Search: “hardware”

SparLeak: Privacy Leakage from Sparse Attention in LLM Inference on Shared GPUs

arXiv:2609.38830v1 Announce Type: new Abstract: Sparse attention is widely used to accelerate long-context inference in modern large language models (LLMs), but its input-dependent execution behavior

1 Oct 2026

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Hardware

Sparse-WAM: Accelerating World Action Models via Action-Guided Sparse Imagination

DGX agent

arXiv:2609.38984v1 Announce Type: cross Abstract: World-action models (WAMs) leverage pretrained video models to improve generalization in robot control by jointly predicting future visual states and

hardwarearxiv-cs-ai

1 Oct 2026

When Masking Helps or Hurts Robustness in Compressed CLIP: A Pre-Deployment Diagnostic

arXiv:2609.39704v1 Announce Type: cross Abstract: This paper demonstrate that whether masking-based token pruning helps or hurts worst-group robustness can be predicted before deployment, without labe

1 Oct 2026

AI as a Compiler: Compiling Triton kernels without the Triton compiler

arXiv:2609.36800v1 Announce Type: new Abstract: Compiler backends are expensive to build and maintain as programming models, workloads, and accelerators evolve. We investigate whether large language m

BASE: Batch-Aware Selection of Experts Using Predicted Removal Error for Efficient MoE Decoding

arXiv:2609.36222v1 Announce Type: new Abstract: Large language models are increasingly expensive to serve. In large-scale serving systems, autoregressive decoding is often bottlenecked by transferring

BiFE: Search-Efficient Discovery of CPU-Only Branching Policies via LLM-based Bi-Fidelity Evolution

arXiv:2609.36735v1 Announce Type: new Abstract: In branch-and-bound (B&B) for mixed-integer linear programming (MILP), branching variable selection critically impacts efficiency. Existing neural branc

DIET: Deletion-response Expert Trimming for Video Diffusion Transformers

arXiv:2609.37829v1 Announce Type: new Abstract: Video diffusion transformers (DiTs) increasingly adopt mixture-of-experts (MoE) architectures to reduce active computation, but their full expert storag

FastVR: Efficient Streaming Video Restoration with One-Step Diffusion

arXiv:2609.36757v1 Announce Type: new Abstract: Diffusion-based video restoration recovers realistic details, but its practical deployment is limited by two efficiency bottlenecks: costly VAE encoding

Federation of Experts: Communication Efficient Distributed Inference for Large Language Models

arXiv:2605.06206v2 Announce Type: replace Abstract: Mixture of experts has emerged as the primary mechanism for making Large Language Models (LLMs) computationally efficient. However, in distributed s

GEM-KMeans: Memory-Efficient and Accurate Clustering on Massive Scale with GPU Optimization

arXiv:2609.36074v1 Announce Type: new Abstract: Memory-efficient scaling on clustering problems without sacrificing statistical accuracy is of central interest for large-scale data analysis and machin

HelixWorld: A Real-time Interactive Audio-Visual World Model

arXiv:2609.38123v1 Announce Type: new Abstract: World simulation is inherently multisensory, demanding synchronized visual and acoustic dynamics in real time. Yet prevailing interactive world models r

Label Less, Learn More: Resource-Efficient Active Semi-Supervised Learning for Onboard Satellite Image Annotation

arXiv:2609.37481v1 Announce Type: new Abstract: Large-scale pervasive sensing increasingly relies on high-resolution satellite imagery, yet task-specific onboard vision is constrained by costly annota

One from Infinity: Actualizing Futures from Pretrained World Models into Robot Actions

arXiv:2609.36413v1 Announce Type: new Abstract: A pretrained video world model admits many plausible futures for a scene, but a robot must realize the exact task-conditioned one. To turn world models

Probe-Space Preconditioning for Fast and Stable Zero-Order Training

arXiv:2609.38095v1 Announce Type: new Abstract: Backpropagation (BP) dominates deep learning but imposes a massive memory tax. For example, training OPT-30B with Adam requires approx 600GB of GPU memo

RelayVSR: Large-Small Model Collaboration for Efficient Real-World Video Super-Resolution

arXiv:2609.37850v1 Announce Type: new Abstract: Large generative models can recover realistic detail in real-world video super-resolution (VSR), but processing an entire video with them is computation

STEPQuant: When and Where Errors Matter in Delta-Rule Recurrent State Quantization

arXiv:2609.38169v1 Announce Type: cross Abstract: Linear attention replaces growing KV caches with fixed-size recurrent states, yet these persistent states can become a substantial memory bottleneck u

