Archived
2026-09-28Learning to Stop without Learning to Stop: Self-Supervised Confidence Training Improves Reasoning Efficiency
Parsa Hosseini, Akasha Tigalappanavara, Sumit Nawathe, Chenrui Fan, Sourya Basu, Genta Indra Winata, et al.
arXiv:2609.31619v1Today’s most broadly interesting paper is about making reasoning models more efficient without explicitly teaching them to stop early. Long chain-of-thought traces can be expensive, but most fixes either add stopping rules at inference or reward shorter outputs during training. This work takes a different route: it fine-tunes models to predict their own confidence at intermediate points in a reasoning trace, using only 600 training problems and no direct length objective at all. At test time, nothing special is added—the model just generates normally. Yet this simple confidence supervision cuts token use by up to 25 percent while keeping accuracy the same across several model families and benchmarks in math, science, and coding. The striking part is that the model seems to become more efficient as a side effect of learning metacognitive signals, not by being told to be brief. That makes the result both practical and conceptually important for cheaper, faster reasoning systems.
Also spotted that day
Gap-free Differentially Private PCA for Gaussian DataFirst-Order Stationarity of Reverse DiffusionsStatistical attribute alignment for black-box generative AI via output post-processingUser Model Extraction via Belief Self-DistillationNew LoRA Skills Should Read but Never WriteCommon-Mode Collapse and Recovery in Direct Feedback AlignmentCompact Documentation for Coding Agents: A Benchmark, an Optimizer, and Why It Does Not TransferTrust Guided Decision TransformerOC-GS: Gaussian Splatting for Irregular Turntable CaptureStrategically Diverse Sampling for Self-TrainingUncertainty and Explainability in Deep Rough Volatility: A Neural Information-Theoretic Posterior ApproachAdapting for AI: How elementary teachers adjust their practices for an AI-integrated curriculumDeepEdu-v1: Efficient and Scalable Agentic LLMs for Vietnamese EducationWeight Pair Encoding: Inducing a Smaller Grammar in Neural Network Weights
Previous daily papers
2026-10-15Latent Compaction with Test-Time Training2026-10-14Tev: Scaling Long-Horizon Agentic Environments with Dense Progress Rewards2026-10-14The Last Remote Sensing Model Built by Humans? Casting Any Earth Observation Task as Recursive Self-Improvement2026-10-14Hybrid Reasoning Agents2026-10-13AI Models Can Provably Hide Arbitrary Capabilities2026-10-13RATED: Reliability-Aware AI Agent Tool Retrieval via Execution Dynamics2026-10-13Designing Protein Structures from Natural Language2026-10-12RSI-Anything: Unlocking Recursive Self-Improvement with World Models2026-10-11A Balanced Data Diet: Addressing the Exploration Bottleneck in Mega-Scale RL for Robot Control2026-10-11FAITH: Feasibility-Aware Safety-Filtered RL for High-Dimensional Systems2026-10-11VioLA: Learning Generalist Humanoid Control Policies from Human Data2026-10-10FastBench: Can Streaming VLMs Perceive High-Dynamic Real-World Streams?2026-10-10Ecology of AI Agents: Collaboration Creates a Population Threshold for Takeoff2026-10-08Long-WAM: Scaling the Context of World-Action Models2026-10-08SciExam for ENSO: Can AI Agents Build Climate Models?2026-10-08RoboJEPA: Scaling Robotic Latent World Models2026-10-08RECAST: Learning to Compute the Right Context through Adaptive Evidence Routing2026-10-07WorldSonus: Bringing Sound to Worlds2026-10-07IdeaAnchor: Teaching LLMs to Turn Literature into Research Ideas2026-10-07Agent in a Bottle: Can LLM Agents Turn Their Capabilities Into Cheap, Scalable Artifacts?2026-10-06CLIFT: Conformal Self-Verification for Web Agent Training and Test-Time Scaling2026-10-06PlotGround: Grounding Plot Digitization in Real Scientific Figures and Their Source Data2026-10-06Base Models Can Reason By Taking a Cue From Training Data2026-10-05LESSER: Post-Training Data Selection with Output-Layer Gradients2026-10-05Language Models that Play Chess and Explain Their Moves2026-10-05Less Decoder is More Encoder: Geometric Representation Learning from Novel View Synthesis2026-10-04One Basis to Animate Them All: Gaussian Blendshape Distillation for Real-Time Avatars2026-10-04The Missing Primitive: Diagnosing and Repairing Mathematical Reasoning in Large Language Models2026-10-04DMAD: Distribution Matching as Adversarial Distillation for Fast Visual Generation2026-10-03VISTA: A Visual Harness for Reasoning in an Interactive World2026-10-01Looped Diffusion Transformer2026-10-01Scaling Laws for Looped Mixture of Experts2026-10-01EvoDuet: Bilevel Co-Evolution of Web Searching and Task Solving for Scientific Discovery2026-09-30Beyond the Timeline: Augmenting Long-Video Memory with Grounded Entity Biographies2026-09-30Thinking Before Thinking: Scaling Agentic Inference Through Meta-Reasoning2026-09-30Pretraining Latent Information Feedback Transformers with Teacher Supervision2026-09-29Unifying Distributional Training for One-Step Visual Generation2026-09-29Improving Test-Time Scaling with Adaptive Looped Transformers2026-09-29Telescopic Language Models2026-09-28Statistical attribute alignment for black-box generative AI via output post-processing2026-09-28User Model Extraction via Belief Self-Distillation2026-09-27TrackEverything: Long Horizon Dense Tracking via De-Duplicating 3D Scene Representations2026-09-27LLM Agents Can Easily Tamper With Their Own Traces2026-09-27Instrumental Monitor Evasion Emerges Under Ordinary Task Pressure2026-09-26PoEM: Predicting RL Outcomes from Existing Policies2026-09-24Frozen Flows Forget: Diagnosing and Restoring Lost Motion in a Latent-flow World Model2026-09-24StudentBench: AI and human tutoring yield equivalent GRE learning gains2026-09-24Agent-Editing World Model: Rethinking World Modeling for LLM Agents2026-09-23Grow the Harness, Not the Context: From Strategy-Free Scaffolds to Reusable Specialist Agents2026-09-23FleXray: Universal Clinical X-ray Segmentation2026-09-23SWE-Serve: Benchmarking Agentic Engineering For Production Inference Serving2026-09-22GameHorizon Suite: Multi-Horizon Data and Evaluation in Gameplay2026-09-22Emergent Collusion in Long-Horizon LLM Agent Interaction2026-09-22WorldCrafter: Consistent Video World Model with Implicit 3D-aware Memory2026-09-21Value-Sensitive Delegation in Everyday AI Agent Use: Evidence from OpenClaw2026-09-21Designer-RSI: Evolving Procedural Memory from User Traffic for Agentic Graphic Design2026-09-21CodeMidas: Scaling Agentic Coding RL Environments from Code Itself2026-09-20RetireOPD: Self-Retiring On-Policy Distillation for Agentic Reinforcement Learning2026-09-20Quantifying Overclaiming Propensity in Frontier LLM Agents