Looped Transformers aren't magic, just a deeper network, researcher argues
savvyRL · x · 2026-09-03
Researcher savvyRL pushes back on the Looped Transformers hype: it's not magical, doesn't add real recurrence to an inherently feedforward network, and the same idea was already explored by Dehghani et al. in 2019. The effect is simply making the model deeper, equivalent to stacking more layers from the start.
Related event: Looped Transformers Are Just Weight-Tied Deeper Networks, Researchers Argue(4 posts)→
More from Research
- Counterfactual Debugging: causal attribution over 1M steps to localize sim2real gaps in world-model agents — MichaelD1729 · 2026-09-03
- DeepMind's 83-Page Study: Autonomous Research Agents Fabricate 90% of Findings — williamtp · 2026-09-03
- How Google's RT-2 triggered the robotics boom: Understanding AI explains VLA models — binarybits · 2026-09-03
- Goodfire chief scientist Tom McGrath on interpretability: SAEs may fracture what networks really learn — Machine Learning Street Talk · 2026-09-03
- CBAI opens Fall AI Safety fellowship: $15k stipend, 10 weeks in Boston — benno_krojer · 2026-09-03
- What 12 million empirical research results can teach us — RexDouglass · 2026-09-03