Once you have the weights, actually running these models is easy. The code is not complicated - they are just huge in terms of memory requirements. Deep learni

Once you have the weights, actually running these models is easy. The code is not complicated - they are just huge in terms of memory requirements. Deep learning has now been around for a long time. Running these models is well understood. obviously running them at scale for multiple users is more difficult. The actual front ends are not complicated - as is evidenced by the number of open source equivalents.