LDMv0.1

DISCOVERY DATA

Evidence is what closes the loop.

LDM learns from complete search trajectories: what was proposed, what was evaluated, what changed, and why the next move became worth trying.

01

Programs

Candidate training programs paired with fixed-time validation feedback.

02

Proteins

CDRH3 sequence proposals paired with structural-simulator binding scores.

03

Molecules

SMILES proposals paired with docking and predicted-activity objectives.