Cover Image

Pruning the Unsurprising: Efficient LLM Reasoning via First-Token Surprisal

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Zeng, Wenhao, Wang, Yaoning, Hu, Chao, Shi, Yuling, Wan, Chengcheng, Zhang, Hongyu, Gu, Xiaodong
Format:	Preprint
Published:	2025
Subjects:	Machine Learning Software Engineering
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Similar Items

Between Lines of Code: Unraveling the Distinct Patterns of Machine and Human Programmers
by: Shi, Yuling, et al.
Published: (2024)

In Line with Context: Repository-Level Code Generation via Context Inlining
by: Hu, Chao, et al.
Published: (2026)

ClassEval-Pro: A Cross-Domain Benchmark for Class-Level Code Generation
by: Chen, Yeheng, et al.
Published: (2026)

From Code to Correctness: Closing the Last Mile of Code Generation with Hierarchical Debugging
by: Shi, Yuling, et al.
Published: (2024)

LongCodeZip: Compress Long Context for Code Language Models
by: Shi, Yuling, et al.
Published: (2025)

CodeOCR: On the Effectiveness of Vision Language Models in Code Understanding
by: Shi, Yuling, et al.
Published: (2026)

EffiSkill: Agent Skill Based Automated Code Efficiency Optimization
by: Wang, Zimu, et al.
Published: (2026)

Rethinking Code Complexity Through the Lens of Large Language Models
by: Xie, Chen, et al.
Published: (2026)

Test vs Mutant: Adversarial LLM Agents for Robust Unit Test Generation
by: Chang, Pengyu, et al.
Published: (2026)

On the Effectiveness of Large Language Models in Domain-Specific Code Generation
by: Gu, Xiaodong, et al.
Published: (2023)

Rethinking the Value of Agent-Generated Tests for LLM-Based Software Engineering Agents
by: Chen, Zhi, et al.
Published: (2026)

SWE-Pruner: Self-Adaptive Context Pruning for Coding Agents
by: Wang, Yuhang, et al.
Published: (2026)

Readability-Robust Code Summarization via Meta Curriculum Learning
by: Zeng, Wenhao, et al.
Published: (2026)

Code Less, Align More: Efficient LLM Fine-tuning for Code Generation with Data Pruning
by: Tsai, Yun-Da, et al.
Published: (2024)

Cross-Domain Deep Code Search with Meta Learning
by: Chai, Yitian, et al.
Published: (2022)

SECRET: Towards Scalable and Efficient Code Retrieval via Segmented Deep Hashing
by: Gu, Wenchao, et al.
Published: (2024)

SWE-Debate: Competitive Multi-Agent Debate for Software Issue Resolution
by: Li, Han, et al.
Published: (2025)

When to Stop? Towards Efficient Code Generation in LLMs with Excess Token Prevention
by: Guo, Lianghong, et al.
Published: (2024)

Improving Random Testing via LLM-powered UI Tarpit Escaping for Mobile Apps
by: Xu, Mengqian, et al.
Published: (2026)

CodeCoR: An LLM-Based Self-Reflective Multi-Agent Framework for Code Generation
by: Pan, Ruwei, et al.
Published: (2025)

SWE-Exp: Experience-Driven Software Issue Resolution
by: Chen, Silin, et al.
Published: (2025)

Adversarial Feature Map Pruning for Backdoor
by: Huang, Dong, et al.
Published: (2023)

Neuron Patching: Semantic-based Neuron-level Language Model Repair for Code Generation
by: Gu, Jian, et al.
Published: (2023)

A Semantic-based Optimization Approach for Repairing LLMs: Case Study on Code Generation
by: Gu, Jian, et al.
Published: (2025)

TeleResilienceBench: Quantifying Resilience for LLM Reasoning in Telecommunications
by: Gajjar, Pranshav, et al.
Published: (2026)

Assessing the Impact of Requirement Ambiguity on LLM-based Function-Level Code Generation
by: Yang, Di, et al.
Published: (2026)

SWE-QA: Can Language Models Answer Repository-level Code Questions?
by: Peng, Weihan, et al.
Published: (2025)

Zero-Shot Code Representation Learning via Prompt Tuning
by: Cui, Nan, et al.
Published: (2024)

FLARE: Agentic Coverage-Guided Fuzzing for LLM-Based Multi-Agent Systems
by: Hui, Mingxuan, et al.
Published: (2026)

OptLLM: Optimal Assignment of Queries to Large Language Models
by: Liu, Yueyue, et al.
Published: (2024)

EfficientUICoder: Efficient MLLM-based UI Code Generation via Input and Output Token Compression
by: Xiao, Jingyu, et al.
Published: (2025)

How Efficient is LLM-Generated Code? A Rigorous & High-Standard Benchmark
by: Qiu, Ruizhong, et al.
Published: (2024)

PRIMG : Efficient LLM-driven Test Generation Using Mutant Prioritization
by: Bouafif, Mohamed Salah, et al.
Published: (2025)

LLM-Guided Runtime Parameter Optimization for Energy-Efficient Model Inference
by: Crumpacker, Katelyn, et al.
Published: (2026)

SparseCoder: Advancing Source Code Analysis with Sparse Attention and Learned Token Pruning
by: Yang, Xueqi, et al.
Published: (2023)

Engineering Resource-constrained Software Systems with DNN Components: a Concept-based Pruning Approach
by: Formica, Federico, et al.
Published: (2026)

Optimizing Token Consumption in LLMs: A Nano Surge Approach for Code Reasoning Efficiency
by: Hu, Junwei, et al.
Published: (2025)

EvoC2Rust: A Skeleton-guided Framework for Project-Level C-to-Rust Translation
by: Wang, Chaofan, et al.
Published: (2025)

On Reasoning-Centric LLM-based Automated Theorem Proving
by: Sun, Yican, et al.
Published: (2026)

Detecting Proxy Gaming in RL and LLM Alignment via Evaluator Stress Tests
by: Shihab, Ibne Farabi, et al.
Published: (2025)