Image de couverture de livre

Extracting and Understanding the Superficial Knowledge in Alignment

Fuente: arXiv

Enregistré dans:

Détails bibliographiques
Auteurs principaux:	Chen, Runjin, Perin, Gabriel Jacob, Chen, Xuxi, Chen, Xilun, Han, Yan, Hirata, Nina S. T., Hong, Junyuan, Kailkhura, Bhavya
Format:	Preprint
Publié:	2025
Sujets:	Computation and Language Artificial Intelligence
Accès en ligne:	Acceder al recurso
Tags:	Ajouter un tag Pas de tags, Soyez le premier à ajouter un tag!

Documents similaires

LoX: Low-Rank Extrapolation Robustifies LLM Safety Against Fine-tuning
par: Perin, Gabriel J., et autres
Publié: (2025)

SEAL: Steerable Reasoning Calibration of Large Language Models for Free
par: Chen, Runjin, et autres
Publié: (2025)

LLMs Can Get "Brain Rot": A Pilot Study on Twitter/X
par: Xing, Shuo, et autres
Publié: (2025)

The Fragility of Chain-of-Thought Monitoring Across Typologically Diverse Languages
par: Onyame, Eric, et autres
Publié: (2026)

UProp: Investigating the Uncertainty Propagation of LLMs in Multi-Step Agentic Decision-Making
par: Duan, Jinhao, et autres
Publié: (2025)

STAR-1: Safer Alignment of Reasoning LLMs with 1K Data
par: Wang, Zijun, et autres
Publié: (2025)

LLM Unlearning Reveals a Stronger-Than-Expected Coreset Effect in Current Benchmarks
par: Pal, Soumyadeep, et autres
Publié: (2025)

FLAME: Factuality-Aware Alignment for Large Language Models
par: Lin, Sheng-Chieh, et autres
Publié: (2024)

Revisiting the Superficial Alignment Hypothesis
par: Raghavendra, Mohit, et autres
Publié: (2024)

Improving Long Text Understanding with Knowledge Distilled from Summarization Model
par: Liu, Yan, et autres
Publié: (2024)

FACTORY: A Challenging Human-Verified Prompt Set for Long-Form Factuality
par: Chen, Mingda, et autres
Publié: (2025)

Low-rank finetuning for LLMs: A fairness perspective
par: Das, Saswat, et autres
Publié: (2024)

Decoding Compressed Trust: Scrutinizing the Trustworthiness of Efficient LLMs Under Compression
par: Hong, Junyuan, et autres
Publié: (2024)

DERA: Dense Entity Retrieval for Entity Alignment in Knowledge Graphs
par: Wang, Zhichun, et autres
Publié: (2024)

Layer-Level Self-Exposure and Patch: Affirmative Token Mitigation for Jailbreak Attack Defense
par: Ouyang, Yang, et autres
Publié: (2025)

Superficial Safety Alignment Hypothesis
par: Li, Jianwei, et autres
Publié: (2024)

Knowledgeable Preference Alignment for LLMs in Domain-specific Question Answering
par: Zhang, Yichi, et autres
Publié: (2023)

Two Heads Are Better Than One: Integrating Knowledge from Knowledge Graphs and Large Language Models for Entity Alignment
par: Yang, Linyao, et autres
Publié: (2024)

GTBench: Uncovering the Strategic Reasoning Limitations of LLMs via Game-Theoretic Evaluations
par: Duan, Jinhao, et autres
Publié: (2024)

Towards Understanding and Improving Knowledge Distillation for Neural Machine Translation
par: Zhang, Songming, et autres
Publié: (2023)

Krul: Efficient State Restoration for Multi-turn Conversations with Dynamic Cross-layer KV Sharing
par: Wen, Junyi, et autres
Publié: (2025)

LLM-Align: Utilizing Large Language Models for Entity Alignment in Knowledge Graphs
par: Chen, Xuan, et autres
Publié: (2024)

SudoLM: Learning Access Control of Parametric Knowledge with Authorization Alignment
par: Liu, Qin, et autres
Publié: (2024)

From Superficial Patterns to Semantic Understanding: Fine-Tuning Language Models on Contrast Sets
par: Petrov, Daniel
Publié: (2025)

CrossIn: An Efficient Instruction Tuning Approach for Cross-Lingual Knowledge Alignment
par: Lin, Geyu, et autres
Publié: (2024)

Can LLMs Estimate Student Struggles? Human-AI Difficulty Alignment with Proficiency Simulation for Item Difficulty Prediction
par: Li, Ming, et autres
Publié: (2025)

Shifting Attention to Relevance: Towards the Predictive Uncertainty Quantification of Free-Form Large Language Models
par: Duan, Jinhao, et autres
Publié: (2023)

Learning Facts at Scale with Active Reading
par: Lin, Jessy, et autres
Publié: (2025)

RA-DIT: Retrieval-Augmented Dual Instruction Tuning
par: Lin, Xi Victoria, et autres
Publié: (2023)

Safety Alignment Can Be Not Superficial With Explicit Safety Signals
par: Li, Jianwei, et autres
Publié: (2025)

Condor: Enhance LLM Alignment with Knowledge-Driven Data Synthesis and Refinement
par: Cao, Maosong, et autres
Publié: (2025)

When Style Breaks Safety: Defending LLMs Against Superficial Style Alignment
par: Xiao, Yuxin, et autres
Publié: (2025)

Inverse Scaling in Test-Time Compute
par: Gema, Aryo Pradipta, et autres
Publié: (2025)

TruthPrInt: Mitigating Large Vision-Language Models Object Hallucination Via Latent Truthful-Guided Pre-Intervention
par: Duan, Jinhao, et autres
Publié: (2025)

Post-training an LLM for RAG? Train on Self-Generated Demonstrations
par: Finlayson, Matthew, et autres
Publié: (2025)

Large Knowledge Model: Perspectives and Challenges
par: Chen, Huajun
Publié: (2023)

Understanding Layer Significance in LLM Alignment
par: Shi, Guangyuan, et autres
Publié: (2024)

Knowledge-Level Consistency Reinforcement Learning: Dual-Fact Alignment for Long-Form Factuality
par: Li, Junliang, et autres
Publié: (2025)

SKA-Bench: A Fine-Grained Benchmark for Evaluating Structured Knowledge Understanding of LLMs
par: Liu, Zhiqiang, et autres
Publié: (2025)

Scaling Textual Gradients via Sampling-Based Momentum
par: Ding, Zixin, et autres
Publié: (2025)