Search: “inference”

When Does Multi-Agent RL Improve LLM Workflows? Workflow, Scale, and Policy-Sharing Tradeoffs

arXiv:2605.24202v1 Announce Type: new Abstract: Multi-agent LLM workflows route inference through specialized roles to lift end-task accuracy, but jointly training those roles with reinforcement learn

safetyarxiv-cs-ai

26 May 2026

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Research

A Simple Plug-in for Improving Eviction-Based KV Cache Compression

DGX agent

arXiv:2605.23258v1 Announce Type: new Abstract: KV cache growth is a major bottleneck for long-context inference in large language models. Existing methods are often dominated by binary eviction or re

researcharxiv-cs-lg

25 May 2026

Adaptive Mass-Segmented KV Compression for Long-Context Reasoning

arXiv:2605.23200v1 Announce Type: cross Abstract: The linear growth of the Key-Value (KV) cache is a critical bottleneck in long-form LLM inference. Existing KV compression methods mitigate this by ev

researcharxiv-cs-ai

Dynamic Weight-based Temporal Aggregation for Low-light Video Enhancement Under Extreme Noise

arXiv:2510.09450v2 Announce Type: replace Abstract: Low-light video enhancement (LLVE) is challenging due to noise, low contrast, and color degradation. While learning-based methods enable fast infere

safetyarxiv-cs-cv

Encrypted Neural Networks without Overflows

arXiv:2605.23096v1 Announce Type: cross Abstract: Fully homomorphic encryption (FHE) enables private inference by evaluating neural networks on encrypted data. In this way, we can delegate the computa

One Policy, Infinite NPCs: Persona-Traceable Shared RL Policies for Scalable Game Agents

arXiv:2605.23652v1 Announce Type: new Abstract: On a 300-persona life-simulation benchmark, pcsp achieves compositional zero-shot persona identification up to 17x above chance, Spearman rho approx 0.7

PhotoFlow: Agentic 3D Virtual Photography Missions

arXiv:2605.23771v1 Announce Type: cross Abstract: Virtual photography asks an agent to enter a prepared 3D scene with no preselected camera pose or reference image, infer a suitable shot from scene in

Test-Time Training Undermines Safety Guardrails

arXiv:2605.22984v1 Announce Type: cross Abstract: Test-Time Training (TTT) is an emerging paradigm that enables models to adapt their parameters during inference, improving performance on tasks such a

safetyarxiv-cs-ai

Training-Free Looped Transformers

arXiv:2605.23872v1 Announce Type: new Abstract: We introduce training-free looped transformers, in which a lightweight inference-time wrapper loops a contiguous mid-stack block of layers of a frozen c

Live Music Diffusion Models: Efficient Fine-Tuning and Post-Training of Interactive Diffusion Music Generators

arXiv:2605.22717v1 Announce Type: cross Abstract: Interactive streaming music generation promises the use of generative models for live performance and co-creation that is impossible with offline mode

23 May 2026

Applications

Provable Robustness against Backdoor Attacks via the Primal-Dual Perspective on Differential Privacy

arXiv:2605.21780v1 Announce Type: new Abstract: Randomized smoothing is a powerful tool for certifying robustness to adversarial perturbations, including poisoning attacks via randomized training and

applicationsarxiv-cs-lg

23 May 2026

Richer Bayesian Last Layers with Subsampled NTK Features

arXiv:2602.01279v2 Announce Type: replace Abstract: Bayesian Last Layers (BLLs) provide a convenient and computationally efficient way to estimate uncertainty in neural networks. However, they underes

23 May 2026

CoRMA: Contrastive RMA for Contact-Rich Meta-Adaptation

model-releasesarxiv-cs-ro

arXiv:2605.22082v1 Announce Type: new Abstract: We present CoRMA(Contrastive Robotic Motor Adaptation), a context-based meta-adaptation framework that modifies RMA for force-dominant assembly. CoRMA r

Dual-Integrated Low-Latency Single-Lens Infrared Computational Imaging for Object Detection

model-releasesarxiv-cs-cv

arXiv:2605.21964v1 Announce Type: new Abstract: Computational imaging enables compact infrared systems, but deep-learning pipelines that combine image reconstruction and object detection often introdu

Flow-based Gaussian Splatting for Continuous-Scale Remote Sensing Image Super-Resolution

arXiv:2605.22147v1 Announce Type: new Abstract: High-resolution remote sensing images (RSIs) are crucial for Earth observation applications, yet acquiring them is often limited by sensor constraints a

