Alby M.

Same here. I just changed it not to link Bump on i386. Ugly, but it's the best I got. OSX 10.7 Xcode 4.6.3

@tewks: Any news on this? I like the underlying technology of the Bump API, but the API itself is driving me insane between its inability to compile for the simulator...

@4ndrey : No kidding. No updates, no support. Hope they still work at all...

@Jon889, @aravindkumar , @xtreme-jbyung : Nothing at all. I got to the point of starting to write my own server and library to do this, I will probably release it...

@Plasma : Interesting. I would have used that, but I'm so far into my own replacement after I couldn't get it to work that it's easier for me to finish...

Thanks. I'll try using that. On Wednesday, October 2, 2013, Andrey Toropchin wrote: > Hi folks, > > I have merged the following: > — i386 + armv6 libs from...

`model.eval()` doesn't do anything except set a boolean flag on `model`, so whether you do it inside or outside the context manager shouldn't matter. (See https://pytorch.org/docs/stable/_modules/torch/nn/modules/module.html#Module.eval, `def train(`.) Re distributed,...

Hi @Devoe-97, I think you'd want to implement a Lightning [`Callback`](https://pytorch-lightning.readthedocs.io/en/latest/api/pytorch_lightning.callbacks.Callback.html) that runs the appropriate `ema.update()`, `ema.store(); ema.copy_to()`, and `ema.restore()` operations at the right points in the training loop. Which...

I don't know if it helps, but you can see `pydpc` being used in some of my code here: https://github.com/Linux-cpp-lisp/sitator/blob/master/sitator/site_descriptors/SiteTypeAnalysis.py#L96

Do you get the same issue if you try with, for example, only 10k of your data points? If you try with the last 10k or 1k or whatever or...