tk26eng

@patrick-su Yes. That's right. But maybe simply moving document is not enough because `debug` directory will be empty after moving `keep_me.md`. Empty directory (`debug` directory) will be removed automatically by...

@iizukak @hadusam Maybe there's no method to recover it after FPGA gets stucked. But avoiding it is more important than recovering it. We might need to implement the way to...

@iizukak I discussed with @primenumber about this directly. This feature to change tolerance is acceptable for him. But an warning should be shown when the user changes the tolerance because...

@lm-jira The problem might come from lack of amount of memory. If you use -j2 option, two compilations are running at the same time. The problem might not occur when...

@hadusam I tried to add the document about `CUDA_VISIBLE_DEVICES=-1`. But now I am thinking that it might be better to set CUDA_VISIBLE_DEVICES=-1 as default on Docker. Then the convert command...

FYI, `PerImageStandardization` might be also necessary for MNIST dataset. About MNIST dataset: When I used `DivideBy255`, an accuracy for test data was very unstable. But I used `PerImageStandardization`, the result...

@patrick-su This might come from lack of cleaning cache. Please let me check it again after cleaning cache for docker.

@kalpitthakkar-lm > @tk26eng @primenumber > > Just a related question: > Is there a possibility of using multiple FPGAs for inference? > (Basically if the model is large, two FPGAs...

@primenumber How about using threshold like this ? ``` float val_gt; // ground truth float val_calc; // calculated value // Certain value is set to val_gt and val_calc here abs_tol...

Is compiling openmpi really required ? It takes most part of whole time for building. If it's possible to remove it, it's better.