Trending Papers

VibeVoice synthesizes long-form multi-speaker speech using next-token diffusion and a highly efficient continuous speech tokenizer, achieving superior performance and fidelity.

A multi-agent framework using large language models for stock trading simulates real-world trading firms, improving performance metrics like cumulative returns and Sharpe ratio.

GenericAgent is a self-evolving large language model agent system that maximizes context information density through hierarchical memory, reusable SOPs, and efficient compression to overcome long-horizon limitations.

Kronos, a specialized pre-training framework for financial K-line data, outperforms existing models in forecasting and synthetic data generation through a unique tokenizer and autoregressive pre-training on a large dataset.

RAG-Anything is a unified framework that enhances multimodal knowledge retrieval by integrating cross-modal relationships and semantic matching, outperforming existing methods on complex benchmarks.

LingBot-Map is a feed-forward 3D foundation model that reconstructs scenes from video streams using a geometric context transformer architecture with specialized attention mechanisms for coordinate grounding, dense geometric cues, and long-range drift correction, achieving stable real-time performance at 20 FPS.

World-R1 framework improves video generation by incorporating 3D constraints through reinforcement learning and specialized text datasets while maintaining visual quality and scalability.

Tuna-2 is a unified multimodal model that performs visual understanding and generation directly from pixel embeddings without pretrained vision encoders, achieving state-of-the-art performance in multimodal benchmarks.

OpenGame is an open-source agentic framework for end-to-end web game creation that uses specialized code models and evaluation benchmarks to overcome challenges in interactive application development.

Researchers developed TIDE, a framework for cross-architecture distillation of diffusion large language models that improves performance through specialized modules for distillation strength modulation, context enrichment, and cross-tokenizer objectives.

PagedAttention algorithm and vLLM system enhance the throughput of large language models by efficiently managing memory and reducing waste in the key-value cache.

AutoDev is an AI-driven software development framework that automates complex engineering tasks within a secure Docker environment, achieving high performance in code and test generation.

MinerU2.5, a 1.2B-parameter document parsing vision-language model, achieves state-of-the-art recognition accuracy with computational efficiency through a coarse-to-fine parsing strategy.

A large language model adapted for time-series forecasting achieves near-optimal zero-shot performance on diverse datasets across different time scales and granularities.

Mem0, a memory-centric architecture with graph-based memory, enhances long-term conversational coherence in LLMs by efficiently extracting, consolidating, and retrieving information, outperforming existing memory systems in terms of accuracy and computational efficiency.

DeepSeek-V3 is a parameter-efficient Mixture-of-Experts language model using MLA and DeepSeekMoE architectures, achieving high performance with efficient training and minimal computational cost.

OpenDevin is a platform for developing AI agents that interact with the world by writing code, using command lines, and browsing the web, with support for multiple agents and evaluation benchmarks.

PaddleOCR-VL, a vision-language model combining NaViT-style dynamic resolution and ERNIE, achieves state-of-the-art performance in document parsing and element recognition with high efficiency.

LightRAG improves Retrieval-Augmented Generation by integrating graph structures for enhanced contextual awareness and efficient information retrieval, achieving better accuracy and response times.

LLaDA2.0-Uni is a unified discrete diffusion language model that integrates multimodal understanding and generation through a semantic discrete tokenizer, MoE-based backbone, and diffusion decoder, achieving performance comparable to specialized vision-language models while enabling efficient inference and high-fidelity image generation.

SmolDocling is a compact vision-language model that performs end-to-end document conversion with robust performance across various document types using 256M parameters and a new markup format.

Vision-Language-Action models present unique safety challenges due to their embodied nature, requiring unified approaches across multiple domains to address threats from data poisoning to adversarial attacks and ensuring secure deployment across various applications.

Sapiens2 is a high-resolution transformer model family for human-centric vision that achieves superior performance through combined pretraining objectives, large-scale human image datasets, and architectural improvements enabling detailed dense prediction and semantic understanding.

We study allowing large language models (LLMs) to process arbitrarily long prompts through the lens of inference-time scaling. We propose Recursive Language Models (RLMs), a general inference strategy that treats long prompts as part of an external environment and allows the LLM to programmatically examine, decompose, and recursively call itself over snippets of the prompt. We find that RLMs successfully handle inputs up to two orders of magnitude beyond model context windows and, even for shorter prompts, dramatically outperform the quality of base LLMs and common long-context scaffolds across four diverse long-context tasks, while having comparable (or cheaper) cost per query.

RF-DETR, a light-weight detection transformer, uses weight-sharing NAS to optimize accuracy and latency for real-time detection across diverse datasets.

World models are categorized into three capability levels and four law regimes to better understand and develop predictive environment models for AI agents across diverse domains.

The study analyzes Claude Code's architecture, identifying five motivating human values and tracing them through thirteen design principles to specific implementation choices, including a core while-loop architecture and supporting systems for safety, context management, and extensibility.

