DASH: Divergence-Adaptive Supervision Horizons for On-Policy Self-Distillation of Reasoning Models

• URL: http://arxiv.org/abs/2608.06243v2
• Date: Fri, 07 Aug 2026 03:26:12 GMT
• Title: DASH: Divergence-Adaptive Supervision Horizons for On-Policy Self-Distillation of Reasoning Models
• Abstract summary: Reinforcement learning with verifiable rewards improves the reasoning capabilities of large language models.<n>On-policy self-distillation mitigates signal sparsity by querying a privileged teacher at student-visited prefixes.<n>Standard OPSD still underexploits the temporal structure of the rollout.<n>DASH maps the gap between each local distillation signal and the sequence-level mean to an adaptive propagation gate.
• Score: 26.283978881046107
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Reinforcement learning with verifiable rewards (RLVR) improves the reasoning capabilities of large language models using automatically verifiable outcome signals, but these signals are typically sparse and at the sequence-level. On-policy self-distillation (OPSD) mitigates this sparsity by querying a privileged teacher at student-visited prefixes and providing dense token-level distributional supervision. Although this dense supervision alleviates signal sparsity, we find that standard OPSD still underexploits the temporal structure of the rollout. It assigns every local divergence the same coefficient, regardless of its position or the divergence sequence in which it occurs. In on-policy autoregressive generation, the same divergence magnitude can follow different discrepancy histories, reflecting different evolutions of the mismatch between the teacher and student. Since the local scalar alone cannot distinguish these temporal contexts, standard OPSD cannot adapt its token-level weights to the realized discrepancy sequence. To address this limitation, we propose Divergence-Adaptive Supervision Horizons (DASH). DASH maps the gap between each local distillation signal and the sequence-level mean to an adaptive propagation gate and then uses these gates to control backward multi-step aggregation. By doing so, DASH adjusts token-level supervision weights according to how local divergences evolve during generation. Experiments on three mathematical reasoning benchmarks across three model scales show that DASH improves over our matched vanilla OPSD reruns on every benchmark at all three scales. DASH reuses the teacher and student distributions that OPSD already computes, so the gains require no additional teacher or student forward pass. Code: https://github.com/DBtxy/DASH-OPSD

