Search: “benchmark”

Bridging SFT and RL: Dynamic Policy Optimization for Robust Reasoning

arXiv:2604.08926v1 Announce Type: new Abstract: Post-training paradigms for Large Language Models (LLMs), primarily Supervised Fine-Tuning (SFT) and Reinforcement Learning (RL), face a fundamental dil

safetyarxiv-cs-lg

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Model Releases

CatalogStitch: Dimension-Aware and Occlusion-Preserving Object Compositing for Catalog Image Generation

DGX agent

arXiv:2604.08836v1 Announce Type: new Abstract: Generative object compositing methods have shown remarkable ability to seamlessly insert objects into scenes. However, when applied to real-world catalo

model-releasesarxiv-cs-cv

13 Apr 2026

CausalVAD: De-confounding End-to-End Autonomous Driving via Causal Intervention

arXiv:2603.18561v2 Announce Type: replace Abstract: Planning-oriented end-to-end driving models show great promise, yet they fundamentally learn statistical correlations instead of true causal relatio

safetyarxiv-cs-cv

ChipSeek: Optimizing Verilog Generation via EDA-Integrated Reinforcement Learning

arXiv:2507.04736v2 Announce Type: replace Abstract: Large Language Models have emerged as powerful tools for automating Register-Transfer Level (RTL) code generation, yet they face critical limitation

Conformal Prediction in Hierarchical Classification with Constrained Representation Complexity

arXiv:2501.19038v3 Announce Type: replace-cross Abstract: Conformal prediction has emerged as a widely used framework for constructing valid prediction sets in classification and regression tasks. In

Continuous Orthogonal Mode Decomposition: Haptic Signal Prediction in Tactile Internet

arXiv:2604.09446v1 Announce Type: cross Abstract: The Tactile Internet demands sub-millisecond latency and ultra-high reliability, as high latency or packet loss could lead to haptic control instabili

researcharxiv-cs-lg

Cross-Lingual Attention Distillation with Personality-Informed Generative Augmentation for Multilingual Personality Recognition

model-releasesarxiv-cs-cl

arXiv:2604.08851v1 Announce Type: new Abstract: While significant work has been done on personality recognition, the lack of multilingual datasets remains an unresolved challenge. To address this, we

DeFakeQ: Enabling Real-Time Deepfake Detection on Edge Devices via Adaptive Bidirectional Quantization

model-releasesarxiv-cs-cv

arXiv:2604.08847v1 Announce Type: new Abstract: Deepfake detection has become a fundamental component of modern media forensics. Despite significant progress in detection accuracy, most existing metho

Dictionary-Aligned Concept Control for Safeguarding Multimodal LLMs

arXiv:2604.08846v1 Announce Type: cross Abstract: Multimodal Large Language Models (MLLMs) have been shown to be vulnerable to malicious queries that can elicit unsafe responses. Recent work uses prom

Distilling Genomic Models for Efficient mRNA Representation Learning via Embedding Matching

arXiv:2604.08574v1 Announce Type: cross Abstract: Large Genomic Foundation Models have recently achieved remarkable results and in-vivo translation capabilities. However these models quickly grow to o

researcharxiv-cs-ai

Distributionally Robust Token Optimization in RLHF

arXiv:2604.08577v1 Announce Type: cross Abstract: Large Language Models (LLMs) tend to respond correctly to prompts that align to the data they were trained and fine-tuned on. Yet, small shifts in wor

researcharxiv-cs-ai

Drift-Aware Online Dynamic Learning for Nonstationary Multivariate Time Series: Application to Sintering Quality Prediction

arXiv:2604.09358v1 Announce Type: new Abstract: Accurate prediction of nonstationary multivariate time series remains a critical challenge in complex industrial systems such as iron ore sintering. In

Dual Mamba for Node-Specific Representation Learning: Tackling Over-Smoothing with Selective State Space Modeling

arXiv:2511.06756v3 Announce Type: replace Abstract: Over-smoothing remains a fundamental challenge in deep Graph Neural Networks (GNNs), where repeated message passing causes node representations to b

researcharxiv-cs-lg

EchoTrail-GUI: Building Actionable Memory for GUI Agents via Critic-Guided Self-Exploration

arXiv:2512.19396v3 Announce Type: replace Abstract: Contemporary GUI agents, while increasingly capable due to advances in Large Vision-Language Models (VLMs), often operate with a critical limitation

agentsarxiv-cs-ai

Efficient Hierarchical Implicit Flow Q-learning for Offline Goal-conditioned Reinforcement Learning

arXiv:2604.08960v1 Announce Type: new Abstract: Offline goal-conditioned reinforcement learning (GCRL) is a practical reinforcement learning paradigm that aims to learn goal-conditioned policies from

Applications

Efficient Spatial-Temporal Focal Adapter with SSM for Temporal Action Detection

arXiv:2604.09164v1 Announce Type: new Abstract: Temporal human action detection aims to identify and localize action segments within untrimmed videos, serving as a pivotal task in video understanding.

