Unstable Features, Reproducible Subspaces: Understanding Seed Dependence in Sparse Autoencoders - 惯性聚合

推荐订阅源

The Blog of Author Tim Ferriss

人人都是产品经理

博客园 - 叶小钗

博客园_首页

Help Net Security

aimingoo的专栏

Fortinet All Blogs

DataBreaches.Net

罗磊的独立博客

Kaspersky official blog

Cyber Attacks, Cyber Crime and Cyber Security

Palo Alto Networks Blog

Know Your Adversary

Security Affairs

Engineering at Meta

Recent Commits to openclaw:main

The Exploit Database - CXSecurity.com

LINUX DO - 热门话题

Threat Research - Cisco Blogs

Threat Intelligence Blog | Flashpoint

Privacy International News Feed

Cisco Talos Blog

Tor Project blog

Simon Willison's Weblog

Help Net Security

OSCHINA 社区最新新闻

有赞技术团队

cs.AI updates on arXiv.org

Vulnerabilities – Threatpost

The Hacker News

博客园 - 聂微东

Schneier on Security

Recent Announcements

Darknet – Hacking Tools, Hacker News & Cyber Security

cs.CL updates on arXiv.org

DharmaOCR: Specialized Small Language Models for Structured OCR that outperform Open-Source and Commercial Baselines Learning Adaptive Reasoning Paths for Efficient Visual Reasoning AIM: Asymmetric Information Masking for Visual Question Answering Continual Learning RaTA-Tool: Retrieval-based Tool Selection with Multimodal Large Language Models MM-WebAgent: A Hierarchical Multimodal Web Agent for Webpage Generation Neuro-Oracle: A Trajectory-Aware Agentic RAG Framework for Interpretable Epilepsy Surgical Prognosis The Cost of Language: Centroid Erasure Exposes and Exploits Modal Competition in Multimodal Language Models Rethinking Patient Education as Multi-turn Multi-modal Interaction Knowing When Not to Answer: Evaluating Abstention in Multimodal Reasoning Systems Reasoning Dynamics and the Limits of Monitoring Modality Reliance in Vision-Language Models ADAPT: Benchmarking Commonsense Planning under Unspecified Affordance Constraints OpenMobile: Building Open Mobile Agents with Task and Trajectory Synthesis One RL to See Them All: Visual Triple Unified Reinforcement Learning VisRet: Visualization Improves Knowledge-Intensive Text-to-Image Retrieval HERMES: KV Cache as Hierarchical Memory for Efficient Streaming Video Understanding Counting Without Numbers and Finding Without Words KnowRL: Exploring Knowledgeable Reinforcement Learning for Factuality POP: Prefill-Only Pruning for Efficient Large Model Inference ConfLayers: Adaptive Confidence-based Layer Skipping for Self-Speculative Decoding LongAct: Harnessing Intrinsic Activation Patterns for Long-Context Reinforcement Learning What Is the Minimum Architecture for Prolepsis? Early Irrevocable Commitment Across Tasks in Small Transformers AdaSplash-2: Faster Differentiable Sparse Attention Can Large Language Models Detect Methodological Flaws? Evidence from Gesture Recognition for UAV-Based Rescue Operation Based on Deep Learning Decoupling Scores and Text: The Politeness Principle in Peer Review Correcting Suppressed Log-Probabilities in Language Models with Post-Transformer Adapters Attention to Mamba: A Recipe for Cross-Architecture Distillation Dive into Claude Code: The Design Space of Today's and Future AI Agent Systems Three-Phase Transformer Retrieve, Then Classify: Corpus-Grounded Automation of Clinical Value Set Authoring CURaTE: Continual Unlearning in Real Time with Ensured Preservation of LLM Knowledge Comparison of Modern Multilingual Text Embedding Techniques for Hate Speech Detection Task Route to Rome Attack: Directing LLM Routers to Expensive Models via Adversarial Suffix Optimization IUQ: Interrogative Uncertainty Quantification for Long-Form Large Language Model Generation Context Over Content: Exposing Evaluation Faking in Automated Judges Diagnosing LLM Judge Reliability: Conformal Prediction Sets and Transitivity Violations Similarity-Distance-Magnitude Activations AccelOpt: A Self-Improving LLM Agentic System for AI Accelerator Kernel Optimization Improving Language Models with Intentional Analysis Towards Bridging the Reward-Generation Gap in Direct Alignment Algorithms Social Story Frames: Contextual Reasoning about Narrative Intent and Reception Adaptive Layer Selection for Layer-Wise Token Pruning in LLM Inference De-Anonymization at Scale via Tournament-Style Attribution Large Language Models for Math Education in Low-Resource Languages: A Study in Sinhala and Tamil IROSA: Interactive Robot Skill Adaptation using Natural Language Large Language Model Post-Training: A Unified View of Off-Policy and On-Policy Learning Compressed-Sensing-Guided, Inference-Aware Structured Reduction for Large Language Models MemGround: Long-Term Memory Evaluation Kit for Large Language Models in Gamified Scenarios HUOZIIME: An On-Device LLM-enhanced Input Method for Deep Personalization SeaAlert: Critical Information Extraction From Maritime Distress Communications with Large Language Models How to Fine-Tune a Reasoning Model? A Teacher-Student Cooperation Framework to Synthesize Student-Consistent SFT Data