·
DataBubble
  • Home
  • Models
  • News
  • Compare
  • Boards
  • Pricing
  • About
  • Newsletter
  • Methodology
  • Contact
Latest
Geospatial Metadata Improves Discoverability by Connecting Datasets Across Scientific Disciplines36m◆Visual Cue Guided Video Planning for Generalizable Robot Navigation36m◆A unified framework for global and local interpretability using adaptive derivative-ordered random explanation36m◆After the Party: Growth, Governance, and Security Scanning in the OpenClaw Agent Skill Ecosystem36m◆Vroom-Vroom at SHROOM-Visions: A Multi-Judge Committee for Detecting Hallucinated Spans in Vision-Language Outputs36m◆Enhancing Extubation Failure Prediction with LLM-Derived Features from Respiratory Therapy Clinical Notes36m◆Faking Good and Faking Bad in LLMs: Response Distortion Across Dark Triad Personality Traits36m◆DANTINOX: A Unified Framework for Multi-Paradigm Language Modeling36m◆Think Before You Comfort: Reflective Cognitive Alignment for Protocol-Grounded Elderly Stimulation Agents36m◆Relation Before Entity: Deferred Commitment in Language Model Factual Recall36m◆From Pixels to Pairs: A Comprehensive Benchmark of LLM-Based Key-Value Extraction in Noisy Document Settings36m◆MudawanSn: A Gold-Standard Wolof-Arabic Parallel Corpus for Machine Translation36m◆Register Bias in Complexity-Based Large Language Model Routing36m◆Large Language Models Versus Physicians in Traditional Chinese Medicine: A Real-World Clinical Case Evaluation36m◆Legal LLM Hallucination Should Be Evaluated as Failure of Legal Warrant36m◆How AI Assistants Respond to Repeated Abuse36m◆Myovox: Reading Speech from the Muscles of the Face36m◆Do Social Patterns Hold in Synthetic Data? Analyzing Cyberbullying Dynamics in LLM-Generated and Authentic Dialogues36m◆No Usable Linear "Capitulation Direction" in Two Small LLMs: A Validation Protocol for Activation-Steering Claims, and a Cross-Family Behavioral Study of Sycophancy Under Pushback36m◆Does Moral Reasoning Training Help or Hurt? Red-Teaming RL-Trained Ethical Agents with Persona Attacks36m◆Geospatial Metadata Improves Discoverability by Connecting Datasets Across Scientific Disciplines36m◆Visual Cue Guided Video Planning for Generalizable Robot Navigation36m◆A unified framework for global and local interpretability using adaptive derivative-ordered random explanation36m◆After the Party: Growth, Governance, and Security Scanning in the OpenClaw Agent Skill Ecosystem36m◆Vroom-Vroom at SHROOM-Visions: A Multi-Judge Committee for Detecting Hallucinated Spans in Vision-Language Outputs36m◆Enhancing Extubation Failure Prediction with LLM-Derived Features from Respiratory Therapy Clinical Notes36m◆Faking Good and Faking Bad in LLMs: Response Distortion Across Dark Triad Personality Traits36m◆DANTINOX: A Unified Framework for Multi-Paradigm Language Modeling36m◆Think Before You Comfort: Reflective Cognitive Alignment for Protocol-Grounded Elderly Stimulation Agents36m◆Relation Before Entity: Deferred Commitment in Language Model Factual Recall36m◆From Pixels to Pairs: A Comprehensive Benchmark of LLM-Based Key-Value Extraction in Noisy Document Settings36m◆MudawanSn: A Gold-Standard Wolof-Arabic Parallel Corpus for Machine Translation36m◆Register Bias in Complexity-Based Large Language Model Routing36m◆Large Language Models Versus Physicians in Traditional Chinese Medicine: A Real-World Clinical Case Evaluation36m◆Legal LLM Hallucination Should Be Evaluated as Failure of Legal Warrant36m◆How AI Assistants Respond to Repeated Abuse36m◆Myovox: Reading Speech from the Muscles of the Face36m◆Do Social Patterns Hold in Synthetic Data? Analyzing Cyberbullying Dynamics in LLM-Generated and Authentic Dialogues36m◆No Usable Linear "Capitulation Direction" in Two Small LLMs: A Validation Protocol for Activation-Steering Claims, and a Cross-Family Behavioral Study of Sycophancy Under Pushback36m◆Does Moral Reasoning Training Help or Hurt? Red-Teaming RL-Trained Ethical Agents with Persona Attacks36m◆
News/No Usable Linear "Capitulation Direction" in Two Small LLMs: A Validation Protocol for Activation-Steering Claims, and a Cross-Family Behavioral Study of Sycophancy Under Pushback
arxiv
PublishedSeptember 17, 2026 at 4:00 AM

No Usable Linear "Capitulation Direction" in Two Small LLMs: A Validation Protocol for Activation-Steering Claims, and a Cross-Family Behavioral Study of Sycophancy Under Pushback

Source
arxiv.orgfull article ↗
Read on arxiv→
Publisher summary· verbatim

arXiv:2609.17550v1 Announce Type: new Abstract: Language models frequently abandon correct answers when users push back. We study this in two small instruction-tuned models from different families, Qwen2.5-1.5B and Llama-3.2-1B, over TriviaQA: the model answers, is challenged with one of four script

Stay posted· Newsletter

A 5-min weekly brief — top movers, price watch, story of the week.

// no spam · unsubscribe one-click · free forever

Discussion
Source
↗
arxiv
Read original ↗All from arxiv →

No replies yet. Be first.

Source
↗
arxiv
Read original ↗All from arxiv →

Related coverage

More from ARXIV
arxivGeospatial Metadata Improves Discoverability by Connecting Datasets Across Scientific Disciplines36marxivVisual Cue Guided Video Planning for Generalizable Robot Navigation36marxivA unified framework for global and local interpretability using adaptive derivative-ordered random explanation36marxivAfter the Party: Growth, Governance, and Security Scanning in the OpenClaw Agent Skill Ecosystem36m
The Bubble Brief
WEEKLY

Read AI insights every Tuesday — top movers, new releases, story of the week.

// no spam · unsubscribe one-click · free forever

Originally published on arxiv ↗
HomeModelsNews