plus more about Hyperloop Transformer, Qwen-3.5 Omni, and Scaling Self-Play with Self-Guidance
plus more about Looped Transformers, Nexus, RNN with Memory, and more
plus more about In-Place TTT, TriAttention, and Interleaved Head Attention.
plus more on Path-Constrained MoE, HISA, and Screening is not enough
plus more on Claudini, Composer 2, and self-distillation
plus more about V-JEPA 2.1, Mamba 3, and latent planning
and more about GLM-OCR, pre-pre-training on NCA, IndexCache, and neural thickets
and more about Speculative Speculative Decoding, SWE-CI, and Beyond Language Modeling
plus more on Learning Without Training and The Geometry of Noise
plus more about Experiential RL, GLM-5 Report, and Attention Matching
plus more on Evolving Agents via Recursive Skill-Augmented RL and Low Hanging Fruits in Vision Transformers
an insane big week in AI reseasrch