Hierarchical Variational Policies for Reward-Guided Diffusion

arXiv:2605.21661v1 Announce Type: cross Abstract: Adapting pretrained diffusion models to downstream objectives such as inverse problems often requires expensive test-time guidance or optimization. We

safetyarxiv-cs-cv

IdioLink: Retrieving Meaning Beyond Words Across Idiomatic and Literal Expressions

model-releasesarxiv-cs-cl

arXiv:2605.22247v1 Announce Type: new Abstract: Idioms pose a fundamental challenge for language models, as their meaning cannot be inferred from surface form alone. Understanding such expressions, th

InnerQ: Hardware-Aware Tuning-Free Quantization of KV Cache for Large Language Models

model-releasesarxiv-cs-cl

arXiv:2602.23200v2 Announce Type: replace-cross Abstract: When transformer-based language models are deployed for text generation, most of the inference time is spent in the decoding stage, where outp

Local Ai

Intelligence per Watt: Measuring Intelligence Efficiency of Local AI

arXiv:2511.07885v4 Announce Type: replace-cross Abstract: Large language model (LLM) queries are predominantly processed by frontier models in centralized cloud infrastructure. Demand growth strains t

local-aiarxiv-cs-cl

Steins;Gate Drive: Semantic Safety Arbitration over Structured Futures for Latency-Decoupled LLM Planning

model-releasesarxiv-cs-ro

arXiv:2605.22456v1 Announce Type: new Abstract: Cloud-hosted LLM driver agents provide useful semantic judgments, but their inference latency exceeds stepwise vehicle-control windows. Learned world mo

Ultra-High-Definition Image Quality Assessment via Graph Representation Learning

model-releasesarxiv-cs-cv

arXiv:2605.22192v1 Announce Type: new Abstract: Blind image quality assessment (BIQA) for ultrahighdefinition (UHD) images remains challenging because native-resolution inference is computationally ex

Video-o3: Native Interleaved Clue Seeking for Long Video Multi-Hop Reasoning

arXiv:2601.23224v2 Announce Type: replace Abstract: Existing multimodal large language models for long-video understanding predominantly rely on uniform sampling and single-turn inference, limiting th

A Free Lunch in LLM Compression: Revisiting Retraining after Pruning

model-releasesarxiv-cs-lg

arXiv:2510.14444v3 Announce Type: replace Abstract: Post-training pruning can substantially reduce LLM inference costs, but it often degrades quality unless the remaining weights are adapted. Since gl

ArchSIBench: Benchmarking the Architectural Spatial Intelligence of Vision-Language Models

model-releasesarxiv-cs-cv

arXiv:2605.20837v1 Announce Type: new Abstract: Architectural spatial intelligence, the ability to recognize and infer architectural space, is fundamental to tasks such as robot navigation, embodied i

Compute Only Once: UG-Separation for Efficient Large Recommendation Models

arXiv:2602.10455v2 Announce Type: replace-cross Abstract: Driven by scaling laws, recommender systems increasingly rely on larger-scale models to capture complex feature interactions and user behavior

Conflict-Aware Additive Guidance for Flow Models under Compositional Rewards

arXiv:2605.20758v1 Announce Type: cross Abstract: Inference-time guided sampling steers state-of-the-art diffusion and flow models without fine-tuning by interpreting the generation process as a contr

DASH: Fast Differentiable Architecture Search for Hybrid Attention in Minutes on a Single GPU

model-releasesarxiv-cs-cl

arXiv:2605.20936v1 Announce Type: cross Abstract: Hybrid attention architectures are becoming an increasingly important paradigm for improving LLM inference efficiency while preserving model quality,

Disentangling Generation and Regression in Stochastic Interpolants for Controllable Image Restoration

arXiv:2605.21381v1 Announce Type: new Abstract: Recent advances in Image Restoration (IR) have been largely driven by generative methods such as Diffusion Models and Flow Matching, which excel in synt

Focus-then-Context: Subject-Centric Progressive Visual Token Reduction for Vision-Language Models

arXiv:2605.20950v1 Announce Type: new Abstract: Vision-Language Models (VLMs) face a bottleneck of prohibitive computational costs arising from massive visual token sequences during inference. Existin

Agents

LiteViLNet: Lightweight Vision-LiDAR Fusion Network for Efficient Road Segmentation

arXiv:2605.21007v1 Announce Type: new Abstract: Road segmentation is a fundamental perception task for autonomous driving and intelligent robotic systems, requiring both high accuracy and real-time in

agentsarxiv-cs-cv

OCTOPUS: Optimized KV Cache for Transformers via Octahedral Parametrization Under optimal Squared error quantization

arXiv:2605.21226v1 Announce Type: new Abstract: The key-value (KV) cache dominates memory bandwidth and footprint in long-context autoregressive inference. Recent rotation-preconditioned codecs (Turbo

Praxium: Diagnosing Cloud Anomalies with AI-based Telemetry and Dependency Analysis

arXiv:2603.23890v2 Announce Type: replace-cross Abstract: As the modern microservice architecture for cloud applications grows in popularity, cloud services are becoming increasingly complex and more

