MOPD improves on-policy distillation for LLMs by using peer successes for positive patterns and failures for negative examples to create more informative teacher signals.
hub
Self-Distillation Enables Continual Learning
35 Pith papers cite this work. Polarity classification is still indexing.
35
Pith papers citing it
abstract
Continual learning, enabling models to acquire new skills and knowledge without degrading existing capabilities, remains a fundamental challenge for foundation models. While on-policy reinforcement learning can reduce forgetting, it requires explicit reward functions that are often unavailable. Learning from expert demonstrations, the primary alternative, is dominated by supervised fine-tuning (SFT), which is inherently off-policy. We introduce Self-Distillation Fine-Tuning (SDFT), a simple method that enables on-policy learning directly from demonstrations. SDFT leverages in-context learning by using a demonstration-conditioned model as its own teacher, generating on-policy training signals that preserve prior capabilities while acquiring new skills. Across skill learning and knowledge acquisition tasks, SDFT consistently outperforms SFT, achieving higher new-task accuracy while substantially reducing catastrophic forgetting. In sequential learning experiments, SDFT enables a single model to accumulate multiple skills over time without performance regression, establishing on-policy distillation as a practical path to continual learning from demonstrations.
hub tools
claims ledger
- abstract Continual learning, enabling models to acquire new skills and knowledge without degrading existing capabilities, remains a fundamental challenge for foundation models. While on-policy reinforcement learning can reduce forgetting, it requires explicit reward functions that are often unavailable. Learning from expert demonstrations, the primary alternative, is dominated by supervised fine-tuning (SFT), which is inherently off-policy. We introduce Self-Distillation Fine-Tuning (SDFT), a simple method that enables on-policy learning directly from demonstrations. SDFT leverages in-context learning
co-cited works
years
2026 35representative citing papers
ActGuide-RL uses human action data as plan-style guidance in mixed-policy RL to overcome exploration barriers in LLM agents, matching SFT+RL performance on search benchmarks without cold-start training.
Self-distillation token rewards measure input-response-feedback pointwise mutual information, and CREDIT extracts the input-specific component with contrastive baselines to improve LLM reasoning performance.
Distillation signals align better with ideal updates on incorrect student rollouts than correct ones, with optimal teacher context depending on student capacity and task.
RLRT augments GRPO by reinforcing tokens on correct student rollouts that the teacher would not have predicted, outperforming standard self-distillation and exploration baselines on Qwen3 models.
TRACE improves math reasoning by distilling only on annotator-marked critical spans with forward KL on correct key spans, optional reverse KL on errors, and GRPO elsewhere, gaining 2.76 points over GRPO while preserving OOD performance.
TAD improves the accuracy-parallelism trade-off in diffusion LLMs via temporal-aware self-distillation that applies hard labels to soon-to-be-decoded tokens and soft supervision to future tokens.
OPHSD uses harness-augmented models as teachers to distill reasoning capabilities into base LLMs, yielding strong standalone performance on classification and math tasks.
vOPD stabilizes on-policy distillation gradients by subtracting a closed-form per-token negative reverse KL baseline as a detached control variate, preserving unbiasedness while lowering variance and matching expensive full-vocabulary methods.
PBSD derives a reward-reweighted teacher distribution as the analytic optimum of a reward-regularized objective, yielding better stability and performance than KL-based self-distillation on math reasoning and tool-use tasks.
GUI-SD introduces on-policy self-distillation with visually enriched privileged context and entropy-guided weighting, outperforming GRPO and naive OPSD on six GUI grounding benchmarks while improving training efficiency.
NPO uses a policy's own near-future checkpoint as auxiliary trajectories to maximize effective learning signal S = Q/V, improving performance from 57.88 to 63.15 on Qwen3-VL-8B-Instruct with GRPO while accelerating convergence.
A single LLM improves its own reasoning by self-distilling from privileged verified traces as teacher to its question-only student policy, outperforming off-policy distillation and RL on math benchmarks with better token efficiency.
Emergent and subliminal misalignment in LLMs arise from data structure interactions and transfer via benign distillation data, with stronger effects under shared functional structure and on-policy settings.
SORT turns all-wrong prompts into selective learning signals by weighting tokens more predictable under plan guidance from reference solutions, improving over GRPO on reasoning benchmarks especially for weaker models.
ATESD makes teacher exposure to reference reasoning a learnable control variable via a Beta-policy optimized on future student improvement, yielding gains of up to +2.33 points over fixed-exposure self-distillation on AIME and HMMT math benchmarks.
SOD reweights on-policy distillation strength step-by-step using divergence to stabilize tool use in small language model agents, yielding up to 20.86% gains and 26.13% on AIME 2025 for a 0.6B model.
LiteGUI trains 2B/3B-scale GUI agents via SFT-free guided on-policy distillation and multi-solution dual-level GRPO to reach SOTA lightweight performance and compete with larger models.
D-OPSD enables continuous supervised fine-tuning of few-step diffusion models via on-policy self-distillation where the model acts as both teacher (multimodal context) and student (text-only context) on its own roll-outs.
