dLLM: Simple Diffusion Language Modeling
Paper • 2602.22661 • Published • 152
idea
-
-
OpenSeeker: Democratizing Frontier Search Agents by Fully Open-Sourcing Training Data
Paper • 2603.15594 • Published • 148 -
Qianfan-OCR: A Unified End-to-End Model for Document Intelligence
Paper • 2603.13398 • Published • 152 -
Penguin-VL: Exploring the Efficiency Limits of VLM with LLM-based Vision Encoders
Paper • 2603.06569 • Published • 117 -
Beyond Language Modeling: An Exploration of Multimodal Pretraining
Paper • 2603.03276 • Published • 102 -
Demystifing Video Reasoning
Paper • 2603.16870 • Published • 367 -
Strategic Navigation or Stochastic Search? How Agents and Humans Reason Over Document Collections
Paper • 2603.12180 • Published • 64 -
LoGeR: Long-Context Geometric Reconstruction with Hybrid Memory
Paper • 2603.03269 • Published • 62 -
Omni-Diffusion: Unified Multimodal Understanding and Generation with Masked Discrete Diffusion
Paper • 2603.06577 • Published • 48 -
Online Experiential Learning for Language Models
Paper • 2603.16856 • Published • 57 -
HiAR: Efficient Autoregressive Long Video Generation via Hierarchical Denoising
Paper • 2603.08703 • Published • 32 -
Reading, Not Thinking: Understanding and Bridging the Modality Gap When Text Becomes Pixels in Multimodal LLMs
Paper • 2603.09095 • Published • 28 -
S2D2: Fast Decoding for Diffusion LLMs via Training-Free Self-Speculation
Paper • 2603.25702 • Published • 6 -
MemMA: Coordinating the Memory Cycle through Multi-Agent Reasoning and In-Situ Self-Evolution
Paper • 2603.18718 • Published • 10 -
Pixel-level Scene Understanding in One Token: Visual States Need What-is-Where Composition
Paper • 2603.13904 • Published • 4 -
CUA-Suite: Massive Human-annotated Video Demonstrations for Computer-Use Agents
Paper • 2603.24440 • Published • 93 -
GradMem: Learning to Write Context into Memory with Test-Time Gradient Descent
Paper • 2603.13875 • Published • 34 -
Dynamic Chunking Diffusion Transformer
Paper • 2603.06351 • Published • 14 -
MemSkill: Learning and Evolving Memory Skills for Self-Evolving Agents
Paper • 2602.02474 • Published • 62 -
MSign: An Optimizer Preventing Training Instability in Large Language Models via Stable Rank Restoration
Paper • 2602.01734 • Published • 32 -
Scaling Embeddings Outperforms Scaling Experts in Language Models
Paper • 2601.21204 • Published • 102 -
VersatileFFN: Achieving Parameter Efficiency in LLMs via Adaptive Wide-and-Deep Reuse
Paper • 2512.14531 • Published • 15 -
MIDUS: Memory-Infused Depth Up-Scaling
Paper • 2512.13751 • Published • 9 -
UI-Voyager: A Self-Evolving GUI Agent Learning via Failed Experience
Paper • 2603.24533 • Published • 46 -
Understanding the Challenges in Iterative Generative Optimization with LLMs
Paper • 2603.23994 • Published • 26 -
MinerU-Diffusion: Rethinking Document OCR as Inverse Rendering via Diffusion Decoding
Paper • 2603.22458 • Published • 131 -
SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning
Paper • 2603.23483 • Published • 60 -
CanViT: Toward Active-Vision Foundation Models
Paper • 2603.22570 • Published • 11 -
Look Where It Matters: High-Resolution Crops Retrieval for Efficient VLMs
Paper • 2603.16932 • Published • 85 -
Repurposing Geometric Foundation Models for Multi-view Diffusion
Paper • 2603.22275 • Published • 46 -
Generalized Discrete Diffusion from Snapshots
Paper • 2603.21342 • Published • 11 -
MemDLM: Memory-Enhanced DLM Training
Paper • 2603.22241 • Published • 3 -
HopChain: Multi-Hop Data Synthesis for Generalizable Vision-Language Reasoning
Paper • 2603.17024 • Published • 107 -
Reasoning as Compression: Unifying Budget Forcing via the Conditional Information Bottleneck
Paper • 2603.08462 • Published • 21 -
BEAVER: A Training-Free Hierarchical Prompt Compression Method via Structure-Aware Page Selection
Paper • 2603.19635 • Published • 11 -
Teaching an Agent to Sketch One Part at a Time
Paper • 2603.19500 • Published • 5 -
From Masks to Pixels and Meaning: A New Taxonomy, Benchmark, and Metrics for VLM Image Tampering
Paper • 2603.20193 • Published • 1 -
FASTER: Rethinking Real-Time Flow VLAs
Paper • 2603.19199 • Published • 57 -
Memento-Skills: Let Agents Design Agents
Paper • 2603.18743 • Published • 56 -
Efficient Reasoning with Balanced Thinking
Paper • 2603.12372 • Published • 143 -
Efficient Exploration at Scale
Paper • 2603.17378 • Published • 13 -
Out of Sight but Not Out of Mind: Hybrid Memory for Dynamic Video World Models
Paper • 2603.25716 • Published • 147 -
Beyond Single Tokens: Distilling Discrete Diffusion Models via Discrete MMD
Paper • 2603.20155 • Published • 8 -
Agentic Proposing: Enhancing Large Language Model Reasoning via Compositional Skill Synthesis
Paper • 2602.03279 • Published -
GEMS: Agent-Native Multimodal Generation with Memory and Skills
Paper • 2603.28088 • Published • 79 -
AdaptToken: Entropy-based Adaptive Token Selection for MLLM Long Video Understanding
Paper • 2603.28696 • Published • 5 -
Density-aware Soft Context Compression with Semi-Dynamic Compression Ratio
Paper • 2603.25926 • Published • 7 -
Universal YOCO for Efficient Depth Scaling
Paper • 2604.01220 • Published • 13 -
Adaptive Loops and Memory in Transformers: Think Harder or Know More?
Paper • 2603.08391 • Published