arXiv:2602.21529v3 Announce Type: replace
Abstract: Rug pull is a critical attack in the world of blockchain technology. Despite this, the absence of sufficient time-bound and well-structured datasets is considered one of the significant issues faced while identifying early detection. Existing datasets do not provide the solution to this challenge because of temporal leakage or use of post-collapse indicators, insufficient modality coverage, and confusing or partial labels, especially with regards to DeFi tokens. To solve these problems, we present a highly curated and strictly time-bound dataset called TM-RugPull containing 1,000 projects, which include DeFi, meme, NFT, and celebrity token projects. We achieve temporal validation of the dataset by acquiring all three modalities, namely on-chain behavior, smart contract metadata, and OSINT signals. The project labels are provided based on manual investigation for the entire project's lifespan and its collapse. Also, we make our dataset publicly available together with its codebase for data acquisition and feature extraction.
Science Journals
arXiv:2603.00631v2 Announce Type: replace
Abstract: LiTS is a modular Python framework for LLM reasoning via tree search. It decomposes tree search into three reusable components (Policy, Transition, and RewardModel) that plug into algorithms like MCTS and BFS. A decorator-based registry enables domain experts to extend to new domains by registering components, and algorithmic researchers to implement custom search algorithms. We demonstrate composability on MATH500 (language reasoning), Crosswords (environment planning), and MapEval (tool use), showing that components and algorithms are orthogonal: components are reusable across algorithms within each task type, and algorithms work across all components and domains. We also report a mode-collapse finding: in infinite action spaces, LLM policy diversity (not reward quality) is the bottleneck for effective tree search. A demonstration video is available at https://youtu.be/nRGX43YrR3I. The package is released under the Apache 2.0 license at https://github.com/xinzhel/lits-llm, including installation instructions and runnable examples that enable users to reproduce the demonstrated workflows.
arXiv:2603.00952v2 Announce Type: replace
Abstract: High-fidelity reconstruction of dynamic scenes is an important yet challenging problem. While recent 4D Gaussian Splatting (4DGS) has demonstrated the ability to model temporal dynamics, it couples Gaussian motion and geometric attributes within a single covariance formulation, which limits its expressiveness for complex motions and often leads to visual artifacts. To address this, we propose VeGaS, a novel velocity-based 4D Gaussian Splatting framework that decouples Gaussian motion and geometry. Specifically, we introduce a Galilean shearing matrix that explicitly incorporates time-varying velocity to flexibly model complex non-linear motions, while strictly isolating the effects of Gaussian motion from the geometry-related conditional Gaussian covariance. Furthermore, a Geometric Deformation Network is introduced to refine Gaussian shapes and orientations using spatio-temporal context and velocity cues, enhancing temporal geometric modeling. Extensive experiments on public datasets demonstrate that VeGaS achieves state-of-the-art performance.
arXiv:2506.11502v3 Announce Type: replace
Abstract: With the advent of digital transformation, organisations are increasingly generating large volumes of data through the execution of various processes across disparate systems. By integrating data from these heterogeneous sources, it becomes possible to derive new insights essential for tasks such as monitoring and analysing process performance. Typically, this information is extracted during a data pre-processing or engineering phase. However, this step is often performed in an ad-hoc manner and is time-consuming and labour-intensive. To streamline this process, we introduce a reference model and a collection of patterns designed to enrich production event data. The reference model provides a standard way for storing and extracting production event data. The patterns describe common information extraction tasks and how such tasks can be automated effectively. The reference model is developed by combining the ISA-95 industry standard with the Event Knowledge Graph formalism. The patterns are developed based on empirical observations from event data sets originating in manufacturing processes and are formalised using the reference model. We evaluate the relevance and applicability of these patterns by demonstrating their application to use cases.
