Copertina

Multi-Layer Attention is the Amplifier of Demonstration Effectiveness

Fuente: arXiv

Salvato in:

Dettagli Bibliografici
Autori principali:	Wang, Dingzirui, Zhang, Xuangliang, Xu, Keyan, Zhu, Qingfu, Che, Wanxiang, Deng, Yang
Natura:	Preprint
Pubblicazione:	2025
Soggetti:	Computation and Language Machine Learning
Accesso online:	Acceder al recurso
Tags:	Aggiungi Tag Nessun Tag, puoi essere il primo ad aggiungerne!!

Documenti analoghi

Scaling Laws for Agent Harnesses via Effective Feedback Compute
di: Zhang, Xuanliang, et al.
Pubblicazione: (2026)

Abacus-SQL: A Text-to-SQL System Empowering Cross-Domain and Open-Domain Database Retrieval
di: Xu, Keyan, et al.
Pubblicazione: (2025)

RoT: Enhancing Table Reasoning with Iterative Row-Wise Traversals
di: Zhang, Xuanliang, et al.
Pubblicazione: (2025)

MULTITAT: Benchmarking Multilingual Table-and-Text Question Answering
di: Zhang, Xuanliang, et al.
Pubblicazione: (2025)

How Do Language Models Understand Tables? A Mechanistic Analysis of Cell Location
di: Zhang, Xuanliang, et al.
Pubblicazione: (2026)

AlignEvoSkill: Towards Knowledge-Aware and Task-Aligned Agent Skill Evolution
di: Wang, Dingzirui, et al.
Pubblicazione: (2025)

Bounds of Chain-of-Thought Robustness: Reasoning Steps, Embed Norms, and Beyond
di: Wang, Dingzirui, et al.
Pubblicazione: (2025)

When Does Context Help? Error Dynamics of Contextual Information in Large Language Models
di: Wang, Dingzirui, et al.
Pubblicazione: (2026)

Improving Demonstration Diversity by Human-Free Fusing for Text-to-SQL
di: Wang, Dingzirui, et al.
Pubblicazione: (2024)

In-Context Transfer Learning: Demonstration Synthesis by Transferring Similar Tasks
di: Wang, Dingzirui, et al.
Pubblicazione: (2024)

CRVQ: Channel-Relaxed Vector Quantization for Extreme Compression of LLMs
di: Xu, Yuzhuang, et al.
Pubblicazione: (2024)

Learning-to-Context Slope: Evaluating In-Context Learning Effectiveness Beyond Performance Illusions
di: Wang, Dingzriui, et al.
Pubblicazione: (2025)

MURRE: Multi-Hop Table Retrieval with Removal for Open-Domain Text-to-SQL
di: Zhang, Xuanliang, et al.
Pubblicazione: (2024)

ProxyAttn: Guided Sparse Attention via Representative Heads
di: Wang, Yixuan, et al.
Pubblicazione: (2025)

Enhancing Numerical Reasoning with the Guidance of Reliable Reasoning Processes
di: Wang, Dingzirui, et al.
Pubblicazione: (2024)

DAC: Decomposed Automation Correction for Text-to-SQL
di: Wang, Dingzirui, et al.
Pubblicazione: (2024)

A Survey of Table Reasoning with Large Language Models
di: Zhang, Xuanliang, et al.
Pubblicazione: (2024)

CAMERA: Multi-Matrix Joint Compression for MoE Models via Micro-Expert Redundancy Analysis
di: Xu, Yuzhuang, et al.
Pubblicazione: (2025)

SCITAT: A Question Answering Benchmark for Scientific Tables and Text Covering Diverse Reasoning Types
di: Zhang, Xuanliang, et al.
Pubblicazione: (2024)

Make Some Noise: Unlocking Language Model Parallel Inference Capability through Noisy Training
di: Wang, Yixuan, et al.
Pubblicazione: (2024)

FLEXTAF: Enhancing Table Reasoning with Flexible Tabular Formats
di: Zhang, Xuanliang, et al.
Pubblicazione: (2024)

ChartREG++: Towards Benchmarking and Improving Chart Referring Expression Grounding under Diverse referring clues and Multi-Target Referring
di: Niu, Tianhao, et al.
Pubblicazione: (2026)

Turning Trash into Treasure: Accelerating Inference of Large Language Models with Token Recycling
di: Luo, Xianzhen, et al.
Pubblicazione: (2024)

Manager: Aggregating Insights from Unimodal Experts in Two-Tower VLMs and MLLMs
di: Xu, Xiao, et al.
Pubblicazione: (2025)

Exploring Multi-Grained Concept Annotations for Multimodal Large Language Models
di: Xu, Xiao, et al.
Pubblicazione: (2024)

PowerAttention: Exponentially Scaling of Receptive Fields for Effective Sparse Attention
di: Chen, Lida, et al.
Pubblicazione: (2025)

Mechanism and Emergence of Stacked Attention Heads in Multi-Layer Transformers
di: Musat, Tiberiu
Pubblicazione: (2024)

Semi-Instruct: Bridging Natural-Instruct and Self-Instruct for Code Large Language Models
di: Luo, Xianzhen, et al.
Pubblicazione: (2024)

Seer Self-Consistency: Advance Budget Estimation for Adaptive Test-Time Scaling
di: Ji, Shiyu, et al.
Pubblicazione: (2025)

EchoKV: Efficient KV Cache Compression via Similarity-Based Reconstruction
di: Ji, Shiyu, et al.
Pubblicazione: (2026)

BridgeTower: Building Bridges Between Encoders in Vision-Language Representation Learning
di: Xu, Xiao, et al.
Pubblicazione: (2022)

LayerBoost: Layer-Aware Attention Reduction for Efficient LLMs
di: Souibgui, Mohamed Ali, et al.
Pubblicazione: (2026)

Paying Attention to Facts: Quantifying the Knowledge Capacity of Attention Layers
di: Wong, Liang Ze
Pubblicazione: (2025)

Multi-matrix Factorization Attention
di: Hu, Jingcheng, et al.
Pubblicazione: (2024)

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

High-Layer Attention Pruning with Rescaling
di: Liu, Songtao, et al.
Pubblicazione: (2025)

AutoTriton: Automatic Triton Programming with Reinforcement Learning in LLMs
di: Li, Shangzhan, et al.
Pubblicazione: (2025)

Attention Sink Forges Native MoE in Attention Layers: Sink-Aware Training to Address Head Collapse
di: Fu, Zizhuo, et al.
Pubblicazione: (2026)

Think Before You Accept: Semantic Reflective Verification for Faster Speculative Decoding
di: Wang, Yixuan, et al.
Pubblicazione: (2025)

S$^3$-Attention:Attention-Aligned Endogenous Retrieval for Memory-Bounded Long-Context Inference
di: Ma, Qingsen, et al.
Pubblicazione: (2026)