Re: java.lang.AssertionError

[Erin LeDell]

Rogerio,
Sending your question to the open source h2ostream community...

On 9/7/15 10:57 AM, Rogerio Normand wrote:
> Dear Sirs,
>
> I am facing this error (java.lang.AssertionError) when running H2O
> 3.0. I have installed the latest version of H2O as well restarted
> program/computer.
>
> Please see attached the error files.
>
> Kind Regards,
>
>
> --
> Rogério Normand
> YNRo Advisory Services
> +351 913 883 600

[Rogerio Normand]

Thanks Erin.

It also happened in a different machine/OS.

Kind Regards,

[SriSatish Ambati]

Rogerio,

Can you add more info on how it happens so we can reproduce this in-house? 

https://0xdata.atlassian.net/browse/PUBDEV-2043

Sorry you ran into an issue with our product.

thanks, Sri 

--
You received this message because you are subscribed to the Google Groups "H2O Open Source Scalable Machine Learning - h2ostream" group.
To unsubscribe from this group and stop receiving emails from it, send an email to h2ostream+...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

--

culture . code . customer

c o m m u n i t y, h2o.ai, 

+1-408.316.8192

[Rogerio Normand]

Hi Sri,

In the last 72h I have run multiple tests and now it is hard to remember exactly that one.

As far memory serves me, I was trying to adapt my program to https://gist.github.com/ledell/e1abae4410330e44b36d suggested by Erin. I had the h2o.init inside a function that was executed many times. The dataset was not big as a file had to be read at each cycle.

I though that the log files should be enough. If it would repeat again, I will forward the full pack to you.

Kind Regards,

[arno....@gmail.com]

Rogerio,
Yes, please rerun and send the full logs. Alternatively, you can run without assertions, h2o.init(assertion = F, threads = -1) to work around this bug, it's not a critical assertion.
Thanks,
Arno

[Rogerio Normand]

Dear Sri,

Please see attached an error like the previous one that happened today.

Kind Regards,

RN

> rdl(c7,8,30) |==================================================================================================| 100% |============================= | 30%
Got exception 'class java.lang.AssertionError', with msg 'null' java.lang.AssertionError at hex.deeplearning.DeepLearning$DeepLearningDriver.trainModel(DeepLearning.java:393) at hex.deeplearning.DeepLearning$DeepLearningDriver.buildModel(DeepLearning.java:248) at hex.deeplearning.DeepLearning$DeepLearningDriver.compute2(DeepLearning.java:171) at water.H2O$H2OCountedCompleter.compute(H2O.java:1005) at jsr166y.CountedCompleter.exec(CountedCompleter.java:429) at jsr166y.ForkJoinTask.doExec(ForkJoinTask.java:263) at jsr166y.ForkJoinPool$WorkQueue.runTask(ForkJoinPool.java:974) at jsr166y.ForkJoinPool.runWorker(ForkJoinPool.java:1477) at jsr166y.ForkJoinWorkerThread.run(ForkJoinWorkerThread.java:104)

Hide Traceback Rerun with Debug Error: 'null' 14 stop(m, call. = FALSE) 13 doTryCatch(return(expr), name, parentenv, handler) 12 tryCatchOne(expr, names, parentenv, handlers[[1L]]) 11 tryCatchList(expr, classes, parentenv, handlers) 10 tryCatch({ while (keepRunning) { myJobUrlSuffix <- paste0(.h2o.__JOBS, "/", job_key) rawResponse <- .h2o.doSafeGET(conn, urlSuffix = myJobUrlSuffix) ... 9 .h2o.__waitOnJob(object@conn, object@job_key) 8 h2o.getFutureModel(.h2o.startModelJob(conn, algo, params, h2oRestApiVersion)) 7 .h2o.createModel(training_frame@conn, "deeplearning", parms) 6 h2o.deeplearning(x = 2:(ncol(tmpX) - 1), y = 1, training_frame = train, validation_frame = valid, activation = "Tanh", hidden = c(2^hid, 2^hid), balance_classes = T, epochs = epo) at h2o Error.R#88 5 rdl(c7, 8, 30) at h2o Error.R#97 4 eval(expr, envir, enclos) 3 eval(ei, envir) 2 withVisible(eval(ei, envir)) 1 source("~/Google Drive YNRo/Neuro Studies/DrAL/h2o Error.R", echo = TRUE)

[Arno Candel]

Hi Rogerio,

Thanks for catching this.

It wasn’t re-scoring the model after stopping as soon as the training error reached 0%, and after replacing the final model with the best model so far based on validation error.

The model itself was correct, just the metrics weren’t up-to-date, and you would have to call h2o.predict(model, frame) to see the updated metrics.  The assertion was checking that the final model was self-consistent (which it wasn’t, so it was a good check).

1) You can work around that by either setting

classification_stop = -1

or 

overwrite_with_best_model = F

2) Alternatively, you can just start H2O without assertions via h2o.init(assertion=F).

3) The next nightly build will have the fix, so you can just wait a few hours :)

Thanks,

Arno

You received this message because you are subscribed to a topic in the Google Groups "H2O Open Source Scalable Machine Learning - h2ostream" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/h2ostream/4ai-VdZtU6k/unsubscribe.
To unsubscribe from this group and all its topics, send an email to h2ostream+...@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

<c7DB.RData><h2o Error.R>

[Rogerio Normand]

Thank you Arno.

Glad to collaborate.

I will wait a few hours. It is already late here.

Best Regards,