arXiv:2603.02218v2 Announce Type: replace
Abstract: Large language models (LLMs) make it plausible to build systems that improve through self-evolving loops, but many existing proposals are better understood as self-play and often plateau quickly. A central failure mode is that the loop synthesises more data without increasing learnable information for the next iteration. Through experiments on a self-play coding task, we reveal that sustainable self-evolution requires a self-synthesised data pipeline with learnable information that increases across iterations. We identify triadic roles that self-evolving LLMs play: the Proposer, which generates tasks; the Solver, which attempts solutions; and the Verifier, which provides training signals, and we identify three system designs that jointly target learnable information gain from this triadic roles perspective. Asymmetric co-evolution closes a weak-to-strong-to-weak loop across roles. Capacity growth expands parameter and inference-time budgets to match rising learnable information. Proactive information seeking introduces external context and new task sources that prevent saturation. Together, these modules provide a measurable, system-level path from brittle self-play dynamics to sustained self-evolution.
arXiv:2506.12119v2 Announce Type: replace
Abstract: Mixture-of-Experts (MoE) language models dramatically expand model capacity and achieve remarkable performance without increasing per-token compute. However, can MoEs surpass dense architectures under strictly equal resource constraints -- that is, when the total parameter count, training compute, and data budget are identical? This question remains under-explored despite its significant practical value and potential. In this paper, we propose a novel perspective and methodological framework to study this question thoroughly. First, we comprehensively investigate the architecture of MoEs and achieve an optimal model design that maximizes the performance. Based on this, we subsequently find that an MoE model with activation rate in an optimal region is able to outperform its dense counterpart under the same total parameter, training compute and data resource. More importantly, this optimal region remains consistent across different model sizes. Although additional amount of data turns out to be a trade-off for enhanced performance, we show that this can be resolved via reusing data. We validate our findings through extensive experiments, training nearly 200 language models at 2B scale and over 50 at 7B scale, cumulatively processing 50 trillion tokens. All model checkpoints are publicly available.
arXiv:2603.08462v2 Announce Type: replace
Abstract: \ac{CoT} prompting improves LLM accuracy on complex tasks but often increases token usage and inference cost. Existing ``Budget Forcing'' methods reduce cost via fine-tuning with heuristic length penalties, suppressing both essential reasoning and redundant filler. We recast efficient reasoning as a lossy compression problem under the \ac{IB} principle, and identify a key theoretical gap when applying naive \ac{IB} to transformers: attention violates the Markov property between prompt, reasoning trace, and response. To resolve this issue, we model \ac{CoT} generation under the \ac{CIB} principle, where the reasoning trace $Z$ acts as a computational bridge that contains only the information about the response $Y$ that is not directly accessible from the prompt $X$. This yields a general Reinforcement Learning objective: maximize task reward while compressing completions under a prior over reasoning traces, subsuming common heuristics (e.g., length penalties) as special cases (e.g., uniform priors). In contrast to naive token-counting approaches, we introduce a semantic prior that measures token cost by surprisal under a language model. Crucially, the prior is queried only for token-level log-probabilities, adding negligible overhead to the training loop. Empirically, our \ac{CIB} objective prunes reasoning redundancy while preserving fluency and logic, improving accuracy at moderate compression and enabling aggressive compression with minimal accuracy drop. These gains generalize across model families and task domains, confirming \ac{CIB} as a domain-agnostic CoT compression framework.
arXiv:2603.12243v4 Announce Type: replace
Abstract: Mastering dexterous manipulation with multi-fingered hands has been a grand challenge in robotics for decades. Despite its potential, the difficulty of collecting high-quality data remains a primary bottleneck for high-precision tasks. While reinforcement learning and simulation-to-real-world transfer offer a promising alternative, the transferred policies often fail for tasks demanding millimeter-scale precision, such as bimanual piano playing. In this work, we introduce HandelBot, a framework that combines a simulation policy and rapid adaptation through a two-stage pipeline. Starting from a simulation-trained policy, we first apply a structured refinement stage to correct spatial alignments by adjusting lateral finger joints based on physical rollouts. Next, we use residual reinforcement learning to autonomously learn fine-grained corrective actions. Through extensive hardware experiments across five recognized songs, we demonstrate that HandelBot can successfully perform precise bimanual piano playing. Our system outperforms direct simulation deployment by a factor of 1.8x and requires only 30 minutes of physical interaction data.
