Meta AI

Multimodal protein representation model that iteratively fuses a sequence language model with a 3D structure encoder through a shared learnable token.

Surface-EMG wristband models that decode hand gestures, handwriting, and wrist movement, generalizing across users without per-person calibration.

Latent diffusion model generating 3D drug-like molecules and inorganic crystals from one shared all-atom autoencoder and Transformer denoiser.

Brain-to-text decoder that reconstructs typed sentences from non-invasive MEG and EEG brain recordings using a CNN, transformer, and language model.

Breast cancer recurrence risk read from an H&E slide and six routine clinical variables, using a frozen pan-cancer pathology encoder.

Scientific large language model trained on 48 million papers, textbooks, and reference works to store, combine, and reason about scientific knowledge.

Protein structure prediction from a single sequence, with no multiple sequence alignment. Folds a 384-residue protein in 14.2 seconds on one GPU.

Protein language model family from 8M to 15B parameters, used as a frozen sequence encoder whose representations encode atomic-level structure.

Inverse folding model predicting protein sequence from backbone coordinates, trained on 12 million AlphaFold2-predicted structures.

Protein language model for zero-shot variant effect prediction, scoring mutations by log-odds from evolutionary sequences with no MSA or assay data.

Transformer protein language model trained on 250 million protein sequences that learns structural and functional representations without supervision.