Keyword

looped transformer

Appears in 4 articles·Latest Nov 3, 2026

Nov 3, 2026

What Is a Looped Transformer? NVIDIA Dory Explained

A looped transformer reruns a shared block of layers per token to gain depth without more weights. Here is how NVIDIA's Dory does it, and what it costs.

looped transformerwhat is a looped transformerNVIDIA Dory

Diagram-like illustration of a transformer layer stack where a signal ribbon loops back through the same group of layers a second time, showing layer reuse for added depth without more stored parameters.

Nov 3, 2026

How Does SenseNova-Loop Get More Depth Without More Parameters?

SenseNova-Loop-9B-Preview reuses 24 layers twice for 56 layer executions from 32 stored layers, with two-stream mHC and 3:1 linear-to-full attention.

looped transformerSenseNova-Loopmanifold-constrained hyper-connections

Diagram-like illustration of a 24-layer transformer stack with a two-strand path looping through it twice, alongside a smaller stored-weights stack and a long context strip.

Nov 3, 2026

How Does SenseNova-Loop-9B-Preview's Looped Transformer Work?

SenseNova-Loop-9B-Preview reuses 24 layers twice with two-stream mHC and 3:1 hybrid attention: ~9B stored parameters, ~14.6B at inference, 256K context.

SenseNova-Loop-9B-Previewlooped transformermHC hyper-connections

Abstract illustration of a recurrent AI architecture shown as a single looping band of identical segments passing repeatedly through one processing gate instead of a tall stack of distinct layers.

Sep 7, 2026

What Is Recurrent Architecture AI? GPT-6 Astra's Looping Explained

Recurrent depth loops shared transformer layers instead of stacking new ones. Here's how it powers GPT-6 Astra and why it raises governance concerns.

recurrent architecture AIrecurrent depthGPT-6 Astra