arXiv:2603.17706v2 Announce Type: replace
Abstract: We prove that Balanced Biclique Reconfiguration on bipartite graphs is PSPACE-complete. This implies the PSPACE-completeness of the spanning variant of Subgraph Reconfiguration under the token jumping rule for the property "a graph is an $(i, j)$-complete bipartite graph," which was previously known only to be NP-hard [Hanaka et al. TCS 2020]. Using our result, we also show that Connected Components Reconfiguration with two connected components is PSPACE-complete under all previously studied rules, resolving an open problem of Nakahata [COCOON 2025] in the negative.
arXiv:2501.01046v4 Announce Type: replace
Abstract: Dataset deduplication is widely recognized as a crucial preprocessing step that enhances data quality and improves the performance of large language models. A commonly used method for this process is the MinHash Locality-Sensitive Hashing (LSH) algorithm. Recently, GPU-accelerated frameworks such as NVIDIA NeMo Curator have been introduced to handle large-scale corpora; however, they remain suboptimal due to high communication overhead from physical data shuffling and underutilization of GPU resources. In this paper, we propose SEDD, a high-performance GPU-accelerated deduplication framework optimized for distributed cluster environments. SEDD introduces a computationally efficient, partially reusable hash function, alongside highly optimized GPU kernels and a hardware-aware automatic parameter selection mechanism. By replacing traditional data shuffling with a streaming-based approach, SEDD significantly mitigates communication bottlenecks. Our framework outperforms the CPU-based deduplication tool in SlimPajama by up to 158$\times$ and the GPU-based tool in NVIDIA NeMo Curator by up to 7.8$\times$ when processing 30 million documents on a node with four GPUs. Notably, SEDD dramatically accelerates the previously time-consuming MinHash signature generation phase, achieving speedups of up to 375$\times$ over the CPU baseline. Despite these gains in efficiency, SEDD maintains high deduplication fidelity, with duplicate document sets achieving Jaccard similarities of over 0.95 compared to those identified by the standard MinHash algorithm. In large-scale experiments, the deduplication of 1.2 trillion tokens is completed in just 3 hours on an 8-node 32-GPU V100 cluster. The related code is publicly available on GitHub (https://github.com/mcrl/SEDD).
arXiv:2507.01533v2 Announce Type: replace
Abstract: We prove consistency of a recently proposed scheme that evaluates expected values by composing a learned transport map with Clenshaw--Curtis sparse-grid quadrature on a tractable product source. Our analysis hinges on the structural fact that composition of a $C^k_{\mathrm{mix}}$-regular function -- which carries the fast quadrature rate $m^{-k}(\log m)^{(d-1)(k+1)}$ -- with a $C^1$-diffeomorphism can only be guaranteed to be $C^k_{\mathrm{mix}}$ itself, if the diffeomorphism is diagonal up to a permutation of coordinates. The fast rate is therefore available exclusively for product targets, and the analysis splits into two regimes. In the general regime of arbitrary targets, we learn the transport as the time-one flow of a $\mathrm{ReLU}^{k+1}$-neural ODE trained by maximum likelihood. The resulting flow lies in the isotropic space $C^k$ and yields the rate $m^{-k/d}(\log m)^{(d-1)(k/d+1)}$, with raising the density smoothness $k$ and the matched activation order $k+1$ mitigating the curse of dimensionality at the cost of harder optimization. In the diagonal regime of product targets, the Knothe--Rosenblatt map is itself diagonal and we estimate it pointwise via empirical quantile transport, a lightweight alternative that recovers the full mixed-regularity rate. In both regimes, the resulting LtI estimator is PAC (probably approximately correct) consistent. With high probability the numerical integral approximates the true value to arbitrary accuracy as both the sample size $n$ and the quadrature budget $m$ tend to infinity.
