·
DataBubble
  • Home
  • Models
  • News
  • Compare
  • Boards
  • Pricing
  • About
  • Newsletter
  • Methodology
  • Contact
Latest
Bridging the Gap: Converting Read Text to Conversational Dialogue5h◆Sharp and Stable Label-Free Calibration Certificates from Logical Coherence5h◆Backward Compatibility in Zero-Bit Neural Audio Watermarking5h◆Disagreement-Guided Weak-to-Strong Routing5h◆One Is Not Enough: How People Use Multiple AI Models in Everyday Life5h◆Compression Does Not End at Selection: Post-Selection Positional Geometry for Sparse KV Caches5h◆Unfragmenting Nawatl NMT: Single-Variant Specialization for Masewalkopa5h◆Do LLMs have core beliefs?5h◆L2G-SR: Learning to Guide LLM-based Symbolic Regression5h◆When Reality Outpaces Cognition: A New Scaling Principle for Reliable Intelligence5h◆ARIA: Runtime Inspection and Amendment for Multi-Agent Systems5h◆Confidence-Calibrated Inference Expansion for Evaluator-Guided Test-Time Reasoning5h◆KnowGPO: Knowledge-Grouped Policy Optimization for Search Agents5h◆Reference-Guided Turn-Taking Style Transfer for Full-Duplex Spoken Language Models5h◆Wolof-Adapted Industry-Ready Open-Weight Language Models5h◆CoTranslator: Reverse to Forward Chain-of-Thought Translator for Reasoning Distillation5h◆Amortizing Sequential Experimental Design in Reasoning LLMs5h◆Probing Negation Understanding in Vision-Language Models With Object Attention5h◆SHRED: Self-Play for Heterogeneous Road-Users without Expert Demonstrations5h◆Wrong Votes, Right Gradients: Why Majority-Vote Rewards Track Verifiable RL5h◆Bridging the Gap: Converting Read Text to Conversational Dialogue5h◆Sharp and Stable Label-Free Calibration Certificates from Logical Coherence5h◆Backward Compatibility in Zero-Bit Neural Audio Watermarking5h◆Disagreement-Guided Weak-to-Strong Routing5h◆One Is Not Enough: How People Use Multiple AI Models in Everyday Life5h◆Compression Does Not End at Selection: Post-Selection Positional Geometry for Sparse KV Caches5h◆Unfragmenting Nawatl NMT: Single-Variant Specialization for Masewalkopa5h◆Do LLMs have core beliefs?5h◆L2G-SR: Learning to Guide LLM-based Symbolic Regression5h◆When Reality Outpaces Cognition: A New Scaling Principle for Reliable Intelligence5h◆ARIA: Runtime Inspection and Amendment for Multi-Agent Systems5h◆Confidence-Calibrated Inference Expansion for Evaluator-Guided Test-Time Reasoning5h◆KnowGPO: Knowledge-Grouped Policy Optimization for Search Agents5h◆Reference-Guided Turn-Taking Style Transfer for Full-Duplex Spoken Language Models5h◆Wolof-Adapted Industry-Ready Open-Weight Language Models5h◆CoTranslator: Reverse to Forward Chain-of-Thought Translator for Reasoning Distillation5h◆Amortizing Sequential Experimental Design in Reasoning LLMs5h◆Probing Negation Understanding in Vision-Language Models With Object Attention5h◆SHRED: Self-Play for Heterogeneous Road-Users without Expert Demonstrations5h◆Wrong Votes, Right Gradients: Why Majority-Vote Rewards Track Verifiable RL5h◆
News/Seeing What Should Be Heard: Diagnosing and Repairing Cross-Modal Shortcuts in Omni-Modal LLMs
arxiv
PublishedOctober 1, 2026 at 4:00 AM

Seeing What Should Be Heard: Diagnosing and Repairing Cross-Modal Shortcuts in Omni-Modal LLMs

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

arXiv:2609.36798v1 Announce Type: cross Abstract: Omni-modal large language models (LLMs) are expected to answer a question using the modality it explicitly refers to. However, existing training paradigms rarely verify whether models actually follow this modality, because multimodal inputs from the

Stay posted· Newsletter

Three short emails a week: Monday, Wednesday and Friday. Top movers, new models and the most-referenced story.

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

Your email is used only to send The Bubble Brief. Privacy policy

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
arxivBridging the Gap: Converting Read Text to Conversational Dialogue5harxivSharp and Stable Label-Free Calibration Certificates from Logical Coherence5harxivBackward Compatibility in Zero-Bit Neural Audio Watermarking5harxivDisagreement-Guided Weak-to-Strong Routing5h
The Bubble Brief
MON · WED · FRI

Read AI insights every Monday, Wednesday and Friday — top movers, new models, story of the week.

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

Your email is used only to send The Bubble Brief. Privacy policy

Originally published on arxiv ↗
PrivacyTermsRefundsLegal notice|AboutMethodologyContact

© 2026 Databubble

Built by Marouane Gazouzi · Readyo, AI interview coach

HomeModelsNews