AI & ML interests

foundation models for small molecule discovery

Recent Activity

devcayer  new activity about 11 hours ago
omtx/lula-1.1:Demo for this model on Spaces
devcayer  published a dataset about 11 hours ago
omtx/lula-data-public-train-test
devcayer  updated a dataset about 13 hours ago
omtx/lula-data-public-train-test
View all activity

Organization Card
Protein and ligand visualization
+ + +

omtx.ai

Decoding the language of small molecule discovery

Open-weight protein–ligand scoring for small molecule discovery.

LULA models score protein sequence plus ligand SMILES, then return binding scores and ranks. Run public weights locally, fine-tune on your own data, or use Om-hosted workflows to score, order, and validate molecules.

Sequence-only Local inference Fine-tuning Molecule ordering

Visit Om LULA-2 LULA-1.1 LULA-1 GitHub

01

Open-weight LULA

Download LULA-1, LULA-1.1, and LULA-2 for local scoring and fine-tuning, with runnable examples and checksum verification.

02

Hosted scoring

Use Om-hosted workflows when you want managed scoring, tracking, and scale behind one API.

03

Molecule ordering

Score candidates, select molecules, and order compounds through Om Wallet Credits when you are ready to test.

+

Release status

Open-weight LULA releases are live on Hugging Face.

Start with local scoring from public weights, fine-tune on your own labeled data, or use Om to score and order molecules through hosted workflows.

LULA-2 — live LULA-1.1 — live LULA-1 — live Local scoring — SDK Ordering — Om Wallet Credits