Trident: Unifying Guarded Dispatch and Host Execution for PyTorch Triton Workloads

arXiv:2609.37241v1 Announce Type: new Abstract: User-written Triton kernels enable high-performance GPU computation within PyTorch, but their end-to-end latency can remain dominated by host-side orche

ADPTNet: Adaptive with Prescriptive Timescales Non-Linear SSM for Sequence Modelling

arXiv:2609.34034v1 Announce Type: cross Abstract: A central aim of neuromorphic computing is to provide a viable alternative to highly energy-intensive Transformer-based AI. However, efficient alterna

Model Releases

Autonomous Research Project Management as an Agent Skill: A Case Study in Exact Spectral Spatial Regression

model-releasesarxiv-cs-lg

arXiv:2609.31683v1 Announce Type: new Abstract: This work presents an end-to-end demonstration of autonomous machine learning research conducted by an agent skill on consumer hardware. The demonstrati

Does Execution Require Target KV Fidelity? A Mixed-Fidelity KV Runtime for LLM Serving

arXiv:2609.33536v1 Announce Type: new Abstract: Large language model (LLM) serving is increasingly constrained by the GPU memory consumed by key-value (KV) caches. Existing compression, eviction, and

Efficient World Action Model Inference with Adaptive Intermediate States

arXiv:2609.34608v1 Announce Type: new Abstract: World Action Models (WAMs) enable future-aware control by jointly modeling actions and environment dynamics. However, iterative diffusion or flow infere

Fill2SR: Repurposing Inpainting Diffusion Transformers for Real-World Super-Resolution

arXiv:2609.33582v1 Announce Type: new Abstract: Recent real-world image super-resolution (SR) methods often adapt text-to-image (T2I) backbones with ControlNet-style branches or spatial conditioning t

FLARE: Flow Matching with Local Axis-Angle Representations for Stochastic Micromagnetic Evolution

arXiv:2609.34070v1 Announce Type: new Abstract: Long-horizon micromagnetic simulation remains expensive because conventional and learned solvers typically propagate Landau--Lifshitz--Gilbert (LLG) dyn

FoldAttention: Declared-Reference Softmax for Fast Decode and Deterministic Backward

arXiv:2609.33410v1 Announce Type: new Abstract: Autoregressive decode repeatedly streams a growing KV cache, making attention a major cost at long context. Existing high-performance kernels use online

In-Flight KV Cache with Clean Anchors for Faster Autoregressive Video Diffusion

arXiv:2609.32540v1 Announce Type: new Abstract: Few-step autoregressive video diffusion generates a long video by splitting the video into temporal chunks and generating chunk-by-chunk, each through a

JEPA Learns What the Mask Leaves Unrecoverable

arXiv:2609.32481v1 Announce Type: new Abstract: Joint-embedding predictive architectures are unusually sensitive to how the input is masked: block masks work, scattered masks do not, and the explanati

KV-streams for Efficient Compaction in Agentic Reinforcement Learning

arXiv:2609.35750v1 Announce Type: new Abstract: Scaling the horizon of agentic LLMs is bottlenecked by the need to fit ever longer context traces in GPU memory. Context compaction has been the most po

KVCMAS: Efficient KV cache Correction for Shared Context in Multi-Agent Systems

arXiv:2609.34060v1 Announce Type: new Abstract: Prompt-specialized multi-agent systems enable multiple agents to share a model while performing complementary roles to solve complex tasks. However, age

Learned Preconditioning for a Primal-Dual Interior-Point Method

arXiv:2609.35665v1 Announce Type: cross Abstract: Interior-point methods (IPMs) are among the most widely used algorithms for constrained optimization, yet their Newton-based search directions require

LionMuon: Alternating Spectral and Sign Descent for Efficient Training

arXiv:2609.35297v1 Announce Type: new Abstract: Pretraining a language model takes enormous compute, and the right optimizer can save a good part of it. Muon's spectral step gives a stronger direction

MarsLab: A Martian Rover Simulator for Planetary Rover Autonomous Navigation

arXiv:2609.34702v1 Announce Type: new Abstract: Future Mars missions will require rover autonomy that can operate across unstructured terrain, changing illumination, atmospheric dust, and limited comm

NOSA: Native and Offloadable Sparse Attention

arXiv:2510.13602v3 Announce Type: replace-cross Abstract: Decoding throughput improvements from larger inference batches are limited by GPU memory, which is largely consumed by the key-value (KV) cach

