Pith. sign in

REVIEW 5 cited by

Rethinking Reward Signals in Video GRPO: When Scores Become Targets

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2511.19356 v4 pith:D75O2IKG submitted 2025-11-24 cs.CV

Rethinking Reward Signals in Video GRPO: When Scores Become Targets

classification cs.CV
keywords optimizationrewardgrpovideoconsistentdirectionsfidelitygeneration
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
read the original abstract

Group Relative Policy Optimization (GRPO) enables stable and preference-oriented updates via group-wise comparisons for post-training video generation. However, GRPO directly optimizes reward-induced advantages. Under sustained optimization, the reward score can lose fidelity as a proxy for true video quality, consistent with the phenomenon described by Goodhart's Law. This leads to two recurring issues: (i) shortcut-driven optimization under composite objectives and (ii) reward saturation within prompt groups. To address these issues, we introduce TaRoS, a Target-Robust Reward Signaling framework for Video generation GRPO. TaRoS leverages component level performance assessment together with intra-group sparsity to organize multi-aspect rewards towards optimization objectives. In addition, it adaptively downweights components that exhibit saturation, thereby preserving effective optimization directions and mitigating redundancy. This maintains meaningful optimization directions and preserves within-group ranking separation, thereby preventing reward hacking and leading to more reliable policy updates. Extensive experiments show consistent improvements in visual fidelity, motion coherence, and text-video alignment over strong baselines.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 5 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Learning to Credit the Right Steps: Objective-aware Process Optimization for Visual Generation

    cs.CV 2026-04 unverdicted novelty 7.0

    OTCA improves GRPO training for visual generation by estimating step importance in trajectories and adaptively weighting multiple reward objectives.

  2. Physics-Grounded Fluid Video Generation with a Simulation Dataset and Dual-Stream Optical-Flow Supervision

    cs.AI 2026-07 conditional novelty 6.0

    Adding a dual-stream optical-flow decoder and a simulation+real fluid dataset to a frozen video diffusion model improves the physical plausibility of generated pours and splashes.

  3. Twins: Learn to Predict Unified Representations with Focal Loss

    cs.CV 2026-07 conditional novelty 6.0

    Channel-wise concatenation of SigLIP2 and Flux VAE features into one token, trained with a focal-style flow-matching loss, yields a unified representation with 1.59 gFID on ImageNet 256 and VAE-level reconstruction.

  4. Video Models Can Reason with Verifiable Rewards

    cs.CV 2026-05 unverdicted novelty 6.0

    VideoRLVR uses SDE-GRPO optimization, dense decomposed rewards, and Early-Step Focus to train video diffusion models on verifiable reasoning tasks, outperforming supervised fine-tuning and other video generators on Ma...

  5. CogniVerse: Revolutionizing Multi-Modal Retrieval-Augmented Generation with Cognitive Reflection and Geometric Reasoning

    cs.CV 2026-05 unverdicted novelty 3.0

    CogniVerse is a proposed MMRAG framework that combines cognitive reflection for retrieval filtering, Riemannian manifold alignment plus spectral graphs for retrieval, and optimal transport loss for generation, claimin...