Hello,
I want to train the STL Geometry example from modulus using a A100 or H100 GPU, but I have a limited GPU budget and I am not sure how long it will take to run 1.2M iterations on this hardware. Are there any resources that point to time/cost of training these modulus models on different geometries?
Thank you!
One other thing - I saw in the example it says “The case using 10 M points… implies that the network is overfitting. Increasing the points to 20 M solves that problem and the flow field is generalized to a better resolution.”.
Did you increase the points sampled on the tessellated geometry? How did you do so?
This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.
