Models, methods, and the machinery of intelligence.
From the archiveYou are reading the September 29, 2026 edition. Read the latest
arXiv
Verdict: NotableReinforcement learning can turn one language model into several specialists, each excellent at a single skill such as mathematics, coding or following instructions, but users need one model with all of these skills.
Symbolic models make melody, harmony, rhythm, and form explicit but typically stop before a finished recording; audio models produce complete songs while leaving composition implicit.
Vision-language-action (VLA) and world-action (WAM) models map observations and instructions directly to robot actions.
Looped transformers have demonstrated promising parameter efficiency by reusing layers for latent computation.
Scaling test-time computation is a powerful way to improve language-model reasoning, and is particularly appealing for small reasoning models (sRMs) that are cheap to serve.
FullAttn often assigns negligible normalized mass to much of the causal score space, yet dense kernels execute the complete post-score path after forming each QK tile.
An agent can complete a task while exhibiting undesirable behavior during execution.
FullAttn repeatedly exposes the complete causal history to every attention head, creating substantial redundant computation and memory traffic even with IO-efficient dense kernels.
Interactive world models require responding in real time to versatile controls and maintaining long-horizon consistency.