PrefLUT: Reusable and Refinable Personalized Color Editing from Pairwise Preferences

arXiv:2609.34133v1 Announce Type: new Abstract: Photographic color editing is inherently personal: the same image can appear too warm, too muted, or already satisfactory to different users. Most looku

PulseInfer: I/O-Centric Sparse KV Cache Offloading for Efficient Long-Context LLM Decoding

arXiv:2609.34555v1 Announce Type: new Abstract: Long-context LLM serving is increasingly bottlenecked by decode, where large KV caches limit batch size and underutilize GPUs. Sparse KV cache offloadin

SANTA++: Sampling Attention through Representative Keys

arXiv:2609.35629v1 Announce Type: new Abstract: Attention often concentrates on a small subset of tokens in the context, but which subset matters changes from one query to the next. To exploit this ch

Scope-WM: Scoped Computation for Efficient Visual World Models

arXiv:2609.33218v1 Announce Type: new Abstract: Visual world models enable robotic planning by predicting future observations, but dense latent-state propagation and sample-intensive trajectory optimi

Single-Layer MeMo as a Randomized Hamming-Kernel Classifier

arXiv:2609.34562v1 Announce Type: new Abstract: MeMo (Zanzotto et al., 2025) is a recent language-model architecture that stores associations between token contexts and next tokens in a correlation ma

SparSP: Exploiting Communication Sparsity for Sequence-Parallel Video DiTs

arXiv:2609.32197v1 Announce Type: cross Abstract: Diffusion Transformers have become the dominant architecture for video generation. Their substantial computational cost motivates scaling inference ac

Spexis: Speculative Lookahead Scheduling for LLM Inference

arXiv:2609.34370v1 Announce Type: new Abstract: Spexis is a multi-GPU LLM inference framework that improves the efficiency of pipeline and tensor parallelism through speculative parallelism. Rather th

STAMP: Predicting Out-of-Distribution Generalization without Target Data

arXiv:2609.32672v1 Announce Type: new Abstract: Predicting whether a trained model will generalize under distribution shift remains difficult, especially when target-domain data are unavailable. We in

TempoKV: Timely Staging of LLM KV Caches for Memory-Semantic Flash

arXiv:2609.35065v1 Announce Type: cross Abstract: Reusable prefix key-value (KV) caches can outgrow GPU memory in large language model (LLM) serving. A memory-semantic flash hierarchy offers SSD-backe

Terrain-Aware Autonomous Planetary Exploration for Exteroceptive-Proprioceptive Mapping with Quadruped Scouts

arXiv:2609.35493v1 Announce Type: new Abstract: Autonomous planetary exploration requires robots to navigate unknown, uneven terrain while assessing risk, traversability, and energetic cost. Quadruped

TopoEP: Topology-Aware Load Balancing for Expert-Parallel MoE Training

arXiv:2609.35481v1 Announce Type: cross Abstract: Dynamic routing creates severe load imbalance in large-scale expert-parallel Mixture-of-Experts (MoE) training, turning GPUs that host hot experts int

Towards Kinematic Actionable Infeasibility Detection in Motion Planning

arXiv:2609.32806v1 Announce Type: new Abstract: Motion planning in robotics requires not only computing collision-free paths but also certifying infeasibility when no such path exists. Complete method

UnStep: Training-Free Acceleration of Causal Video Diffusion with Fewer Steps Than Distillation

arXiv:2609.32518v1 Announce Type: new Abstract: Distilling bidirectional multi-step video diffusion transformers into few-step causal models has become a common approach for streaming video generation

WorldAttention: An Efficient Attention Architecture for Interactive Video World Models

arXiv:2609.34606v1 Announce Type: new Abstract: Leveraging the paradigm of autoregressive diffusion, text-conditioned interactive video world models aim to simulate temporally coherent environments gu

AutoResearch at Production Scale: Failure Modes and a Multi-Agent Framework

arXiv:2609.30541v1 Announce Type: new Abstract: Optimizing embedding systems for production recommendation pipelines demands systematic exploration that consumes disproportionate engineering effort at

28 Sept 2026

Compress What You See, Not What You Say: Anchored Context Distillation for Latent-Observation Software Engineering Agents

arXiv:2609.31430v1 Announce Type: new Abstract: Tool observations dominate the context of software-engineering agents, making long interaction histories costly to maintain. Existing context compressio