arXiv:2604.00634v2 Announce Type: replace
Abstract: Panoptic segmentation is a key enabler for robotic perception, as it unifies semantic understanding with object-level reasoning. However, the increasing complexity of state-of-the-art models makes them unsuitable for deployment on resource-constrained platforms such as mobile robots. We propose a novel approach called LiPS that addresses the challenge of efficient-to-compute panoptic segmentation with a lightweight design that retains query-based decoding while introducing a streamlined feature extraction and fusion pathway. It aims at providing a strong panoptic segmentation performance while substantially lowering the computational demands. Evaluations on standard benchmarks demonstrate that LiPS attains accuracy comparable to much heavier baselines, while providing up to 4.5 higher throughput, measured in frames per second, and requiring nearly 6.8 times fewer computations. This efficiency makes LiPS a highly relevant bridge between modern panoptic models and real-world robotic applications.
arXiv:2604.01674v2 Announce Type: replace
Abstract: Model merging aims to integrate multiple expert models into a single model that inherits their complementary strengths without incurring the inference-time cost of ensembling. Recent progress has shown that merging can be highly effective when all source models are \emph{homogeneous}, i.e., derived from the same pretrained backbone and therefore share aligned parameter coordinates or compatible task vectors. Yet this assumption is increasingly unrealistic in open model ecosystems, where useful experts are often built on different families such as Llama, Qwen, and Mistral. In such \emph{heterogeneous} settings, direct weight-space fusion becomes ill-posed due to architectural mismatch, latent basis misalignment, and amplified cross-source conflict. We address this problem with \texttt{HeteroFusion} for heterogeneous language model fusion, which consists of two key components: topology-based alignment that transfers knowledge across heterogeneous backbones by matching functional module structures instead of raw tensor coordinates, and conflict-aware denoising that suppresses incompatible or noisy transfer signals during fusion. We further provide analytical justification showing that preserving the target adapter basis while predicting structured updates leads to a stable and well-conditioned transfer process. Across heterogeneous transfer, multi-source fusion, noisy-source robustness, and cross-family generalization settings, \texttt{HeteroFusion} consistently outperforms strong merging, fusion, and ensemble baselines.
arXiv:2604.04202v2 Announce Type: replace
Abstract: AI agents deployed as persistent assistants must maintain correct beliefs as their information environment evolves. In practice, evidence is scattered across heterogeneous sources that often contradict one another, new information can invalidate earlier conclusions, and user preferences surface through corrections rather than explicit instructions. Existing benchmarks largely assume static, single-authority settings and do not evaluate whether agents can keep up with this complexity. We introduce ClawArena, a benchmark for evaluating AI agents in evolving information environments. Each scenario maintains a complete hidden ground truth while exposing the agent only to noisy, partial, and sometimes contradictory traces across multi-channel sessions, workspace files, and staged updates. Evaluation is organized around three coupled challenges: multi-source conflict reasoning, dynamic belief revision, and implicit personalization, whose interactions yield a 14-category question taxonomy. Two question formats, multi-choice (set-selection) and shell-based executable checks, test both reasoning and workspace grounding. ClawArena comprises 12 multi-turn scenarios spanning 337 evaluation rounds with 45 dynamic updates, evaluated across five agent frameworks and 18 language models from proprietary, community-accessible, and self-hosted sources. Experiments show that model capability accounts for a 29-point score range across models while framework design accounts for up to a 24-point range, that MetaClaw's skill overlay reliably improves score without degrading accuracy, and that belief revision difficulty is determined by update design strategy rather than update volume. Code is available at https://github.com/aiming-lab/ClawArena.
