mlsanityData to 14 Oct 2026

Trust Functions: Near-Lossless Weak-to-Strong Generalization by Learning When to Trust the Weak Teacher

arXiv 2606.01000 ↗First posted 31 May 2026cs.LG cs.CLcited by 0

Not on the map. Why?

Abstract (arXiv)

Weak-to-strong generalization studies how to improve a strong student using supervision from a weaker teacher when reliable labels are scarce. We view this primarily as a data selection problem, where the key challenge is to identify which weak labels are reliable enough to serve as a training signal. To address this, we introduce trust functions that assign each weak label a scalar trust score and use these scores to filter weak supervision. Across several domains, including world knowledge, quantitative reasoning, and strategy games, trust filtering yields students that match and sometimes surpass ground-truth supervision, achieving near-lossless weak-to-strong generalization. Moreover, trust functions enable an iterative weak-to-strong chain that compounds gains by training a student and reusing it as the next teacher, amplifying the gains. There are several mechanisms to which advantage of trust functions can be attributed.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 44 of its 57 references that are in the map. First posted on arXiv 31 May 2026.

Builds on

44 of 57 references.

References · 44

  1. 2015Deep Residual Learning for Image Recognitioncited by 52,096
  2. 2021Training Verifiers to Solve Math Word Problemscited by 10,109
  3. 2024DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Modelscited by 9,379
  4. 2022Self-Consistency Improves Chain of Thought Reasoning in Language Modelscited by 5,708
  5. 2021Measuring Mathematical Problem Solving With the MATH Datasetcited by 5,371
  6. 2016SGDR: Stochastic Gradient Descent with Warm Restartscited by 5,045
  7. 2017On Calibration of Modern Neural Networkscited by 4,960
  8. 2018Think you have Solved Question Answering? Try ARC, the AI2 Reasoning Challengecited by 4,886
  9. 2023FlashAttention-2: Faster Attention with Better Parallelism and Work Partitioningcited by 2,598
  10. 2018Can a Suit of Armor Conduct Electricity? A New Dataset for Open Book Question Answeringcited by 2,164
  11. 2024HybridFlow: A Flexible and Efficient RLHF Frameworkcited by 2,057
  12. 2020GLU Variants Improve Transformercited by 1,804
32 more
  1. 2019Root Mean Square Layer Normalizationcited by 1,480
  2. 2024Qwen2.5-Math Technical Report: Toward Mathematical Expert Model via Self-Improvementcited by 1,090
  3. 2016Deep Networks with Stochastic Depthcited by 979
  4. 2023Semantic Uncertainty: Linguistic Invariances for Uncertainty Estimation in Natural Language Generationcited by 819
  5. 2023Just Ask for Calibration: Strategies for Eliciting Calibrated Confidence Scores from Language Models Fine-Tuned with Human Feedbackcited by 812
  6. 2017Crowdsourcing Multiple Choice Science Questionscited by 765
  7. 2023The Internal State of an LLM Knows When It's Lyingcited by 659
  8. 2024Interpretable Preferences via Multi-Objective Reward Modeling and Mixture-of-Expertscited by 376
  9. 2025Spurious Rewards: Rethinking Training Signals in RLVRcited by 209
  10. 2025Reasoning Models Know When They're Right: Probing Hidden States for Self-Verificationcited by 159
  11. 2024Easy-to-Hard Generalization: Scalable Alignment Beyond Human Supervisioncited by 96
  12. 2020Fine-Tuning Pre-trained Language Model with Weak Supervision: A Contrastive-Regularized Self-Training Approachcited by 83
  13. 2024Theoretical Analysis of Weak-to-Strong Generalizationcited by 48
  14. 2024Quantifying the Gain in Weak-to-Strong Generalizationcited by 41
  15. 2025CompassVerifier: A Unified and Robust Verifier for LLMs Evaluation and Outcome Rewardcited by 40
  16. 2024Co-Supervised Learning: Improving Weak-to-Strong Generalization with Hierarchical Mixture of Expertscited by 28
  17. 2024Superhuman performance of a large language model on the reasoning tasks of a physiciancited by 28
  18. 2024I-SHEEP: Self-Alignment of LLM from Scratch through an Iterative Self-Enhancement Paradigmcited by 27
  19. 2024Your Weak LLM is Secretly a Strong Teacher for Alignmentcited by 20
  20. 2022Training Subset Selection for Weak Supervisioncited by 19
  21. 2024Provable Weak-to-Strong Generalization via Benign Overfittingcited by 16
  22. 2024Optimizing Language Model's Reasoning Abilities with Weak Supervisioncited by 15
  23. 2025Weak-to-Strong Generalization Even in Random Feature Networks, Provablycited by 15
  24. 2024A transfer learning framework for weak-to-strong generalizationcited by 14
  25. 2024Improving Weak-to-Strong Generalization with Reliability-Aware Alignmentcited by 14
  26. 2017Learning to Learn from Weak Supervision by Full Supervisioncited by 8
  27. 2025Incentivizing Strong Reasoning from Weak Supervisioncited by 8
  28. 2026The Impact of Post-training on Data Contaminationcited by 5
  29. 2025EnsemW2S: Enhancing Weak-to-Strong Generalization with Large Language Model Ensemblescited by 4
  30. 2025W2S-AlignTree: Weak-to-Strong Inference-Time Alignment for Large Language Models via Monte Carlo Tree Searchcited by 4
  31. 2025Weak-to-Strong Generalization with Failure Trajectories: A Tree-based Approach to Elicit Optimal Policy in Strong Modelscited by 3
  32. 2024Reward Modeling with Weak Supervision for Language Modelscited by 1

Built on by

Not cited yet.