Cover Image

Logit Arithmetic Elicits Long Reasoning Capabilities Without Training

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Zhang, Yunxiang, Khalifa, Muhammad, Zhang, Lechen, Liu, Xin, Lee, Ayoung, Zhang, Xinliang Frederick, Bayat, Farima Fatahi, Wang, Lu
Format:	Preprint
Published:	2025
Subjects:	Computation and Language Artificial Intelligence
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Similar Items

Logit Arithmetic Elicits Long Reasoning Capabilities Without Training
by: Zhang, Yunxiang, et al.
Published: (2025)

FactBench: A Dynamic Benchmark for In-the-Wild Language Model Factuality Evaluation
by: Bayat, Farima Fatahi, et al.
Published: (2024)

Enhancing Language Model Factuality via Activation-Based Confidence Calibration and Guided Decoding
by: Liu, Xin, et al.
Published: (2024)

From Proof to Program: Characterizing Tool-Induced Reasoning Hallucinations in Large Language Models
by: Bayat, Farima Fatahi, et al.
Published: (2025)

Enhanced Language Model Truthfulness with Learnable Intervention and Uncertainty Expression
by: Bayat, Farima Fatahi, et al.
Published: (2024)

TSUBASA: Improving Long-Horizon Personalization via Evolving Memory and Self-Learning with Context Distillation
by: Zhang, Xinliang Frederick, et al.
Published: (2026)

Skill-Aware Data Selection and Fine-Tuning for Data-Efficient Reasoning Distillation
by: Zhang, Lechen, et al.
Published: (2026)

Narrative-of-Thought: Improving Temporal Reasoning of Large Language Models via Recounted Narratives
by: Zhang, Xinliang Frederick, et al.
Published: (2024)

LiveOIBench: Can Large Language Models Outperform Human Contestants in Informatics Olympiads?
by: Zou, Kaijian, et al.
Published: (2025)

PRIME: Large Language Model Personalization with Cognitive Dual-Memory and Personalized Thought Process
by: Zhang, Xinliang Frederick, et al.
Published: (2025)

GRACE: Discriminator-Guided Chain-of-Thought Reasoning
by: Khalifa, Muhammad, et al.
Published: (2023)

Gaming the Judge: Unfaithful Chain-of-Thought Can Undermine Agent Evaluation
by: Khalifa, Muhammad, et al.
Published: (2026)

Small Language Models Need Strong Verifiers to Self-Correct Reasoning
by: Zhang, Yunxiang, et al.
Published: (2024)

Elicit and Enhance: Advancing Multimodal Reasoning in Medical Scenarios
by: Huang, Zhongzhen, et al.
Published: (2025)

LitCab: Lightweight Language Model Calibration over Short- and Long-form Responses
by: Liu, Xin, et al.
Published: (2023)

How Numerical Precision Affects Arithmetical Reasoning Capabilities of LLMs
by: Feng, Guhao, et al.
Published: (2024)

Are Your LLMs Capable of Stable Reasoning?
by: Liu, Junnan, et al.
Published: (2024)

Source-Aware Training Enables Knowledge Attribution in Language Models
by: Khalifa, Muhammad, et al.
Published: (2024)

Exploring the System 1 Thinking Capability of Large Reasoning Models
by: Zhang, Wenyuan, et al.
Published: (2025)

Eliciting In-context Retrieval and Reasoning for Long-context Large Language Models
by: Qiu, Yifu, et al.
Published: (2025)

VeriFact: Enhancing Long-Form Factuality Evaluation with Refined Fact Extraction and Reference Facts
by: Liu, Xin, et al.
Published: (2025)

API Is Enough: Conformal Prediction for Large Language Models Without Logit-Access
by: Su, Jiayuan, et al.
Published: (2024)

Divide, Reweight, and Conquer: A Logit Arithmetic Approach for In-Context Learning
by: Huang, Chengsong, et al.
Published: (2024)

Diversity of Thought Elicits Stronger Reasoning Capabilities in Multi-Agent Debate Frameworks
by: Hegazy, Mahmood
Published: (2024)

Logic-Regularized Verifier Elicits Reasoning from LLMs
by: Wang, Xinyu, et al.
Published: (2026)

RAT: Retrieval Augmented Thoughts Elicit Context-Aware Reasoning in Long-Horizon Generation
by: Wang, Zihao, et al.
Published: (2024)

A Training-Free Length Extrapolation Approach for LLMs: Greedy Attention Logit Interpolation (GALI)
by: Li, Yan, et al.
Published: (2025)

Self-Training Elicits Concise Reasoning in Large Language Models
by: Munkhbat, Tergel, et al.
Published: (2025)

Learning to Reason via Program Generation, Emulation, and Search
by: Weir, Nathaniel, et al.
Published: (2024)

CLASH: Evaluating Language Models on Judging High-Stakes Dilemmas from Multiple Perspectives
by: Lee, Ayoung, et al.
Published: (2025)

Tokenization Constraints in LLMs: A Study of Symbolic and Arithmetic Reasoning Limits
by: Zhang, Xiang, et al.
Published: (2025)

Oolong: Evaluating Long Context Reasoning and Aggregation Capabilities
by: Bertsch, Amanda, et al.
Published: (2025)

AI Scientists Fail Without Strong Implementation Capability
by: Zhu, Minjun, et al.
Published: (2025)

LongMamba: Enhancing Mamba's Long Context Capabilities via Training-Free Receptive Field Enlargement
by: Ye, Zhifan, et al.
Published: (2025)

Eliciting Instruction-tuned Code Language Models' Capabilities to Utilize Auxiliary Function for Code Generation
by: Lee, Seonghyeon, et al.
Published: (2024)

Beyond Elicitation: Provision-based Prompt Optimization for Knowledge-Intensive Tasks
by: Xu, Yunzhe, et al.
Published: (2025)

Eliciting Reasoning in Language Models with Cognitive Tools
by: Ebouky, Brown, et al.
Published: (2025)

SPRIG: Improving Large Language Model Performance by System Prompt Optimization
by: Zhang, Lechen, et al.
Published: (2024)

MedReason: Eliciting Factual Medical Reasoning Steps in LLMs via Knowledge Graphs
by: Wu, Juncheng, et al.
Published: (2025)

Let's Reason Formally: Natural-Formal Hybrid Reasoning Enhances LLM's Math Capability
by: Wang, Ruida, et al.
Published: (2025)