Searcharxiv⌕ Search

arXiv subjects

Ku Onoda

Publications and source records attributed to Ku Onoda.

4 recordsLinked to original sources

How Many Exposures Does a World Model Need? Learning Unseen Game Mechanics at Test Time

World models enable agents to plan and learn from imagined experience, but in an unfamiliar environment they must learn its mechanics from limited interaction. A mechanic is a rule for how an action or an interaction changes part of the environment. Each occurrence of a mechanic that the model observes is one exposure. We ask how many exposures a world model needs to predict the effects of a new environment's mechanics. We address this question on ARC-AGI-3, a benchmark of interactive games defined by their programs. We introduce MechEval, which labels 69 recurring mechanics across 20 games from these programs. It scores each prediction on the cells the mechanic changes at each exposure. As a test bed, we build WARC (World model for ARC-AGI-3), a 35M-parameter vision transformer that is pretrained on other games and updates all of its parameters after each transition. On held-out games, WARC learns to predict four in five mechanics with at least 80% accuracy on the cells they change and keeps that accuracy to the end of the recordings. Half of them reach that accuracy within about 80 exposures, in a single play of the game, and without updates only one in seven does at all. In closed-loop play on the public games, WARC-Agent, which pairs WARC with a fixed graph explorer, triples the explorer's official efficiency score from 0.15 to 0.46, mostly through test-time training. On the hidden games of the Kaggle leaderboard, our entry scores 0.20.

cs.LG↗

Multi-Axis Max@K Reinforcement Learning for Representative Diversity in Text-to-Image Generation

Text-to-image (T2I) models produce realistic, prompt-aligned images, but samples for the same prompt often cover only a few visually distinct modes. For person-centric prompts, this lack of diversity can reflect or amplify demographic skew. We formalize representative diversity as target-mode coverage: how well a set of samples covers predefined modes, such as colors or perceived-race categories. We then propose MultiMaxPO, a reinforcement learning post-training method that improves this coverage in diffusion T2I models by optimizing a multi-axis max@K objective. Given one score per mode for each sample, multi-axis max@K takes the maximum over samples for each mode and then sums these maxima. A sample therefore receives credit for a mode only when it raises that mode's maximum, so different samples are rewarded for different modes. We verify this credit assignment on a synthetic Gaussian mixture and on Stable Diffusion 3.5 Medium with pixel-based color rewards, and then apply it to perceived-appearance fairness. On held-out prompts and under three automatic evaluators, MultiMaxPO improves the Fairness Score over the base model by 0.22-0.36 on Stable Diffusion 3.5 Medium and by 0.07-0.19 on FLUX.1-dev, outperforms matched RL baselines, and maintains image quality and text alignment. Code is available at https://github.com/KuOnoda/multi-axis-maxk.

cs.LG↗

Does "Do Differentiable Simulators Give Better Policy Gradients?" Give Better Policy Gradients?

In policy gradient reinforcement learning, access to a differentiable model enables 1st-order gradient estimation that accelerates learning compared to relying solely on derivative-free 0th-order estimators. However, discontinuous dynamics cause bias and undermine the effectiveness of 1st-order estimators. Prior work addressed this bias by constructing a confidence interval around the REINFORCE 0th-order gradient estimator and using these bounds to detect discontinuities. However, the REINFORCE estimator is notoriously noisy, and we find that this method requires task-specific hyperparameter tuning and has low sample efficiency. This paper asks whether such bias is the primary obstacle and what minimal fixes suffice. First, we re-examine standard discontinuous settings from prior work and introduce DDCG, a lightweight test that switches estimators in nonsmooth regions; with a single hyperparameter, DDCG achieves robust performance and remains reliable with small samples. Second, on differentiable robotics control tasks, we present IVW-H, a per-step inverse-variance implementation that stabilizes variance without explicit discontinuity detection and yields strong results. Together, these findings indicate that while estimator switching improves robustness in controlled studies, careful variance control often dominates in practical deployments.

cs.LG↗

AutoRef: Harness Optimization for Agentic Multi-Reference Image Generation

Recent image generation models can take multiple reference images as input and combine them into a new image. However, multi-reference image generation remains challenging: models may omit or duplicate subjects from the references, or produce images in which multiple subjects appear unnaturally pasted. Recent work has proposed image generation agents that combine image generation models, reasoning models, and a harness, which is an executable program that specifies how reference images are interpreted, how generation is performed, how outputs are diagnosed, and how the final image is selected. In multi-reference generation, however, references play different roles and outputs must satisfy many criteria at once, such as fidelity to each reference and the naturalness of the whole image, so many parts of the harness could be improved, from how references are processed to how outputs are diagnosed. This makes it hard to predict which changes will improve performance and by how much, and good harnesses difficult to design by hand; indeed, human-written harnesses vary widely in performance. We therefore propose AutoRef, which optimizes the harness automatically while keeping both models frozen: a coding agent iteratively rewrites the harness code. AutoRef separates the tasks whose feedback informs proposals from the tasks used to select candidates, and continues the search from a beam of the top-ranked harnesses on the selection tasks. Using this procedure, we discover AutoRef-Harness, which improves the open-weight FLUX.2 [klein] 4B from 5.72 to 7.37 on held-out four-reference tasks of the MultiBanana benchmark, matching or exceeding proprietary models including Nano Banana Pro and GPT-Image-1.5. Without re-optimization, the same harness also improves results when the generator, number of references, benchmark, evaluator, or reasoning model differs from those used in the search.

cs.CV↗