LTX-2: Efficient Joint Audio-Visual Foundation Model
Paper • 2601.03233 • Published • 187
Papers I'm going to read
-
-
MHLA: Restoring Expressivity of Linear Attention via Token-Level Multi-Head
Paper • 2601.07832 • Published • 53 -
Motion Attribution for Video Generation
Paper • 2601.08828 • Published • 72 -
Post-LayerNorm Is Back: Stable, ExpressivE, and Deep
Paper • 2601.19895 • Published • 27 -
Elastic Attention: Test-time Adaptive Sparsity Ratios for Efficient Transformers
Paper • 2601.17367 • Published • 33 -
Advancing Open-source World Models
Paper • 2601.20540 • Published • 135 -
Why Attention Patterns Exist: A Unifying Temporal Perspective Analysis
Paper • 2601.21709 • Published • 3 -
ERNIE 5.0 Technical Report
Paper • 2602.04705 • Published • 271 -
FASA: Frequency-aware Sparse Attention
Paper • 2602.03152 • Published • 154 -
LLaDA2.1: Speeding Up Text Diffusion via Token Editing
Paper • 2602.08676 • Published • 71 -
MOVA: Towards Scalable and Synchronized Video-Audio Generation
Paper • 2602.08794 • Published • 159 -
OPUS: Towards Efficient and Principled Data Selection in Large Language Model Pre-training in Every Iteration
Paper • 2602.05400 • Published • 355 -
When to Memorize and When to Stop: Gated Recurrent Memory for Long-Context Reasoning
Paper • 2602.10560 • Published • 31 -
Towards Autonomous Mathematics Research
Paper • 2602.10177 • Published • 36 -
MOSS-Audio-Tokenizer: Scaling Audio Tokenizers for Future Audio Foundation Models
Paper • 2602.10934 • Published • 50 -
Experiential Reinforcement Learning
Paper • 2602.13949 • Published • 76 -
BitDance: Scaling Autoregressive Generative Models with Binary Tokens
Paper • 2602.14041 • Published • 56 -
STAPO: Stabilizing Reinforcement Learning for LLMs by Silencing Rare Spurious Tokens
Paper • 2602.15620 • Published • 3 -
SLA2: Sparse-Linear Attention with Learnable Routing and QAT
Paper • 2602.12675 • Published • 59 -
VESPO: Variational Sequence-Level Soft Policy Optimization for Stable Off-Policy LLM Training
Paper • 2602.10693 • Published • 222 -
Avey-B
Paper • 2602.15814 • Published • 3 -
Decoding as Optimisation on the Probability Simplex: From Top-K to Top-P (Nucleus) to Best-of-K Samplers
Paper • 2602.18292 • Published • 13 -
Test-Time Training with KV Binding Is Secretly Linear Attention
Paper • 2602.21204 • Published • 32 -
Memory Caching: RNNs with Growing Memory
Paper • 2602.24281 • Published • 14 -
Timer-S1: A Billion-Scale Time Series Foundation Model with Serial Scaling
Paper • 2603.04791 • Published • 21 -
LoGeR: Long-Context Geometric Reconstruction with Hybrid Memory
Paper • 2603.03269 • Published • 63 -
ReMix: Reinforcement routing for mixtures of LoRAs in LLM finetuning
Paper • 2603.10160 • Published • 26 -
V_{0.5}: Generalist Value Model as a Prior for Sparse RL Rollouts
Paper • 2603.10848 • Published • 17 -
Video-Based Reward Modeling for Computer-Use Agents
Paper • 2603.10178 • Published • 43 -
Attention Residuals
Paper • 2603.15031 • Published • 193 -
Efficient Exploration at Scale
Paper • 2603.17378 • Published • 15 -
MinerU-Diffusion: Rethinking Document OCR as Inverse Rendering via Diffusion Decoding
Paper • 2603.22458 • Published • 139 -
Why Does Self-Distillation (Sometimes) Degrade the Reasoning Capability of LLMs?
