Using SubsetVertex followed by an EmbeddingLayer throws an exception

[RoiViber]

Name: org.apache.spark.SparkException
Message: Job aborted due to stage failure: Task 461 in stage 14.0 failed 4 times, most recent failure: Lost task 461.3 in stage 14.0 (TID 18314, ip-10-166-9-28.ec2.internal): java.lang.IllegalStateException: Cannot do backward pass: error not set
at org.deeplearning4j.nn.graph.vertex.impl.SubsetVertex.doBackward(SubsetVertex.java:93)
at org.deeplearning4j.nn.graph.ComputationGraph.calcBackpropGradients(ComputationGraph.java:1203)
at org.deeplearning4j.nn.graph.ComputationGraph.computeGradientAndScore(ComputationGraph.java:969)
at org.deeplearning4j.optimize.solvers.BaseOptimizer.gradientAndScore(BaseOptimizer.java:151)
at org.deeplearning4j.optimize.solvers.StochasticGradientDescent.optimize(StochasticGradientDescent.java:54)
at org.deeplearning4j.optimize.Solver.optimize(Solver.java:51)
at org.deeplearning4j.nn.graph.ComputationGraph.fit(ComputationGraph.java:833)
at org.deeplearning4j.nn.graph.ComputationGraph.fit(ComputationGraph.java:740)
at org.deeplearning4j.spark.impl.paramavg.ParameterAveragingTrainingWorker.processMinibatch(ParameterAveragingTrainingWorker.java:205)
at org.deeplearning4j.spark.impl.paramavg.ParameterAveragingTrainingWorker.processMinibatch(ParameterAveragingTrainingWorker.java:43)
at org.deeplearning4j.spark.api.worker.ExecuteWorkerMultiDataSetFlatMap.call(ExecuteWorkerMultiDataSetFlatMap.java:83)
at org.deeplearning4j.spark.api.worker.ExecuteWorkerMultiDataSetFlatMap.call(ExecuteWorkerMultiDataSetFlatMap.java:26)
at org.apache.spark.api.java.JavaRDDLike$$anonfun$fn$4$1.apply(JavaRDDLike.scala:152)
at org.apache.spark.api.java.JavaRDDLike$$anonfun$fn$4$1.apply(JavaRDDLike.scala:152)
at org.apache.spark.rdd.RDD$$anonfun$mapPartitions$1$$anonfun$apply$23.apply(RDD.scala:785)
at org.apache.spark.rdd.RDD$$anonfun$mapPartitions$1$$anonfun$apply$23.apply(RDD.scala:785)
at org.apache.spark.rdd.MapPartitionsRDD.compute(MapPartitionsRDD.scala:38)
at org.apache.spark.rdd.RDD.computeOrReadCheckpoint(RDD.scala:319)
at org.apache.spark.rdd.RDD.iterator(RDD.scala:283)
at org.apache.spark.scheduler.ResultTask.runTask(ResultTask.scala:70)
at org.apache.spark.scheduler.Task.run(Task.scala:86)
at org.apache.spark.executor.Executor$TaskRunner.run(Executor.scala:274)
at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
at java.lang.Thread.run(Thread.java:745)

[AlexDBlack]

@RoiViber can you post the network configuration that you are using here?

Also fyi embedding layers are typically used as the first layer in a network, though, subsetting a set of integer inputs should be OK (I assume that's what you are doing)

[AlexDBlack]

Closing due to inactivity.

[lock]

This thread has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.