Antoine Hue
Antoine Hue
xinput_calibrator could have a "raw" output mode for each output type in which only needed code is outputed and can be redirected to a file. Example: xinput_calibrator --output-type xorg.conf.d >...
Have you tried the code patch attached to this thread as a pull request ? Antoine
Hello Tias, Your bracketed comment is hard and unjustified. I have removed dirty things like #include "xxx.cxx", I have cleaned up includes and headers, I have created functions to handle...
Hello Tias, Thanks for this detailed answer. See below my comments. On 04/21/2011 11:08 PM, tias wrote: > Hello Antoine, > > I'm sorry for the short comment, I did...
Cases depending on layers (before -> after): - Conv->Conv: nothing to do, both go over the input space, sampling is changing only the dimensions - Conv->Flatten->Dense: already handled with pooling...
Main issue is the when and where to compute the activation map since it is used in two locations (two bottom right quadrants) and it requires quite large computation such...
Similar issues as for UpSampling2D (#47)
There are a few situation in which the reshape does not make much sense: - Reshape followed by Flatten (or Reshape) (e.g.: between the generator and the discriminator of a...
Reference doc : - https://conda-forge.org/docs/maintainer/adding_pkgs.html#the-recipe-meta-yaml - https://github.com/conda-forge/staged-recipes
May require support for the recurrent layers (LSTM, GRU)