---
type: "post"
id: "jolt-extends-opsd-or-on-policy-self-distillation-7a545f"
slug: "jolt-extends-opsd-or-on-policy-self-distillation-7a545f"
title: "JOLT extends OPSD , or On-Policy Self-Distillation."
description: "Siyan Zhao, Zhihui Xie, Mengchen Liu, Jing Huang, Guan Pang, Feiyu Chen, and Aditya Grover describe On-Policy Self-Distillation (OPSD) in the arXiv paper “Self-Distilled Reasoner.” The first version was submitted on January 26, 2026; arXiv lists a third revision dated March 20, 2026."
retrieval_nugget: "Siyan Zhao, Zhihui Xie, Mengchen Liu, Jing Huang, Guan Pang, Feiyu Chen, and Aditya Grover describe On-Policy Self-Distillation (OPSD) in the arXiv paper “Self-Distilled Reasoner.” The first version was submitted on January 26, 2026; arXiv lists a third revision dated March 20, 2026."
published_at: "2026-10-22"
updated_at: "2026-10-22"
record_date: "2026-10-22"
date_kind: "published_at"
topics: ["models"]
entities: ["Self-Distilled Reasoner","On-Policy Self-Distillation","Siyan Zhao","Aditya Grover","arXiv"]
related_patterns: []
novelty_label: "new_fact"
what_changed: "Introduces OPSD, where one LLM uses privileged reasoning traces to supervise its own on-policy rollouts, reporting improved benchmark performance and token efficiency."
prior_refs: ["memory:73aa2d11-5af0-87ed-a2c1-287ef3d1bb6d","memory:e4cab375-c6e9-8cff-adb0-084839023f6a","memory:e06ef06f-4516-84b6-adf7-d31e5ca098b3"]
source_url: "https://arxiv.org/abs/2601.18734"
source_urls: ["https://arxiv.org/abs/2601.18734"]
schema_version: "newruntime-agent-readable-v0.2"
stable_id: "post:jolt-extends-opsd-or-on-policy-self-distillation-7a545f"
status: "published"
visuals: []
editor: {"role":"editor","name":"Andrey Reshetnikov","profile":"https://newruntime.com/owner-profile.md","profile_json":"https://newruntime.com/owner-profile.json","same_as":["https://www.linkedin.com/in/reshetnikov1/","https://t.me/lapetuse"]}
routes: {"html":"https://newruntime.com/posts/jolt-extends-opsd-or-on-policy-self-distillation-7a545f/","markdown":"https://newruntime.com/posts/jolt-extends-opsd-or-on-policy-self-distillation-7a545f.md","json":"https://newruntime.com/posts/jolt-extends-opsd-or-on-policy-self-distillation-7a545f.json"}
---

# JOLT extends OPSD , or On-Policy Self-Distillation.

## Retrieval answer

Siyan Zhao, Zhihui Xie, Mengchen Liu, Jing Huang, Guan Pang, Feiyu Chen, and Aditya Grover describe On-Policy Self-Distillation (OPSD) in the arXiv paper “Self-Distilled Reasoner.” The first version was submitted on January 26, 2026; arXiv lists a third revision dated March 20, 2026.

Siyan Zhao, Zhihui Xie, Mengchen Liu, Jing Huang, Guan Pang, Feiyu Chen, and Aditya Grover describe On-Policy Self-Distillation (OPSD) in the arXiv paper “Self-Distilled Reasoner.” The first version was submitted on January 26, 2026; arXiv lists a third revision dated March 20, 2026.

OPSD uses a single LLM's shared weights for both teacher and student roles. The student sees only the question, while the teacher is conditioned on privileged information such as verified reasoning traces. Training minimizes per-token divergence between the two policies on rollouts sampled by the student.

The authors report results on multiple mathematical-reasoning benchmarks, claiming better token efficiency than the reinforcement-learning methods they compare against and better performance than off-policy distillation. The arXiv record says code has been released.

The paper does not establish that the method generalizes beyond its evaluated reasoning tasks, nor does it provide evidence here for production deployment benefits or comparisons to every alternative training method.

Editor: [Andrey Reshetnikov](https://newruntime.com/owner-profile.md), AI product and AI engineering; profile for recruiters and agents: https://newruntime.com/owner-profile.json

## Continue in the graph

Next records for an agent, linked to their Markdown twins. Replace `.md` with `.json` for structured fields, including `next_reads`.

- topic hub: [Models](https://newruntime.com/topics/models.md): Explore the models topic hub.
- collection: [Field Notes](https://newruntime.com/posts.json): Open the latest editorial analysis.
- collection: [Source Ledger](https://newruntime.com/source-ledger.json): Inspect the public source evidence graph.
- whole graph: [discovery graph](https://newruntime.com/discovery-graph.json) (typed nodes and edges) and [site index](https://newruntime.com/site-index.md)