Fable's predicted protein contact map vs ground truth: partly correct

sokrypton · x · 2026-11-18

sokrypton continues his experiment: he simply asks the model to return a protein contact map, posting Fable's output against the ground-truth reference. Verdict: "it got some parts correct." Part of his series testing whether LLMs genuinely understand long-range protein contacts.

Related event: Researcher tests whether Claude Fable can predict protein contact maps from sequence alone(12 posts)→

Original post →

More from Models

Models channel →