But I started to feel the poor habits the point and click interface teaches. And it does not scale for more complicated ML designs. So now I just use pytorch directly with vim.
@jmw150 Interesting. One of the things I do in dayjob is deploy JupyterHub in compute environments that are generally beefier than normal lab equipment.
@jmw150 Interesting. One of the things I do in dayjob is deploy JupyterHub in compute environments that are generally beefier than normal lab equipment.