AIToday
Large Language ModelsAI Safety & AlignmentLessWrong AIPublished: Sep 11, 2026, 06:00 JST1 min read

Simplex team maps telescoping cones in transformer activations

Simplex team maps telescoping cones in transformer activations

Kyle J. Ray, Paul M. Riechers, and Adam S. Shai of Simplex and the Astera Institute recovered "telescoping cones" from a transformer's residual-stream activations using linear regression, and report that each component grows and shrinks in accordance with in-context evidence.

Ask the AI about this article →

Summaries like this, in your inbox every morning.

Get the latest Large Language Models news every morning

For example, today's edition would include:

  • Dynatrace acquires Arize AI as observability shifts to actionSiliconANGLE AI · 5h ago
  • Shared base cuts 100 fine-tunes from 1.5 TB to 19.3 GBDaily Dose of Data Science · 5h ago
  • OpenAI agents hit RubyGems, undisclosed since May 12thSimon Willison's Weblog · 5h ago

AI-summarized, only the topics you pick — one digest a day via Email, Slack, or Discord.

Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →

Ask AI

Ask AI anything about this article. Q&As are published on this page for other readers too.

Related Articles

Next articlenocall raises 約8億円 to link AI calls to other systems