paper-with-me

홈 › Papers

PromptEcho: Annotation-Free Reward from Vision-Language Models for Text-to-Image Reinforcement Learning

2026-04-14 · Jinlong Liu, Wanggui He, Peng Zhang, Mushui Liu, Hao Jiang, Pipei Huang arxiv

Reinforcement learning (RL) can improve the prompt following capability of text-to-image (T2I) models, yet obtaining high-quality reward signals remains challenging: CLIP Score is too coarse-grained, while VLM-based reward models (e.g., RewardDance) require costly human-annotated preference data and additional fine-tuning. We propose PromptEcho, a reward construction method that requires \emph{no} annotation and \emph{no} reward model training. Given a generated image and a guiding query, PromptEcho computes the token-level cross-entropy loss of a frozen VLM with the original prompt as the label, directly extracting the image-text alignment knowledge encoded during VLM pretraining. The reward is deterministic, computationally efficient, and improves automatically as stronger open-source VLMs become available. For evaluation, we develop DenseAlignBench, a benchmark of concept-rich dense captions for rigorously testing prompt following capability. Experimental results on two state-of-the-art T2I models (Z-Image and QwenImage-2512) demonstrate that PromptEcho achieves substantial improvements on DenseAlignBench (+26.8pp / +16.2pp net win rate), along with consistent gains on GenEval, DPG-Bench, and TIIFBench without any task-specific training. Ablation studies confirm that PromptEcho comprehensively outperforms inference-based scoring with the same VLM, and that reward quality scales with VLM size. We will open-source the trained models and the DenseAlignBench.

📄 PDF Abstract BibTeX arXiv:2604.12652

Code (0)

등록된 구현이 없습니다.

Tasks

Reinforcement Learning

Similar Papers 제목 키워드 기반

PISCES: Annotation-free Text-to-Video Post-Training via Optimal Transport-Aligned Rewards

2026-02-02 · Minh-Quan Le, Gaurav Mittal, Cheng Zhao, David Gu 외 arxiv

Text-to-video (T2V) generation aims to synthesize videos with high visual quality and temporal consistency that are semantically aligned with input text. Reward-based post-training has emerged as a promising direction to…

Reinforcement LearningVideo Generation

Neuron-Aware Data Selection for Annotation-Free LLM Self-Distillation

2026-07-02 · Zhuowei Chen, Xiang Lorraine Li arxiv

Post-training large language models (LLMs) without real-world interaction feedback or human-labeled supervision remains challenging, particularly in specialized domains where expert annotations are costly to obtain. Rece…

Domain GeneralizationOffline RL

SR-GRPO: Stable Rank as an Intrinsic Geometric Reward for Large Language Model Alignment

2025-12-02 · Yixuan Tang, Yi Yang arxiv

Aligning Large Language Models (LLMs) with human preferences typically relies on external supervision, which faces critical limitations: human annotations are scarce and subjective, reward models are vulnerable to reward…

Reinforcement LearningMathematical Reasoning

LIV: Language-Image Representations and Rewards for Robotic Control

2023-06-01 · Yecheng Jason Ma, William Liang, Vaidehi Som, Vikash Kumar 외

We present Language-Image Value learning (LIV), a unified objective for vision-language representation and reward learning from action-free videos with text annotations. Exploiting a novel connection between dual reinfor…

Contrastive LearningImitation Learning

Towards Annotation-Free Validation of MLLMs: A Vision-Language Logical Consistency Metric

2026-05-07 · Ying Gu, Mei Chee Leong, Hui Li Tan, Shangbo Mao 외 arxiv

Dominant accuracy evaluation might reward unwarranted guessing of Large Language Models, and it might not be applicable to novel tasks for model validation without ground-truth (gt) annotation. Based on basic logic princ…