| Sep 11, 2026 | Gated Delta Networks: Improving Mamba2 with Delta Rule |
| Sep 09, 2026 | DeltaNet Explained (Part I) |
| Sep 06, 2026 | ExploitGym: Can AI Agents Turn Security Vulnerabilities into Real Attacks? |
| Sep 02, 2026 | Training a Misaligned Reward Seeker |
| Sep 02, 2026 | Teaching Claude Why |
| Aug 21, 2026 | R-lens: Making J-lens More Faithful on Early Layers |
| Aug 19, 2026 | Verbalizable Representations Form a Global Workspace in Language Models |
| Aug 11, 2026 | Scaling RL training in the age of multi-agent systems |
| Aug 09, 2026 | InstructPix2Pix: Learning to Follow Image Editing Instructions |
| Jul 26, 2026 | Video Generation Models are General-Purpose Vision Learners |
| Jul 26, 2026 | Self-Supervised Flow Matching for Scalable Multi-Modal Synthesis |
| Jul 14, 2026 | AIDE²: The First Evidence of Recursive Self-Improvement |
| May 18, 2026 | Training Agents Inside of Scalable World Models |
| Apr 13, 2026 | What’s in the Image? A Deep-Dive into the Vision of Vision Language Models |
| Apr 08, 2026 | Generative Modeling via Drifting |
| Apr 04, 2026 | FLUX.1 Kontext: Flow Matching for In-Context Image Generation and Editing in Latent Space |
| Apr 01, 2026 | MaxRL: Maximum Likelihood via Reinforcement Learning |
| Mar 22, 2026 | Circuit Mechanisms for Spatial Relation Generation in Diffusion Transformers |
| Mar 21, 2026 | Same Task, Different Circuits: Disentangling Modality-Specific Mechanisms in VLMs |
| Mar 20, 2026 | Position-aware Automatic Circuit Discovery |
| Mar 19, 2026 | Arithmetic Without Algorithms: Language Models Solve Math With a Bag of Heuristics |
| Mar 18, 2026 | Neural Thickets: Diverse Task Experts Are Dense Around Pretrained Weights |
| Mar 17, 2026 | Attention Residuals |
| Mar 16, 2026 | Fine-Tuning Enhances Existing Mechanisms: A Case Study on Entity Tracking |
| Mar 15, 2026 | Causal Abstractions of Neural Networks |
| Feb 26, 2026 | End-to-End Test-Time Training for Long Context |
| Feb 26, 2026 | Interpreting Physics in Video World Models |
| Feb 21, 2026 | Language Models use Lookbacks to Track Beliefs |
| Feb 20, 2026 | Fast KV Compaction via Attention Matching |
| Feb 17, 2026 | BRIDGE: Predicting Human Task Completion Time From Model Performance |
| Feb 16, 2026 | HunyuanVideo: A Systematic Framework For Large Video Generative Models |
| Feb 14, 2026 | Exploring Multimodal Diffusion Transformers for Enhanced Prompt-based Image Editing |
| Feb 14, 2026 | Unraveling MMDiT Blocks: Training-free Analysis and Enhancement of Text-conditioned Diffusion |
| Feb 13, 2026 | Stable Flow: Vital Layers for Training-Free Image Editing |
| Feb 13, 2026 | Localizing Knowledge in Diffusion Transformers |
| Feb 12, 2026 | Tutorial on Diffusion Models for Imaging and Vision |
| Feb 11, 2026 | Learning to Discover at Test Time |
| Feb 10, 2026 | ConceptAttention: Diffusion Transformers Learn Highly Interpretable Features |
| Feb 09, 2026 | Recursive Language Models |
| Feb 09, 2026 | FLUX.2: Analyzing and Enhancing the Latent Space of FLUX – Representation Comparison |
| Feb 08, 2026 | Scaling Rectified Flow Transformers for High-Resolution Image Synthesis |
| Feb 08, 2026 | Scalable Diffusion Models with Transformers |
| Feb 07, 2026 | Flow Straight and Fast: Learning to Generate and Transfer Data with Rectified Flow |
| Jan 24, 2026 | The Adolescence of Technology |
| Jan 20, 2026 | Diffusion Meets Flow Matching: Two Sides of the Same Coin |
| Jan 12, 2026 | mHC: Manifold-Constrained Hyper-Connections |
| Jan 10, 2026 | A Rosetta Stone for AI Benchmarks |
| Jan 06, 2026 | Measuring AI Ability to Complete Long Software Tasks |
| Jan 06, 2026 | On scalable oversight with weak LLMs judging strong LLMs |
| Jan 05, 2026 | Reliable and Efficient Amortized Model-based Evaluation |