![]() Note that only NVIDIA devices are supported. To enable the training and inference on a GPU, please read this TensorFlow GPU Support pageĪnd install the CUDA framework to allow calculations on a GPU device. Running inference with ONNX models on Androidįor more inspiration, take a look at the code examples in this repository and Sample Android App.Running inference with ONNX models on JVM.We are working on including extensive documentation to help you get started.Īt this point, please feel free to check out the following tutorials we have prepared: You do not need prior experience with Deep Learning to use KotlinDL. Introduction to Deep Learning with KotlinDL (Zinoviev Alexey at Kotlin Budapest User Group 2021, slides). ![]() Deep Learning with KotlinDL (Zinoviev Alexey at Huawei Developer Group HDG UK 2021, slides).This table shows the mapping between KotlinDL, TensorFlow, ONNX Runtime, Compile SDK for Android and minimum supported Java versions. KotlinDL, ONNX Runtime, Android, and JDK versions Implementation ( ":kotlin-deeplearning-onnx: ")įor more details, please refer to the Quick Start Guide. To use KotlinDL in your project, ensure that mavenCentral is added to the repositories list: How to configure KotlinDL in your project Modules kotlin-deeplearning-tensorflow and kotlin-deeplearning-dataset are only available for desktop JVM, while other artifacts could also be used on Android. kotlin-deeplearning-dataset dataset classes.kotlin-deeplearning-visualization visualization utilities.kotlin-deeplearning-tensorflow learning and inference with TensorFlow.kotlin-deeplearning-onnx inference with ONNX Runtime.kotlin-deeplearning-impl implementation classes and utilities.kotlin-deeplearning-api api interfaces and classes.KotlinDL, ONNX Runtime, Android, and JDK versions.Working with KotlinDL in Jupyter Notebook.Working with KotlinDL in Android projects.How to configure KotlinDL in your project.Val accuracy = it.evaluate(dataset = test, batchSize = TEST_BATCH_SIZE).metrics ![]() It.fit(dataset = train, epochs = EPOCHS, batchSize = TRAINING_BATCH_SIZE) Optimizer = Adam(clipGradient = ClipGradientByValue( 0.1f)), Private const val EPOCHS = 3 private const val TRAINING_BATCH_SIZE = 1000 private const val NUM_CHANNELS = 1L private const val IMAGE_SIZE = 28L private const val SEED = 12L private const val TEST_BATCH_SIZE = 1000 private val lenet5Classic = Sequential.of( ![]()
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |