arXiv:2606. 19162v1 Announce Type: new Abstract: Score- and flow-matching models often rely on preference-based reinforcement learning for two purposes: aligning with subjective preferences and, surprisingly, recovering properties such as visual realism and coherent object structure that matching-based training is intended to learn from the data itself.
The paper introduces SatisDive, a training‑free inference method that balances reward and diversity in text‑to‑image diffusion by enforcing a reward floor for each image and a diversity cutoff for the batch. By adjusting the reward floor, the method traces a Pareto frontier between worst‑candidate reward and batch diversity. Experiments on Pick‑a‑Pic show that SatisDive consistently outperforms FK steering, improving worst‑candidate reward by up to 0.70 in some settings and Pareto‑dominating FK steering across overlapping DreamSim ranges.
arXiv:2606. 17979v1 Announce Type: new Abstract: Existing RL post-training methods for text-to-image generation usually convert the final-image reward into a single scalar advantage and apply it with the same strength to the entire generative trajectory.
arXiv:2608.21425v1 Announce Type: cross Abstract: Video generation is central to AI-powered content creation. Aligning generated videos with human preferences is a key criterion for evaluating genera...
arXiv:2608. 09385v1 Announce Type: cross Abstract: Generative AI models are primarily designed to imitate the data distribution, an objective that neither corrects diversity lost by a learned generator nor defines how generation should extend beyond the diversity of the data itself.
arXiv:2601. 12401v2 Announce Type: replace-cross Abstract: Reinforcement learning (RL) has emerged as a powerful paradigm for fine-tuning large-scale generative models, such as diffusion and flow models, to align with complex human preferences and user-specified tasks.
The paper presents a post‑training approach for text‑to‑image models that combines a preference reward, trained on large human preference data, with rubric‑based rewards that assess prompt faithfulness and other desirable traits. The authors show that a simple reward composition strategy outperforms a naive weighted average, leading to significant Elo gains on the Arena leaderboard for models like Flux2dev and Ideogram‑4. They also release Arena‑T2I‑Training, a 1K subset of data to aid reproducible research in post‑training.
T-DIG is a text-conditioned diversity framework that enables users to generate multiple images from a single prompt and then request variations along specific axes—such as background, style, or clothing—using natural language. The system employs an LLM policy to produce K axis-conditioned prompts or edit instructions for a frozen image generator or editor, trained with GRPO and a diversity reward. T‑DIG works with open‑weight, distilled, and proprietary models, and the authors release a new dataset of human‑written axes and real user interactions to benchmark text‑conditioned diversity.
Text-to-image (T2I) platforms typically return several images per prompt so users can explore different visual interpretations and choose one that matches their intent. However, T2I models are usually...
Practical image generation requires both human preference alignment and few-step efficiency, yet combining reinforcement learning (RL) with distillation is often multi-stage and may erode RL gains We...
Generative AI models are primarily designed to imitate the data distribution, an objective that neither corrects diversity lost by a learned generator nor defines how generation should extend beyond the diversity of the data itself. We introduce Imaginative Generative AI (IGA), a framework that makes diversity part of the target-distribution design problem: among distributions close to a reference, IGA selects one whose spectral diversity reaches a prescribed level.
arXiv:2606. 03937v1 Announce Type: new Abstract: While token-level entropy is commonly recognized as effective for credit assignment in text-only reinforcement learning with verifiable rewards (RLVR), it remains unclear whether this mechanism still holds in visual reasoning.