·
DataBubble
  • Home
  • Models
  • News
  • Compare
  • Boards
  • Pricing
  • About
  • Newsletter
  • Methodology
  • Contact
Latest
Announcing SimReady Central: An Open Library of Simulation-Ready 3D Assets for Robotics and AI Factories59m◆An AI Security Agent for University ACMIS: Multi-Vector Threat Detection and Automated Response7h◆Learning Transferable Latent User Preferences for Human-Aligned Decision Making7h◆Repair the Core, Not the Symptom: Planning-Graph Repair for Long-Horizon Agents7h◆IndustryForge-27B: A Domain-Enhanced Multimodal Foundation Model for Industrial CAD7h◆Ethical LLM-Assisted Research: A Framework for Responsible Delegation, Verification, and Epistemic Value7h◆PALATE: Peculiar Application of the Law of Total Expectation to Enhance the Evaluation of Deep Generative Models7h◆WoW-Bench: Evaluating Fine-Grained Acoustic Perception in Audio-Language Models via Marine Mammal Vocalizations7h◆When Synthetic Speech Is All You Have: Better Call GRPO7h◆Reducing belief in conspiracy theories as they unfold using large language models7h◆Harnessing Magnitude-Only and Complex Measurements for Improved Dynamic MRI Reconstruction with Learned Priors7h◆EDISCO: Equivariant DIScrete Diffusion for Euclidean Combinatorial Optimization7h◆Composing What Each Teacher Learned: Delta Distillation for Multi-Teacher On-Policy Learning7h◆CRISP: Fixing Flying Pixels in Latent LiDAR Generation via Diffusion Decoding7h◆WeGR: Towards Unified Generative Retrieval via Behavior-aware Tokenization and Preference Alignment7h◆StartLux-Decision: AutoThink Decision Models with Reinforcement Learning for Calibrated Deliberative Decisions7h◆Many Steps into One: Reasoning Contraction and the Limits of Safe Memory Reuse for LLM Agents7h◆Secure Agentic Coding through Counterexample-Grounded Feedback7h◆Policy Search in Code Space for Multi-Agent Pathfinding7h◆TamedFlow: Trajectory-Aware Distillation for Teacher-Assisted Diffusion Sampling7h◆Announcing SimReady Central: An Open Library of Simulation-Ready 3D Assets for Robotics and AI Factories59m◆An AI Security Agent for University ACMIS: Multi-Vector Threat Detection and Automated Response7h◆Learning Transferable Latent User Preferences for Human-Aligned Decision Making7h◆Repair the Core, Not the Symptom: Planning-Graph Repair for Long-Horizon Agents7h◆IndustryForge-27B: A Domain-Enhanced Multimodal Foundation Model for Industrial CAD7h◆Ethical LLM-Assisted Research: A Framework for Responsible Delegation, Verification, and Epistemic Value7h◆PALATE: Peculiar Application of the Law of Total Expectation to Enhance the Evaluation of Deep Generative Models7h◆WoW-Bench: Evaluating Fine-Grained Acoustic Perception in Audio-Language Models via Marine Mammal Vocalizations7h◆When Synthetic Speech Is All You Have: Better Call GRPO7h◆Reducing belief in conspiracy theories as they unfold using large language models7h◆Harnessing Magnitude-Only and Complex Measurements for Improved Dynamic MRI Reconstruction with Learned Priors7h◆EDISCO: Equivariant DIScrete Diffusion for Euclidean Combinatorial Optimization7h◆Composing What Each Teacher Learned: Delta Distillation for Multi-Teacher On-Policy Learning7h◆CRISP: Fixing Flying Pixels in Latent LiDAR Generation via Diffusion Decoding7h◆WeGR: Towards Unified Generative Retrieval via Behavior-aware Tokenization and Preference Alignment7h◆StartLux-Decision: AutoThink Decision Models with Reinforcement Learning for Calibrated Deliberative Decisions7h◆Many Steps into One: Reasoning Contraction and the Limits of Safe Memory Reuse for LLM Agents7h◆Secure Agentic Coding through Counterexample-Grounded Feedback7h◆Policy Search in Code Space for Multi-Agent Pathfinding7h◆TamedFlow: Trajectory-Aware Distillation for Teacher-Assisted Diffusion Sampling7h◆
News/Router Drift Control: Mitigating Off-Policy Degradation in Mixture-of-Experts Reinforcement Learning
arxiv
PublishedOctober 19, 2026 at 4:00 AM

Router Drift Control: Mitigating Off-Policy Degradation in Mixture-of-Experts Reinforcement Learning

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

arXiv:2610.21682v2 Announce Type: replace-cross Abstract: Reinforcement learning (RL) can become off-policy due to training-inference mismatch, mini-batch updates, and asynchronous training. The resulting mismatch is especially harmful in Mixture-of-Experts (MoE) models, where rollout-training discr

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
arxivAn AI Security Agent for University ACMIS: Multi-Vector Threat Detection and Automated Response7harxivLearning Transferable Latent User Preferences for Human-Aligned Decision Making7harxivRepair the Core, Not the Symptom: Planning-Graph Repair for Long-Horizon Agents7harxivIndustryForge-27B: A Domain-Enhanced Multimodal Foundation Model for Industrial CAD7h
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