David Warde-Farley
David Warde-Farley
This should now be addressable.
Once we are fully over to the new conv interface we can maybe remove the "hackability" of the Convolutional class (the static methods) as it really won't be necessary anymore.
@bartvm Is this still relevant?
Does blocks-continue work? On Thu, Dec 17, 2015 at 7:05 PM, Dzmitry Bahdanau [email protected] wrote: > When I unpickle an iterator iter over an out-of-memory H5PyDataset which > was pickled...
@orhanf thoughts on this?
Looks like this is caused by being a bit too clever and handling "before_first_epoch" as `before_epoch` with a predicate. At the very least this needs to be documented. I'm also...
@rizar @dmitriy-serdyuk Any thoughts on this issue? if my fix sounds reasonable I'll go ahead and implement it.
Sorry for letting this drop. @rizar `before_training` is not what I need. I need a callback that fires _after_ setup has been completed as it operates on `StepRule` buffers, which...
Is this still needed @rizar?
I like it a lot. Current syntax gets quite hairy. Is it going to be easy to do this without breaking literally everyone's code? I imagine we could make the...