Fix minor code warnings

Author: juliabeliaeva

api/src/main/kotlin/org/jetbrains/kotlinx/dl/api/core/GraphTrainableModel.kt

@@ -412,7 +412,7 @@ public abstract class GraphTrainableModel(vararg layers: Layer) : TrainableModel
                     // TODO: probably I should it by name, not by type
                     val validationLossValue = evaluationResult.lossValue
                     epochTrainingEvent.valLossValue = validationLossValue
-                    epochTrainingEvent.valMetricValues = validationMetricValues!!
+                    epochTrainingEvent.valMetricValues = validationMetricValues
                     logger.info { "epochs: $i loss: $avgLossValue metric: ${avgTrainingMetricValue.contentToString()} val loss: $validationLossValue val metrics: $validationMetricValues" } // TODO: check printing for validation
                 } else {
                     logger.info { "epochs: $i loss: $avgLossValue metric: ${avgTrainingMetricValue.contentToString()}" }

api/src/main/kotlin/org/jetbrains/kotlinx/dl/api/core/Sequential.kt

@@ -112,9 +112,7 @@ public class Sequential(vararg layers: Layer) : GraphTrainableModel(*layers) {
                         "It is generated during Sequential model saving with SavingFormat.JSON_CONFIG_CUSTOM_VARIABLES."
             )
 
-            return org.jetbrains.kotlinx.dl.api.inference.keras.loadSequentialModelConfiguration(
-                configuration, inputShape
-            )
+            return loadSequentialModelConfiguration(configuration, inputShape)
         }
 
         /**

api/src/main/kotlin/org/jetbrains/kotlinx/dl/api/core/TrainableModel.kt

@@ -1,5 +1,5 @@
 /*
- * Copyright 2020 JetBrains s.r.o. and Kotlin Deep Learning project contributors. All Rights Reserved.
+ * Copyright 2020-2022 JetBrains s.r.o. and Kotlin Deep Learning project contributors. All Rights Reserved.
  * Use of this source code is governed by the Apache 2.0 license that can be found in the LICENSE.txt file.
  */
 
@@ -476,10 +476,6 @@ public abstract class TrainableModel : TensorFlowInferenceModel() {
         )
     }
 
-    public override fun close() {
-        super.close()
-    }
-
     /**
      * Returns model summary.
      *

api/src/main/kotlin/org/jetbrains/kotlinx/dl/api/core/callback/EarlyStopping.kt

@@ -1,5 +1,5 @@
 /*
- * Copyright 2020 JetBrains s.r.o. and Kotlin Deep Learning project contributors. All Rights Reserved.
+ * Copyright 2020-2022 JetBrains s.r.o. and Kotlin Deep Learning project contributors. All Rights Reserved.
  * Use of this source code is governed by the Apache 2.0 license that can be found in the LICENSE.txt file.
  */
 
@@ -125,9 +125,7 @@ public class EarlyStopping(
         if ((monitorOp ?: return).apply(current.toDouble() - minDelta, best)) {
             best = current.toDouble()
             wait = 0
-            if (restoreBestWeights) {
-                // TODO this.bestWeights = this.model.getWeights();
-            }
+            // TODO if (restoreBestWeights) this.bestWeights = this.model.getWeights();
         } else {
             wait++
             if (wait > patience) {

api/src/main/kotlin/org/jetbrains/kotlinx/dl/api/core/layer/merge/Dot.kt

@@ -33,8 +33,8 @@ public class Dot(
         require(input.size == 2) { "A `Dot` layer should be called on exactly 2 input. 'Received: input=${input}" }
         var x1 = input[0]
         var x2 = input[1]
-        val axes: IntArray = IntArray(2)
-        val scope: Scope = tf.scope()
+        val axes = IntArray(2)
+        val scope = tf.scope()
         for (i in 0 until 2) {
             if (axis[i] < 0) {
                 axes[i] = axis[i] % input[i].asOutput().shape().numDimensions()
@@ -89,7 +89,7 @@ public fun batchDot(scope: Scope?, x: Operand<Float>, y: Operand<Float>, axis: I
     val xDim = x.asOutput().shape().numDimensions()
     val yDim = y.asOutput().shape().numDimensions()
     val diff: Int
-    var x2: Operand<Float> = x;
+    var x2: Operand<Float> = x
     var y2: Operand<Float> = y
     if (xDim > yDim) {
         diff = xDim - yDim
@@ -120,10 +120,10 @@ public fun batchDot(scope: Scope?, x: Operand<Float>, y: Operand<Float>, axis: I
     val x2Dim = x2.asOutput().shape().numDimensions()
     val y2Dim = y2.asOutput().shape().numDimensions()
     if (x2Dim == 2 && y2Dim == 2) {
-        if (axis[0] == axis[1]) {
-            out = ReduceSum.create(scope, Mul.create(scope, x2, y2), Constant.create(scope, axis[0]))
+        out = if (axis[0] == axis[1]) {
+            ReduceSum.create(scope, Mul.create(scope, x2, y2), Constant.create(scope, axis[0]))
         } else {
-            out = ReduceSum.create(
+            ReduceSum.create(
                 scope,
                 Mul.create(scope, Transpose.create(scope, x2, Constant.create(scope, intArrayOf(1, 0))), y2),
                 Constant.create(scope, axis[1])
@@ -142,10 +142,10 @@ public fun batchDot(scope: Scope?, x: Operand<Float>, y: Operand<Float>, axis: I
     }
     val idx: Float
     if (diff != 0) {
-        if (xDim > yDim) {
-            idx = (xDim + yDim - 3).toFloat()
+        idx = if (xDim > yDim) {
+            (xDim + yDim - 3).toFloat()
         } else {
-            idx = (xDim - 1).toFloat()
+            (xDim - 1).toFloat()
         }
         out = Squeeze.create(scope, Constant.create(scope, FloatArray(diff) { it + idx }))
     }

