Fig. 5: Training curve for the pre-training of ProtBFN. | Nature Communications

Fig. 5: Training curve for the pre-training of ProtBFN.

From: Protein sequence modelling with Bayesian flow networks

Fig. 5: Training curve for the pre-training of ProtBFN.The alternative text for this image may have been generated using AI.

The loss is averaged over every 1000 consecutive steps to produce a smooth plot. The dashed line indicates the step at which the weighting of clusters is changed. Source data are provided as a Source Data file.

Back to article page