# Live-OPSD: On-Policy Self-Distillation for Proactive Streaming Large Video Models

DevFeed: [Live-OPSD: On-Policy Self-Distillation for Proactive Streaming Large Video Models](<https://devfeed.tech/articles/live-opsd-on-policy-self-distillation-for-proactive-streaming-large-video-models-104025.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2610.36959>)

Author: Kaixuan Du, Xin Wan, Dai Guan, Ming Chen, Xiaoxi Jiang, Guanjun Jiang

Published: 2026-10-28T04:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [LLM Techniques](<https://devfeed.tech/topics/llm-techniques.md>), [Streaming](<https://devfeed.tech/topics/streaming.md>), [Post-training optimization](<https://devfeed.tech/topics/post-training-optimization.md>), [qwen](<https://devfeed.tech/topics/qwen.md>)

Tags: [abstract](<https://devfeed.tech/tags/abstract.md>), [arxiv](<https://devfeed.tech/tags/arxiv.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [distillation](<https://devfeed.tech/tags/distillation.md>), [large](<https://devfeed.tech/tags/large.md>), [live](<https://devfeed.tech/tags/live.md>), [models](<https://devfeed.tech/tags/models.md>), [policy](<https://devfeed.tech/tags/policy.md>), [proactive](<https://devfeed.tech/tags/proactive.md>), [streaming](<https://devfeed.tech/tags/streaming.md>), [video](<https://devfeed.tech/tags/video.md>)

## AI overview

Live-OPSD trains streaming assistants to decide when and how to respond by distilling guidance from the model's own continuations. During training, the policy first generates from streaming context alone; the same model then re-evaluates that continuation using a privileged hint about relevant evidence, and those distributions guide the unhinted policy. On Qwen3-Omni-30B-A3B, it scored 59.64 on LiveProBench, 5.26 points above the strongest continued-SFT baseline, and scored 65.84 on OVO-Bench, where it remained competitive. The evidence here is an abstract excerpt.

## Source excerpt

arXiv:2610.36959v1 Announce Type: new Abstract: Streaming multimodal assistants must make response decisions while events are still unfolding. They need to recognize when the available evidence warrants a response, produce an appropriate continuation, and remain silent otherwise. Supervised fine-tuning learns from recorded assistant turns, but does not directly train on the response prefixes produced by the evolving policy itself. We introduce Live-OPSD, an on-policy self-distillation method that uses privileged hints available only during training to supervise the policy on its own outputs. The policy first generates a continuation from the streaming context alone. The same model then re-evaluates that continuation with a hint describing the relevant evidence, and the resulting distributions guide the unhinted policy. On Qwen3-Omni-30B-A3B, Live-OPSD attains an average score of 59.64 on LiveProBench, 5.26 points above the strongest continued-SFT baseline, and stays competitive on OVO-Bench (65.84). These results indicate that guiding the policy on its own outputs with privileged hints improves both when and how a streaming assistant responds.