Open Jhsmit opened 10 months ago
it seems that fitting deltaG's through the web interface where dask fit the pytorch model is (much) worse then running directly without dask
https://pytorch.org/docs/stable/multiprocessing.html
it seems that fitting deltaG's through the web interface where dask fit the pytorch model is (much) worse then running directly without dask