T2VTextBench: A Human Evaluation Benchmark for Textual Control in Video Generation Models

Thanks to recent advancements in scalable deep architectures and large-scale pretraining, text-to-video generation has achieved unprecedented capabilities in producing high-fidelity, instruction-following content across a wide range of styles, enabling applications in advertising, entertainment, and education. However, these models' ability to render precise on-screen text, such as captions or mathematical formulas, remains largely untested, posing significant challenges for applications requiring exact textual accuracy. In this work, we introduce T2VTextBench, the first human-evaluation benchmark dedicated to evaluating on-screen text fidelity and temporal consistency in text-to-video models. Our suite of prompts integrates complex text strings with dynamic scene changes, testing each model's ability to maintain detailed instructions across frames. We evaluate ten state-of-the-art systems, ranging from open-source solutions to commercial offerings, and find that most struggle to generate legible, consistent text. These results highlight a critical gap in current video generators and provide a clear direction for future research aimed at enhancing textual manipulation in video synthesis.

T2VPhysBench: A First-Principles Benchmark for Physical Consistency in Text-to-Video Generation

Text-to-video generative models have made significant strides in recent years, producing high-quality videos that excel in both aesthetic appeal and accurate instruction following, and have become central to digital art creation and user engagement online. Yet, despite these advancements, their ability to respect fundamental physical laws remains largely untested: many outputs still violate basic constraints such as rigid-body collisions, energy conservation, and gravitational dynamics, resulting in unrealistic or even misleading content. Existing physical-evaluation benchmarks typically rely on automatic, pixel-level metrics applied to simplistic, life-scenario prompts, and thus overlook both human judgment and first-principles physics. To fill this gap, we introduce \textbf{T2VPhysBench}, a first-principled benchmark that systematically evaluates whether state-of-the-art text-to-video systems, both open-source and commercial, obey twelve core physical laws including Newtonian mechanics, conservation principles, and phenomenological effects. Our benchmark employs a rigorous human evaluation protocol and includes three targeted studies: (1) an overall compliance assessment showing that all models score below 0.60 on average in each law category; (2) a prompt-hint ablation revealing that even detailed, law-specific hints fail to remedy physics violations; and (3) a counterfactual robustness test demonstrating that models often generate videos that explicitly break physical rules when so instructed. The results expose persistent limitations in current architectures and offer concrete insights for guiding future research toward truly physics-aware video generation.

Attention Mechanism, Max-Affine Partition, and Universal Approximation

We establish the universal approximation capability of single-layer, single-head self- and cross-attention mechanisms with minimal attached structures. Our key insight is to interpret single-head attention as an input domain-partition mechanism that assigns distinct values to subregions. This allows us to engineer the attention weights such that this assignment imitates the target function. Building on this, we prove that a single self-attention layer, preceded by sum-of-linear transformations, is capable of approximating any continuous function on a compact domain under the $L_\infty$-norm. Furthermore, we extend this construction to approximate any Lebesgue integrable function under $L_p$-norm for $1\leq p <\infty$. Lastly, we also extend our techniques and show that, for the first time, single-head cross-attention achieves the same universal approximation guarantees.

Discriminator-Free Direct Preference Optimization for Video Diffusion

Direct Preference Optimization (DPO), which aligns models with human preferences through win/lose data pairs, has achieved remarkable success in language and image generation. However, applying DPO to video diffusion models faces critical challenges: (1) Data inefficiency. Generating thousands of videos per DPO iteration incurs prohibitive costs; (2) Evaluation uncertainty. Human annotations suffer from subjective bias, and automated discriminators fail to detect subtle temporal artifacts like flickering or motion incoherence. To address these, we propose a discriminator-free video DPO framework that: (1) Uses original real videos as win cases and their edited versions (e.g., reversed, shuffled, or noise-corrupted clips) as lose cases; (2) Trains video diffusion models to distinguish and avoid artifacts introduced by editing. This approach eliminates the need for costly synthetic video comparisons, provides unambiguous quality signals, and enables unlimited training data expansion through simple editing operations. We theoretically prove the framework's effectiveness even when real videos and model-generated videos follow different distributions. Experiments on CogVideoX demonstrate the efficiency of the proposed method.

Do LLMs trust AI regulation? Emerging behaviour of game-theoretic LLM agents

There is general agreement that fostering trust and cooperation within the AI development ecosystem is essential to promote the adoption of trustworthy AI systems. By embedding Large Language Model (LLM) agents within an evolutionary game-theoretic framework, this paper investigates the complex interplay between AI developers, regulators and users, modelling their strategic choices under different regulatory scenarios. Evolutionary game theory (EGT) is used to quantitatively model the dilemmas faced by each actor, and LLMs provide additional degrees of complexity and nuances and enable repeated games and incorporation of personality traits. Our research identifies emerging behaviours of strategic AI agents, which tend to adopt more "pessimistic" (not trusting and defective) stances than pure game-theoretic agents. We observe that, in case of full trust by users, incentives are effective to promote effective regulation; however, conditional trust may deteriorate the "social pact". Establishing a virtuous feedback between users' trust and regulators' reputation thus appears to be key to nudge developers towards creating safe AI. However, the level at which this trust emerges may depend on the specific LLM used for testing. Our results thus provide guidance for AI regulation systems, and help predict the outcome of strategic LLM agents, should they be used to aid regulation itself.

