keras-sharp
keras-sharp copied to clipboard
Still active?
If this project is no longer active, what is the reason?
Was there not enough community support? Did a new project supersede it?
Thank you
Anyone?
@vosmith @MCGPPeters @cesarsouza I think we can continue to do this project. The author stuck at the Gradient
of TensorFlow as this issue. The good news is I've made it work as this issue. Now we can move forward. The bad news is we have to change the backend from TensorFlowSharp
to TensorFlow.NET
Hey @Oceania2018 ! Thanks a lot for the update. Would you like to help contributing to this project? I have been a bit short on time for the past year as it could be seen. However, I am more than willing to have other people jumping the project so we could continue it together.
If this project is no longer active, what is the reason?
Was there not enough community support? Did a new project supersede it?
Thank you
Hi @vosmith,
Thanks a lot for the interest in the project. The fact is that when I started the project I immediately stumbled into multiple issues in TensorFlow C/C++ bindings, TensorFlowSharp, and CNTK that prevented me from going much further at the time. This, together with the fact that I had migrated from Keras to PyTorch some time ago, started a new job and moved between countries, did not help. However, I know that the project can still be useful to others.
If anyone would like to fork this project, please feel free - or even better, please let me know so I can grant commit access to those who think would like to contribute to the project directly.
Also, sorry for the delay in answering.
Regards, Cesar
@cesarsouza I'll PR if I made changes, we have a similar project like keras-sharp SiaNet.
@cesarsouza I'll PR if I made changes, we have a similar project like keras-sharp SiaNet.
Hi @Oceania2018, have your project obtained better success moving forward with a Keras implementation for dot-net? If yes please let me know and I will link your implementation in the main page of this project.
Cheers! Cesar
@cesarsouza We're working on the TensorFlow.Keras that only works for TensorFlow, I think we can make KearsSharp to introduce TensorFlow.Keras as one backend. What do you think about it?
Hi @Oceania2018,
I mean, if you've managed to achieve a project that works with TensorFlow, that's more than what this project has ever achieved. If you want, let me know and I can link your project in the main readme as a likely replacement!
Cheers, Cesar
I thought KerasSharp
should support multiple backends like KerasSharp.Backends.TensorFlowBackend
and KerasSharp.Backends.CNTKBackend
. Our TensorFlow.Keras
only support TensorFlow
ad backend. If people who are only seeking a Keras for TensorFlow, you can point it out in the main readme.