GATES: Self-Distillation under Privileged Context with Consensus Gating

• URL: http://arxiv.org/abs/2602.20574v1
• Date: Tue, 24 Feb 2026 05:56:20 GMT
• Title: GATES: Self-Distillation under Privileged Context with Consensus Gating
• Authors: Alex Stein, Furong Huang, Tom Goldstein,
• Abstract summary: We study self-distillation in settings where supervision is unreliable.<n>We focus on document-grounded question answering with asymmetric context.<n>We derive supervision online from tutor consensus by sampling multiple document-grounded reasoning traces.
• Score: 89.62339954332248
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: We study self-distillation in settings where supervision is unreliable: there are no ground truth labels, verifiable rewards, or external graders to evaluate answers. We focus on document-grounded question answering with asymmetric context, where a single model serves as both tutor (with access to a relevant source document during training) and student (answering from the question alone at test time). Rather than assuming tutor correctness, we derive supervision online from tutor consensus by sampling multiple document-grounded reasoning traces and using agreement to gate learning. Conditioned on this reliability signal, we distill knowledge through full tutor reasoning trajectories (not just final answers), providing a dense and stable learning signal. Empirically, this consensus-gated trajectory distillation substantially improves transfer to the document-free student. Held-out in-domain accuracy under asymmetric evaluation improves from 46.0\% to 62.0\%, and average (maj@8) accuracy on public document-free math benchmarks improves from 20.2\% to 35.4\%.

Related papers

• DocVAL: Validated Chain-of-Thought Distillation for Grounded Document VQA [1.580774794371876]
  Document visual question answering (DocVQA) requires models to jointly reason over textual content and spatial layout.<n>Current systems exhibit a sharp accuracy--efficiency trade-off: large teacher models achieve strong grounding but are too expensive for deployment.<n>We propose DocVAL, a validated chain-of-thought distillation framework that transfers the spatial reasoning ability of a large teacher into a deployable student VLM.
  arXiv  Detail & Related papers  (2025-11-27T15:00:58Z)
• Look As You Think: Unifying Reasoning and Visual Evidence Attribution for Verifiable Document RAG via Reinforcement Learning [55.232400251303794]
  Look As You Think (LAT) is a reinforcement learning framework that trains models to produce verifiable reasoning paths with consistent attribution.<n>LAT consistently improves the vanilla model in both single- and multi-image settings, yielding average gains of 8.23% in soft exact match (EM) and 47.0% in IoU@0.5.
  arXiv  Detail & Related papers  (2025-11-15T02:50:23Z)
• CLUE: Non-parametric Verification from Experience via Hidden-State Clustering [64.50919789875233]
  We show that correctness of a solution is encoded as a geometrically separable signature within the trajectory of hidden activations.<n>ClUE consistently outperforms LLM-as-a-judge baselines and matches or exceeds modern confidence-based methods in reranking candidates.
  arXiv  Detail & Related papers  (2025-10-02T02:14:33Z)
• Beyond Agreement: Rethinking Ground Truth in Educational AI Annotation [1.8434042562191815]
  We argue that overreliance on human inter-rater reliability (IRR) as a gatekeeper for annotation quality hampers progress in classifying data.<n>We highlight five examples of complementary evaluation methods, such as multi-label annotation schemes, expert-based approaches, and close-the-loop validity.<n>We call on the field to rethink annotation quality and ground truth--prioritizing validity and educational impact over consensus alone.
  arXiv  Detail & Related papers  (2025-07-31T20:05:26Z)
• Lie Detector: Unified Backdoor Detection via Cross-Examination Framework [68.45399098884364]
  We propose a unified backdoor detection framework in the semi-honest setting.<n>Our method achieves superior detection performance, improving accuracy by 5.4%, 1.6%, and 11.9% over SoTA baselines.<n> Notably, it is the first to effectively detect backdoors in multimodal large language models.
  arXiv  Detail & Related papers  (2025-03-21T06:12:06Z)
• Unsupervised Pretraining for Fact Verification by Language Model Distillation [4.504050940874427]
  We propose SFAVEL (Self-supervised Fact Verification via Language Model Distillation), a novel unsupervised pretraining framework. It distils self-supervised features into high-quality claim-fact alignments without the need for annotations. This is enabled by a novel contrastive loss function that encourages features to attain high-quality claim and evidence alignments.
  arXiv  Detail & Related papers  (2023-09-28T15:53:44Z)
• Faithful Knowledge Distillation [75.59907631395849]
  We focus on two crucial questions with regard to a teacher-student pair: (i) do the teacher and student disagree at points close to correctly classified dataset examples, and (ii) is the distilled student as confident as the teacher around dataset examples? These are critical questions when considering the deployment of a smaller student network trained from a robust teacher within a safety-critical setting.
  arXiv  Detail & Related papers  (2023-06-07T13:41:55Z)
• MDFlow: Unsupervised Optical Flow Learning by Reliable Mutual Knowledge Distillation [12.249680550252327]
  Current approaches impose an augmentation regularization term for continual self-supervision. We propose a novel mutual distillation framework to transfer reliable knowledge back and forth between the teacher and student networks. Our approach, termed MDFlow, achieves state-of-the-art real-time accuracy and generalization ability on challenging benchmarks.
  arXiv  Detail & Related papers  (2022-11-11T05:56:46Z)
• PRover: Proof Generation for Interpretable Reasoning over Rules [81.40404921232192]
  We propose a transformer-based model that answers binary questions over rule-bases and generates the corresponding proofs. Our model learns to predict nodes and edges corresponding to proof graphs in an efficient constrained training paradigm. We conduct experiments on synthetic, hand-authored, and human-paraphrased rule-bases to show promising results for QA and proof generation.
  arXiv  Detail & Related papers  (2020-10-06T15:47:53Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.