ROM: Real-time Overthinking Mitigation via Streaming Detection and Intervention
cs.LG, cs.AI, cs.CL
Submitted: 2026-03-23
Updated: 2026-10-29
Code: https://github.com/SaFo-Lab/ROM
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Terminology
Sources
- Do NOT Think That Much for 2+3=? On the Overthinking of o1-Like LLMs
- Training Verifiers to Solve Math Word Problems
- Gemma 4 Technical Report
- The Llama 3 Herd of Models
- Mitigating Overthinking in Large Reasoning Language Models via Reasoning Path Deviation Monitoring
- Disentangled Safety Adapters Enable Efficient Guardrails and Flexible Inference-Time Alignment
- SyncThink: A Training-Free Strategy to Align Inference Termination with Reasoning Saturation
- Kelp: A Streaming Safeguard for Large Models via Latent Dynamics-Guided Risk Detection
- From Judgment to Interference: Early Stopping LLM Harmful Outputs via Streaming Content Monitoring
- O1-Pruner: Length-Harmonizing Fine-Tuning for O1-Like Reasoning Pruning
- Stop When Reasoning Converges: Semantic-Preserving Early Exit for Reasoning Models
- TERMINATOR: Learning Optimal Exit Points for Early Stopping in Chain-of-Thought Reasoning
- Real-Time Detection of Hallucinated Entities in Long-Form Generation
- GPT-4 Technical Report
- GPQA: A Graduate-Level Google-Proof Q&A Benchmark
- Constitutional Classifiers: Defending against Universal Jailbreaks across Thousands of Hours of Red Teaming
- Stop Overthinking: A Survey on Efficient Reasoning for Large Language Models
- Entropy After </Think> for reasoning model early exiting
- MMLU-Pro: A More Robust and Challenging Multi-Task Language Understanding Benchmark
- Chain-of-Thought Prompting Elicits Reasoning in Large Language Models
Related papers
- OptiMind: Improving Small Language Models for Optimization Modeling through Formulation Guidance and Multi-Turn Learning
- Staying in the Sweet Spot: Boost Reasoning Evolution via Responsive Rollout Difficulty Manipulation
- Examining the Impact of Data and Training Methods on Emergent and Subliminal Misalignment
- Rethinking Black-Box On-Policy Distillation: A Rubric-based Analysis
- (Mis)generalization of Helpful-Only Fine-tuning
- Provably Learning Diffusion Models under the Manifold Hypothesis: Collapse and Refine