Hardware

PulseCol: Periodically Refreshed Column-Sparse Attention for Accelerating Diffusion Language Models

arXiv:2605.20813v1 Announce Type: new Abstract: Inference in diffusion large language models (dLLMs) is computationally expensive, as full self-attention must be repeatedly executed at each step of th

hardwarearxiv-cs-cl

Runtime-Certified Bounded-Error Quantized Attention

model-releasesarxiv-cs-lg

arXiv:2605.20868v1 Announce Type: new Abstract: KV cache quantization reduces the memory cost of long-context LLM inference, but introduces approximation error that is typically validated only empiric

TabPFN-MT: A Natively Multitask In-Context Learner for Tabular Data

arXiv:2605.20234v1 Announce Type: new Abstract: Prior-Data Fitted networks (PFNs) have been very successful in tabular contexts, handling prediction tasks in context. However, they are designed for si

Tutorials

Task-Routed Mixture-of-Experts with Cognitive Appraisal for Implicit Sentiment Analysis

arXiv:2605.20916v1 Announce Type: new Abstract: Implicit sentiment analysis is challenging because sentiment toward an aspect is often inferred from events rather than expressed through explicit opini

tutorialsarxiv-cs-cl

Applications

The Generation-Recognition Asymmetry: Six Dimensions of a Fundamental Divide in Formal Language Theory

arXiv:2603.10139v2 Announce Type: replace Abstract: Every formal grammar defines a language and can in principle be used in three ways: to generate strings (production), to recognize them (parsing), o

applicationsarxiv-cs-cl

Time-Prompt: Integrated Heterogeneous Prompts for Unlocking LLMs in Time Series Forecasting

arXiv:2506.17631v4 Announce Type: replace Abstract: Time series forecasting aims to model temporal dependencies among variables for future state inference, holding significant importance and widesprea

Tracing the ongoing emergence of human-like reasoning in Large Language Models

arXiv:2605.21299v1 Announce Type: new Abstract: Humans effortlessly go beyond literal meanings: If you mow the lawn, I will give you fifty dollars, is typically understood as implying that the speaker

researcharxiv-cs-cl

Trusted Weights, Treacherous Optimizations? Optimization-Triggered Backdoor Attacks on LLMs

arXiv:2605.20641v1 Announce Type: cross Abstract: Inference optimization is a vital technique for deploying LLMs at scale. Compilation is the most widely adopted optimization technique for LLMs. While

UniT: Unified Geometry Learning with Group Autoregressive Transformer

arXiv:2605.21131v1 Announce Type: new Abstract: Recent feed-forward models have significantly advanced geometry perception for inferring dense 3D structure from sensor observations. However, its essen

Directed Acyclic Graph Convolutional Networks

arXiv:2506.12218v2 Announce Type: replace-cross Abstract: Directed acyclic graphs (DAGs) are central to science and engineering applications including causal inference, scheduling, and neural architec

Distribution-Free Uncertainty Quantification for Continuous AI Agent Evaluation

arXiv:2605.19779v1 Announce Type: new Abstract: We adapt split conformal prediction and adaptive conformal inference (ACI) to continuous AI agent evaluation, providing distribution-free coverage guara

Local Ai

Draft Less, Retrieve More: Hybrid Tree Construction for Speculative Decoding

arXiv:2605.20104v1 Announce Type: cross Abstract: Speculative decoding (SD) accelerates large language model inference by leveraging a draft-then-verify paradigm. To maximize the acceptance rate, rece

local-aiarxiv-cs-ai

Exploring and Developing a Pre-Model Safeguard with Draft Models

arXiv:2605.19321v1 Announce Type: cross Abstract: Large Language Model (LLM) alignment remains vulnerable to jailbreak attacks that elicit unsafe responses, motivating pre-model and post-model guards.

safetyarxiv-cs-ai

FlexDraft: Flexible Speculative Decoding via Attention Tuning and Bonus-Guided Calibration

arXiv:2605.20022v1 Announce Type: new Abstract: Speculative decoding accelerates memory-bound LLM inference without quality degradation by using a fast drafter to propose multiple candidate tokens and

researcharxiv-cs-cl

From Sparsity to Simplicity: Enabling Simpler Sequential Replacements via Sparse Attention Distillation

arXiv:2605.18865v1 Announce Type: cross Abstract: Self-attention serves as the core foundation of large-scale transformer pretraining, but its quadratic token interaction cost makes inference expensiv

Agents

Hybrid Training for Vision-Language-Action Models

arXiv:2510.00600v2 Announce Type: replace-cross Abstract: Using Large Language Models to produce intermediate thoughts, a.k.a. Chain-of-thought (CoT), before providing an answer has been a successful

agentsarxiv-cs-ai