# Claim-Level Rubric Rewards for Video Caption Reinforcement Learning

arXiv: https://arxiv.org/abs/2607.05150 · HTML: https://mlsanity.com/paper/2607.05150/

## Key facts

- Cited by 3 papers in the map (data through 14 Oct 2026).
- Placed in Diffusion and multimodal model alignment, one of 10 directions in LLM alignment / RLHF / preference optimisation.
- Builds on 23 of its 49 references that are in the map.
- First posted on arXiv 6 Jul 2026.

## Placed in

- LLM alignment / RLHF / preference optimisation / [Diffusion and multimodal model alignment](https://mlsanity.com/alignment/diffusion-and-multimodal-model-alignment/)

## Builds on (most cited first, 23 of 23)

- [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://mlsanity.com/paper/2402.03300/) (2024)
- [Qwen3 Technical Report](https://mlsanity.com/paper/2505.09388/) (2025)
- [Qwen2.5-VL Technical Report](https://mlsanity.com/paper/2502.13923/) (2025)
- [Qwen3-VL Technical Report](https://mlsanity.com/paper/2511.21631/) (2025)
- [HybridFlow: A Flexible and Efficient RLHF Framework](https://mlsanity.com/paper/2409.19256/) (2024)
- [InternVL3.5: Advancing Open-Source Multimodal Models in Versatility, Reasoning, and Efficiency](https://mlsanity.com/paper/2508.18265/) (2025)
- [LLaVA-Video: Video Instruction Tuning With Synthetic Data](https://mlsanity.com/paper/2410.02713/) (2024)
- [Rubrics as Rewards: Reinforcement Learning Beyond Verifiable Domains](https://mlsanity.com/paper/2507.17746/) (2025)
- [The False Promise of Imitating Proprietary LLMs](https://mlsanity.com/paper/2305.15717/) (2023)
- [A Survey of On-Policy Distillation for Large Language Models](https://mlsanity.com/paper/2604.00626/) (2026)
- [RM-R1: Reward Modeling as Reasoning](https://mlsanity.com/paper/2505.02387/) (2025)
- [Tarsier: Recipes for Training and Evaluating Large Video Description Models](https://mlsanity.com/paper/2407.00634/) (2024)
- [Molmo2: Open Weights and Data for Vision-Language Models with Video Understanding and Grounding](https://mlsanity.com/paper/2601.10611/) (2026)
- [OpenRubrics: Towards Scalable Synthetic Rubric Generation for Reward Modeling and LLM Alignment](https://mlsanity.com/paper/2510.07743/) (2025)
- [Tarsier2: Advancing Large Vision-Language Models from Detailed Video Description to Comprehensive Video Understanding](https://mlsanity.com/paper/2501.07888/) (2025)
- [EventHallusion: Diagnosing Event Hallucinations in Video LLMs](https://mlsanity.com/paper/2409.16597/) (2024)
- [VideoCap-R1: Enhancing MLLMs for Video Captioning via Structured Thinking](https://mlsanity.com/paper/2506.01725/) (2025)
- [Auto-Rubric: Learning From Implicit Weights to Explicit Rubrics for Reward Modeling](https://mlsanity.com/paper/2510.17314/) (2025)
- [Reinforcing Chain-of-Thought Reasoning with Self-Evolving Rubrics](https://mlsanity.com/paper/2602.10885/) (2026)
- [OwlCap: Harmonizing Motion-Detail for Video Captioning via HMD-270K and Caption Set Equivalence Reward](https://mlsanity.com/paper/2508.18634/) (2025)
- [BLIP3-KALE: Knowledge Augmented Large-Scale Dense Captions](https://mlsanity.com/paper/2411.07461/) (2024)
- [RubiCap: Rubric-Guided Reinforcement Learning for Dense Image Captioning](https://mlsanity.com/paper/2603.09160/) (2026)
- [AVC-DPO: Aligned Video Captioning via Direct Preference Optimization](https://mlsanity.com/paper/2507.01492/) (2025)

## Built on by (most cited first, 3 of 3)

- [LEDGERMIND: Provenance-Constrained Multimodal Agentic Reasoning with a Structured Evidence Ledger](https://mlsanity.com/paper/2607.28374/) (2026)
- [OmniReasoner: Thinking with Long Audio-Video via Native Tool Use](https://mlsanity.com/paper/2607.19339/) (2026)
- [Learning Multimodal Embeddings with Evidence-Aligned Readout](https://mlsanity.com/paper/2609.33659/) (2026)

## Abstract (arXiv)

In this paper, we introduce Claim-Level Rubric Rewards (CuRe), a structured reward framework designed to address the reward-design bottleneck in reinforcement learning for dense video captioning. Existing reward designs generally fall into two categories: holistic response-level judgment across heterogeneous criteria, or alignment-based evaluation against reference captions. However, both paradigms suffer from fundamental limitations. Holistic rewards struggle to ensure factual accuracy and are prone to stylistic reward hacking, while reference-based rewards overly rely on rigid textual alignment, failing to preserve the completeness and diversity inherent to open-ended generation tasks. To address these challenges, CuRe reformulates reward modeling as fine-grained claim-level verification. Specifically, CuRe decomposes captions into category-aware atomic claims through a structured rubric, converting holistic evaluation into simpler and more reliable claim-level verification.