Slice op: 'starts' and 'ends' values resulted in a negative dimension

[pengwa]

sometime slice need -1 in starts or ends, onnx doc actually allow https://github.com/onnx/onnx/blob/master/docs/Changelog.md#slice-1

starts and ends attributes to specify the start and end dimension for each axis in the list of axes, it uses this information to slice the input data tensor. If a negative value is passed for any of the start or end indices, it represent number of elements before the end of that dimension.

So I think this is a bug. This is blocking a real model conversion, could anybody familiar with this take a look? Thanks a lot.

[pengwa]

Scott McKay (@skottmckay) helped clarify why my case wrong.

basically, I am using starts = [-1], ends=[0].

in original tf.stride_sclice, this is allowed since another parameter stride=[1].

in onnx, -1 will become a concrete valid positive numb before running. ends will not. So I got this error.

seems not a bug, more likely I am using in an inproper way. closing this.

thanks for the investigation.