Cover Image

Soft Head Selection for Injecting ICL-Derived Task Embeddings

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Park, Jungwon, Kim, Jimyeong, Choi, Changin, Rhee, Wonjong
Format:	Preprint
Published:	2025
Subjects:	Computation and Language
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Similar Items

Selectively Informative Description can Reduce Undesired Embedding Entanglements in Text-to-Image Personalization
by: Kim, Jimyeong, et al.
Published: (2024)

Orthogonal Negative Guidance in Attention Feature Space for Text-to-Image Generation
by: Ko, Jungmin, et al.
Published: (2026)

When Confidence Misleads: Suffix Anchoring and Anchor-Proximity Confidence Modulation for Diffusion Language Models
by: Park, Jungwon, et al.
Published: (2026)

DOS: Directional Object Separation in Text Embeddings for Multi-Object Image Generation
by: Byun, Dongnam, et al.
Published: (2025)

An Image Grid Can Be Worth a Video: Zero-shot Video Question Answering Using a VLM
by: Kim, Wonkyun, et al.
Published: (2024)

ReFlex: Text-Guided Editing of Real Images in Rectified Flow via Mid-Step Feature Extraction and Attention Adaptation
by: Kim, Jimyeong, et al.
Published: (2025)

Unveiling Key Aspects of Fine-Tuning in Sentence Embeddings: A Representation Rank Analysis
by: Jung, Euna, et al.
Published: (2024)

Progressive Multimodal Search and Reasoning for Knowledge-Intensive Visual Question Answering
by: Choi, Changin, et al.
Published: (2025)

Enhancing Retrieval-Augmented Audio Captioning with Generation-Assisted Multimodal Querying and Progressive Learning
by: Changin, Choi, et al.
Published: (2024)

Harmonizing Visual and Textual Embeddings for Zero-Shot Text-to-Image Customization
by: Song, Yeji, et al.
Published: (2024)

Task-Specific Preconditioner for Cross-Domain Few-Shot Learning
by: Kang, Suhyun, et al.
Published: (2024)

Selective Aggregation of Attention Maps Improves Diffusion-Based Visual Interpretation
by: Park, Jungwon, et al.
Published: (2026)

Evaluating Feature Attribution Methods for Electrocardiogram
by: Suh, Jangwon, et al.
Published: (2022)

Cross-Attention Head Position Patterns Can Align with Human Visual Concepts in Text-to-Image Generative Models
by: Park, Jungwon, et al.
Published: (2024)

Task Diversity Shortens the ICL Plateau
by: Kim, Jaeyeon, et al.
Published: (2024)

Towards a Better Evaluation of Out-of-Domain Generalization
by: Hwang, Duhun, et al.
Published: (2024)

Enhancing Contrastive Learning with Efficient Combinatorial Positive Pairing
by: Kim, Jaeill, et al.
Published: (2024)

STARE at the Structure: Steering ICL Exemplar Selection with Structural Alignment
by: Li, Jiaqian, et al.
Published: (2025)

UniICL: An Efficient Unified Framework Unifying Compression, Selection, and Generation
by: Gao, Jun, et al.
Published: (2024)

CrossICL: Cross-Task In-Context Learning via Unsupervised Demonstration Transfer
by: Gao, Jinglong, et al.
Published: (2025)

Improving Multi-lingual Alignment Through Soft Contrastive Learning
by: Park, Minsu, et al.
Published: (2024)

Bridging the Missing-Modality Gap: Improving Text-Only Calibration of Vision Language Models
by: Kim, Mingyeong, et al.
Published: (2026)

Refract ICL: Rethinking Example Selection in the Era of Million-Token Models
by: Akula, Arjun R., et al.
Published: (2025)

ParaICL: Towards Parallel In-Context Learning
by: Li, Xingxuan, et al.
Published: (2024)

Stress-Testing Long-Context Language Models with Lifelong ICL and Task Haystack
by: Xu, Xiaoyue, et al.
Published: (2024)

KID: Knowledge-Injected Dual-Head Learning for Knowledge-Grounded Harmful Meme Detection
by: Li, Yaocong, et al.
Published: (2026)

Opt-ICL at LeWiDi-2025: Maximizing In-Context Signal from Rater Examples via Meta-Learning
by: Sorensen, Taylor, et al.
Published: (2025)

An Efficient Task-Oriented Dialogue Policy: Evolutionary Reinforcement Learning Injected by Elite Individuals
by: Zhao, Yangyang, et al.
Published: (2025)

C-ICL: Contrastive In-context Learning for Information Extraction
by: Mo, Ying, et al.
Published: (2024)

Explore, Select, Derive, and Recall: Augmenting LLM with Human-like Memory for Mobile Task Automation
by: Lee, Sunjae, et al.
Published: (2023)

Bayesian Multi-Task Transfer Learning for Soft Prompt Tuning
by: Lee, Haeju, et al.
Published: (2024)

MateICL: Mitigating Attention Dispersion in Large-Scale In-Context Learning
by: Ahmed, Murtadha, et al.
Published: (2025)

ICL CIPHERS: Quantifying "Learning" in In-Context Learning via Substitution Ciphers
by: Fang, Zhouxiang, et al.
Published: (2025)

On-Off Pattern Encoding and Path-Count Encoding as Deep Neural Network Representations
by: Jung, Euna, et al.
Published: (2024)

Vector-ICL: In-context Learning with Continuous Vector Representations
by: Zhuang, Yufan, et al.
Published: (2024)

Can Editing LLMs Inject Harm?
by: Chen, Canyu, et al.
Published: (2024)

Improving Forward Compatibility in Class Incremental Learning by Increasing Representation Rank and Feature Richness
by: Kim, Jaeill, et al.
Published: (2024)

Garbage Attention in Large Language Models: BOS Sink Heads and Sink-aware Pruning
by: Sok, Jaewon, et al.
Published: (2026)

Conflict-Aware Soft Prompting for Retrieval-Augmented Generation
by: Choi, Eunseong, et al.
Published: (2025)

SoftEDA: Rethinking Rule-Based Data Augmentation with Soft Labels
by: Choi, Juhwan, et al.
Published: (2024)