Chiyuan Zhang
Chiyuan Zhang
@Evizero You are very welcome to run a profiling and locate the bottleneck in the computation! :) My guess is that the optimizer is not a bottleneck, and even if...
I think it should be possible with basic operations like matrix multiplication, plus, and exp.
Hi @kasiabozek can you provide more details (e.g. a short snippet that we can run to reproduce this error)?
@antinucleon Looking at the code for upsampling, I think the doc [here](https://github.com/dmlc/mxnet/blob/master/src/operator/upsampling-inl.h#L42) should be updated to say that `num_filters` is only used by bilinear upsampling, instead of nearest neighbor upsampling....
GC is really unpredictable, I guess the generational GC is retaining some of the objects because they are still young? Maybe you can try to explicitly call the destructor like...
Probably cannot call it directly. How about calling `finalize(exec.handle)`?
For the "Neural networks" section: I'm recently writing a neural network package for julia: [Mocha](https://github.com/pluskid/Mocha.jl) (this name because it is deeply inspired by the very popular C++ deep learning framework...
@lindahua Thanks! Mocha looks like Caffe (general architecture, layers, blobs, solvers, etc.) but they are not completely the same. For example, Mocha uses HDF5/JLD to store snapshot of model, while...
After looking at the examples more closely, it seems `rewrite` already have per-function control, which looks very cool.
@MichaelHatherly Thanks! That would be really great! :+1: