| 2026-11-20 | 2611.24809 | On-Policy Morphing into Hybrid Attention Models | Disen Lan et al. | arXiv | 12 | |
| 2026-11-09 | 2610.41711 | Hierarchical Residuals: A Minimalist Decoupling of Residual Width from Compute Width | Ziteng Wang et al. | arXiv | 8 | |
| 2026-11-05 | 2611.08089 | SAE Interprets MoE: Sparse AutoEncoders Reveal That MoE Experts Exhibit Human-Interpretable Fine-Grained Specialization | Anyuan Zhuo et al. | arXiv | 7 |  |
| 2026-10-29 | 2610.40362 | Your LR Schedule May Not Be Principled: Effective Learning Rate Scheduling in Weight-Normalized Training | Qifan Yu et al. | arXiv | 7 | |
| 2026-10-23 | 2610.31305 | Cola2: Co-Adaptive Hierarchical Language Modeling with Continuous Semantic Compression and Balanced Discrete Generation | Hongcan Guo et al. | arXiv | 21 | |
| 2026-10-15 | 2610.20399 | Cortex: Consolidating Context into Parametric Memory | Jintian Zhang et al. | arXiv | 70 | |
| 2026-10-12 | 2610.12102 | Few-Step Generation via Data-Space Iteration | Shanchuan Lin et al. | arXiv | 3 | |
| 2026-10-01 | 2609.30840 | Aligning One-Step Generative Models with Reward-Weighted Transport Distillation | Austin Wang et al. | arXiv | 2 | |
| 2026-09-30 | 2609.36322 | Periodic Weak Spots: Phase Sensitivity from Chunked KV-Cache Compression | Xingyu Zhu et al. | arXiv | 36 | |
| 2026-09-28 | 2609.31093 | Block Sparse Attention with Log-Linear Complexity | Bohao Tang et al. | arXiv | 32 | |
| 2026-09-22 | 2609.22870 | Towards Full Pipeline FP8 Reinforcement Learning for LLMs | Fanchao Chen et al. | arXiv | 14 | |
| 2026-09-21 | 2609.20816 | Paint-Anything: Unified Any-Color Control for Image Generation and Editing | Ji Xie et al. | arXiv | 50 | |
| 2026-09-03 | 2609.01437 | HarnessDev: Can LLMs Create and Evolve Their Own Agent Harness? | Yuhao Wu et al. | arXiv | 88 | |
| 2026-09-03 | 2608.31111 | Aspire: Can Models Self-Evolve from Vague Goals? | Yuhao Wu et al. | arXiv | 49 | |
| 2026-09-03 | 2608.31100 | S3Gym: Can LLMs Turn Self-Testing and Self-Judging into Self-Improvement? | Jiajun Shi et al. | arXiv | 37 | |
| 2026-09-02 | 2609.01343 | SMELT: Scaling Laws for Compute-Matched MoE Looped Transformers | Shaowen Wang et al. | arXiv | 54 | |
| 2026-09-01 | 2608.29335 | GenFirst: Generation Before Reconstruction for Stable End-to-End Latent Generative Modeling | Guangting Zheng et al. | arXiv | 72 | |
| 2026-09-01 | 2608.30821 | Lucida: Parse, Generate, and Place for Composable Real-to-Sim Scene Modeling | Minghan Qin et al. | arXiv | 64 | |
| 2026-08-26 | 2608.24646 | On-Policy Self-Distillation in Diffusion Models | Wei Zhou et al. | arXiv | 68 | |
| 2026-08-19 | 2608.17800 | StartupBench: Benchmarking General-Purpose Agents on Market-Validated End-to-End Workflows | Liya Zhu et al. | arXiv | 14 | |
| 2026-08-17 | 2608.14071 | Scaling Domain Data Repetition in LLM Pretraining | Jingwei Li et al. | arXiv | 16 | |
| 2026-08-10 | 2608.07110 | Modular TTT: Rethinking Test-Time Training as Composable Modules | Bohao Tang et al. | arXiv | 11 | |