arXiv:2605.18337v1 Announce Type: new
Abstract: Large-scale news corpora support a wide range of research in Computational Social Science and NLP, yet access remains constrained: commercial archives impose prohibitive costs and licensing restrictions, while open alternatives like Common Crawl's CC-News require terabyte-scale storage and computationally intensive processing. We present Infini-News, a retrieval toolkit and index for the entire CC-News archive from August 2016 to the latest available snapshot. Our contributions are threefold. First, we extract, clean the text, and parse the structured metadata of over 1.35B articles. Second, we enrich the corpus with language detection using three frontier language classifiers (GlotLID, lingua, and CommonLingua), and with multi-source geographic attribution that resolves a country of origin for 83.4% of articles across 222 countries. Third, we construct Infini-gram indexes: suffix-array structures that let researchers search the full archive for arbitrary text patterns in sub-second time. Together, these resources lower the barrier to longitudinal, cross-national media research.
arXiv:2502.18663v3 Announce Type: replace
Abstract: This paper is the second in a series of studies on developing efficient artificial intelligence-based approaches to pathfinding on extremely large graphs (e.g. $10^{70}$ nodes) with a focus on Cayley graphs and mathematical applications. The open-source CayleyPy project is a central component of our research. The present paper proposes a novel combination of a reinforcement learning approach with a more direct diffusion distance approach from the first paper. Our analysis includes benchmarking various choices for the key building blocks of the approach: architectures of the neural network, generators for the random walks and beam search pathfinding. We compared these methods against the classical computer algebra system GAP, demonstrating that they "overcome the GAP" for the considered examples. As a particular mathematical application we examine the Cayley graph of the symmetric group with cyclic shift and transposition generators. We provide strong support for the OEIS-A186783 conjecture that the diameter is equal to n(n-1)/2 by machine learning and mathematical methods. We identify the conjectured longest element and generate its decomposition of the desired length. We prove a diameter lower bound of n(n-1)/2-n/2 and an upper bound of n(n-1)/2+ 3n by presenting the algorithm with given complexity. We also present several conjectures motivated by numerical experiments, including observations on the central limit phenomenon (with growth approximated by a Gumbel distribution), the uniform distribution for the spectrum of the graph, and a numerical study of sorting networks. To stimulate crowdsourcing activity, we create challenges on the Kaggle platform and invite contributions to improve and benchmark approaches on Cayley graph pathfinding and other tasks.
arXiv:2604.07630v2 Announce Type: replace
Abstract: The final size of an earthquake typically cannot be predicted from its ongoing seismic radiation. Expanding observations reveal distinct exceptions, such as slow earthquakes, injection-induced seismicity, and earthquake swarms, in which fault slip has an upper bound. A common thread among these anomalies is the diffusive migration of their active areas. Here, we report a unified scaling relation for these diffusional earthquakes. By tracking prolonged earthquake swarms in Northeast Japan, we constrained the time evolution of their active seismicity areas and cumulative seismic moments. Their moment-duration trajectories coincide with the final states documented for global swarms and induced seismicity across various scales. When plotted as seismic moment versus seismicity area, their trajectories collapse onto those of slow earthquakes, uniformly explained by a diffusional constant-slip model. This constant-slip scaling carves out a unique class of diffusional earthquakes, where the final available seismic energy is predetermined by slip distance.
arXiv:2605.18340v1 Announce Type: new
Abstract: Presently, there is a steady state approach in Computational fluid dynamics (CFD) to obtain a steady solution directly from the steady state governing equations. Whereas, for obtaining a time-periodic flow solution, the present unsteady governing equations-based CFD approach starts from an initial condition and requires a large computational time during the initial non-periodic transient phase before reaching the periodic state. For obtaining the periodic flow directly, without transient simulations that may not be of interest, our objective is to propose a Physics Informed Neural Network (PINN)-based periodic CFD approach. The motivation is a substantial reduction in computational time by a meshless PINN-based periodic CFD solver as compared to the present mesh-based transient-to-periodic solver. Proof-of-concept, for the periodic CFD approach, is demonstrated here for 2D periodic heat diffusion and fluid flow problems. The proposed PINN-based periodic solver primarily focuses on the time-periodic state, optimizing the neural network model's trainable parameters to precisely fit a smaller time window (one time-period) rather than the temporal domain starting from the initial condition. After presenting a verification study, effect of the PINN-related various hyperparameters such as the number of collocation points, neural network architecture, and point spacing for numerical differentiation, on computational time and accuracy are presented. Our results demonstrate that the PINN-based periodic solver takes substantially less computational time to achieve almost same accuracy as that obtained by the traditional transient-to-periodic solver.