MSD enables cross-lingual safety transfer in LLMs via self-distillation with Dual-Perspective Safety Weighting, improving safety in low-resource languages without target response data.
A silicon-native modular system with parallel live distillation and a tight-bottleneck autoencoder achieves parameter isolation, autonomous task discovery, and strong retention across vision and language tasks without a student fidelity gap.
π-Play uses self-generated question construction paths as privileged information in multi-agent self-distillation to convert sparse-reward self-play into a dense-feedback loop, surpassing supervised search agents and improving efficiency 2-3× over standard self-play.
On-policy distillation works when student and teacher models share thinking patterns and the teacher adds new capabilities, with success tied to alignment on a small set of high-probability tokens.
Lightning OPD is an offline on-policy distillation method that matches standard OPD performance at 4x efficiency by enforcing teacher consistency between SFT and distillation phases.
citing papers explorer
-
Multi-Rollout On-Policy Distillation via Peer Successes and Failures
MOPD improves on-policy distillation for LLMs by using peer successes for positive patterns and failures for negative examples to create more informative teacher signals.
-
Learning Agentic Policy from Action Guidance
ActGuide-RL uses human action data as plan-style guidance in mixed-policy RL to overcome exploration barriers in LLM agents, matching SFT+RL performance on search benchmarks without cold-start training.
-
From Generic Correlation to Input-Specific Credit in On-Policy Self Distillation
Self-distillation token rewards measure input-response-feedback pointwise mutual information, and CREDIT extracts the input-specific component with contrastive baselines to improve LLM reasoning performance.
-
Unmasking On-Policy Distillation: Where It Helps, Where It Hurts, and Why
Distillation signals align better with ideal updates on incorrect student rollouts than correct ones, with optimal teacher context depending on student capacity and task.
-
Rebellious Student: Reversing Teacher Signals for Reasoning Exploration with Self-Distilled RLVR
RLRT augments GRPO by reinforcing tokens on correct student rollouts that the teacher would not have predicted, outperforming standard self-distillation and exploration baselines on Qwen3 models.
-
TRACE: Distilling Where It Matters via Token-Routed Self On-Policy Alignment
TRACE improves math reasoning by distilling only on annotator-marked critical spans with forward KL on correct key spans, optional reverse KL on errors, and GRPO elsewhere, gaining 2.76 points over GRPO while preserving OOD performance.
-
TAD: Temporal-Aware Trajectory Self-Distillation for Fast and Accurate Diffusion LLM
TAD improves the accuracy-parallelism trade-off in diffusion LLMs via temporal-aware self-distillation that applies hard labels to soon-to-be-decoded tokens and soft supervision to future tokens.
-
Training with Harnesses: On-Policy Harness Self-Distillation for Complex Reasoning
OPHSD uses harness-augmented models as teachers to distill reasoning capabilities into base LLMs, yielding strong standalone performance on classification and math tasks.
-
KL for a KL: On-Policy Distillation with Control Variate Baseline
vOPD stabilizes on-policy distillation gradients by subtracting a closed-form per-token negative reverse KL baseline as a detached control variate, preserving unbiasedness while lowering variance and matching expensive full-vocabulary methods.
-
Preference-Based Self-Distillation: Beyond KL Matching via Reward Regularization
PBSD derives a reward-reweighted teacher distribution as the analytic optimum of a reward-regularized objective, yielding better stability and performance than KL-based self-distillation on math reasoning and tool-use tasks.
-
Learn where to Click from Yourself: On-Policy Self-Distillation for GUI Grounding
GUI-SD introduces on-policy self-distillation with visually enriched privileged context and entropy-guided weighting, outperforming GRPO and naive OPSD on six GUI grounding benchmarks while improving training efficiency.
-
Near-Future Policy Optimization
NPO uses a policy's own near-future checkpoint as auxiliary trajectories to maximize effective learning signal S = Q/V, improving performance from 57.88 to 63.15 on Qwen3-VL-8B-Instruct with GRPO while accelerating convergence.
-
Self-Distilled Reasoner: On-Policy Self-Distillation for Large Language Models
A single LLM improves its own reasoning by self-distilling from privileged verified traces as teacher to its question-only student policy, outperforming off-policy distillation and RL on math benchmarks with better token efficiency.
-
Emergent and Subliminal Misalignment Through the Lens of Data-Mediated Transfer
Emergent and subliminal misalignment in LLMs arise from data structure interactions and transfer via benign distillation data, with stronger effects under shared functional structure and on-policy settings.
-
Selective Off-Policy Reference Tuning with Plan Guidance
SORT turns all-wrong prompts into selective learning signals by weighting tokens more predictable under plan guidance from reference solutions, improving over GRPO on reasoning benchmarks especially for weaker models.
-
Adaptive Teacher Exposure for Self-Distillation in LLM Reasoning
ATESD makes teacher exposure to reference reasoning a learnable control variable via a Beta-policy optimized on future student improvement, yielding gains of up to +2.33 points over fixed-exposure self-distillation on AIME and HMMT math benchmarks.
-
SOD: Step-wise On-policy Distillation for Small Language Model Agents
SOD reweights on-policy distillation strength step-by-step using divergence to stabilize tool use in small language model agents, yielding up to 20.86% gains and 26.13% on AIME 2025 for a 0.6B model.
