Trending Papers

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.

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.

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.

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.

Hyperagents represent a self-referential framework that integrates task and meta-agents into a single editable program, enabling metacognitive self-modification and open-ended improvement across diverse computational domains.

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.

daVinci-MagiHuman is an open-source audio-video generative model that synchronizes text, video, and audio through a single-stream Transformer architecture, achieving high-quality human-centric content generation with efficient inference capabilities.

We present MiroThinker v1.0, an open-source research agent designed to advance tool-augmented reasoning and information-seeking capabilities. Unlike previous agents that only scale up model size or context length, MiroThinker explores interaction scaling at the model level, systematically training the model to handle deeper and more frequent agent-environment interactions as a third dimension of performance improvement. Unlike LLM test-time scaling, which operates in isolation and risks degradation with longer reasoning chains, interactive scaling leverages environment feedback and external information acquisition to correct errors and refine trajectories. Through reinforcement learning, the model achieves efficient interaction scaling: with a 256K context window, it can perform up to 600 tool calls per task, enabling sustained multi-turn reasoning and complex real-world research workflows. Across four representative benchmarks-GAIA, HLE, BrowseComp, and BrowseComp-ZH-the 72B variant achieves up to 81.9%, 37.7%, 47.1%, and 55.6% accuracy respectively, surpassing previous open-source agents and approaching commercial counterparts such as GPT-5-high. Our analysis reveals that MiroThinker benefits from interactive scaling consistently: research performance improves predictably as the model engages in deeper and more frequent agent-environment interactions, demonstrating that interaction depth exhibits scaling behaviors analogous to model size and context length. These findings establish interaction scaling as a third critical dimension for building next-generation open research agents, complementing model capacity and context windows.

A generalist language model agent system autonomously designs and improves task-specific agents through memory-based reinforcement learning with stateful prompts and skill libraries.

EvoScientist is an adaptive multi-agent framework that enhances scientific discovery by continuously learning from past interactions through persistent memory modules.

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

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

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.

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

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.

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.

Residual connections with PreNorm are standard in modern LLMs, yet they accumulate all layer outputs with fixed unit weights. This uniform aggregation causes uncontrolled hidden-state growth with depth, progressively diluting each layer's contribution. We propose Attention Residuals (AttnRes), which replaces this fixed accumulation with softmax attention over preceding layer outputs, allowing each layer to selectively aggregate earlier representations with learned, input-dependent weights. To address the memory and communication overhead of attending over all preceding layer outputs for large-scale model training, we introduce Block AttnRes, which partitions layers into blocks and attends over block-level representations, reducing the memory footprint while preserving most of the gains of full AttnRes. Combined with cache-based pipeline communication and a two-phase computation strategy, Block AttnRes becomes a practical drop-in replacement for standard residual connections with minimal overhead. Scaling law experiments confirm that the improvement is consistent across model sizes, and ablations validate the benefit of content-dependent depth-wise selection. We further integrate AttnRes into the Kimi Linear architecture (48B total / 3B activated parameters) and pre-train on 1.4T tokens, where AttnRes mitigates PreNorm dilution, yielding more uniform output magnitudes and gradient distribution across depth, and improves downstream performance across all evaluated tasks.

A continual meta-learning framework for large language model agents that jointly evolves policies and reusable behavioral skills while minimizing downtime through opportunistic updates and skill-driven adaptation.

WildWorld is a large-scale dataset for action-conditioned world modeling that provides explicit state annotations from a photorealistic game, enabling better understanding of latent-state dynamics and long-horizon consistency.

OpenClaw-RL framework enables policy learning from diverse next-state signals across multiple interaction modalities using asynchronous training with PRM judges and hindsight-guided distillation.

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.

AutoGaze is a lightweight module that reduces redundant video patches before processing by vision transformers or multi-modal large language models, enabling efficient processing of long, high-resolution videos while maintaining performance.

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.

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.

Fish Audio S2 is an open-source text-to-speech system with multi-speaker capabilities, multi-turn generation, and instruction-following control through natural-language descriptions, utilizing a multi-stage training approach and production-ready inference engine.

The AI Scientist is a framework that enables independent scientific research through automatic idea generation, experimentation, and paper writing using large language models.

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.

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

MinerU-Diffusion is a diffusion-based framework that replaces autoregressive decoding with parallel diffusion denoising for document OCR, improving robustness and decoding speed.

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.

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.

EverMemOS presents a self-organizing memory system for large language models that processes dialogue streams into structured memory cells and scenes to enhance long-term interaction capabilities.

PowerInfer, a high-speed LLM inference engine for personal computers, enhances efficiency using hotspot neuron analysis, GPU-CPU hybrid computation, adaptive predictors, and neuron-aware sparse operators, achieving performance close to server-grade GPUs.

VideoDetective framework improves long video understanding by integrating query-to-segment relevance and inter-segment affinity through visual-temporal graphs and hypothesis verification loops.

The AI Scientist-v2 autonomously proposes hypotheses, performs experiments, analyzes data, and writes peer-reviewed scientific papers, marking the first fully AI-generated paper accepted by a conference.

SpecEyes accelerates agentic multimodal large language models by using a lightweight speculative planner with cognitive gating and heterogeneous parallel processing to reduce latency and improve throughput.

Conditional memory via Engram module enhances Transformer models by enabling efficient knowledge lookup and improving reasoning capabilities through optimized sparsity allocation.

OASIS is a scalable and generalizable social media simulator that models large-scale user interactions and replicates complex social phenomena across platforms.

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.

Logics-Parsing, an end-to-end LVLM model enhanced with reinforcement learning, improves document parsing by optimizing layout analysis and reading order inference, achieving state-of-the-art performance on a diverse benchmark.

A two-expert architecture for pose-free 3D Gaussian Splatting separates geometry estimation from appearance synthesis, achieving superior performance compared to unified monolithic approaches.

Cheers is a unified multimodal model that decouples visual details from semantic representations using a vision tokenizer, LLM-based Transformer, and cascaded flow matching head to achieve efficient joint optimization for both visual understanding and generation tasks.

Agent S2, a compositional framework using Mixture-of-Grounding and Proactive Hierarchical Planning, achieves state-of-the-art performance in computer use automation across various benchmarks and operating systems.

mmGRPO, a multi-module extension of GRPO, enhances accuracy in modular AI systems by optimizing LM calls and prompts across various tasks.

Geometric Latent Diffusion (GLD) framework utilizes geometric foundation models' feature space as latent space for novel view synthesis, achieving superior 2D and 3D performance while reducing training time significantly.

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.

OpenResearcher presents a reproducible pipeline for training deep research agents using offline search environments and synthesized trajectories, achieving improved accuracy on benchmark tasks.

InfiniDepth represents depth as neural implicit fields using a local implicit decoder, enabling continuous 2D coordinate querying for arbitrary-resolution depth estimation and superior performance in fine-detail regions.

The Qwen3-TTS series presents advanced multilingual text-to-speech models with voice cloning and controllable speech generation capabilities, utilizing dual-track LM architecture and specialized speech tokenizers for efficient streaming synthesis.

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.