Supported Applications

Description

is the first flow-matching based protein folding model that uses only general purpose transformer layers, without relying on expensive modules like triangle attention or pair representation biases. The model scales to 3B parameters and is trained on over 8.6M distilled protein structures plus experimental PDB data, making it the largest scale folding model ever developed. SimpleFold achieves competitive performance on standard folding benchmarks and demonstrates strong ensemble prediction capabilities due to its generative training objective.

Usage

Installation