Yannick Hold

Thanks @paffell for the trick. For those of you who don't want to modify PyTorch's source code, you can do the following in your code just before loading the model:...

There is currently no way in the API to do this explicitly. API proposals for this are welcome for discussion. To solve your problem, you could either [add a callback](http://scoop.readthedocs.io/en/latest/api.html#scoop._types.Future.add_done_callback)...

@anandtrex : This is already the case if you employ [`map_as_completed()`](http://scoop.readthedocs.org/en/0.7/api.html#scoop.futures.map_as_completed) function instead of the standard `map()`. Because the standard map ensures a perfect emulation of the serial built-in `map()`,...

Thanks for those, I'll review them this week and merge them as soon as possible.

Sorry you had this issue. I sadly don't have the time to fix it currently. I'll accept any pull request fixing memory accumulation problems.

Thanks for reaching out. I'm happy to see that SCOOP works for you at home. I have a guess about what's happening when you disable your internet connection: SCOOP queries...

This section has changed since the stable version. Does it happens with the development version (the master branch)? If so, please keep me posted.

Hello, The issue with constant deletion is not exactly the propagation assertion, but the synchronization between the tasks. If an API offered constant deletion, a user could want to perform...

I understand your use case, and I like your last example of deleteConst. It is indeed important not to allow the reset of a variable, otherwise non-aware users may simply...

Something could be done by centralizing the locks on the broker, I believe. I'm curious as if this should be offered directly in SCOOP or if users could rely on...