Yuta Oshima

Yuta Oshima, Yusuke Iwasawa, Masahiro Suzuki et al.

Video world models have attracted significant attention for their ability to produce high-fidelity future visual observations conditioned on past observations and navigation actions. Temporally- and spatially-consistent, long-term world modeling has been a long-standing problem, unresolved with even recent state-of-the-art models, due to the prohibitively expensive computational costs for long-context inputs. In this paper, we propose WorldPack, a video world model with efficient compressed memory, which significantly improves spatial consistency, fidelity, and quality in long-term generation despite much shorter context length. Our compressed memory consists of trajectory packing and memory retrieval; trajectory packing realizes high context efficiency, and memory retrieval maintains the consistency in rollouts and helps long-term generations that require spatial reasoning. Our performance is evaluated with LoopNav, a benchmark on Minecraft, specialized for the evaluation of long-term consistency, and we verify that WorldPack notably outperforms strong state-of-the-art models.

Shohei Taniguchi, Keno Harada, Gouki Minegishi et al.

Adam is one of the most popular optimization algorithms in deep learning. However, it is known that Adam does not converge in theory unless choosing a hyperparameter, i.e., $β_2$, in a problem-dependent manner. There have been many attempts to fix the non-convergence (e.g., AMSGrad), but they require an impractical assumption that the gradient noise is uniformly bounded. In this paper, we propose a new adaptive gradient method named ADOPT, which achieves the optimal convergence rate of $\mathcal{O} ( 1 / \sqrt{T} )$ with any choice of $β_2$ without depending on the bounded noise assumption. ADOPT addresses the non-convergence issue of Adam by removing the current gradient from the second moment estimate and changing the order of the momentum update and the normalization by the second moment estimate. We also conduct intensive numerical experiments, and verify that our ADOPT achieves superior results compared to Adam and its variants across a wide range of tasks, including image classification, generative modeling, natural language processing, and deep reinforcement learning. The implementation is available at https://github.com/iShohei220/adopt.

Yuta Oshima, Shohei Taniguchi, Masahiro Suzuki et al.

Given the remarkable achievements in image generation through diffusion models, the research community has shown increasing interest in extending these models to video generation. Recent diffusion models for video generation have predominantly utilized attention layers to extract temporal features. However, attention layers are limited by their computational costs, which increase quadratically with the sequence length. This limitation presents significant challenges when generating longer video sequences using diffusion models. To overcome this challenge, we propose leveraging state-space models (SSMs) as temporal feature extractors. SSMs (e.g., Mamba) have recently gained attention as promising alternatives due to their linear-time memory consumption relative to sequence length. In line with previous research suggesting that using bidirectional SSMs is effective for understanding spatial features in image generation, we found that bidirectionality is also beneficial for capturing temporal features in video data, rather than relying on traditional unidirectional SSMs. We conducted comprehensive evaluations on multiple long-term video datasets, such as MineRL Navigate, across various model sizes. For sequences up to 256 frames, SSM-based models require less memory to achieve the same FVD as attention-based models. Moreover, SSM-based models often deliver better performance with comparable GPU memory usage. Our codes are available at https://github.com/shim0114/SSM-Meets-Video-Diffusion-Models.

Yuta Oshima, Daiki Miyake, Kohsei Matsutani et al.

Recent text-to-image generation models have acquired the ability of multi-reference generation and editing; the ability to inherit the appearance of subjects from multiple reference images and re-render them under new contexts. However, the existing benchmark datasets often focus on the generation with single or a few reference images, which prevents us from measuring the progress on how model performance advances or pointing out their weaknesses, under different multi-reference conditions. In addition, their task definitions are still vague, typically limited to axes such as "what to edit" or "how many references are given", and therefore fail to capture the intrinsic difficulty of multi-reference settings. To address this gap, we introduce $\textbf{MultiBanana}$, which is carefully designed to assesses the edge of model capabilities by widely covering multi-reference-specific problems at scale: (1) varying the number of references, (2) domain mismatch among references (e.g., photo vs. anime), (3) scale mismatch between reference and target scenes, (4) references containing rare concepts (e.g., a red banana), and (5) multilingual textual references for rendering. Our analysis among a variety of text-to-image models reveals their superior performances, typical failure modes, and areas for improvement. MultiBanana will be released as an open benchmark to push the boundaries and establish a standardized basis for fair comparison in multi-reference image generation. Our data and code are available at https://github.com/matsuolab/multibanana .

Yuta Oshima, Masahiro Suzuki, Yutaka Matsuo et al.

The remarkable progress in text-to-video diffusion models enables the generation of photorealistic videos, although the content of these generated videos often includes unnatural movement or deformation, reverse playback, and motionless scenes. Recently, an alignment problem has attracted huge attention, where we steer the output of diffusion models based on some measure of the content's goodness. Because there is a large room for improvement of perceptual quality along the frame direction, we should address which metrics we should optimize and how we can optimize them in the video generation. In this paper, we propose diffusion latent beam search with lookahead estimator, which can select a better diffusion latent to maximize a given alignment reward at inference time. We then point out that improving perceptual video quality with respect to alignment to prompts requires reward calibration by weighting existing metrics. This is because when humans or vision language models evaluate outputs, many previous metrics to quantify the naturalness of video do not always correlate with the evaluation. We demonstrate that our method improves the perceptual quality evaluated on the calibrated reward, VLMs, and human assessment, without model parameter update, and outputs the best generation compared to greedy search and best-of-N sampling under much more efficient computational cost. The experiments highlight that our method is beneficial to many capable generative models, and provide a practical guideline: we should prioritize the inference-time compute allocation into enabling the lookahead estimator and increasing the search budget, rather than expanding the denoising steps.

Yuta Oshima, Masahiro Suzuki, Yutaka Matsuo

Multimodal variational autoencoders (VAEs) aim to capture shared latent representations by integrating information from different data modalities. A significant challenge is accurately inferring representations from any subset of modalities without training an impractical number (2^M) of inference networks for all possible modality combinations. Mixture-based models simplify this by requiring only as many inference models as there are modalities, aggregating unimodal inferences. However, they suffer from information loss when modalities are missing. Alignment-based VAEs address this by aligning unimodal inference models with a multimodal model through minimizing the Kullback-Leibler (KL) divergence but face issues due to amortization gaps, which compromise inference accuracy. To tackle these problems, we introduce multimodal iterative amortized inference, an iterative refinement mechanism within the multimodal VAE framework. This method overcomes information loss from missing modalities and minimizes the amortization gap by iteratively refining the multimodal inference using all available modalities. By aligning unimodal inference to this refined multimodal posterior, we achieve unimodal inferences that effectively incorporate multimodal information while requiring only unimodal inputs during inference. Experiments on benchmark datasets show that our approach improves inference performance, evidenced by higher linear classification accuracy and competitive cosine similarity, and enhances cross-modal generation, indicated by lower FID scores. This demonstrates that our method enhances inferred representations from unimodal inputs.