arXiv:2604.08205v2 Announce Type: replace
Abstract: Payment channel networks (PCNs) are a promising approach to making cryptocurrency transactions faster and more scalable. At their core, PCNs bypass the blockchain by routing transactions through intermediary channels. However, a channel can forward a transaction only if it has the necessary funds: the problem of keeping the channels balanced is a current bottleneck for the PCN's transaction throughput. This paper considers the problem of maximizing the number of transactions accepted by a channel in a PCN. Previous works either considered the associated optimization problem with all transactions known in advance or developed heuristics tested on particular transaction datasets. This work, however, considers the problem in its purely online form where the transactions are arbitrary and revealed one after the other. We show that the problem can be modeled as a new online knapsack variant where the items (transaction proposals) can be either positive or negative depending on the direction of the transaction. The main contribution of this paper is a deterministic online algorithm that is $O(\log B)$-competitive, where $B$ is the knapsack capacity (maximum allowed channel balance). We complement this result with an asymptotically matching lower bound of $\Omega(\log B)$ which holds for any randomized algorithm, demonstrating our algorithm's optimality.
arXiv:2604.08216v3 Announce Type: replace
Abstract: Large Language Models (LLMs) still suffer from severe hallucinations and catastrophic forgetting during causal reasoning over massive, fragmented long contexts. Existing memory mechanisms typically treat retrieval as a static, single-step passive matching process, leading to severe semantic dilution and contextual fragmentation. To overcome these fundamental bottlenecks, we propose MemCoT, a test-time memory scaling framework that redefines the reasoning process by transforming long-context reasoning into an iterative, stateful information search. MemCoT introduces a multi-view long-term memory perception module that enables Zoom-In evidence localization and Zoom-Out contextual expansion, allowing the model to first identify where relevant evidence resides and then reconstruct the surrounding causal structure necessary for reasoning. In addition, MemCoT employs a task-conditioned dual short-term memory system composed of semantic state memory and episodic trajectory memory. This short-term memory records historical search decisions and dynamically guides query decomposition and pruning across iterations. Empirical evaluations demonstrate that MemCoT establishes a state-of-the-art performance. Empowered by MemCoT, several open- and closed-source models achieve SOTA performance on the LoCoMo benchmark and LongMemEval-S benchmark.
arXiv:2507.05482v3 Announce Type: replace
Abstract: Training-free diffusion guidance offers a flexible framework for leveraging off-the-shelf classifiers without additional training. Yet, current approaches hinge on posterior approximations via Tweedie's formula, which often yield unreliable guidance, particularly in low-density regions. Stochastic optimal control (SOC), in contrast, enables principled posterior sampling but remains computationally prohibitive for efficient inference. In this work, we reconcile the strengths of these paradigms by introducing Stein Diffusion Guidance (SDG), a novel training-free framework grounded in a surrogate SOC objective. We establish a new theoretical bound on the SOC value function, revealing the necessity of correcting approximate posteriors to reflect true diffusion dynamics. Building on Stein variational inference, SDG computes the steepest descent direction that minimizes the Kullback-Leibler divergence between approximate and true posteriors. By integrating a principled Stein correction mechanism along with a novel running cost functional, SDG enables effective guidance in low-density regions. Our experiments on diverse image-guidance tasks and on challenging small-ligand sampling for protein docking suggest that SDG consistently outperforms standard training-free guidance methods and highlights its potential for broader posterior sampling problems beyond high-density regimes.
arXiv:2605.15586v2 Announce Type: replace
Abstract: Complementary-label learning (CLL) is a weakly supervised paradigm where instances are labeled with classes they do not belong to. Despite a decade of research, CLL methods remain competitive mainly on 10-class classification, with scaling to large label spaces continuing to be an enduring bottleneck. This limitation stems from the common assumption of uniform label generation in traditional methods, which fatally dilutes the learning signal in many-class settings. In this paper, we demonstrate that this long-standing barrier can be overcome by deliberately designing a biased (non-uniform) generation process that restricts complementary labels to a subset of classes. This finding motivates us to propose Bias-Induced Constrained Labeling (BICL), a principled framework spanning data collection to training that leverages this bias. BICL enables effective learning on CIFAR-100 and TinyImageNet-200, achieving more than sevenfold accuracy improvements over traditional methods. Our findings establish a new trajectory for making CLL feasible for many classes in real-world applications.
arXiv:2604.08936v2 Announce Type: replace
Abstract: Medical foundation models (MFMs) aim to learn universal representations from multimodal medical images that can generalize effectively to diverse downstream clinical tasks. However, most existing MFMs suffer from information ambiguity that blends multimodal representations in a single embedding space, leading to the degradation of modality specificity and diversity. In this paper, we propose M-IDoL, a self-supervised MFM that introduces Information Decomposition for multimodal representation Learning via two objectives: i) maximizing inter-modality entropy by dispersing multimodal representations into separable Mixture-of-Experts (MoE) subspaces to achieve representation specificity across modalities; and ii) minimizing intra-modality uncertainty by performing fine-grained semantic discrimination within each MoE subspace to enrich representation diversity per modality. By pre-training on 1.15 million medical images, M-IDoL i) delivers superior generalization across 21 downstream clinical tasks, outperforming 20 foundation models on five imaging modalities (e.g., X-ray, fundus, OCT, dermoscopy and pathology), and ii) learns modality-specific and diverse representations, showing clearer separation of feature clusters across modalities and finer-grained feature discrimination within each modality.
arXiv:2605.18342v1 Announce Type: new
Abstract: This work continues the development of an intensional approach to computability initiated in previous work, in which programs and computations, rather than functions, constitute the primary objects of study. In this setting, models of computation are described as monoid actions on a configuration space, and programs as dynamical systems constrained by this action. Within this framework, we introduce a formal notion of algorithm as a finite directed graph whose edges are labelled by partial maps over an abstract data structure. This definition separates control from data, representing the former as a graph and the latter as an algebra of operations. We then define what it means for a program, in a given model of computation, to implement such an algorithm, by requiring a correspondence between computational steps and labelled transitions that preserves the induced transformations on representations of data. This yields a precise notion of implementation and situates algorithms as abstract partial specifications of computational behaviour.
arXiv:2605.18346v1 Announce Type: new
Abstract: Recent advances in autoregressive video diffusion have enabled sequential and streaming video generation. However, long-horizon generation requires increasingly large KV caches, making efficient compression without sacrificing quality challenging. Existing methods mostly select historical frames based on attention scores, but their context decisions remain coarse. When multiple frames are generated in the same chunk, these methods often apply a shared history selection to the whole chunk, score historical frames solely by attention, and assign head-wise budgets either uniformly or by attention-pattern heuristics rather than explicit head-importance estimation. We show that frames within the same generated chunk can depend on distinct historical frames, that the same historical frame can receive different attention scores as its relative temporal distance to the current frames changes, and that masking different heads induces unequal generation degradation. Motivated by these findings, we propose \textbf{Focused Forcing}, a training-free KV selection method that focuses cached history along both generated-frame and head dimensions. For each generated frame, Focused Forcing preserves the most relevant and distinctive historical frames by combining attention scores with diversity scores of historical frames, while assigning larger budgets to heads with higher estimated importance. Across multiple autoregressive generation paradigms, Focused Forcing achieves up to $\textbf{1.48}\times$ end-to-end acceleration without training, while \textbf{improving visual quality and text alignment}. \textit{Our code will be released on GitHub.}