EviSearch: A Human in the Loop System for Extracting and Auditing Clinical Evidence for Systematic Reviews Hierarchical Retrieval Augmented Generation for Adversarial Technique Annotation in Cyber Threat Intelligence Text Chinese Essay Rhetoric Recognition Using LoRA, In-context Learning and Model Ensemble SAGE Celer 2.6 Technical Card Chronological Knowledge Retrieval: A Retrieval-Augmented Generation Approach to Construction Project Documentation Stateful Evidence-Driven Retrieval-Augmented Generation with Iterative Reasoning Benchmarking Linguistic Adaptation in Comparable-Sized LLMs: A Study of Llama-3.1-8B, Mistral-7B-v0.1, and Qwen3-8B on Romanized Nepali Tug-of-War within A Decade: Conflict Resolution in Vulnerability Analysis via Teacher-Guided Retrieval-Augmented Generations QU-NLP at ArchEHR-QA 2026: Two-Stage QLoRA Fine-Tuning of Qwen3-4B for Patient-Oriented Clinical Question Answering and Evidence Sentence Alignment Listen, Correct, and Feed Back: Spoken Pedagogical Feedback Generation An Underexplored Frontier: Large Language Models for Rare Disease Patient Education and Communication -- A scoping review Internal Knowledge Without External Expression: Probing the Generalization Boundary of a Classical Chinese Language Model The PICCO Framework for Large Language Model Prompting: A Taxonomy and Reference Architecture for Prompt Structure Chinese Language Is Not More Efficient Than English in Vibe Coding: A Preliminary Study on Token Cost and Problem-Solving Rate CROP: Token-Efficient Reasoning in Large Language Models via Regularized Prompt Optimization MEME-Fusion@CHiPSAL 2026: Multimodal Ablation Study of Hate Detection and Sentiment Analysis on Nepali Memes ReviewGrounder: Improving Review Substantiveness with Rubric-Guided, Tool-Integrated Agents EuropeMedQA Study Protocol: A Multilingual, Multimodal Medical Examination Dataset for Language Model Evaluation Tracking the Temporal Dynamics of News Coverage of Catastrophic and Violent Events LLM Predictive Scoring and Validation: Inferring Experience Ratings from Unstructured Text Purging the Gray Zone: Latent-Geometric Denoising for Precise Knowledge Boundary Awareness Faithfulness Serum: Mitigating the Faithfulness Gap in Textual Explanations of LLM Decisions via Attribution Guidance Shuffle the Context: RoPE-Perturbed Self-Distillation for Long-Context Adaptation When PCOS Meets Eating Disorders: An Explainable AI Approach to Detecting the Hidden Triple Burden APEX-MEM: Agentic Semi-Structured Memory with Temporal Reasoning for Long-Term Conversational AI BiCon-Gate: Consistency-Gated De-colloquialisation for Dialogue Fact-Checking Generating Concept Lexicalizations via Dictionary-Based Cross-Lingual Sense Projection The Autocorrelation Blind Spot: Why 42% of Turn-Level Findings in LLM Conversation Analysis May Be Spurious Hierarchical vs. Flat Iteration in Shared-Weight Transformers MARCA: A Checklist-Based Benchmark for Multilingual Web Search Filling in the Mechanisms: How do LMs Learn Filler-Gap Dependencies under Developmental Constraints? Psychological Steering of Large Language Models CobwebTM: Probabilistic Concept Formation for Lifelong and Hierarchical Topic Modeling PeerPrism: Peer Evaluation Expertise vs Review-writing AI Mechanistic Decoding of Cognitive Constructs in LLMs NLP needs Diversity outside of 'Diversity' CausalDetox: Causal Head Selection and Intervention for Language Model Detoxification StoryCoder: Narrative Reformulation for Structured Reasoning in LLM Code Generation Pushing the Boundaries of Multiple Choice Evaluation to One Hundred Options Fact4ac at the Financial Misinformation Detection Challenge Task: Reference-Free Financial Misinformation Detection via Fine-Tuning and Few-Shot Prompting of Large Language Models CURA: Clinical Uncertainty Risk Alignment for Language Model-Based Risk Prediction SPAGBias: Uncovering and Tracing Structured Spatial Gender Bias in Large Language Models Which bird does not have wings: Negative-constrained KGQA with Schema-guided Semantic Matching and Self-directed Refinement CoPA: Benchmarking Personalized Question Answering with Data-Informed Cognitive Factors Dissecting Failure Dynamics in Large Language Model Reasoning Modeling LLM Unlearning as an Asymmetric Two-Task Learning Problem Domain Fine-Tuning FinBERT on Finnish Histopathological Reports: Train-Time Signals and Downstream Correlations MARS$^2$: Scaling Multi-Agent Tree Search via Reinforcement Learning for Code Generation Pangu-ACE: Adaptive Cascaded Experts for Educational Response Generation on EduBench Exploring and Testing Skill-Based Behavioral Profile Annotation: Human Operability and LLM Feasibility under Schema-Guided Execution

Unstable Features, Reproducible Subspaces: Understanding Seed Dependence in Sparse Autoencoders

[Submitted on 10 Jun 2026] · 2026-06-11 · via cs.CL updates on arXiv.org

此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。原文来自 — 版权归原作者所有。