (2026)

Latent Concept-based Explanation of NLP Models
por: Yu, Xuemin, et al.
Publicado: (2024)

Improved Paraphrase Generation via Controllable Latent Diffusion
por: Zou, Wei, et al.
Publicado: (2024)

Beyond Words: A Latent Memory Approach to Internal Reasoning in LLMs
por: Orlicki, José I.
Publicado: (2025)

MixRea: Benchmarking Explicit-Implicit Reasoning in Large Language Models
por: Cai, Yuanqing, et al.
Publicado: (2026)

Diversity of Thought Improves Reasoning Abilities of LLMs
por: Naik, Ranjita, et al.
Publicado: (2023)

Fair In-Context Learning via Latent Concept Variables
por: Bhaila, Karuna, et al.
Publicado: (2024)

A Survey on Latent Reasoning
por: Zhu, Rui-Jie, et al.
Publicado: (2025)

Exploiting LLMs' Reasoning Capability to Infer Implicit Concepts in Legal Information Retrieval
por: Nguyen, Hai-Long, et al.
Publicado: (2024)

Latent Reasoning via Sentence Embedding Prediction
por: Hwang, Hyeonbin, et al.
Publicado: (2025)

Latent Visual Reasoning
por: Li, Bangzheng, et al.
Publicado: (2025)