Learning from human feedback has been shown to improve text-to-image models. These techniques first learn a reward function that captures what humans care about in the task and then improve the models based on the learned reward function. Even though relatively simple approaches (e.g., rejection sampling based on reward scores) have been investigated, fine-tuning text-to-image models with the reward function remains challenging. In this work, we propose using online reinforcement learning (RL) to fine-tune text-to-image models. We focus on diffusion models, defining the fine-tuning task as an RL problem, and updating the pre-trained text-to-image diffusion models using policy gradient to maximize the feedbacktrained reward. Our approach, coined DPOK, integrates policy optimization with KL regularization. We conduct an analysis of KL regularization for both RL fine-tuning and supervised fine-tuning. In our experiments, we show that DPOK is generally superior to supervised fine-tuning with respect to both image-text alignment and image quality.

Delayed feedback is a critical problem in dynamic recommender systems. In practice, the feedback result often depends on the frequency of recommendation. Most existing online learning literature fails to consider optimization of the recommendation frequency, and regards the reward from each successfully recommended message as equal. In this paper, we consider a novel cascading bandits setting, where individual messages from a selected list are sent to a user periodically. Whenever a user does not like a message, she may abandon the system with a probability positively correlated with the recommendation frequency.

AI models surpassed the highest score recorded for a human test taker in this year's University of Tokyo entrance exam, a new study shows. If an artificial intelligence model such as ChatGPT had taken the entrance exams for Japan's top university in 2026, it would have been assessed as top of the class and admitted for scoring higher than any human test takers, a study by AI startup LifePrompt has found. The research used three major AI models -- ChatGPT 5.2 Thinking by OpenAI, Gemini 3 Pro Preview by Google and Claude Opus 4.5 by Anthropic -- and had them take the actual entrance exam used by the University of Tokyo in February 2026 to assess candidates for courses set to start in April. The university's category 3 science exam, often taken by those who want to enter the institution's medical school, is considered the most difficult exam to pass in Japan. In a time of both misinformation and too much information, quality journalism is more crucial than ever.