H4dr1en

> @H4dr1en I haven't yet checked the code but I agree that this seems useless to wrap the model by DataParallel in case of 1 device. [...] I think this...

Thanks for clarifying @vfdev-5 ! Indeed I can observe that I am training on 4 GPUs with DataParallel by default. I understand the logic of idist.auto_model, my confusion comes from...

> By the way, I still do not quite understand what happens exactly if nproc_per_node > 1, what happens with the subprocess entering into __main__, config has nproc_per_node=4 by default,...

> Yes, it makes perfectly sense, but how ignite can know that you have nproc_per_node=1 ? Can we maybe check the world_size? > In addition, there can be (old) cases...

Bringing here a specific use case that this FR could solve: Having the `RunningAverage` metric being able to restore the state would allow to not "forget" previous iterations scores when...

In my case, `trace = mobile_optimizer.optimize_for_mobile(traced_model)` works but `coremltools.convert(traced_model)` fails with the same error describe above, `AttributeError: 'torch._C.Node' object has no attribute 'ival'` python 3.7, pytorch 1.11 and coremltools 6.0b1

> Actually we are working on accelerating pip install, in this issue you can see the full potential , and the initial PR. I'm hoping that after 21.1 is released...

This is _awesome_, thanks a lot @bmartinn and the team!! I am testing that right away 🤩

With `trains-agent==0.14.2rc2` it also fails: ``` Collecting Cython Using cached Cython-0.29.17-cp37-cp37m-manylinux1_x86_64.whl (2.1 MB) Installing collected packages: Cython Successfully installed Cython-0.29.17 Collecting torch==1.3.1+cpu File was already downloaded /home/H4dr1en/.trains/pip-download-cache/cu0/torch-1.3.1+cpu-cp37-cp37m-linux_x86_64.whl Successfully downloaded torch...

> Regrading using pypi with torch, the problem is, this is unstabe, for example there is no way of knowing whether the torchvision on pypi is the CPU or the...