serser

``` } ^ /home/hadoop-basecv/.local/lib/python3.6/site-packages/torch/include/ATen/core/dispatch/Dispatcher.h: In member function ‘Return c10::Dispatcher::callUnboxedOnly(const c10::OperatorHandle&, Args ...) const [with Return = at::Tensor; Args = {const at::Tensor&, c10::ArrayRef, c10::ArrayRef, c10::ArrayRef, c10::ArrayRef}]’: /home/hadoop-basecv/.local/lib/python3.6/site-packages/torch/include/ATen/core/dispatch/Dispatcher.h:203:1: warning: control reaches end...

Updated to gcc7 and it worked. It seems gcc 4.8.5 doesn't go with it.

Hi, it will be useful to save the intermediate inputs to reproduce what has lead to nan results (a try catch will do). I've met a similar problem w/ ciou...

@liqier Are you referring to the supernet performance (train_search) vs. subnet performance (best cell)? Supernet is a tool for us to find the best cell (what we care), we shall...

Commenting out `SM_30` in the Makefile works.

but met https://github.com/NVIDIA/MinkowskiEngine/issues/177 on cuda11. :(

> Slice_79requires bool I/O but node can not be handled by Myelin.) In my case, torch boolean (i.e. typically an output of NOT node) is treated as INT64 in ONNX...

@kcyu2014 Thanks for the detailed reply. I deleted the second question by mistake but I don't know how to find it back. I see that once the topology is chosen...

Thanks. But my question still remains unanswered, and could you please give me the searching logs?

Thanks for the response. I've tested a couple of them, and it unfortunately seems to deteriorate the performance (-1.2% drop on val). Tweaks on bbox fusion even harm it further....