Trending Papers

Moonshine, an encoder-decoder transformer architecture for speech recognition, uses Rotary Position Embedding, reducing compute requirements without decreasing accuracy.

Monolingual ASR models trained on a balanced mix of high-quality, pseudo-labeled, and synthetic data outperform multilingual models for small model sizes, achieving superior error rates and enabling on-device ASR for underrepresented languages.

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.

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.

_paperbanana is an agentic framework that automates the creation of publication-ready academic illustrations using advanced vision-language models and image generation techniques.

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.

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

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.

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.

A suite of open-source music foundation models is introduced, featuring components for audio-text alignment, lyric recognition, music coding, and large language model-based song generation with controllable attributes and scalable parameterization.

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.

Agent Lightning is a flexible RL framework for training LLMs in various agents, using a hierarchical RL algorithm and decoupling execution from training to handle complex interactions.

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.

IMO-Bench, a suite of advanced reasoning benchmarks, evaluates mathematical reasoning capabilities of foundation models using IMO-level problems and detailed grading guidelines, achieving gold-level performance with Gemini Deep Think.

SkillRL enables LLM agents to improve through hierarchical skill discovery and recursive policy evolution, achieving superior performance on complex tasks while reducing computational overhead.

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.

SceneSmith is a hierarchical agentic framework that generates simulation-ready indoor environments from natural language prompts through multiple stages involving VLM agents and integrated asset generation techniques.

UI-Venus-1.5 is a unified GUI agent with improved performance through mid-training stages, online reinforcement learning, and model merging techniques.

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.

AI-Trader presents the first fully automated live benchmark for evaluating large language models in financial decision-making across multiple markets with autonomous information processing.

A unified Vision-Language Model, dots.ocr, achieves state-of-the-art performance on document layout parsing by jointly learning layout detection, text recognition, and relational understanding, validated on OmniDocBench and XDocParse benchmarks.

DeepCode, a fully autonomous framework, addresses the challenges of document-to-codebase synthesis by optimizing information flow through source compression, structured indexing, knowledge injection, and error correction, achieving state-of-the-art performance and surpassing human experts.

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.

MiniCPM-V 4.5, a 8B parameter multimodal large language model, achieves high performance and efficiency through a unified 3D-Resampler architecture, a unified learning paradigm, and a hybrid reinforcement learning strategy.

A lightweight 5B unified multimodal model achieves competitive performance through hierarchical feature extraction, learnable think tokens, and progressive training strategies including alignment pre-training, joint supervised fine-tuning, and reinforcement learning with MR-GRPO.

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.

Large language model agents trained in synthetic environments with code-driven simulations and database-backed state transitions demonstrate superior out-of-distribution generalization compared to traditional benchmark-specific approaches.

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.

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.

Financial markets are noisy and non-stationary, making alpha mining highly sensitive to noise in backtesting results and sudden market regime shifts. While recent agentic frameworks improve alpha mining automation, they often lack controllable multi-round search and reliable reuse of validated experience. To address these challenges, we propose QuantaAlpha, an evolutionary alpha mining framework that treats each end-to-end mining run as a trajectory and improves factors through trajectory-level mutation and crossover operations. QuantaAlpha localizes suboptimal steps in each trajectory for targeted revision and recombines complementary high-reward segments to reuse effective patterns, enabling structured exploration and refinement across mining iterations. During factor generation, QuantaAlpha enforces semantic consistency across the hypothesis, factor expression, and executable code, while constraining the complexity and redundancy of the generated factor to mitigate crowding. Extensive experiments on the China Securities Index 300 (CSI 300) demonstrate consistent gains over strong baseline models and prior agentic systems. When utilizing GPT-5.2, QuantaAlpha achieves an Information Coefficient (IC) of 0.1501, with an Annualized Rate of Return (ARR) of 27.75% and a Maximum Drawdown (MDD) of 7.98%. Moreover, factors mined on CSI 300 transfer effectively to the China Securities Index 500 (CSI 500) and the Standard & Poor's 500 Index (S&P 500), delivering 160% and 137% cumulative excess return over four years, respectively, which indicates strong robustness of QuantaAlpha under market distribution shifts.

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

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

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.

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.

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.

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

A fully end-to-end Transformer-based audio tokenizer architecture achieves high-fidelity reconstruction across diverse audio domains and enables superior text-to-speech and automatic speech recognition performance.

MOVA is an open-source model that generates synchronized audio-visual content using a Mixture-of-Experts architecture with 32 billion parameters, supporting image-text to video-audio generation tasks.

IndexTTS, an enhanced text-to-speech system combining XTTS and Tortoise models, offers improved naturalness, enhanced voice cloning, and controllable usage through hybrid character-pinyin modeling and optimized vector quantization.

Code2Worlds enables 4D dynamic scene generation by formulating it as language-to-simulation code generation with a dual-stream architecture and physics-aware closed-loop refinement.

Audio Language Models (ALM) have emerged as the dominant paradigm for speech and music generation by representing audio as sequences of discrete tokens. Yet, unlike text tokens, which are invertible, audio tokens are extracted from lossy codecs with a limited bitrate. As a consequence, increasing audio quality requires generating more tokens, which imposes a trade-off between fidelity and computational cost. We address this issue by studying Continuous Audio Language Models (CALM). These models instantiate a large Transformer backbone that produces a contextual embedding at every timestep. This sequential information then conditions an MLP that generates the next continuous frame of an audio VAE through consistency modeling. By avoiding lossy compression, CALM achieves higher quality at lower computational cost than their discrete counterpart. Experiments on speech and music demonstrate improved efficiency and fidelity over state-of-the-art discrete audio language models, facilitating lightweight, high-quality audio generation. Samples are available at https://continuous-audio-language-models.github.io

A survey of self-evolving AI agents that adapt to dynamic environments through automatic enhancement based on interaction data and feedback.

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

Region-to-Image Distillation enables fine-grained visual perception in MLLMs by training models to internally perform iterative zooming during inference, eliminating the need for repeated tool calls and visual re-encoding while maintaining high performance across multiple benchmarks.

Kimi K2.5 is an open-source multimodal agentic model that enhances text and vision processing through joint optimization techniques and introduces Agent Swarm for parallel task execution.

Z-Image, a 6B-parameter Scalable Single-Stream Diffusion Transformer (S3-DiT) model, achieves high-performance image generation with reduced computational cost, offering sub-second inference and compatibility with consumer hardware.

A vision-language-action model enhanced with world model-based reinforcement learning demonstrates improved performance and long-horizon execution capabilities for robotic manipulation tasks.

A vision-language-action model for robotics combines large-scale pretraining with specialized training techniques to enable real-time execution and high-performance manipulation tasks.

Single-stream Policy Optimization (SPO) improves policy-gradient training for Large Language Models by eliminating group-based issues and providing a stable, low-variance learning signal, leading to better performance and efficiency.

Researchers developed a discrete audio codec called ReasoningCodec that separates audio into reasoning and reconstruction tokens for improved understanding and generation, and created UniAudio 2.0, a unified autoregressive model trained on large-scale text and audio data that shows strong performance across various audio tasks and generalizes well in few-shot and zero-shot scenarios.