Skip to main content
. Author manuscript; available in PMC: 2024 Sep 1.
Published in final edited form as: J Comput Phys. 2023 May 9;488:112152. doi: 10.1016/j.jcp.2023.112152

Figure 10:

Figure 10:

The value of the loss as the training progresses for several replicates. We add data adaptively every 100 epochs and halt training when the loss goes below zero. Synchronized spikes in the loss function result from adding data where the loss is high.