Paper • 2603.24472 • Published • 57 -
Voxtral TTS
Paper • 2603.25551 • Published • 63 -
Out of Sight but Not Out of Mind: Hybrid Memory for Dynamic Video World Models
Paper • 2603.25716 • Published • 157 -
LongCat-Next: Lexicalizing Modalities as Discrete Tokens
Paper • 2603.27538 • Published • 150 -
TriAttention: Efficient Long Reasoning with Trigonometric KV Compression
Paper • 2604.04921 • Published • 117 -
Think in Strokes, Not Pixels: Process-Driven Image Generation via Interleaved Reasoning
Paper • 2604.04746 • Published • 73 -
ELT: Elastic Looped Transformers for Visual Generation
Paper • 2604.09168 • Published • 24 -
Matrix-Game 3.0: Real-Time and Streaming Interactive World Model with Long-Horizon Memory
Paper • 2604.08995 • Published • 51 -
Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation
Paper • 2604.10098 • Published • 82 -
Continuous Adversarial Flow Models
Paper • 2604.11521 • Published • 12 -
Efficient RL Training for LLMs with Experience Replay
Paper • 2604.08706 • Published • 23 -
RAD-2: Scaling Reinforcement Learning in a Generator-Discriminator Framework
Paper • 2604.15308 • Published • 29 -
Elucidating the SNR-t Bias of Diffusion Probabilistic Models
Paper • 2604.16044 • Published • 72 -
Extending One-Step Image Generation from Class Labels to Text via Discriminative Text Representation
Paper • 2604.18168 • Published • 95 -
Near-Future Policy Optimization
Paper • 2604.20733 • Published • 77 -
Tuna-2: Pixel Embeddings Beat Vision Encoders for Multimodal Understanding and Generation
Paper • 2604.24763 • Published • 71 -
V-GRPO: Online Reinforcement Learning for Denoising Generative Models Is Easier than You Think
Paper • 2604.23380 • Published • 4 -
Representation Fréchet Loss for Visual Generation
Paper • 2604.28190 • Published • 32 -
HERMES++: Toward a Unified Driving World Model for 3D Scene Understanding and Generation
Paper • 2604.28196 • Published • 74 -
Stream-T1: Test-Time Scaling for Streaming Video Generation
Paper • 2605.04461 • Published • 109 -
Continuous Latent Diffusion Language Model
Paper • 2605.06548 • Published • 87 -
Flow-OPD: On-Policy Distillation for Flow Matching Models
Paper • 2605.08063 • Published • 102 -
Listwise Policy Optimization: Group-based RLVR as Target-Projection on the LLM Response Simplex
Paper • 2605.06139 • Published • 69 -
Mean Mode Screaming: Mean--Variance Split Residuals for 1000-Layer Diffusion Transformers
Paper • 2605.06169 • Published • 238 -
Rethinking State Tracking in Recurrent Models Through Error Control Dynamics
Paper • 2605.07755 • Published • 24 -
δ-mem: Efficient Online Memory for Large Language Models
Paper • 2605.12357 • Published • 133 -
Qwen-Image-VAE-2.0 Technical Report
Paper • 2605.13565 • Published • 62 -
Asymmetric Flow Models
Paper • 2605.12964 • Published • 23 -
LongLive-2.0: An NVFP4 Parallel Infrastructure for Long Video Generation
Paper • 2605.18739 • Published • 116 -
Anti-Self-Distillation for Reasoning RL via Pointwise Mutual Information
Paper • 2605.11609 • Published • 196 -
Delta Attention Residuals
Paper • 2605.18855 • Published • 8 -
Stable Audio 3
Paper • 2605.17991 • Published • 21 -
Toto 2.0: Time Series Forecasting Enters the Scaling Era
Paper • 2605.20119 • Published • 39 -
Gated DeltaNet-2: Decoupling Erase and Write in Linear Attention
Paper • 2605.22791 • Published • 33 -
LocateAnything: Fast and High-Quality Vision-Language Grounding with Parallel Box Decoding
Paper • 2605.27365 • Published • 146 -
The MiniMax-M2 Series: Mini Activations Unleashing Max Real-World Intelligence
Paper • 2605.26494 • Published • 41 -
DVAO: Dynamic Variance-adaptive Advantage Optimization for Multi-reward Reinforcement Learning
Paper • 2605.25604 • Published • 139 -
ProRL: Effective Reinforcement Learning for Proactive Recommendation via Rectified Policy Gradient Estimation
Paper • 2605.28293 • Published • 89 -