api/src/main/kotlin/org/jetbrains/kotlinx/dl/dataset/embeddedDatasets.kt

@@ -189,7 +189,7 @@ public fun freeSpokenDigits(
     val dataset = File(path)
         .listFiles()?.flatMap(::extractWavFileSamples)
         ?: throw IllegalStateException("Cannot find Free Spoken Digits Dataset files in $path")
-    val maxDataSize = dataset.map { it.first.size }.maxOrNull()
+    val maxDataSize = dataset.maxOfOrNull { it.first.size }
         ?: throw IllegalStateException("Empty Free Spoken Digits Dataset")
     check(maxDataSize <= FSDD_SOUND_DATA_SIZE) {
         "Sound data should be limited to $FSDD_SOUND_DATA_SIZE values but has $maxDataSize"

dataset/src/main/kotlin/org/jetbrains/kotlinx/dl/dataset/OnHeapDataset.kt

@@ -242,7 +242,7 @@ public class OnHeapDataset internal constructor(public val x: Array<FloatArray>,
                     return FloatArray(xFiles.size) { Float.NaN }
                 }
                 else -> {
-                    throw UnsupportedOperationException("Unknown label generator: ${labelGenerator.toString()}") // TODO: labelGenerator.apply(...) will be better solution here.
+                    throw UnsupportedOperationException("Unknown label generator: ${labelGenerator}") // TODO: labelGenerator.apply(...) will be better solution here.
                 }
             }
         }

docs/quick_start_guide.md

@@ -7,7 +7,7 @@ first you'll need to set up the appropriate dependencies.
 1. Open a Kotlin project where you want to use Kotlin DL, or create a new Kotlin project in IntelliJ IDEA as described in the [Kotlin documentation](https://kotlinlang.org/docs/tutorials/jvm-get-started.html).
 2. Add the Kotlin DL dependency to your project's build file. 
     * If you're using Gradle as the build system, add the following to the `build.gradle` file:
-    ```kotlin
+    ```groovy
    repositories {
        mavenCentral()
    }

examples/src/main/kotlin/examples/cnn/mnist/advanced/LeNetWithAlternativeLossFunction.kt

@@ -48,24 +48,24 @@ fun lenetWithAlternativeLossFunction() {
 
     val (newTrain, validation) = train.split(0.95)
 
-    lenet5Classic.use {
-        it.compile(
+    lenet5Classic.use { model ->
+        model.compile(
             optimizer = Adam(),
             loss = Losses.HUBER,
             metric = Metrics.ACCURACY
         )
 
-        it.logSummary()
+        model.logSummary()
 
-        val history = it.fit(
+        val history = model.fit(
             trainingDataset = newTrain,
             validationDataset = validation,
             epochs = EPOCHS,
             trainBatchSize = TRAINING_BATCH_SIZE,
             validationBatchSize = TEST_BATCH_SIZE
         )
 
-        val accuracy = it.evaluate(dataset = test, batchSize = TEST_BATCH_SIZE).metrics[Metrics.ACCURACY]
+        val accuracy = model.evaluate(dataset = test, batchSize = TEST_BATCH_SIZE).metrics[Metrics.ACCURACY]
 
         println("Accuracy: $accuracy")
 

examples/src/main/kotlin/examples/transferlearning/modelhub/mobilenet/additionalTrainingForMobileNet.kt

@@ -49,7 +49,7 @@ private const val TRAIN_TEST_SPLIT_RATIO = 0.7
  */
 fun mobilenetWithAdditionalTraining() {
     val modelHub = TFModelHub(cacheDirectory = File("cache/pretrainedModels"))
-    var modelType = TFModels.CV.MobileNet()
+    val modelType = TFModels.CV.MobileNet()
     val model = modelHub.loadModel(modelType)
 
     val preprocessing: Preprocessing = preprocess {