Agentic coding tools receive goals written in natural language as input, break them down into specific tasks, and write or execute the actual code with minimal human intervention. Central to this process are agent context files ("READMEs for agents") that provide persistent, project-level instructions. In this paper, we conduct the first large-scale empirical study of 2,303 agent context files from 1,925 repositories to characterize their structure, maintenance, and content. We find that these files are not static documentation but complex, difficult-to-read artifacts that evolve like configuration code, maintained through frequent, small additions. Our content analysis of 16 instruction types shows that developers prioritize functional context, such as build and run commands (62.3%), implementation details (69.9%), and architecture (67.7%). We also identify a significant gap: non-functional requirements like security (14.5%) and performance (14.5%) are rarely specified. These findings indicate that while developers use context files to make agents functional, they provide few guardrails to ensure that agent-written code is secure or performant, highlighting the need for improved tooling and practices.

Bitnet.cpp enhances edge inference for ternary LLMs using a novel mixed-precision matrix multiplication library, achieving significant speed improvements over baselines.

LlamaFactory is a unified framework enabling efficient fine-tuning of large language models across various tasks using a web-based user interface.

Vista4D presents a video reshooting framework that uses 4D point cloud representation to synthesize scenes from new viewpoints while maintaining 4D consistency and camera control.

A novel GPT-based model, OmniFlatten, enables real-time natural full-duplex spoken dialogue through a multi-stage post-training technique that integrates speech and text without altering the original model's architecture.

GigaWorld-Policy introduces an action-centered World-Action Model that improves robotic policy learning by decoupling visual and motion representations, enabling faster inference and better task performance through dual supervision from action prediction and video generation.

LeWorldModel presents a stable end-to-end JEPA framework that trains efficiently from raw pixels using minimal loss terms while maintaining competitive performance in control tasks and encoding meaningful physical structures.

Zep, a memory layer service, outperforms MemGPT in the DMR benchmark and LongMemEval by excelling in dynamic knowledge integration and temporal reasoning, critical for enterprise use cases.

Enhancements to the AgentScope platform improve scalability, efficiency, and ease of use for large-scale multi-agent simulations through distributed mechanisms, flexible environments, and user-friendly tools.

AutoResearchBench is a benchmark for autonomous scientific literature discovery that evaluates AI agents' ability to conduct deep and wide research tasks with high difficulty, achieving low accuracy rates even among powerful LLMs.

AgentScope enhances agentic applications by providing flexible tool-based interactions, unified interfaces, and advanced infrastructure based on the ReAct paradigm, supporting efficient and safe development and deployment.

A self-supervised framework optimizes prompts for both closed and open-ended tasks by evaluating LLM outputs without external references, reducing costs and required data.

Training data can be structured as source code and evaluated through unit testing, enabling systematic debugging and improvement of language models' domain-specific capabilities.

RecursiveMAS extends recursive scaling principles from single models to multi-agent systems, enabling collaborative reasoning through iterative latent-space computations with improved efficiency and accuracy.

MemOS, a memory operating system for Large Language Models, addresses memory management challenges by unifying plaintext, activation-based, and parameter-level memories, enabling efficient storage, retrieval, and continual learning.

DFlash is a speculative decoding framework that uses a lightweight block diffusion model for parallel token drafting, achieving significant speedup over existing autoregressive methods while maintaining high-quality outputs.

Tequila, a trapping-free quantization method, enhances ternary quantization of Large Language Models by repurposing deadzone-trapped weights as dynamic biases, improving accuracy and inference speed.

Real-world data visualization (DV) requires native environmental grounding, cross-platform evolution, and proactive intent alignment. Yet, existing benchmarks often suffer from code-sandbox confinement, single-language creation-only tasks, and assumption of perfect intent. To bridge these gaps, we introduce DV-World, a benchmark of 260 tasks designed to evaluate DV agents across real-world professional lifecycles. DV-World spans three domains: DV-Sheet for native spreadsheet manipulation including chart and dashboard creation as well as diagnostic repair; DV-Evolution for adapting and restructuring reference visual artifacts to fit new data across diverse programming paradigms and DV-Interact for proactive intent alignment with a user simulator that mimics real-world ambiguous requirements. Our hybrid evaluation framework integrates Table-value Alignment for numerical precision and MLLM-as-a-Judge with rubrics for semantic-visual assessment. Experiments reveal that state-of-the-art models achieve less than 50% overall performance, exposing critical deficits in handling the complex challenges of real-world data visualization. DV-World provides a realistic testbed to steer development toward the versatile expertise required in enterprise workflows. Our data and code are available at https://github.com/DA-Open/DV-World{this project page}.

GEPA, a prompt optimizer using natural language reflection, outperforms RL methods like GRPO and MIPROv2 with fewer rollouts by learning high-level rules from trial and error.

LTX-2 is an open-source audiovisual diffusion model that generates synchronized video and audio content using a dual-stream transformer architecture with cross-modal attention and classifier-free guidance.

ReVSI addresses flaws in current spatial intelligence evaluation by creating a validated benchmark with improved annotations and controlled frame sampling conditions.

The PyTorch distributed data parallel module optimizes large-scale model training using techniques like gradient bucketing, computation-communication overlap, and selective synchronization to achieve near-linear scalability.

PDFMathTranslate enables layout-preserving scientific document translation using large language models and precise layout detection, offering improved precision, flexibility, and efficiency.