Can it boost model training on CPU (currently using pytorch lightning)
For example, NeuralProphet uses Pytorch Lightning for model fitting. I'm doing grid search to find the best possible hyper-parameters, in parrallel using Dask on 16 CPU threads. I have a single 2080 GPU but it can't be utilized in parallel so there isn't much speedup.
So I was wondering, would it be possible to speed up the grid search process if, somehow we could bring in Max / mojo to the workload?
#model-fitting #NeuralProphet #pytorch-lightning #cpu
0 Replies