Imagen de Portada

Mid-Think: Training-Free Intermediate-Budget Reasoning via Token-Level Triggers

Fuente: arXiv

Guardado en:

Detalles Bibliográficos
Autores principales:	Yang, Wang, Ganguly, Debargha, Li, Xinpeng, Song, Chaoda, Wang, Shouren, Singh, Vikash, Chaudhary, Vipin, Han, Xiaotian
Formato:	Preprint
Publicado:	2026
Materias:	Computation and Language Artificial Intelligence Machine Learning
Acceso en línea:	Acceder al recurso
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

Ejemplares similares

Path-Lock Expert: Separating Reasoning Mode in Hybrid Thinking via Architecture-Level Separation
por: Wang, Shouren, et al.
Publicado: (2026)

AgentCE-Bench: Agent Configurable Evaluation with Scalable Horizons and Controllable Difficulty under Lightweight Environments
por: Yang, Wang, et al.
Publicado: (2026)

When Domains Interact: Asymmetric and Order-Sensitive Cross-Domain Effects in Reinforcement Learning for Reasoning
por: Yang, Wang, et al.
Publicado: (2026)

Demystifying Hybrid Thinking: Can LLMs Truly Switch Between Think and No-Think?
por: Wang, Shouren, et al.
Publicado: (2025)

Toward Guarantees for Clinical Reasoning in Vision Language Models via Formal Verification
por: Singh, Vikash, et al.
Publicado: (2026)

$K^4$: Online Log Anomaly Detection Via Unsupervised Typicality Learning
por: Chen, Weicong, et al.
Publicado: (2025)

Grammars of Formal Uncertainty: When to Trust LLMs in Automated Reasoning Tasks
por: Ganguly, Debargha, et al.
Publicado: (2025)

Speculative Thinking: Enhancing Small-Model Reasoning with Large Model Guidance at Inference Time
por: Yang, Wang, et al.
Publicado: (2025)

Proof of Thought : Neurosymbolic Program Synthesis allows Robust and Interpretable Reasoning
por: Ganguly, Debargha, et al.
Publicado: (2024)

Overcoming Dynamics-Blindness: Training-Free Pace-and-Path Correction for VLA Models
por: Zhang, Yanyan, et al.
Publicado: (2026)

Longer Context, Deeper Thinking: Uncovering the Role of Long-Context Ability in Reasoning
por: Yang, Wang, et al.
Publicado: (2025)

Visual Concept Networks: A Graph-Based Approach to Detecting Anomalous Data in Deep Neural Networks
por: Ganguly, Debargha, et al.
Publicado: (2024)

CausalRAG: Integrating Causal Graphs into Retrieval-Augmented Generation
por: Wang, Nengbo, et al.
Publicado: (2025)

CausalGuard: Conformal Inference under Graph Uncertainty
por: Singh, Vikash, et al.
Publicado: (2026)

Reliability-Gated Source Anchoring for Continual Test-Time Adaptation
por: Singh, Vikash, et al.
Publicado: (2026)

Efficient Fine-Grained GPU Performance Modeling for Distributed Deep Learning of LLM
por: Zhang, Biyao, et al.
Publicado: (2025)

Trust The Typical
por: Ganguly, Debargha, et al.
Publicado: (2026)

Forte : Finding Outliers with Representation Typicality Estimation
por: Ganguly, Debargha, et al.
Publicado: (2024)

Privacy Policy Enforcement Guardrails for Data-Sensitive Retrieval-Augmented Generation
por: Zafar, Osama, et al.
Publicado: (2026)

Token-Budget-Aware LLM Reasoning
por: Han, Tingxu, et al.
Publicado: (2024)

100-LongBench: Are de facto Long-Context Benchmarks Literally Evaluating Long-Context Ability?
por: Yang, Wang, et al.
Publicado: (2025)

Thinking Preference Optimization
por: Yang, Wang, et al.
Publicado: (2025)

SELF: Self-Extend the Context Length With Logistic Growth Function
por: Dang, Phat Thanh, et al.
Publicado: (2025)

LABELING COPILOT: A Deep Research Agent for Automated Data Curation in Computer Vision
por: Ganguly, Debargha, et al.
Publicado: (2025)

HugRAG: Hierarchical Causal Knowledge Graph Design for RAG
por: Wang, Nengbo, et al.
Publicado: (2026)

Think Before Refusal : Triggering Safety Reflection in LLMs to Mitigate False Refusal Behavior
por: Si, Shengyun, et al.
Publicado: (2025)

Reasoning in Token Economies: Budget-Aware Evaluation of LLM Reasoning Strategies
por: Wang, Junlin, et al.
Publicado: (2024)

Exploring Efficiency Frontiers of Thinking Budget in Medical Reasoning: Scaling Laws between Computational Resources and Reasoning Quality
por: Bi, Ziqian, et al.
Publicado: (2025)

Adaptive Deep Reasoning: Triggering Deep Thinking When Needed
por: Wang, Yunhao, et al.
Publicado: (2025)

Is It Thinking or Cheating? Detecting Implicit Reward Hacking by Measuring Reasoning Effort
por: Wang, Xinpeng, et al.
Publicado: (2025)

SyncThink: A Training-Free Strategy to Align Inference Termination with Reasoning Saturation
por: Li, Gengyang, et al.
Publicado: (2026)

Reinforcement Mid-Training
por: Tian, Yijun, et al.
Publicado: (2025)

Think Clearly: Improving Reasoning via Redundant Token Pruning
por: Choi, Daewon, et al.
Publicado: (2025)

Asynchronous Reasoning: Training-Free Interactive Thinking LLMs
por: Yakushev, George, et al.
Publicado: (2025)

Think Deep, Not Just Long: Measuring LLM Reasoning Effort via Deep-Thinking Tokens
por: Chen, Wei-Lin, et al.
Publicado: (2026)

Thinking About Thinking: Evaluating Reasoning in Post-Trained Language Models
por: Singla, Pratham, et al.
Publicado: (2025)

Single-Agent LLMs Outperform Multi-Agent Systems on Multi-Hop Reasoning Under Equal Thinking Token Budgets
por: Tran, Dat, et al.
Publicado: (2026)

MeTHanol: Modularized Thinking Language Models with Intermediate Layer Thinking, Decoding and Bootstrapping Reasoning
por: Xi, Ningyuan, et al.
Publicado: (2024)

TwT: Thinking without Tokens by Habitual Reasoning Distillation with Multi-Teachers' Guidance
por: Xu, Jingxian, et al.
Publicado: (2025)

Distilling Token-Trained Models into Byte-Level Models
por: Bao, Zishuo, et al.
Publicado: (2026)