What GC tuning parameters are you setting in addition to the use of ZGC? Things like Xmx, and then anything ZGC specific or
-XX:+DisableExplicitGC etc. ZGC maps the memory multiple times, which is known to inflate the resident set size (https://mail.openjdk.java.net/pipermail/zgc-dev/2018-November/000540.html), throwing off JavaCPP's calculation.
@rnett yeah, that sounds like the culprit, thanks! @Craigacp I think you're absolutely right about that, but 30 gigs for TFJ!? I run two models, one "small" and one "large", but that's an outrageous amount of RAM being used?! The graphics card they trained on had 32 gigs, and used batches!?
Sounds like I'll be reverting some of those recent changes then, thanks!
You're actually staying within the acceptable memory limits, but the RSS based method that JavaCPP uses can't figure that out (because Linux doesn't know that either).
Based on https://docs.oracle.com/en/java/javase/15/gctuning/available-collectors.html#GUID-C7B19628-27BA-4945-9004-EC0F08C76003 it sounds like one should be using the parallel or G1 garbage collectors tuned for max-throughput if you're training a model. That said, ZGC might still make sense in the context of evaluating a pre-trained model.
Hey. I posted a modeling question over at https://ai.stackexchange.com/q/27657/46787. I would appreciate your advice even though it's not TensorFlow-specific.
Hello, we run Tensorflow Java 0.2.0 (TF 2.3.1). And we have a model that produces an image out of the model as a byte[]. This is fine when running the model in python. We can write the bytes to a file. But when trying to do the same thing using TF Java we run into getting too few bytes from the output. I think I have managed to boil it down to unit test with TString.
public void testTFNdArray() throws Exception {
ClassLoader contextClassLoader = Thread.currentThread().getContextClassLoader();
byte[] readAllBytes = Files.readAllBytes(Path.of(contextClassLoader.getResource("img_12.png").getFile()));
NdArray<byte[]> vectorOfObjects = NdArrays.vectorOfObjects(readAllBytes);
Tensor<TString> tensorOfBytes = TString.tensorOfBytes(vectorOfObjects);
TString data = tensorOfBytes.data();
byte[] asBytes = tensorOfBytes.data().asBytes().getObject(0);
System.out.println("Bytes original file: " + readAllBytes.length);
System.out.println("NdArray byte[] length: " + vectorOfObjects.getObject().length);
System.out.println("Tensor numbytes: " + tensorOfBytes.numBytes());
System.out.println("TString size: " + data.size());
System.out.println("Bytes with reading from TString (WRONG): " + asBytes.length);
}This is the problem that I get with running through a real model. How should we be able to get the full byte[] out again?
But running this in a docker linux vm the test pass.
It fails 9 out of 10 times. And dump usually looks like this.
#
# A fatal error has been detected by the Java Runtime Environment:
#
# SIGSEGV (0xb) at pc=0x0000000133cf0fbb, pid=19618, tid=5891
#
# JRE version: OpenJDK Runtime Environment (11.0.2+9) (build 11.0.2+9)
# Java VM: OpenJDK 64-Bit Server VM (11.0.2+9, mixed mode, tiered, compressed oops, g1 gc, bsd-amd64)
# Problematic frame:
# C [libtensorflow_cc.2.dylib+0x8228fbb] TF_TensorData+0xb
#
# No core dump will be written. Core dumps have been disabled. To enable core dumping, try "ulimit -c unlimited" before starting Java again
#
# An error report file with more information is saved as:
# /Users/jakob/proj/tfjava031/hs_err_pid19618.log
#
# If you would like to submit a bug report, please visit:
# http://bugreport.java.com/bugreport/crash.jsp
# The crash happened outside the Java Virtual Machine in native code.
# See problematic frame for where to report the bug.
#
Can you open an issue here - https://github.com/tensorflow/java/issues, we'll move the discussion over to Github.
Will open a ticket
it is strange yes, because we do have unit tests creating string tensors and these are ran on all platforms
It might be a threading issue because it actually succeeded ~1 out of 10 times. Or depending on how many cores the unit test runs on.
1 reply
Hey guys, is there any documentation for https://github.com/tensorflow/java/tree/master/tensorflow-framework beyond the tiny readme file? For example, is there any public Javadoc for it equivalent to what https://javadoc.io/doc/org.tensorflow/tensorflow-core-api/latest/index.html does for the core API?
Also javadoc.io will auto generate the javadoc for an artifact if you try to load it (which I just did for tensorflow-framework)
Out of curiosity, how do I know which version of the API https://www.tensorflow.org/jvm/api_docs/java/org/tensorflow/package-summary is written against? I don't see any versioning information.
The API changes enough that it would be helpful to know if I'm looking at outdated documentation.
Are there any examples that show the usage of the Java framework API for basic workflow like this?
training_images = training_images/255.0
test_images = test_images/255.0
model = tf.keras.models.Sequential([#tf.keras.layers.Flatten(),
tf.keras.layers.Dense(64, activation=tf.nn.relu),
tf.keras.layers.Dense(10, activation=tf.nn.softmax)])
model.compile(optimizer = 'adam',
loss = 'sparse_categorical_crossentropy')
model.fit(training_images, training_labels, epochs=5)
model.evaluate(test_images, test_labels)
classifications = model.predict(test_images)Specifically, I don't see the equivalent of models.Sequential, model.compile, model.fit etc.
Okay, let's go with the flow then... Say I want to train in python and
model.predict() in Java.- What is the easiest way to build a dataset in Java and pass it into python for training? I want to leverage existing Java code to retrieve the input dataset from a database. I don't want to reimplement it in python.
- Is there an easy way for me to load a graph in Java that was constructed + trained in python? Can you point me at any sample/test code?
It's just lower level than Keras.
For 1, you could emit it as TFRecords, but I'm not sure we've got saving support lined up for that. For 2, yes you can load in a SavedModel, see https://github.com/tensorflow/java-models
Training an MLP in Java you can see this - https://github.com/tensorflow/java-models/blob/master/tensorflow-examples/src/main/java/org/tensorflow/model/examples/dense/SimpleMnist.java
Thank you. If one of my placeholders is a matrix of TFloat32, what's the best way to populate the matrix column by column? I assume I can't push values directly into the
Placeholder. Do I invoke Ops.tensorArray() of the same size as the placeholder, then populate that, then eventually invoke Runner.feed(placeholder, array)? Or is there a better way?
Create a
TFloat32 either an empty one, or by copying from some source. Then feed it to the placeholder. Note the empty one returns memory which has not been zeroed yet tensorflow/java#271.
I assume you guys haven't added many of the debugging tools that are present on the Python side, so I think it's best if I construct the model from python, debug
model.fit() until I get something working.... then I can move model.fit() to Java and model.predict() would always sit in Java.
So to some extent it depends how complicated your model is. Tribuo's next release exposes TF models but wraps up all the fitting, evaluation and prediction in it's interface to make it a lot simpler. It's not the same as Keras, it's a little bit more like scikit-learn as we don't have callbacks in Tribuo.
However TF-Java will have this in the future, it's just a lot of stuff to build with a much smaller team than the Keras team.