Adam Pocock

Did you try to port the embedding ops from https://github.com/eaplatanios/tensorflow_scala over to sit on top of TF-Java? It uses it's own JNI binding to the TF C API and so...

At the moment we don't support it, but we'd welcome contributions to add that support. It would need to build the ndarray support and add the types to [this package](https://github.com/tensorflow/java/tree/master/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types)....

Isn't the C# code emitting channels last and the Python code using channels first?

I can add a Java session constructor which accepts a `ByteBuffer`, then users could memory map the file on disk with (https://docs.oracle.com/javase/8/docs/api/java/nio/channels/FileChannel.html#map-java.nio.channels.FileChannel.MapMode-long-long-), pass through the ByteBuffer to JNI where I...

@yuslepukhin please could you review this after the 1.18 release? And #18998 which is also a small straightforward improvement.

I'm the maintainer of the Java layer, I'll have a look at what's going on.

Switch [tensorflow-core-platform-gpu](https://search.maven.org/artifact/org.tensorflow/tensorflow-core-platform-gpu/0.4.1/jar) for your current OS specific platform jar on your classpath and it will default to running on the GPU.

The TF runtime prefers to place computations on the GPU, so if one is available it will use it in preference to the CPU.

No. TF doesn't support that. There is some minimal support for Intel's integrated GPUs via Microsoft's DirectML on Windows in TF, but I wouldn't expect that to work particularly well....

I think this is because you've set all the weights to zero, so when it goes through the LSTM equation then everything is multiplied by a zero vector. Try setting...