Exception: Source array was not long enough. Check srcIndex and length, and the array's lower bounds

[mdabros]

Exception: Source array was not long enough. Check srcIndex and length, and the array's lower bounds

[mdabros]

@stefanks Did you figure out what the problems was?

[stefanks]

Can't replicate, but this is what happened: with identical data, training gradient boosting with parallel option ON (the default), the training was crashing, but with explicitly setting parallel to off it worked as intended. I'll post here if this happens again.

[mdabros]

Ok, thanks for the information. I will try to reproduce the problem and reopen the issue if I can make it fail.

[stefanks]

The exception was
"Source array was not long enough. Check srcIndex and length, and the array's lower bounds"

[mdabros]

@stefanks I am able to reproduce the problem with the code you provided, so I am reopening the issue and will work on a fix as soon as time permits. Thanks for catching and reporting the problem.

[stefanks]

No, the code I submitted fails due to array size mismatch, and this is expected. I realize that it's expected, so I closed the issue.

It's unrelated to the parallel learning problem, which I can't replicate at the moment.

[mdabros]

@stefanks Ahh yes, I see the example has observations with dimensions [rows;cols][1;1000], when it should have been [1000;1] to match the target length. However, the error message from SharpLearning could have been a lot better, so I have added an issue to improve the error messages in case of dimensionality mismath #27.

So I will close this issue again. Please feel free to reopen it if you encounter the parallel learning problem again.