Operational Proto-Introspection in Looped Language Models: Reading Correctness Before the Answer
cs.LG, cs.AI
Submitted: 2026-07-20
Updated: 2026-11-18
Code: https://github.com/VykosMolt/Branching
Project page: https://ouro-
Terminology
Sources
- Reading Human Preference from Looped Transformer Internal States
- Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedback
- Tell me about yourself: LLMs are aware of their learned behaviors
- Looking Inward: Language Models Can Learn About Themselves by Introspection
- Memory-Efficient Looped Transformer: Decoupling Compute from Memory in Looped Language Models
- Unlocking Decoding-time Controllability: Gradient-Free Multi-Objective Alignment with Contrastive Prompts
- Scaling up Test-Time Compute with Latent Reasoning: A Recurrent Depth Approach
- Training Large Language Models to Reason in a Continuous Latent Space
- Measuring Mathematical Problem Solving With the MATH Dataset
- Chain of Thought Monitorability: A New and Fragile Opportunity for AI Safety
- Efficient Memory Management for Large Language Model Serving with PagedAttention
- RewardBench: Evaluating Reward Models for Language Modeling
- Language Models Are Capable of Metacognitive Monitoring and Control of Their Internal Activations
- Let's Verify Step by Step
- Emergent Introspective Awareness in Large Language Models
- Mechanisms of Introspective Awareness
- The Geometry of Truth: Emergent Linear Structure in Large Language Model Representations of True/False Datasets
- SpecInfer: Accelerating Generative Large Language Model Serving with Tree-based Speculative Inference and Verification
- Latent Introspection: Models Can Detect Prior Concept Injections
- Reasoning with Latent Thoughts: On the Power of Looped Transformers
Related papers
- A Physics-Regulated Neural Framework for Learning 3D Grain Growth Dynamics
- FlexMoE: Learning One-for-All Multi-Budget Expert Pruning for Flexible MoE Deployment
- 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