Check failed: b < len (21 vs. 21) slicing with begin[0]=21 exceends limit of 21

[bobpoekert]

It looks like I’m hitting this condition: https://github.com/apache/incubator-mxnet/blob/master/src/operator/tensor/matrix_op-inl.h#L670

Should that ever happen or does that indicate a bug in mxnet? It’s happening consistently but I haven’t narrowed it down to a small test case.

It is a little suspicious that line 664 uses CHECK_GE but line 670 uses CHECK_LT and not CHECK_LE

Here’s the full error message: mxnet.base.MXNetError: [15:30:34] src/operator/tensor/./matrix_op-inl.h:670: Check failed: b < len (21 vs. 21) slicing with begin[0]=21 exceends limit of 21

[lanking520]

hi @bobpoekert could you please bring a minimal example that can reproduce this issue? Which MXNet version you are using? Are you using Python?

[vdantu]

@bobpoekert : Could you please provide us more information about the setup and minimum reproducible script?

[vrakesh]

@bobpoekert Has the issue been resolved for you? If not requesting you to provide us with setup and minimum reproducible script

[leleamol]

@mxnet-label-bot can this question be closed?

[vdantu]

@rahul003 : Could you help with this?

[zhaotun]

I get the same error when train my own dataset. I use VOCLike to load my dataset which described at
https://gluon-cv.mxnet.io/build/examples_datasets/detection_custom.html，and i change the CLASSES to my own classnames ,such as CLASSES=['1','2','3', ... ,'32']，i have 32 classes and the classname is from 1 to 32, then start the train , get the error 'Check failed: b < len (25 vs. 20) slicing with begin[3]=25 exceends limit of 20' , after many tries ,I found that the CLASSES in "C:\ProgramData\Anaconda3\Lib\site-packages\gluoncv\data\pascal_voc\detection.py" also need to be changed, when i changed it to my own classnames , the error disappeared . holp to be helpful.

[kaiserdan]

Thanks for the suggestion @zhaotun. I faced the same problem and your proposed fix solved it.