Chewxy

Took a stab at this. Turns out to be slightly more difficult than expected. Standard solutions to this uses LAPACK's [gesv](https://software.intel.com/en-us/node/520973) implementation to solve (and rather easily too - it's...

Alternatives available: 1. Convert any Tensor to `*tensorf64.Tensor`, and then use gonum/lapack 2. Convert any Tensor to `*gonum.mat64`, which has a Solve method, then convert it back 3. Write a...

it is still open

Hi... wanna shoot me an email @manjunathsc24 ? chewxy [at] gmail.com

Hah. This is not supposed to be the case - the `convnet` example was built for benchmarking against other systems. This means that some changes made so far have degraded...

Working on several CUDA related issues today, will have a quick look on this to figure out where it's gone wrong.

I can confirm that currently Gorgonia is slower than Keras. I also found a few bugs in the `losses`, so fixing those first. On CUDA, the speeds are equivalent. Will...

Hi, a bunch of fixes just went in.. will be benchmarking this in a bit

v0.9.10 if anyone is following along

Hmm, ok looking into it