applicationsarxiv-cs-cv

Envisioning the Future, One Step at a Time

arXiv:2604.09527v1 Announce Type: cross Abstract: Accurately anticipating how complex, diverse scenes will evolve requires models that represent uncertainty, simulate along extended interaction chains

Applications

Evidential Transformation Network: Turning Pretrained Models into Evidential Models for Post-hoc Uncertainty Estimation

arXiv:2604.08627v1 Announce Type: cross Abstract: Pretrained models have become standard in both vision and language, yet they typically do not provide reliable measures of confidence. Existing uncert

applicationsarxiv-cs-ai

EvoLen: Evolution-Guided Tokenization for DNA Language Model

arXiv:2604.08698v1 Announce Type: new Abstract: Tokens serve as the basic units of representation in DNA language models (DNALMs), yet their design remains underexplored. Unlike natural language, DNA

safetyarxiv-cs-lg

EXAONE 4.5 Technical Report

model-releasesarxiv-cs-cl

arXiv:2604.08644v1 Announce Type: new Abstract: This technical report introduces EXAONE 4.5, the first open-weight vision language model released by LG AI Research. EXAONE 4.5 is architected by integr

Fast-dVLM: Efficient Block-Diffusion VLM via Direct Conversion from Autoregressive VLM

arXiv:2604.06832v2 Announce Type: replace Abstract: Vision-language models (VLMs) predominantly rely on autoregressive decoding, which generates tokens one at a time and fundamentally limits inference

agentsarxiv-cs-cl

FDIF: Formula-Driven supervised Learning with Implicit Functions for 3D Medical Image Segmentation

arXiv:2603.23199v2 Announce Type: replace Abstract: Deep learning-based 3D medical image segmentation methods relies on large-scale labeled datasets, yet acquiring such data is difficult due to privac

researcharxiv-cs-cv

Fine-tuning is Not Enough: A Parallel Framework for Collaborative Imitation and Reinforcement Learning in End-to-end Autonomous Driving

arXiv:2603.13842v3 Announce Type: replace-cross Abstract: End-to-end autonomous driving is typically built upon imitation learning (IL), yet its performance is constrained by the quality of human demo

FIRE-CIR: Fine-grained Reasoning for Composed Fashion Image Retrieval

model-releasesarxiv-cs-cv

arXiv:2604.09114v1 Announce Type: new Abstract: Composed image retrieval (CIR) aims to retrieve a target image that depicts a reference image modified by a textual description. While recent vision-lan

FIT-GNN: Faster Inference Time for GNNs that 'FIT' in Memory Using Coarsening

arXiv:2410.15001v5 Announce Type: replace Abstract: Scalability of Graph Neural Networks (GNNs) remains a significant challenge. To tackle this, methods like coarsening, condensation, and computation

FluidFlow: a flow-matching generative model for fluid dynamics surrogates on unstructured meshes

arXiv:2604.08586v1 Announce Type: cross Abstract: Computational fluid dynamics (CFD) provides high-fidelity simulations of fluid flows but remains computationally expensive for many-query applications

Local Ai

From Frames to Events: Rethinking Evaluation in Human-Centric Video Anomaly Detection

arXiv:2604.09327v1 Announce Type: new Abstract: Pose-based Video Anomaly Detection (VAD) has gained significant attention for its privacy-preserving nature and robustness to environmental variations.

local-aiarxiv-cs-cv

From Reasoning to Agentic: Credit Assignment in Reinforcement Learning for Large Language Models

model-releasesarxiv-cs-cl

arXiv:2604.09459v1 Announce Type: new Abstract: Reinforcement learning (RL) for large language models (LLMs) increasingly relies on sparse, outcome-level rewards -- yet determining which actions withi

Ge^ext{2}mS-T: Multi-Dimensional Grouping for Ultra-High Energy Efficiency in Spiking Transformer

arXiv:2604.08894v1 Announce Type: cross Abstract: Spiking Neural Networks (SNNs) offer superior energy efficiency over Artificial Neural Networks (ANNs). However, they encounter significant deficienci

researcharxiv-cs-ai

GeoPAS: Geometric Probing for Algorithm Selection in Continuous Black-Box Optimisation

arXiv:2604.09095v1 Announce Type: new Abstract: Automated algorithm selection in continuous black-box optimisation typically relies on fixed landscape descriptors computed under a limited probing budg

Applications

Identification and Anonymization of Named Entities in Unstructured Information Sources for Use in Social Engineering Detection

arXiv:2604.09016v1 Announce Type: cross Abstract: This study addresses the challenge of creating datasets for cybercrime analysis while complying with the requirements of regulations such as the Gener

applicationsarxiv-cs-ai

Improving Model Performance by Adapting the KGE Metric to Account for System Non-Stationarity

arXiv:2604.03906v2 Announce Type: replace Abstract: Geoscientific systems tend to be characterized by pronounced temporal non-stationarity, arising from seasonal and climatic variability in hydrometeo

