Merge pull request #488 from JohnSnowLabs/context-spell-tf-access

Fixed concurrent access to TF in spell checker

src/main/scala/com/johnsnowlabs/ml/tensorflow/TensorflowSpell.scala

@@ -18,32 +18,29 @@ class TensorflowSpell(
   val lossKey = "Add:0"
   val dropoutRate = "dropout_rate"
 
-  /* returns the loss associated with the last word, given previous history  */
-  def predict(dataset: Array[Array[Int]], cids: Array[Array[Int]], cwids:Array[Array[Int]]) = this.synchronized {
+  // these are the inputs to the graph
+  val wordIds = "batches:0"
+  val contextIds = "batches:1"
+  val contextWordIds = "batches:2"
 
-    val packed = dataset.zip(cids).zip(cwids).map {
-      case ((_ids, _cids), _cwids) => Array(_ids, _cids, _cwids)
-    }
+  /* returns the loss associated with the last word, given previous history  */
+  def predict(dataset: Array[Array[Int]], cids: Array[Array[Int]], cwids:Array[Array[Int]]) = {
 
-    val tensors = new TensorResources()
-    val inputTensor = tensors.createTensor(packed)
-
-    tensorflow.session.runner
-      .feed(inMemoryInput, inputTensor)
-      .addTarget(testInitOp)
-      .run()
+    val tensors = new TensorResources
 
     val lossWords = tensorflow.session.runner
       .feed(dropoutRate, tensors.createTensor(1.0f))
+      .feed(wordIds, tensors.createTensor(dataset.map(_.dropRight(1))))
+      .feed(contextIds, tensors.createTensor(cids.map(_.tail)))
+      .feed(contextWordIds, tensors.createTensor(cwids.map(_.tail)))
       .fetch(lossKey)
       .fetch(validWords)
       .run()
 
     tensors.clearTensors()
 
     val result = extractFloats(lossWords.get(0))
-    val width = inputTensor.shape()(2)
-    result.grouped(width.toInt - 1).map(_.last)
-
+    val width = dataset.head.length
+    result.grouped(width - 1).map(_.last)
   }
 }

src/test/scala/com/johnsnowlabs/nlp/annotators/spell/context/ContextSpellCheckerTestSpec.scala

@@ -11,6 +11,8 @@ import org.apache.hadoop.fs.FileSystem
 import org.apache.spark.ml.Pipeline
 import org.apache.spark.sql.SparkSession
 import org.scalatest._
+import SparkAccessor.spark
+import spark.implicits._
 
 
 class ContextSpellCheckerTestSpec extends FlatSpec {
@@ -99,9 +101,6 @@ class ContextSpellCheckerTestSpec extends FlatSpec {
 
 
   "a Spell Checker" should "work in a pipeline with Tokenizer" in {
-    import SparkAccessor.spark
-    import spark.implicits._
-
     val data = Seq("It was a cold , dreary day and the country was white with smow .",
       "He wos re1uctant to clange .",
       "he is gane .").toDF("text")
@@ -130,8 +129,6 @@ class ContextSpellCheckerTestSpec extends FlatSpec {
 
   }
 
-
-
   "a Spell Checker" should "work in a light pipeline" in {
     import SparkAccessor.spark
     import spark.implicits._
@@ -155,10 +152,7 @@ class ContextSpellCheckerTestSpec extends FlatSpec {
 
     val pipeline = new Pipeline().setStages(Array(documentAssembler, tokenizer, spellChecker)).fit(Seq.empty[String].toDF("text"))
     val lp = new LightPipeline(pipeline)
-    lp.annotate(data)
-    lp.annotate(data)
-    lp.annotate(data)
-
+    lp.annotate(data ++ data ++ data)
   }