Everything at Every Scale: Scale-Invariant Diffusion with Continuous Super-Resolution
Paper • 2605.26032 • Published • 15 -
Trust Region On-Policy Distillation
Paper • 2606.01249 • Published • 48 -
Your UnEmbedding Matrix is Secretly a Feature Lens for Text Embeddings
Paper • 2606.07502 • Published • 100 -
Dynamic Linear Attention
Paper • 2606.10650 • Published • 10 -
Rethinking the Divergence Regularization in LLM RL
Paper • 2606.09821 • Published • 34 -
Redesign Mixture-of-Experts Routers with Manifold Power Iteration
Paper • 2606.12397 • Published • 90 -
MiniMax Sparse Attention
Paper • 2606.13392 • Published • 156 -
Ling and Ring 2.6 Technical Report: Efficient and Instant Agentic Intelligence at Trillion-Parameter Scale
Paper • 2606.15079 • Published • 87 -
Show the Signal, Hide the Noise: Spectral Forcing for Pixel-Space Diffusion
Paper • 2606.15236 • Published • 22 -
Rethinking the Role of Efficient Attention in Hybrid Architectures
Paper • 2606.15378 • Published • 20 -
FLAT: Feedforward Latent Triangle Splatting for Geometrically Accurate Scene Generation
Paper • 2606.24876 • Published • 24 -
PhysiFormer: Learning to Simulate Mechanics in World Space
Paper • 2606.27364 • Published • 11 -
DanceOPD: On-Policy Generative Field Distillation
Paper • 2606.27377 • Published • 81 -
Nemotron-TwoTower: Diffusion Language Modeling with Pretrained Autoregressive Context
Paper • 2606.26493 • Published • 3 -
MultiHashFormer: Hash-based Generative Language Models
Paper • 2606.28057 • Published • 22 -
AsyncOPD: How Stale Can On-Policy Distillation Be?
Paper • 2606.24143 • Published • 32 -
Valdi: Value Diffusion World Models
Paper • 2607.00917 • Published • 15 -
The State-Prediction Separation Hypothesis
Paper • 2607.01218 • Published • 12 -
Morphing into Hybrid Attention Models
Paper • 2606.30562 • Published • 52 -
MANCE: Manifold Aware Concept Erasure
Paper • 2607.03973 • Published • 49 -
Single-Rollout Asynchronous Optimization for Agentic Reinforcement Learning
Paper • 2607.07508 • Published • 30 -
Sparse Delta Memory: Scaling the State of Linear RNNs through Sparsity
Paper • 2607.07386 • Published • 13 -
Linear Attention Architectures: Mechanisms, Trade-offs, and Cross-Layer Routing
Paper • 2607.07953 • Published • 16 -
Trust Region Policy Distillation
Paper • 2607.04751 • Published • 35 -
Flow-ERD: Agent-type Aware Flow Matching with Entropy-Regularized Distillation for Diverse Traffic Simulation
Paper • 2607.06957 • Published • 12 -
Ring-Zero: Scaling Zero RL to a Trillion Parameters for Emergent Reasoning
Paper • 2607.12395 • Published • 100 -
Boogu-Image-0.1: Boosting Open Agentic Multimodal Generation via Understanding under a Minimal Budget
Paper • 2607.13125 • Published • 139 -
Loop the Loopies!
Paper • 2607.16051 • Published • 77 -
VideoRAE: Taming Video Foundation Models for Generative Modeling via Representation Autoencoders
Paper • 2607.14088 • Published • 14 -
xHC: Expanded Hyper-Connections
Paper • 2607.14530 • Published • 56 -
Text Template Tokens Are Implicit Semantic Registers in Diffusion Transformers
Paper • 2607.19139 • Published • 75 -
Kimi K3: Open Frontier Intelligence
Paper • 2607.24653 • Published • 491 -
dRAE: Representation Autoencoder with Hyper-Spherical Codes
Paper • 2607.22148 • Published • 12 -
DiffusionGemma Technical Report
Paper • 2608.00146 • Published • 36 -
AURORA-LM: Autoencoding Unified Representation for Continuous-Latent Diffusion Language Modeling
Paper • 2608.02602 • Published • 79 -
Skaling: Chinchilla's Exponents Meet Kaplan's Coupling
Paper • 2608.07222 • Published • 9 -
SFT Conflicts, RL Coexists: A Theoretical and Empirical Analysis of Multi-Task Learning for LLMs
Paper • 2608.03573 • Published • 52 -
Motif 3: Technical Report
Paper • 2608.09119 • Published • 43 -
The Loss Does Not See the Basis, but Adam Does
Paper • 2608.05136 • Published • 7 -
Full-bandwidth transformer
Paper • 2608.08888 • Published • 14