Inferring Latent Temporal Sparse Coordination Graph for Multi-Agent Reinforcement Learning

arXiv:2403.19253v3 Announce Type: replace Abstract: Effective agent coordination is crucial in cooperative Multi-Agent Reinforcement Learning (MARL). While agent cooperation can be represented by grap

InsEdit: Towards Instruction-based Visual Editing via Data-Efficient Video Diffusion Models Adaptation

arXiv:2604.08646v1 Announce Type: new Abstract: Instruction-based video editing is a natural way to control video content with text, but adapting a video generation model into an editor usually appear

researcharxiv-cs-cv

Interactive ASR: Towards Human-Like Interaction and Semantic Coherence Evaluation for Agentic Speech Recognition

arXiv:2604.09121v1 Announce Type: cross Abstract: Recent years have witnessed remarkable progress in automatic speech recognition (ASR), driven by advances in model architectures and large-scale train

agentsarxiv-cs-ai

Local Ai

LADR: Locality-Aware Dynamic Rescue for Efficient Text-to-Image Generation with Diffusion Large Language Models

arXiv:2603.13450v2 Announce Type: replace-cross Abstract: Discrete Diffusion Language Models have emerged as a compelling paradigm for unified multimodal generation, yet their deployment is hindered b

local-aiarxiv-cs-cl

Learning Vision-Language-Action World Models for Autonomous Driving

arXiv:2604.09059v1 Announce Type: cross Abstract: Vision-Language-Action (VLA) models have recently achieved notable progress in end-to-end autonomous driving by integrating perception, reasoning, and

Lessons Without Borders? Evaluating Cultural Alignment of LLMs Using Multilingual Story Moral Generation

arXiv:2604.08797v1 Announce Type: cross Abstract: Stories are key to transmitting values across cultures, but their interpretation varies across linguistic and cultural contexts. Thus, we introduce mu

Listener-Rewarded Thinking in VLMs for Image Preferences

arXiv:2506.22832v3 Announce Type: replace-cross Abstract: Training robust and generalizable reward models for human visual preferences is essential for aligning text-to-image and text-to-video generat

LMGenDrive: Bridging Multimodal Understanding and Generative World Modeling for End-to-End Driving

arXiv:2604.08719v1 Announce Type: cross Abstract: Recent years have seen remarkable progress in autonomous driving, yet generalization to long-tail and open-world scenarios remains a major bottleneck

Low-Data Supervised Adaptation Outperforms Prompting for Cloud Segmentation Under Domain Shift

model-releasesarxiv-cs-cv

arXiv:2604.08956v1 Announce Type: new Abstract: Adapting vision-language models to remote sensing imagery presents a fundamental challenge: both the visual and linguistic distributions of satellite da

Low Rank Based Subspace Inference for the Laplace Approximation of Bayesian Neural Networks

arXiv:2502.02345v2 Announce Type: replace Abstract: Subspace inference for neural networks assumes that a subspace of their parameter space suffices to produce a reliable uncertainty quantification. I

MAB-DQA: Addressing Query Aspect Importance in Document Question Answering with Multi-Armed Bandits

arXiv:2604.08952v1 Announce Type: new Abstract: Document Question Answering (DQA) involves generating answers from a document based on a user's query, representing a key task in document understanding

safetyarxiv-cs-cl

MAG-3D: Multi-Agent Grounded Reasoning for 3D Understanding

arXiv:2604.09167v1 Announce Type: new Abstract: Vision-language models (VLMs) have achieved strong performance in multimodal understanding and reasoning, yet grounded reasoning in 3D scenes remains un

agentsarxiv-cs-cv

Mamba-Based Graph Convolutional Networks: Tackling Over-smoothing with Selective State Space

arXiv:2501.15461v4 Announce Type: replace Abstract: Graph Neural Networks (GNNs) have shown great success in various graph-based learning tasks. However, it often faces the issue of over-smoothing as

Many-Tier Instruction Hierarchy in LLM Agents

arXiv:2604.09443v1 Announce Type: cross Abstract: Large language model agents receive instructions from many sources-system messages, user prompts, tool outputs, and more-each carrying different level

MATCHA: Efficient Deployment of Deep Neural Networks on Multi-Accelerator Heterogeneous Edge SoCs

arXiv:2604.09124v1 Announce Type: cross Abstract: Deploying DNNs on System-on-Chips (SoC) with multiple heterogeneous acceleration engines is challenging, and the majority of deployment frameworks can

Memory Intelligence Agent

arXiv:2604.04503v3 Announce Type: replace Abstract: Deep research agents (DRAs) integrate LLM reasoning with external tools. Memory systems enable DRAs to leverage historical experiences, which are es

agentsarxiv-cs-ai