I am using Spark 1.6.1 with Deeplearning Spark's dependency: It most frequently gets hit when calling the SparkUtils.balancedRandomSplit method. The method is in executeTrainingPathsHelper in the ParameterAveragingTrainingMaster class. When running sparkNetwork.fit(JavaRDD data), I get the error in different places, but it seems to be centralized in one method. build() SparkDl4jMultiLayer sparkNetwork = new SparkDl4jMultiLayer(SparkConfig.getSparkContext(), conf, tm) repartitionStrategy(RepartitionStrategy.SparkDefault). TrainingMaster tm = new ParameterAveragingTrainingMaster.Builder(32). My code at the moment looks like (although I am messing around to try and get this to work): With that in mind, the below information is all that I could find. I have tried the obvious such as starting/restarting my computer, tried different parameters, and other related things but have had no success.

The actual error I am receiving is this:ġ6/11/20 21:56:42 INFO Executor: Running task 1.0 in stage 16.0 (TID 80) 16/11/20 21:56:42 INFO Executor: Running task 0.0 in stage 16.0 (TID 79) 16/11/20 21:56:42 INFO ShuffleBlockFetcherIterator: Getting 2 non-empty blocks out of 2 blocks 16/11/20 21:56:42 INFO ShuffleBlockFetcherIterator: Getting 2 non-empty blocks out of 2 blocks 16/11/20 21:56:42 INFO ShuffleBlockFetcherIterator: Started 0 remote fetches in 1 ms 16/11/20 21:56:42 INFO ShuffleBlockFetcherIterator: Started 0 remote fetches in 1 ms java(32550,0x12ff08000) malloc: *** error for object 0x7fc6f2a0db88: incorrect checksum for freed object - object was probably modified after being freed. This is tricky to pinpoint the exact issue because it does not come with a stacktrace.

Eclipse for mac yosemite