"m[0..9, 3..4]" returns same tensor too.
The expected tensor is tensor([[ 3, 4], [13, 14], [23, 24], ..., [93, 94]])
The slicing in the row direction work correctly.
Hey @statjapan, thanks for reporting! Should be fixed on master. There was an issue with how the tensor values were converted to Ruby in that situation.
Description With a tensor over two dimensions, slicing does not work properly. In a two-dimensional matrix, column slicing does not work well.
Sample using pry
"m[0..9, 3]" returns same tensor too. The expected tensor is [ 3, 13, 23, 33, 43, 53, 63, 73, 83, 93]. m.select(0, 3) returns expected tensor [30, 31, 32, 33, 34, 35, 36, 37, 38, 39].
"m[0..9, 3..4]" returns same tensor too. The expected tensor is tensor([[ 3, 4], [13, 14], [23, 24], ..., [93, 94]]) The slicing in the row direction work correctly.
OS / Environment