Learning More by Training Less: Uncertainty-Guided Progressive Contraction for Efficient Reasoning Post-Training
cs.AI
Submitted: 2026-10-09
Updated: 2026-10-09
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Sources
- Training Verifiers to Solve Math Word Problems
- DeepSeek-V4: Towards Highly Efficient Million-Token Context Intelligence
- Beyond Benchmarks: MathArena as an Evaluation Platform for Mathematics with LLMs
- DsDm: Model-Aware Dataset Selection with Datamodels
- GLM-5: from Vibe Coding to Agentic Engineering
- The Llama 3 Herd of Models
- OpenThoughts: Data Recipes for Reasoning Models
- Student-in-the-Loop Chain-of-Thought Distillation via Generation-Time Selection
- Measuring Mathematical Problem Solving With the MATH Dataset
- What Matters in On-Policy Distillation? A Perspective on Data Efficiency and Data Selection
- Distilling Step-by-Step! Outperforming Larger Language Models with Less Training Data and Smaller Model Sizes
- Tailoring the Curriculum: Student-Centered Reasoning Distillation via Dynamic Data-Model Compatibility
- Mistral 7B
- The Signal is in the Steps: Local Scoring for Reasoning Data Selection
- Language Models (Mostly) Know What They Know
- Unified Data Selection for LLM Reasoning
- GPQA: A Graduate-Level Google-Proof Q&A Benchmark
- Distilling Reasoning Capabilities into Smaller Language Models
- Step Rejection Fine-Tuning: A Practical Distillation Recipe
- Gemma 4 Technical Report
Related papers
- MAVEN-T: Reinforced Heterogeneous Distillation for Real-Time Multi-Agent Trajectory Prediction
- Model Discovery Agent: LLM-assisted Bayesian experiment design for data-efficient discovery of mechanistic world models
- Agents' Last Exam
- The Clinician's Veto: Navigating Trust, Liability, and Uncertainty in Autonomous AI Prescribing
- MindHelper: Closed-Loop Embodied Mental-State Reasoning for Precision Intervention
- Incumbent Advantage: Brand Bias and Cognitive Manipulation Dynamics in LLM Recommendation Systems