-
LiteGUI: Distilling Compact GUI Agents with Reinforcement Learning
LiteGUI trains 2B/3B-scale GUI agents via SFT-free guided on-policy distillation and multi-solution dual-level GRPO to reach SOTA lightweight performance and compete with larger models.
-
D-OPSD: On-Policy Self-Distillation for Continuously Tuning Step-Distilled Diffusion Models
D-OPSD enables continuous supervised fine-tuning of few-step diffusion models via on-policy self-distillation where the model acts as both teacher (multimodal context) and student (text-only context) on its own roll-outs.
-
Multilingual Safety Alignment via Self-Distillation
MSD enables cross-lingual safety transfer in LLMs via self-distillation with Dual-Perspective Safety Weighting, improving safety in low-resource languages without target response data.
-
Modular Continual Learning via Zero-Leakage Reconstruction Routing and Autonomous Task Discovery
A silicon-native modular system with parallel live distillation and a tight-bottleneck autoencoder achieves parameter isolation, autonomous task discovery, and strong retention across vision and language tasks without a student fidelity gap.
-
$\pi$-Play: Multi-Agent Self-Play via Privileged Self-Distillation without External Data
π-Play uses self-generated question construction paths as privileged information in multi-agent self-distillation to convert sparse-reward self-play into a dense-feedback loop, surpassing supervised search agents and improving efficiency 2-3× over standard self-play.
-
Rethinking On-Policy Distillation of Large Language Models: Phenomenology, Mechanism, and Recipe
On-policy distillation works when student and teacher models share thinking patterns and the teacher adds new capabilities, with success tied to alignment on a small set of high-probability tokens.
-
Lightning OPD: Efficient Post-Training for Large Reasoning Models with Offline On-Policy Distillation
Lightning OPD is an offline on-policy distillation method that matches standard OPD performance at 4x efficiency by enforcing teacher consistency between SFT and distillation phases.
-
Self-Distilled Reinforcement Learning for Co-Evolving Agentic Recommender Systems
CoARS enables co-evolving recommender and user agents by using interaction-derived rewards and self-distilled credit assignment to internalize multi-turn feedback into model parameters, outperforming prior agentic baselines.
-
Self-Improving 4D Perception via Self-Distillation
SelfEvo enables pretrained 4D perception models to self-improve on unlabeled videos via self-distillation, delivering up to 36.5% relative gains in video depth estimation and 20.1% in camera estimation across eight benchmarks.
-
PolicyLong: Towards On-Policy Context Extension
PolicyLong shifts long-context data synthesis to an on-policy loop that re-screens contexts using the evolving model's entropy landscape, producing a self-curriculum that outperforms static offline baselines with larger gains at longer lengths.
-
The Master Key Hypothesis: Unlocking Cross-Model Capability Transfer via Linear Subspace Alignment
The Master Key Hypothesis states that capabilities are low-dimensional directions transferable across models through linear subspace alignment, with UNLOCK demonstrating gains such as 12.1% accuracy improvement on MATH when transferring CoT from 14B to 7B models.
-
On-Policy Distillation with Best-of-N Teacher Rollout Selection
BRTS improves on-policy distillation by sampling multiple teacher rollouts and selecting the best one via a correctness-first then alignment priority rule, yielding gains on AIME and AMC math benchmarks.
-
Self-Play Enhancement via Advantage-Weighted Refinement in Online Federated LLM Fine-Tuning with Real-Time Feedback
SPEAR enables online federated LLM fine-tuning by using feedback-guided self-play to create contrastive pairs trained with maximum likelihood on correct completions and confidence-weighted unlikelihood on incorrect ones, outperforming baselines without ground-truth contexts.
-
UniSD: Towards a Unified Self-Distillation Framework for Large Language Models
UniSD unifies complementary self-distillation mechanisms for autoregressive LLMs and achieves up to +5.4 point gains over base models and +2.8 over baselines across six benchmarks and six models.
-
VISD: Enhancing Video Reasoning via Structured Self-Distillation
VISD adds structured privileged feedback from a judge model and a direction-magnitude decoupling trick to let VideoLLMs learn token-level credit assignment while keeping RL stable, yielding higher accuracy and roughly 2x faster convergence on video reasoning benchmarks.
-
Towards a Data-Parameter Correspondence for LLMs: A Preliminary Discussion
A data-parameter correspondence unifies data-centric and parameter-centric LLM optimizations as dual geometric operations on the statistical manifold via Fisher-Rao metric and Legendre duality.
-
Self-Distillation as a Performance Recovery Mechanism for LLMs: Counteracting Compression and Catastrophic Forgetting
Self-distillation fine-tuning recovers LLM capabilities by aligning the student's high-dimensional hidden-layer manifold with the teacher's, as quantified by CKA correlation with performance gains.
-
Agent Skills for Large Language Models: Architecture, Acquisition, Security, and the Path Forward
The paper surveys agent skills for LLMs across architecture, acquisition, deployment, and security, proposing a four-tier Skill Trust and Lifecycle Governance Framework to address vulnerabilities in community skills.