I see lad, but it depends on tensorflow, which takes a long time to load (3.7 sec on my machine, this is not an acceptable delay in my case).
Does anybody have a solution not dependent on tensorflow (so that it is quick to run even for a small dataset to avoid the long load time of the dependent modules) yet still has a clean interface? (By clean interface, I mean something easy to call, not something in this post. L1 norm instead of L2 norm for cost function in regression model)
