Chiyuan Zhang

Yes the GC problem is generally painful here. I am wondering if there is (or is going to be) any good ways in Julia to manage external resources.

`do ... end` is very limited and not easy to use in many cases. For this particular case, I think it goes to the `else` branch and re-use executor unless...

Oh, I see. Yes, it would be a good idea to add some function like `sync_params` to copy the parameters over. A yet better way is the _module_ system recently...

I'm not very sure if the R API already ported the `Module` interface. With a `Module`, you can run the [score](http://mxnet.io/api/python/module.html#mxnet.module.BaseModule.score) function to evaluate on a dataset.

Interesting. Batch size = 1 should work theoretically (with properly chosen hyperparameters), despite being inefficient. @vchuravy I'm leaving for a flight in 2 minutes, do you have a chance to...

@nowozin Unfortunately it is currently not possible yet. Because Julia GC is not thread safe currently, so that makes things a lot complicated than in the Python case. We are...

There is currently no rand number generator symbolic node. But a workaround is to treat those random numbers as input data and generate from a data provider. For example, the...

Another option which might be easier to implement (though involves quite a lot refactoring but not technically difficult) is the new module system in the mxnet python side. See dmlc/mxnet#1868...

The `_init_default` method should be a fallback that raise an error (I just added it). It seems strange that the label needs to be initialized. Can you post the network...

Both Mac & Linux building fails, though seem to be due to different reasons.