Cover Image

The Power of Power Law: Asymmetry Enables Compositional Reasoning

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Wang, Zixuan, Dang, Xingyu, Lee, Jason D., Lyu, Kaifeng
Format:	Preprint
Published:	2026
Subjects:	Artificial Intelligence Computation and Language Machine Learning
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Similar Items

A Multi-Power Law for Loss Curve Prediction Across Learning Rate Schedules
by: Luo, Kairong, et al.
Published: (2025)

SPA: A Simple but Tough-to-Beat Baseline for Knowledge Injection
by: Tang, Kexian, et al.
Published: (2026)

RNNs are not Transformers (Yet): The Key Bottleneck on In-context Retrieval
by: Wen, Kaiyue, et al.
Published: (2024)

Reasoning with Latent Thoughts: On the Power of Looped Transformers
by: Saunshi, Nikunj, et al.
Published: (2025)

Data Mixing Can Induce Phase Transitions in Knowledge Acquisition
by: Gu, Xinran, et al.
Published: (2025)

How Learning Rate Decay Wastes Your Best Data in Curriculum-Based LLM Pretraining
by: Luo, Kairong, et al.
Published: (2025)

The Expressive Power of Low-Rank Adaptation
by: Zeng, Yuchen, et al.
Published: (2023)

Power Lines: Scaling Laws for Weight Decay and Batch Size in LLM Pre-training
by: Bergsma, Shane, et al.
Published: (2025)

What Makes a Reward Model a Good Teacher? An Optimization Perspective
by: Razin, Noam, et al.
Published: (2025)

Keeping LLMs Aligned After Fine-tuning: The Crucial Role of Prompt Templates
by: Lyu, Kaifeng, et al.
Published: (2024)

When Bias Pretends to Be Truth: How Spurious Correlations Undermine Hallucination Detection in LLMs
by: Wang, Shaowen, et al.
Published: (2025)

Learning to Reason from Feedback at Test-Time
by: Li, Yanyang, et al.
Published: (2025)

PCMind-2.1-Kaiyuan-2B Technical Report
by: Luo, Kairong, et al.
Published: (2025)

Ineq-Comp: Benchmarking Human-Intuitive Compositional Reasoning in Automated Theorem Proving on Inequalities
by: Zhao, Haoyu, et al.
Published: (2025)

Distillation Scaling Laws
by: Busbridge, Dan, et al.
Published: (2025)

Contextual Drag: How Errors in the Context Affect LLM Reasoning
by: Cheng, Yun, et al.
Published: (2026)

Power Transformer Fault Prediction Based on Knowledge Graphs
by: Wang, Chao, et al.
Published: (2024)

Lawma: The Power of Specialization for Legal Annotation
by: Dominguez-Olmedo, Ricardo, et al.
Published: (2024)

From Reasoning Chains to Verifiable Subproblems: Curriculum Reinforcement Learning Enables Credit Assignment for LLM Reasoning
by: Jiang, Xitai, et al.
Published: (2026)

CogGPT: Unleashing the Power of Cognitive Dynamics on Large Language Models
by: Lv, Yaojia, et al.
Published: (2024)

Batched Contextual Reinforcement: A Task-Scaling Law for Efficient Reasoning
by: Yang, Bangji, et al.
Published: (2026)

Deliberative Alignment: Reasoning Enables Safer Language Models
by: Guan, Melody Y., et al.
Published: (2024)

Rare-to-Frequent: Unlocking Compositional Generation Power of Diffusion Models on Rare Concepts with LLM Guidance
by: Park, Dongmin, et al.
Published: (2024)

Compositional Causal Reasoning Evaluation in Language Models
by: Maasch, Jacqueline R. M. A., et al.
Published: (2025)

Compositional Reasoning with Transformers, RNNs, and Chain of Thought
by: Yehudai, Gilad, et al.
Published: (2025)

Examining Gender and Power on Wikipedia Through Face and Politeness
by: Soubki, Adil, et al.
Published: (2024)

To Think or Not to Think: Exploring the Unthinking Vulnerability in Large Reasoning Models
by: Zhu, Zihao, et al.
Published: (2025)

PoTPTQ: A Two-step Power-of-Two Post-training for LLMs
by: Wang, Xinyu, et al.
Published: (2025)

LLM-Powered Ensemble Learning for Paper Source Tracing: A GPU-Free Approach
by: Chen, Kunlong, et al.
Published: (2024)

Online Reasoning Calibration: Test-Time Training Enables Generalizable Conformal LLM Reasoning
by: Zhou, Cai, et al.
Published: (2026)

Power Scheduler: A Batch Size and Token Number Agnostic Learning Rate Scheduler
by: Shen, Yikang, et al.
Published: (2024)

Large Language Models are Powerful Electronic Health Record Encoders
by: Hegselmann, Stefan, et al.
Published: (2025)

AIGS: Generating Science from AI-Powered Automated Falsification
by: Liu, Zijun, et al.
Published: (2024)

DistillSpec: Improving Speculative Decoding via Knowledge Distillation
by: Zhou, Yongchao, et al.
Published: (2023)

Recursive Concept Evolution for Compositional Reasoning in Large Language Models
by: Chaudhry, Sarim
Published: (2026)

ArcMemo: Abstract Reasoning Composition with Lifelong LLM Memory
by: Ho, Matthew, et al.
Published: (2025)

Skywork-Math: Data Scaling Laws for Mathematical Reasoning in Large Language Models -- The Story Goes On
by: Zeng, Liang, et al.
Published: (2024)

Enhancing Hepatopathy Clinical Trial Efficiency: A Secure, Large Language Model-Powered Pre-Screening Pipeline
by: Gui, Xiongbin, et al.
Published: (2025)

Rethinking LLM-as-a-Judge: Representation-as-a-Judge with Small Language Models via Semantic Capacity Asymmetry
by: Li, Zhuochun, et al.
Published: (2026)

GNN-as-Judge: Unleashing the Power of LLMs for Graph Learning with GNN Feedback
by: Xu, Ruiyao, et al.
Published: (2026)