Files

T

Ilya Matveev 016893991a gradle-plugin: Use cinterop to compile stubs

The gradle plugin performed interop processing in two steps:
 1. generate stubs using cinterop tool
 2. compile the stubs generated using konanc.

This led to code duplication since the cinterop tool also
can do the same things. This patch removes two tasks used
different tools and replaces them with one task which uses
citerop tool for both stub generation and compilation

2017-09-13 12:45:38 +03:00

src/main

Fixed samples for the latest changes in Gradle plugin.

2017-06-08 17:21:04 +03:00

build.gradle

gradle-plugin: Use cinterop to compile stubs

2017-09-13 12:45:38 +03:00

build.sh

Minor build script tweaks

2017-06-22 11:03:50 +03:00

downloadTensorflow.sh

Add exec bit on external library download script

2017-06-19 12:47:06 +03:00

README.md

Fixed samples for the latest changes in Gradle plugin.

2017-06-08 17:21:04 +03:00

README.md

TensorFlow demo

Small Hello World calculation on the TensorFlow backend, arranging simple operations into a graph and running it on a session. Like other TensorFlow clients (e. g. for Python), this example is built on top of the TensorFlow C API, showing how a TensorFlow client in Kotlin/Native could look like.

Installation

./downloadTensorflow.sh

will install TensorFlow for C into $HOME/.konan/third-party/tensorflow (if not yet done).

To build use ../gradlew build or ./build.sh.

Then run

../gradlew run

Alternatively you can run artifact directly

./build/konan/bin/Tensorflow/Tensorflow.kexe

You may need to specify LD_LIBRARY_PATH or DYLD_LIBRARY_PATH to $HOME/.konan/third-party/tensorflow/lib if the TensorFlow dynamic library cannot be found.