| 2026-08-07 | 2608.05747 | GST-Bench: Can VLMs Develop Global Spatial Awareness from Video? | Qifeng Zhang et al. | arXiv | 25 | |
| 2026-08-03 | 2607.29679 | Scaling Properties of Text Conditioning in Visual Generation | Zilong Chen et al. | arXiv | 30 | |
| 2026-07-21 | 2607.18227 | FlowMimic: Mask-free Visual Editing and Generation with Pixel-pair Warped Flow Field for Online Video Editing Data Generation and Modality Mimicry | Dingyun Zhang et al. | arXiv | 52 | |
| 2026-07-15 | 2607.11886 | Read It Back: Pretrained MLLMs Are Zero-Shot Reward Models for Text-to-Image Generation | Runhui Huang et al. | arXiv | 57 | |
| 2026-07-10 | 2607.08763 | OpenCoF: Learning to Reason Through Video Generation | Xinyan Chen et al. | arXiv | 24 | |
| 2026-07-10 | 2607.06987 | UP: Unbounded Positive Asymmetric Optimization for Breaking the Exploration-Stability Dilemma | Chongyu Fan et al. | arXiv | 8 | |
| 2026-07-07 | 2607.05155 | EdgeBench: Unveiling Scaling Laws of Learning from Real-World Environments | Deyao Zhu et al. | arXiv | 15 | |
| 2026-07-03 | 2606.30562 | Morphing into Hybrid Attention Models | Disen Lan et al. | arXiv | 49 | |
| 2026-07-02 | 2607.00248 | Seed2.0 Model Card: Towards Intelligence Frontier for Real-World Complexity | Bytedance Seed | arXiv | 30 | |
| 2026-06-29 | 2606.28133 | Translation as a Bridging Action: Transferring Manipulation Skills from Humans to Robots | Sijin Chen et al. | arXiv | 39 | |
| 2026-06-26 | 2606.27377 | DanceOPD: On-Policy Generative Field Distillation | Wei Zhou et al. | arXiv | 82 | |
| 2026-06-25 | 2606.25331 | Improved Large Language Diffusion Models | Shen Nie et al. | arXiv | 43 | |
| 2026-06-24 | 2606.24742 | World Value Models for Robotic Manipulation | Zhihao Wang et al. | arXiv | 8 | |
| 2026-06-10 | 2606.11042 | Workflow-GYM: Towards Long-Horizon Evaluation of Computer-use Agentic tasks in Real-World Professional Fields | Liya Zhu et al. | arXiv | 24 | |
| 2026-06-10 | 2606.10650 | Dynamic Linear Attention | Xin Wang et al. | arXiv | 9 | |
| 2026-06-01 | 2605.31604 | Representation Forcing for Bottleneck-Free Unified Multimodal Models | Yuqing Wang et al. | arXiv | 61 | |
| 2026-06-01 | 2605.31075 | Task-Focused Memorization for Multimodal Agents | Tao Zou et al. | arXiv | 39 | |
| 2026-05-27 | 2605.26895 | Negligible in Size, Significant in Effect: On Scale Vectors in Large Language Models | Mingze Wang et al. | arXiv | 17 | |
| 2026-05-14 | 2605.13831 | Training Long-Context Vision-Language Models Effectively with Generalization Beyond 128K Context | Zhaowei Wang et al. | arXiv | 89 | |
| 2026-05-08 | 2605.06548 | Continuous Latent Diffusion Language Model | Hongcan Guo et al. | arXiv | 89 | |
| 2026-05-06 | 2605.02134 | Video Generation with Predictive Latents | Yian Zhao et al. | arXiv | 22 | |
| 2026-05-04 | 2605.00503 | End-to-End Autoregressive Image Generation with 1D Semantic Tokenizer | Wenda Chu et al. | arXiv | 10 | |
| 2026-05-01 | 2604.27505 | Leveraging Verifier-Based Reinforcement Learning in Image Editing | Hanzhong Guo et al. | arXiv | 25 | |