Error: tensor slicing does not work properly

[statjapan]

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

require "torch"
=> true
m=Torch.arange(0,100).view([10,10])
=> tensor([[ 0,  1,  2,  3,  4,  5,  6,  7,  8,  9],
            [10, 11, 12, 13, 14, 15, 16, 17, 18, 19],
            [20, 21, 22, 23, 24, 25, 26, 27, 28, 29],
            [30, 31, 32, 33, 34, 35, 36, 37, 38, 39],
            [40, 41, 42, 43, 44, 45, 46, 47, 48, 49],
            [50, 51, 52, 53, 54, 55, 56, 57, 58, 59],
            [60, 61, 62, 63, 64, 65, 66, 67, 68, 69],
            [70, 71, 72, 73, 74, 75, 76, 77, 78, 79],
            [80, 81, 82, 83, 84, 85, 86, 87, 88, 89],
            [90, 91, 92, 93, 94, 95, 96, 97, 98, 99]])

m.select(1,3)
=> tensor([ 3,  4,  5,  6,  7,  8,  9, 10, 11, 12])

"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.narrow(1, 3, 2)
=> tensor([[ 3,  4],
          [ 5,  6],
          [ 7,  8],
          [ 9, 10],
          [11, 12],
          [13, 14],
          [15, 16],
          [17, 18],
          [19, 20],
          [21, 22]])

"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

• torch-rb; 0.2.4 (installed by gem)
• libtorch; 1.5.0
• ruby; 2.7.0
• Debian 10

[ankane]

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.