Papers
arxiv:2611.06728

Turns Matter: Advancing Streaming Video RL via Turn-Level Credit

Published on Nov 4
· Submitted by
Xiangnan Wu
on Nov 5
Authors:
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,

Abstract

TurnWise, a turn-level GRPO framework, improves streaming video understanding by aligning learning with individual streaming decisions through turn-level credit assignment and turn-mean policy optimization, reducing redundancy without a learned critic or extra rollouts.

Streaming video understanding requires models to repeatedly decide whether to respond or continue observing. Recently, Group Relative Policy Optimization (GRPO) has been applied to this setting, with existing approaches emphasizing reward design while retaining a single trajectory-level advantage across turns. Such shared credit can reinforce redundant replies after earlier successes and discourage useful follow-ups after earlier failures. We propose TurnWise, a turn-level GRPO framework that aligns learning with individual streaming decisions. TurnWise addresses the mismatch between trajectory-level feedback and turn-level interaction, promoting informative responses and reducing redundancy. Specifically, turn-level credit assignment uses silence-completed prefix scoring to construct group-relative advantages from each decision's remaining return, separating its learning signal from earlier rewards. Turn-mean policy optimization averages token losses within each turn and then across turns, giving response and waiting decisions equal weight regardless of output length. These complementary designs improve streaming policy learning without a learned critic or additional rollouts. Extensive experiments show our method outperforms prior RL approaches for streaming video understanding.

Community

Paper submitter

Streaming video understanding requires models to repeatedly decide whether to respond or continue observing. Recently, Group Relative Policy Optimization (GRPO) has been applied to this setting, with existing approaches emphasizing reward design while retaining a single trajectory-level advantage across turns. Such shared credit can reinforce redundant replies after earlier successes and discourage useful follow-ups after earlier failures. We propose TurnWise, a turn-level GRPO framework that aligns learning with individual streaming decisions. TurnWise addresses the mismatch between trajectory-level feedback and turn-level interaction, promoting informative responses and reducing redundancy. Specifically, turn-level credit assignment uses silence-completed prefix scoring to construct group-relative advantages from each decision's remaining return, separating its learning signal from earlier rewards. Turn-mean policy optimization averages token losses within each turn and then across turns, giving response and waiting decisions equal weight regardless of output length. These complementary designs improve streaming policy learning without a learned critic or additional rollouts. Extensive experiments show our method outperforms prior RL approaches for streaming video understanding.

Shared credit across turns rewarding a redundant reply after an early win is such a relatable flaw. Does TurnWise help most when the stream is quiet for long stretches, or when events come fast?

Sign up or log in to comment

Get this paper in your agent:

hf papers read 2611.06728
Don't have the latest CLI?
curl -LsSf https://hf.co/cli/install.sh | bash

Models citing this paper 1

Datasets citing this paper 0

No dataset linking this paper

Cite arxiv.org/abs/2611.06728 in a dataset README.md to link it from this page.

Spaces citing this paper 0

No Space linking this paper

Cite arxiv.org/abs/2611.06728 in a Space README.md to link it from this page.

Collections including this paper 0

No Collection including this paper

Add this paper to a collection to link it from this page.