Copertina

From Interpretability to Performance: Optimizing Retrieval Heads for Long-Context Language Models

Fuente: arXiv

Salvato in:

Dettagli Bibliografici
Autori principali:	Ma, Youmi, Okazaki, Naoaki
Natura:	Preprint
Pubblicazione:	2026
Soggetti:	Computation and Language
Accesso online:	Acceder al recurso
Tags:	Aggiungi Tag Nessun Tag, puoi essere il primo ad aggiungerne!!

Documenti analoghi

Building a Japanese Document-Level Relation Extraction Dataset Assisted by Cross-Lingual Transfer
di: Ma, Youmi, et al.
Pubblicazione: (2024)

Sampling-based Pseudo-Likelihood for Membership Inference Attacks
di: Kaneko, Masahiro, et al.
Pubblicazione: (2024)

Synthesizing Instruction-Tuning Datasets with Contrastive Decoding
di: Ichinose, Tatsuya, et al.
Pubblicazione: (2026)

Knowledge of Pretrained Language Models on Surface Information of Tokens
di: Hiraoka, Tatsuya, et al.
Pubblicazione: (2024)

Diffusion-State Policy Optimization for Masked Diffusion Language Models
di: Oba, Daisuke, et al.
Pubblicazione: (2026)

Evaluating Gender Bias of Pre-trained Language Models in Natural Language Inference by Considering All Labels
di: Anantaprayoon, Panatchakorn, et al.
Pubblicazione: (2023)

LLM Output Detectability and Task Performance Can be Jointly Optimized
di: Saito, Koshiro, et al.
Pubblicazione: (2026)

Social Bias Evaluation for Large Language Models Requires Prompt Variations
di: Hida, Rem, et al.
Pubblicazione: (2024)

OUTFOX: LLM-Generated Essay Detection Through In-Context Learning with Adversarially Generated Examples
di: Koike, Ryuto, et al.
Pubblicazione: (2023)

Drifting Objectives for Refining Discrete Diffusion Language Models
di: Oba, Daisuke, et al.
Pubblicazione: (2026)

Intent-Aware Self-Correction for Mitigating Social Biases in Large Language Models
di: Anantaprayoon, Panatchakorn, et al.
Pubblicazione: (2025)

Tokenization as Finite-State Transduction
di: Cognetta, Marco, et al.
Pubblicazione: (2024)

Evaluating Gender Bias in Large Language Models via Chain-of-Thought Prompting
di: Kaneko, Masahiro, et al.
Pubblicazione: (2024)

ExaGPT: Example-Based Machine-Generated Text Detection for Human Interpretability
di: Koike, Ryuto, et al.
Pubblicazione: (2025)

A Japanese Benchmark for Evaluating Social Bias in Reasoning Based on Attribution Theory
di: Shiotani, Taihei, et al.
Pubblicazione: (2026)

Bit-level BPE: Below the byte boundary
di: Moon, Sangwhan, et al.
Pubblicazione: (2025)

Distributional Properties of Subword Regularization
di: Cognetta, Marco, et al.
Pubblicazione: (2024)

How You Prompt Matters! Even Task-Oriented Constraints in Instructions Affect LLM-Generated Text Detection
di: Koike, Ryuto, et al.
Pubblicazione: (2023)

Solving NLP Problems through Human-System Collaboration: A Discussion-based Approach
di: Kaneko, Masahiro, et al.
Pubblicazione: (2023)

SAIE Framework: Support Alone Isn't Enough -- Advancing LLM Training with Adversarial Remarks
di: Loem, Mengsay, et al.
Pubblicazione: (2023)

Retrieval Head Mechanistically Explains Long-Context Factuality
di: Wu, Wenhao, et al.
Pubblicazione: (2024)

Why We Build Local Large Language Models: An Observational Analysis from 35 Japanese and Multilingual LLMs
di: Saito, Koshiro, et al.
Pubblicazione: (2024)

Multi-modal, Multi-task, Multi-criteria Automatic Evaluation with Vision Language Models
di: Ohi, Masanari, et al.
Pubblicazione: (2024)

Likelihood-based Mitigation of Evaluation Bias in Large Language Models
di: Oi, Masanari, et al.
Pubblicazione: (2024)

Building Instruction-Tuning Datasets from Human-Written Instructions with Open-Weight Large Language Models
di: Ma, Youmi, et al.
Pubblicazione: (2025)

QuantumBench: A Benchmark for Quantum Problem Solving
di: Minami, Shunya, et al.
Pubblicazione: (2025)

Constructing Multimodal Datasets from Scratch for Rapid Development of a Japanese Visual Language Model
di: Sasagawa, Keito, et al.
Pubblicazione: (2024)

Two Counterexamples to Tokenization and the Noiseless Channel
di: Cognetta, Marco, et al.
Pubblicazione: (2024)

Aligning Tree-Search Policies with Fixed Token Budgets in Test-Time Scaling of LLMs
di: Miyamoto, Sora, et al.
Pubblicazione: (2026)

ZigzagAttention: Efficient Long-Context Inference with Exclusive Retrieval and Streaming Heads
di: Liu, Zhuorui, et al.
Pubblicazione: (2025)

Query-Focused Retrieval Heads Improve Long-Context Reasoning and Re-ranking
di: Zhang, Wuwei, et al.
Pubblicazione: (2025)

DuoAttention: Efficient Long-Context LLM Inference with Retrieval and Streaming Heads
di: Xiao, Guangxuan, et al.
Pubblicazione: (2024)

Literary Evidence Retrieval via Long-Context Language Models
di: Thai, Katherine, et al.
Pubblicazione: (2025)

Improving Contextual Faithfulness of Large Language Models via Retrieval Heads-Induced Optimization
di: Huang, Lei, et al.
Pubblicazione: (2025)

WAON: A Large-Scale Japanese Image-Text Dataset for Cultural Adaptation in Contrastive Vision-Language Models
di: Sugiura, Issa, et al.
Pubblicazione: (2025)

VERA: Identifying and Leveraging Visual Evidence Retrieval Heads in Long-Context Understanding
di: Pei, Rongcan, et al.
Pubblicazione: (2026)

Stronger Baselines for Retrieval-Augmented Generation with Long-Context Language Models
di: Laitenberger, Alex, et al.
Pubblicazione: (2025)

Stopping Computation for Converged Tokens in Masked Diffusion-LM Decoding
di: Oba, Daisuke, et al.
Pubblicazione: (2026)

Exploring Fine-Tuning for In-Context Retrieval and Efficient KV-Caching in Long-Context Language Models
di: Molfese, Francesco Maria, et al.
Pubblicazione: (2026)

Decoding-Free Sampling Strategies for LLM Marginalization
di: Pohl, David, et al.
Pubblicazione: (2025)