The OpenFold consortium used Weights & Biases to track and reproduce AlphaFold 2 as an open, trainable model.

Reproducing DeepMind's AlphaFold 2 meant reverse-engineering sparse documentation and coordinating many experiments without shared tracking. The team used Weights & Biases for experiment tracking, run comparison and collaboration, including ablation studies. The page says it helped open-source a trainable AlphaFold implementation and inspired derivative projects; it gives no numeric results.

Weights & Biases is an MLOps platform for experiment tracking and model monitoring, used by biotech and hospital AI teams building protein and clinical models

This sits in the Build stage of the AI use-case lifecycle: Named examples from the foundation-model labs and platforms pharma teams use to actually construct, train and generate the model or application.