These are chat archives for dmlc/mxnet

24th
Feb 2018
Sam Hodge
@samhodge
Feb 24 2018 03:40
I think I do, I will change it to out_shap
out_shape is an empty tuple in a list
how can I use that?
Sam Hodge
@samhodge
Feb 24 2018 19:46
@anirudh2290
Anirudh Subramanian
@anirudh2290
Feb 24 2018 19:46
whatsup ?
Sam Hodge
@samhodge
Feb 24 2018 19:46
Do you think it is possible to port the network to resolution independent
In C++
Anirudh Subramanian
@anirudh2290
Feb 24 2018 19:48
i am not sure i understand. when you call the hybridize you basically get all the advantages of running a symbolic model in C++ backend
you are able to hybridize your model correct ?
Sam Hodge
@samhodge
Feb 24 2018 19:48
Well yes you can
but the model needs to know the resolution of the input data
before you can serialise it
let me show you what I mean
this commit kind of spells it out: samhodge/incubator-mxnet@1d72d60
Anirudh Subramanian
@anirudh2290
Feb 24 2018 19:52
looking
Sam Hodge
@samhodge
Feb 24 2018 19:53
I think I may know the solution to my problem
I need to save parts of the model and put the pieces together in C++
Anirudh Subramanian
@anirudh2290
Feb 24 2018 19:53
i know that for backend symbolic api one uses mx.sym.variable and then binds it to the model
Sam Hodge
@samhodge
Feb 24 2018 19:53
that way I can have a resolution independent model
Anirudh Subramanian
@anirudh2290
Feb 24 2018 19:53
i have to see how to do that in gluon
I can just reshape the appropriate symbole
Anirudh Subramanian
@anirudh2290
Feb 24 2018 19:57
yes, but you need something equivalent to mx.sym.variable in gluon correct ?
Sam Hodge
@samhodge
Feb 24 2018 19:58
I think so
it feels like I am trying to rough out an idea at this stage
Sam Hodge
@samhodge
Feb 24 2018 20:14
Actually dont worry about that, I think I need to continue on with a fixed resolution model first and simply try that out with the feedforward in C++ and then build it up from there, I am a little confused about if it is possible to build the gram matrix and the inspiration network in C++, but as a start it might be worth just loading up the model and params that I have for a decent trained model.
I just worry that I am delaying a decision
Anirudh Subramanian
@anirudh2290
Feb 24 2018 20:17
so the link that you pointed to https://github.com/samhodge/incubator-mxnet/blob/master/cpp-package/include/mxnet-cpp/symbol.hpp is just the cpp package, its just another frontend like python.
Sam Hodge
@samhodge
Feb 24 2018 20:18
Yup, but C++ is easier to deploy than Python
Anirudh Subramanian
@anirudh2290
Feb 24 2018 20:19
i am just wondering why you can not keep weight and height as symbol variables and bind it at runtime.
width*
Sam Hodge
@samhodge
Feb 24 2018 20:20
Sounds great
but I am not sure if I understand how to do that
how can you load on the params if the width and height are not known
so when you do x = mx.sym.var('data')
y = style_model(x)
Sam Hodge
@samhodge
Feb 24 2018 20:25
can you also do some sort of magic like a= mx.sym.var('width'), b=mx.sym.var('height') y=style_model(x,height=a,width=b)
is that what you mean?
Anirudh Subramanian
@anirudh2290
Feb 24 2018 20:25
yes
Sam Hodge
@samhodge
Feb 24 2018 20:25
I am honestly a bit of a noob
Anirudh Subramanian
@anirudh2290
Feb 24 2018 20:26
i am also a noob with gluon
Sam Hodge
@samhodge
Feb 24 2018 20:26
its all good you learn by making mistakes
let me try this out.
Anirudh Subramanian
@anirudh2290
Feb 24 2018 20:26
but with symbolic api you just use placeholders for data, and then bind it to the data at the end, i think this should be possible with gluon too
Anirudh Subramanian
@anirudh2290
Feb 24 2018 20:40
apache/incubator-mxnet#6087 you probably need something like this
Sam Hodge
@samhodge
Feb 24 2018 20:49
I am hoping I can work out how to get the width and height as symbols to the network