arXiv subjects

Vyacheslav Zhdanovskiy

Publications and source records attributed to Vyacheslav Zhdanovskiy.

5 recordsLinked to original sources

arXiv2026

LLMs are General Asynchronous Agents

Modern LLMs are increasingly capable as autonomous agents, but they follow sequential interaction cycles: read, think, reply or call tools, repeat. Many real-world use cases are not sequential: voice assistants, embodied agents, and monitoring systems receive new inputs while they think or perform another task. Modern LLMs address this with specialized architectures for voice interaction and video streams, VLAs for robot control, asynchronous tool calling for API usage, and others. In this work, we generalize from different asynchronous tasks to general asynchronous agents that can adapt to different types of concurrency. To achieve this, we develop an asynchronous LLM framework that lets users (or the agents themselves) define inference coroutines with overlapping memory states. We showcase that Qwen 3.x models are capable of asynchronous operation for streaming video understanding, videogames, and monitoring, without task-specific training.

cs.LG↗

arXiv2026

KV Cache Offloading for Context-Intensive Tasks

With the growing demand for long-context LLMs across a wide range of applications, the key-value (KV) cache has become a critical bottleneck for both latency and memory usage. Recently, KV-cache offloading has emerged as a promising approach to reduce memory footprint and inference latency while preserving accuracy. Prior evaluations have largely focused on tasks that do not require extracting large amounts of information from the context. In this work, we study KV-cache offloading on context-intensive tasks: problems where the solution requires looking up a lot of information from the input prompt. We create and release the Text2JSON benchmark, a highly context-intensive task that requires extracting structured knowledge from raw text. We evaluate modern KV offloading on Text2JSON and other context-intensive tasks and find significant performance degradation on both Llama 3 and Qwen 3 models. Our analysis identifies two key reasons for poor accuracy: low-rank projection of keys and unreliable landmarks, and proposes a simpler alternative strategy that significantly improves accuracy across multiple LLM families and benchmarks. These findings highlight the need for a comprehensive and rigorous evaluation of long-context compression techniques.

cs.LG↗

arXiv2026

On Efficient Scaling of GNNs via IO-Aware Layers Implementations

Graph Neural Networks (GNNs) are bottlenecked by sparse, irregular memory access. Popular frameworks such as DGL and PyTorch Geometric support general message passing, but complex layers often materialize edge-wise intermediates, increasing memory traffic and limiting scalability on large graphs. We take an I/O- and arithmetic-intensity--centric view and show that widely used layers fall into three kernel families: SpMM-based convolutions, reduction-based aggregations, and attention-based layers (GATv2/Graph Transformer). For each family, we develop GPU kernels that reduce data movement, improve locality, and remain robust across realistic graphs. We also study graph reordering and find that its impact depends on the kernel mapping: it benefits neighbor-parallel (gather-dominated) kernels more consistently than feature-parallel designs. Empirically, our fused attention kernels reach up to $\textbf{3.9}\times$ speedup for Graph Transformer (median $\textbf{1.6}\times$), with Tensor Core (block-sparse) variants up to $\textbf{7.3}\times$ on locally dense graphs; for GATv2 we reach up to $\textbf{8.5}\times$ speedup (median $\textbf{2.0}\times$) while reducing peak memory by up to $\textbf{76}\times$ (median $\textbf{6}\times$). Our degree-aware reduction kernels achieve up to $\textbf{10}\times$ speedup (median $\textbf{2.6}\times$). For SpMM-based layers, properly cached cuSPARSE achieves up to $\textbf{8}\times$ speedup over DGL and outperforms evaluated custom baselines in the majority of evaluations. We release our implementations as drop-in replacements to support reproducible, hardware-aware GNN acceleration.

cs.LG↗

arXiv2026

Asynchronous Reasoning: Training-Free Interactive Thinking LLMs

Many state-of-the-art LLMs are trained to think before giving their answer. Reasoning can greatly improve language model capabilities, but it also makes them less interactive: given a new input, a model must stop thinking before it can respond. Real-world use cases such as voice-based or embodied assistants require an LLM agent to respond and adapt to additional information in real time, which is incompatible with sequential interactions. In contrast, humans can listen, think, and act asynchronously: we begin thinking about the problem while reading it and continue thinking while formulating the answer. In this work, we augment LLMs capable of reasoning to operate in a similar way without additional training. Our method uses the properties of positional embeddings to enable LLMs built for sequential generation to simultaneously think, listen, and write outputs. We evaluate our approach on math, commonsense, and safety reasoning: it allows models to generate accurate thinking-augmented answers while reducing time to first non-thinking token from minutes to ${\le}$ 5s and the overall delays by up to $12{\times}$.

cs.LG↗

arXiv2022

Predicting Performance of Heterogeneous AI Systems with Discrete-Event Simulations

In recent years, artificial intelligence (AI) technologies have found industrial applications in various fields. AI systems typically possess complex software and heterogeneous CPU/GPU hardware architecture, making it difficult to answer basic questions considering performance evaluation and software optimization. Where is the bottleneck impeding the system? How does the performance scale with the workload? How the speed-up of a specific module would contribute to the whole system? Finding the answers to these questions through experiments on the real system could require a lot of computational, human, financial, and time resources. A solution to cut these costs is to use a fast and accurate simulation model preparatory to implementing anything in the real system. In this paper, we propose a discrete-event simulation model of a high-load heterogeneous AI system in the context of video analytics. Using the proposed model, we estimate: 1) the performance scalability with the increasing number of cameras; 2) the performance impact of integrating a new module; 3) the performance gain from optimizing a single module. We show that the performance estimation accuracy of the proposed model is higher than 90%. We also demonstrate, that the considered system possesses a counter-intuitive relationship between workload and performance, which nevertheless is correctly inferred by the proposed simulation model.

cs.SE↗