# Flux-OPD: On-Policy Distillation with Evolving Contexts

Page: https://stenobird.com/podcast/daily-paper-cast-7079649/flux-opd-on-policy-distillation-with-evolving-contexts
Text version: https://stenobird.com/podcast/daily-paper-cast-7079649/flux-opd-on-policy-distillation-with-evolving-contexts.md
Podcast: [Daily Paper Cast](https://stenobird.com/podcast/daily-paper-cast-7079649)
Published: 2026-08-01T03:37:16+00:00
Episode link: https://share.transistor.fm/s/c0b436c9
Audio file: https://media.transistor.fm/c0b436c9/ae2079c6.mp3
Processing state: not_requested
JSON: https://stenobird.com/v1/public/podcasts/daily-paper-cast-7079649/episodes/flux-opd-on-policy-distillation-with-evolving-contexts
Duration seconds: 1238

## Resource

🤗 Upvotes: 38 | cs.LG, cs.AI Authors: Yuran Wang, Zekun Wang, Bohan Zeng, Ruixu Zhang, Wenxuan Liu, Liu Yang, Yifan Dai, Yang Shi, Bozhou Li, Chengzhuo Tong, Daili Hua, Yuanxing Zhang, Wentao Zhang Title: Flux-OPD: On-Policy Distillation with Evolving Contexts Arxiv: http://arxiv.org/abs/2607.28022v1 Abstract: Large language model training in open-ended domains lacks verifiable rewards, making task preferences difficult to formalize as effective supervision. Contexts can convey such preferences, yet provide little additional supervision once distilled into the student, motivating contexts that evolve with student performance. However, directly using evolving contexts as in-training supervision results in an unstable distillation target and conflicting distributions, requiring mechanisms to stabilize target and downweight conflicts. In this paper, we analyze the effect of contexts through a decomposition of the reverse KL objective, revealing two findings: the student is distilled toward the geometric mean of context-conditioned teachers, and the objective contains a conflict term that measures conflicts among these teachers. Based on this decomposition, we propose Flux-OPD, an OPD paradigm that uses evolving contexts as in-training supervision to capture task preferences in open-ended domains. Flux-OPD treats the differences between context-conditioned and context-free teachers as contextual difference signals, injects them as contextual corrections into the context-free teacher anchor, and weights their correction strength using the conflict term as an indicator. Experiments on open-ended tasks show that Flux-OPD outperforms existing OPD paradigms, highlighting the potential to combine teacher supervision with evolving contexts.

## Actions

- request_transcript: `POST https://stenobird.com/v1/public/podcasts/daily-paper-cast-7079649/episodes/flux-opd-on-policy-distillation-with-evolving-contexts/transcription-requests` — Idempotently request low-priority transcript generation for this episode.
- read_markdown: `GET https://stenobird.com/podcast/daily-paper-cast-7079649/flux-opd-on-policy-distillation-with-evolving-contexts.md` — Read the agent-friendly Markdown representation of this episode resource.

A page view does not enqueue transcription. Agents should invoke `request_transcript` explicitly when they need this episode processed.

## Transcript

Full transcripts are not published on public pages unless there is a clear rights basis.