Zhiyuan Guo

Huimin Qiang, Zhiyuan Guo, Shiyuan Xie et al.

Incorporating the dynamics knowledge into the model is critical for achieving accurate trajectory prediction while considering the spatial and temporal characteristics of the vessel. However, existing methods rarely consider the underlying dynamics knowledge and directly use machine learning algorithms to predict the trajectories. Intuitively, the vessel's motions are following the laws of dynamics, e.g., the speed of a vessel decreases when turning a corner. Yet, it is challenging to combine dynamic knowledge and neural networks due to their inherent heterogeneity. Against this background, we propose MSTFormer, a motion inspired vessel trajectory prediction method based on Transformer. The contribution of this work is threefold. First, we design a data augmentation method to describe the spatial features and motion features of the trajectory. Second, we propose a Multi-headed Dynamic-aware Self-attention mechanism to focus on trajectory points with frequent motion transformations. Finally, we construct a knowledge-inspired loss function to further boost the performance of the model. Experimental results on real-world datasets show that our strategy not only effectively improves long-term predictive capability but also outperforms backbones on cornering data.The ablation analysis further confirms the efficacy of the proposed method. To the best of our knowledge, MSTFormer is the first neural network model for trajectory prediction fused with vessel motion dynamics, providing a worthwhile direction for future research.The source code is available at https://github.com/simple316/MSTFormer.

Zijian He, Adrian Sampson, Yiying Zhang et al.

Modern GPUs increasingly rely on specialized and asynchronous hardware units to deliver high performance. Yet these units are often underutilized because today's GPU software stacks still organize programming and execution around a monolithic kernel model that mismatches asynchronous hardware. To address this issue, Virtual Decoupled Engines (VDCores) presents a new decoupled programming and execution model for asynchronous GPUs. VDCores abstracts asynchronous hardware execution units as resource isolated virtual cores and represents workloads as dependency-connected micro-operations (micro-ops). this abstraction removes static orchestration from the programmer, enables automatic overlap of memory and compute based on dependency and resource readiness, and thereby improves utilization of asynchronous hardware resources. Realizing such a decoupled abstraction efficiently on today's GPUs is itself challenging, VDCores addresses this through a GPU-specialized programming model and GPU runtime design that preserves the flexibility while minimizing implementation overhead. Across four LLM inference workloads on GH200, H100, and RTX 6000 Pro GPUs, VDCores significantly improves decoding throughput by 24% on average and by up to 77% under dynamic inputs, while reducing kernel programming and specialization effort by 90%. We have open sourced VDCores at https://github.com/vdcores/vdcores.

Jingqi Tong, Jixin Tang, Hangcheng Li et al.

Vision-language reinforcement learning (RL) has primarily focused on narrow domains (e.g. geometry or chart reasoning). This leaves broader training scenarios and resources underexplored, limiting the exploration and learning of Vision Language Models (VLMs) through RL. We find video games inherently provide rich visual elements and mechanics that are easy to verify. To fully use the multimodal and verifiable reward in video games, we propose Game-RL, constructing diverse game tasks for RL training to boost VLMs general reasoning ability. To obtain training data, we propose Code2Logic, a novel approach that adapts game code to synthesize game reasoning task data, thus obtaining the GameQA dataset of 30 games and 158 tasks with controllable difficulty gradation. Unexpectedly, RL training solely on GameQA enables multiple VLMs to achieve performance improvements across 7 diverse vision-language benchmarks, demonstrating the value of Game-RL for enhancing VLMs' general reasoning. Furthermore, this suggests that video games may serve as valuable scenarios and resources to boost general reasoning abilities. Our code, dataset and models are available at the GitHub repository.

Shouxu Lin, Zhiyuan Guo, Jiaxin Lin

LLM inference is constrained by GPU memory capacity and bandwidth. Tiered memory architectures mitigate this by allowing the GPU to offload memory to the remote tier. However, existing memory offloading frameworks rely on prefetching data into local GPU HBM. This approach underutilizes system resources by introducing HBM contention, squandering memory capacity, and creating pipeline bubbles. We show that enabling direct GPU access to remote memory significantly outperforms prefetching, achieving optimal aggregate system bandwidth. We propose DAK, an end-to-end direct-access memory offloading framework that repurposes the Tensor Memory Accelerator (TMA) to asynchronously fetch offloaded weights and KV caches directly from remote memory into GPU shared memory (SMEM). To maximize remote access performance, DAK introduces a greedy algorithm to determine optimal per-operation offloading ratios, alongside active congestion control and TMA multicast to eliminate interconnect bottlenecks and read amplification. Evaluations across diverse architectures show that DAK achieves near-optimal bandwidth aggregation, with up to 3$\times$ performance gains on NVLink-C2C and 1.8$\times$ on PCIe systems compared to state-of-the-art memory offloading baselines.

Yutong Huang, Zhiyuan Guo, Yiying Zhang

Memory prefetching has long boosted CPU caches and is increasingly vital for far-memory systems, where large portions of memory are offloaded to cheaper, remote tiers. While effective prefetching requires accurate prediction of future accesses, prior ML approaches have been limited to simulation or small-scale hardware. We introduce FarSight, the first Linux-based far-memory system to leverage deep learning by decoupling application semantics from runtime memory layout. This separation enables offline-trained models to predict access patterns over a compact ordinal vocabulary, which are resolved at runtime through lightweight mappings. Across four data-intensive workloads, FarSight delivers up to 3.6x higher performance than the state-of-the-art.

Zhiyuan Guo, Yuexin Li, Guo Chen et al.

Spoken dialogue systems such as Siri and Alexa provide great convenience to people's everyday life. However, current spoken language understanding (SLU) pipelines largely depend on automatic speech recognition (ASR) modules, which require a large amount of language-specific training data. In this paper, we propose a Transformer-based SLU system that works directly on phones. This acoustic-based SLU system consists of only two blocks and does not require the presence of ASR module. The first block is a universal phone recognition system, and the second block is a Transformer-based language model for phones. We verify the effectiveness of the system on an intent classification dataset in Mandarin Chinese.

Sairamvinay Vijayaraghavan, Ye Wang, Zhiyuan Guo et al.

This is a paper for exploring various different models aiming at developing fake news detection models and we had used certain machine learning algorithms and we had used pretrained algorithms such as TFIDF and CV and W2V as features for processing textual data.