·
DataBubble
  • Home
  • Models
  • News
  • Compare
  • Boards
  • Pricing
  • About
  • Newsletter
  • Methodology
  • Contact
Latest
SciReC: Diagnostic Evaluation of Multimodal, Multi-Turn Relational Reasoning with Adaptive Interaction5h◆UIC-AIHealth4All at ArchEHR-QA 2026: Answer-First Evidence Grounding for Clinical Question Answering5h◆Select, Don't Train: The Benefits of Modular Entity Disambiguation with LLM-Based Selection5h◆INSPIRE: An Internalize-Then-Improve Approach for Example-Driven Mathematical Reasoning5h◆How Do Linear Probes Emerge? A Circuit-Tracing Framework with Concept-Targeted Attribution5h◆Trajectory-Level Speculative Decoding for Diffusion Language Models5h◆Below the Noise Floor: Bimodal Seed Collapse and Distinct Failure Modes in Small-Model Knowledge Distillation5h◆Load-Bearing Context: The Question Damage Score for Evaluating Context Reliance in Linguistic Reasoning5h◆Informational Antilocality and the Locality Bias in LLMs5h◆EvoHarmBench: Breaking Content Moderation with Iterative Human-Like Evasion5h◆OpenStamp: A Watermark for Open-Source Language Models5h◆Lexically conditioned realization ambiguity in Korean predicate morphology5h◆QUORUM: QUality-Optimized Routing Using Multiple annotators5h◆Predicting Turn-Taking Outcomes in Multi-Party Conversation: Interpretable Modelling of Speech and Gaze Dynamics with Interpersonal Closeness5h◆When Linguistic and Internal Confidence Diverge in Large Language Models5h◆CultureConverse: A Multilingual Multi-turn Simulation Harness for Culturally Grounded Assistance in East and Southeast Asia5h◆A Unified Framework to Elicit Structured Feedback for Interpretable Multi-Trait Essay Scoring5h◆PACE: Publisher-Adaptive Content Extraction via Agentic Automation5h◆Fidelity Is Not Enough: Dispatch-Level Instrumentation for Agentic Datasheet Extraction5h◆Quantifying Affective Bias in Low-Resource Media: Large-Scale Emotion Profiling of Bengali Headlines5h◆SciReC: Diagnostic Evaluation of Multimodal, Multi-Turn Relational Reasoning with Adaptive Interaction5h◆UIC-AIHealth4All at ArchEHR-QA 2026: Answer-First Evidence Grounding for Clinical Question Answering5h◆Select, Don't Train: The Benefits of Modular Entity Disambiguation with LLM-Based Selection5h◆INSPIRE: An Internalize-Then-Improve Approach for Example-Driven Mathematical Reasoning5h◆How Do Linear Probes Emerge? A Circuit-Tracing Framework with Concept-Targeted Attribution5h◆Trajectory-Level Speculative Decoding for Diffusion Language Models5h◆Below the Noise Floor: Bimodal Seed Collapse and Distinct Failure Modes in Small-Model Knowledge Distillation5h◆Load-Bearing Context: The Question Damage Score for Evaluating Context Reliance in Linguistic Reasoning5h◆Informational Antilocality and the Locality Bias in LLMs5h◆EvoHarmBench: Breaking Content Moderation with Iterative Human-Like Evasion5h◆OpenStamp: A Watermark for Open-Source Language Models5h◆Lexically conditioned realization ambiguity in Korean predicate morphology5h◆QUORUM: QUality-Optimized Routing Using Multiple annotators5h◆Predicting Turn-Taking Outcomes in Multi-Party Conversation: Interpretable Modelling of Speech and Gaze Dynamics with Interpersonal Closeness5h◆When Linguistic and Internal Confidence Diverge in Large Language Models5h◆CultureConverse: A Multilingual Multi-turn Simulation Harness for Culturally Grounded Assistance in East and Southeast Asia5h◆A Unified Framework to Elicit Structured Feedback for Interpretable Multi-Trait Essay Scoring5h◆PACE: Publisher-Adaptive Content Extraction via Agentic Automation5h◆Fidelity Is Not Enough: Dispatch-Level Instrumentation for Agentic Datasheet Extraction5h◆Quantifying Affective Bias in Low-Resource Media: Large-Scale Emotion Profiling of Bengali Headlines5h◆
News/How Do Linear Probes Emerge? A Circuit-Tracing Framework with Concept-Targeted Attribution
arxiv
PublishedAugust 31, 2026 at 4:00 AM
—neutral

How Do Linear Probes Emerge? A Circuit-Tracing Framework with Concept-Targeted Attribution

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

arXiv:2608.27510v1 Announce Type: new Abstract: Transcoder attribution graphs are usually trained to explain why a model assigns high probability to a particular next token. We introduce Concept-Targeted Attribution (CTA), which instead trains attribution graphs with respect to a linear probe direct

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
arxivSciReC: Diagnostic Evaluation of Multimodal, Multi-Turn Relational Reasoning with Adaptive Interaction5harxivUIC-AIHealth4All at ArchEHR-QA 2026: Answer-First Evidence Grounding for Clinical Question Answering5harxivSelect, Don't Train: The Benefits of Modular Entity Disambiguation with LLM-Based Selection5harxivINSPIRE: An Internalize-Then-Improve Approach for Example-Driven Mathematical Reasoning5h
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