Provable Failure of Language Models in Learning Majority Boolean Logic via Gradient Descent

Recent advancements in Transformer-based architectures have led to impressive breakthroughs in natural language processing tasks, with models such as GPT-4, Claude, and Gemini demonstrating human-level reasoning abilities. However, despite their high performance, concerns remain about the inherent limitations of these models, especially when it comes to learning basic logical functions. While complexity-theoretic analyses indicate that Transformers can represent simple logic functions (e.g., $\mathsf{AND}$, $\mathsf{OR}$, and majority gates) by its nature of belonging to the $\mathsf{TC}^0$ class, these results assume ideal parameter settings and do not account for the constraints imposed by gradient descent-based training methods. In this work, we investigate whether Transformers can truly learn simple majority functions when trained using gradient-based methods. We focus on a simplified variant of the Transformer architecture and consider both $n=\mathrm{poly}(d)$ and $n=\exp(\Omega(d))$ number of training samples, where each sample is a $d$-size binary string paired with the output of a basic majority function. Our analysis demonstrates that even after $\mathrm{poly}(d)$ gradient queries, the generalization error of the Transformer model still remains substantially large, growing exponentially with $d$. This work highlights fundamental optimization challenges in training Transformers for the simplest logical reasoning tasks and provides new insights into their theoretical limitations.

Can You Count to Nine? A Human Evaluation Benchmark for Counting Limits in Modern Text-to-Video Models

Generative models have driven significant progress in a variety of AI tasks, including text-to-video generation, where models like Video LDM and Stable Video Diffusion can produce realistic, movie-level videos from textual instructions. Despite these advances, current text-to-video models still face fundamental challenges in reliably following human commands, particularly in adhering to simple numerical constraints. In this work, we present T2VCountBench, a specialized benchmark aiming at evaluating the counting capability of SOTA text-to-video models as of 2025. Our benchmark employs rigorous human evaluations to measure the number of generated objects and covers a diverse range of generators, covering both open-source and commercial models. Extensive experiments reveal that all existing models struggle with basic numerical tasks, almost always failing to generate videos with an object count of 9 or fewer. Furthermore, our comprehensive ablation studies explore how factors like video style, temporal dynamics, and multilingual inputs may influence counting performance. We also explore prompt refinement techniques and demonstrate that decomposing the task into smaller subtasks does not easily alleviate these limitations. Our findings highlight important challenges in current text-to-video generation and provide insights for future research aimed at improving adherence to basic numerical constraints.

Exploring the Limits of KV Cache Compression in Visual Autoregressive Transformers

A fundamental challenge in Visual Autoregressive models is the substantial memory overhead required during inference to store previously generated representations. Despite various attempts to mitigate this issue through compression techniques, prior works have not explicitly formalized the problem of KV-cache compression in this context. In this work, we take the first step in formally defining the KV-cache compression problem for Visual Autoregressive transformers. We then establish a fundamental negative result, proving that any mechanism for sequential visual token generation under attention-based architectures must use at least $\Omega(n^2 d)$ memory, when $d = \Omega(\log n)$, where $n$ is the number of tokens generated and $d$ is the embedding dimensionality. This result demonstrates that achieving truly sub-quadratic memory usage is impossible without additional structural constraints. Our proof is constructed via a reduction from a computational lower bound problem, leveraging randomized embedding techniques inspired by dimensionality reduction principles. Finally, we discuss how sparsity priors on visual representations can influence memory efficiency, presenting both impossibility results and potential directions for mitigating memory overhead.

Theoretical Foundation of Flow-Based Time Series Generation: Provable Approximation, Generalization, and Efficiency

Recent studies suggest utilizing generative models instead of traditional auto-regressive algorithms for time series forecasting (TSF) tasks. These non-auto-regressive approaches involving different generative methods, including GAN, Diffusion, and Flow Matching for time series, have empirically demonstrated high-quality generation capability and accuracy. However, we still lack an appropriate understanding of how it processes approximation and generalization. This paper presents the first theoretical framework from the perspective of flow-based generative models to relieve the knowledge of limitations. In particular, we provide our insights with strict guarantees from three perspectives: $\textbf{Approximation}$, $\textbf{Generalization}$ and $\textbf{Efficiency}$. In detail, our analysis achieves the contributions as follows: $\bullet$ By assuming a general data model, the fitting of the flow-based generative models is confirmed to converge to arbitrary error under the universal approximation of Diffusion Transformer (DiT). $\bullet$ Introducing a polynomial-based regularization for flow matching, the generalization error thus be bounded since the generalization of polynomial approximation. $\bullet$ The sampling for generation is considered as an optimization process, we demonstrate its fast convergence with updating standard first-order gradient descent of some objective.

Limits of KV Cache Compression for Tensor Attention based Autoregressive Transformers

The key-value (KV) cache in autoregressive transformers presents a significant bottleneck during inference, which restricts the context length capabilities of large language models (LLMs). While previous work analyzes the fundamental space complexity barriers in standard attention mechanism [Haris and Onak, 2025], our work generalizes the space complexity barriers result to tensor attention version. Our theoretical contributions rely on a novel reduction from communication complexity and deduce the memory lower bound for tensor-structured attention mechanisms when $d = \Omega(\log n)$. In the low dimensional regime where $d = o(\log n)$, we analyze the theoretical bounds of the space complexity as well. Overall, our work provides a theoretical foundation for us to understand the compression-expressivity tradeoff in tensor attention mechanisms and offers more perspectives in developing more memory-efficient transformer architectures.