search

h2oai / deepwater

Deep Learning in H2O using Native GPU Backends
Apache License 2.0
282 stars 93 forks source link

[TensorFlow]3rd/Javacpp-preset is too old #25

Closed DjangoPeng closed 7 years ago

DjangoPeng commented 7 years ago

The thirdparty/javacpp-preset reference a fork instead of the original project of javacpp-preset. What's the matter is that the TensorFlow version is too old(v0.10), compared to original javacpp-preset, which is updated to TensorFlow v1.0.0-alpha.

Is there any problem to update TensorFlow Version? I have a look at the TensorFlow Demos MNIST and Lenet I think it's a good idea to use Serializable and Deserialize for decoupling front-end from backend. So, I'd like to join in you to contribute and maintain the project. BTW, I'm an active contributor of TensorFlow and Kubernetes communities.

At a word, I plan to make a PR to update the TensorFlow version and the dependent branch of javacpp-preset.

arnocandel commented 7 years ago

Hi Django - thanks for your interest in contributing to the TensorFlow bindings of Deep Water! We're currently in the progress of finishing the integration work using the JNI bindings provided by TensorFlow without any need for JavaCPP, see the features/tensorflow-jni branch. Please let us know if you'd like to contribute there as well. Thanks, Arno

DjangoPeng commented 7 years ago

Yep, I'm willing to join you and work around that. Please let me know your progress and which part of it that I can contribute?

mdymczyk commented 7 years ago

Closing as we're not using javacpp presets anymore.