Related papers

• TTSD-FAR: Test-Time Self-Distillation with Fisher-Anchored Restoration for Missing-Modality Emotion Recognition in LVLMs [83.78708832901194]
  Large video-language models (LVLMs) have shown remarkable performance on multimodal tasks like multimodal emotion recognition (ER) in the wild.<n>However, real-world deployment remains challenging: modalities may be missing or noisy at test time.<n>We propose a solution in which a frozen teacher guides an adaptive low-rank student via self-distillation.<n>Experiments on MELD, DFEW, and BAH under 0%-50% missing modalities show that this unified adaptation-restoration design consistently outperforms entropy-based adaptation, RAG, and perplexity-based generation.
  arXiv  Detail & Related papers  (2026-08-18T23:40:28Z)
• DASH-OPD: Discrepancy-Aware Switching with Hysteresis for On-Policy Distillation [9.897015521661915]
  On-policy distillation (OPD) trains student models on their own rollouts to reduce exposure bias.<n>Existing curriculum learning methods regulate how much teacher support is used according to training progress, but cannot determine when it is needed.<n>We propose DASH-OPD, a new agentic OPD method that can switch executors adaptively and bidirectionally.
  arXiv  Detail & Related papers  (2026-07-31T06:57:17Z)
• H$^2$SD: Hybrid Hindsight Self-Distillation [61.71131241473028]
  Reinforcement learning with verifiable rewards (RLVR) provides reliable outcome supervision for language model reasoning.<n>Existing self-distillation methods add a privileged teacher but typically assign it a fixed role.<n>We introduce Hybrid Hindsight Self-Distillation ($mathrmH2mathrmSD$), which jointly adapts teacher context and update strategy to trajectory correctness.
  arXiv  Detail & Related papers  (2026-07-21T10:47:27Z)
• dOPSD: On-Policy Self-Distillation for Diffusion Language Models [52.60302464420127]
  Diffusion large language models generate text by iteratively denoising a masked sequence.<n>Supervised fine-tuning is off-policy and suffers from exposure bias, while reinforcement learning gives only sparse, sequence-level rewards.<n>On-policy self-distillation (OPSD) offers a promising alternative, using one model as both student and teacher.
  arXiv  Detail & Related papers  (2026-07-05T17:47:29Z)
• DemoPSD: Disagreement-Modulated Policy Self-Distillation [49.30809363804017]
  On-policy self-distillation has emerged as a practical method for training large language models.<n>DemoPSD steers the student toward a weighted geometric combination of the teacher and student distributions.
  arXiv  Detail & Related papers  (2026-07-02T17:58:29Z)
• CRAFT: Counterfactual Credit Assignment from Free Sibling Rollouts for Self-Distilled Agentic Reinforcement Learning [4.663685189987781]
  Self-distilled agentic reinforcement learning augments trajectory-level reward with a token-level distillation loss.<n>The prevailing recipe gates this loss by a single scalar, the teacher-student log-probability gap.<n>We introduce CRAFT, a threepillar credit-assignment scheme that addresses both limitations.
  arXiv  Detail & Related papers  (2026-06-28T16:11:47Z)
• Self-Distilled Policy Gradient [51.95967256628261]
  On-policy self-distillation is a promising source of dense supervision for sparse-reward reinforcement learning.<n>We propose SDPG, a self-distilled policy-gradient framework that combines group-relative verifier advantages with normalized standard deviation.
  arXiv  Detail & Related papers  (2026-06-02T02:31:13Z)
• Respecting Self-Uncertainty in On-Policy Self-Distillation for Efficient LLM Reasoning [43.9367673156851]
  On-policy self-distillation trains a reasoning model on its own rollouts while a teacher provides dense token-level supervision.<n>We propose EGRSD (Entropy-Guided Reinforced Self-Distillation), which unifies token-level updates through three signals.<n> CL-EGRSD, a causal-lookratio variant that distinguishes sustained high-entropy spans from transient high-entropy positions.
  arXiv  Detail & Related papers  (2026-05-13T09:38:20Z)
• Hidden States Know Where Reasoning Diverges: Credit Assignment via Span-Level Wasserstein Distance [8.66471442661456]
  We show that hidden-state distributions contain a useful signal for local reasoning quality.<n>Within each GRPO group, the Wasserstein distance between span-level hidden state distributions increases around regions where their local reasoning quality diverges.<n>We propose textbfSpan-level textbfHidden state textbfEd textbfAdvantage textbfReweighting.
  arXiv  Detail & Related papers  (2026-04-25T14:11:23Z)
• Self-Distilled RLVR [57.37526213765131]
  We show that learning signals solely derived from the privileged teacher result in severe information leakage and unstable long-term training.<n>We propose textbfRLSD (textbfRLVR with textbfSelf-textbfDistillation)<n>This enables RLSD to simultaneously harness the strengths of both RLVR and OPSD, achieving a higher convergence ceiling and superior training stability.
  arXiv  Detail & Related papers  (2026-04-03T15:50:07Z)
• Revisiting On-Policy Distillation: Empirical Failure Modes and Simple Fixes [31.95045602299568]
  On-policy distillation (OPD) is appealing for large language model (LLM) post-training because it evaluates teacher feedback on student-generated rollouts rather than fixed teacher traces.<n>We revisit OPD from the estimator and implementation sides.<n>We identify three failure modes of sampled-token OPD: an imbalanced one-token signal, unreliable teacher guidance on student-generated prefixes, and distortions caused by tokenizer or special-token mismatch.
  arXiv  Detail & Related papers  (2026-03-26T15:35:59Z)
• Unsupervised Controllable Generation with Self-Training [90.04287577605723]
  controllable generation with GANs remains a challenging research problem. We propose an unsupervised framework to learn a distribution of latent codes that control the generator through self-training. Our framework exhibits better disentanglement compared to other variants such as the variational autoencoder.
  arXiv  Detail & Related papers  (2020-07-17T21:50:35Z)

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.