tensor([[[ 0.0526, -0.0758, -0.0244, -0.1297,  0.2833,  0.1173, -0.0380,
           0.1630,  0.0250,  0.0383, -0.1318, -0.0832,  0.0126,  0.0375,
          -0.0613,  0.0979,  0.1162,  0.0810, -0.1895,  0.0198,  0.0891,
          -0.0294,  0.0407,  0.1235,  0.0691,  0.1792,  0.2678, -0.0591,
          -0.0358,  0.2023, -0.1086,  0.0628, -0.0803,  0.0492,  0.0776,
          -0.0088,  0.1205, -0.0913, -0.1340,  0.2783, -0.0764,  0.0185,
           0.0267,  0.0804,  0.1727,  0.0197, -0.2012,  0.1461,  0.0741,
           0.0027,  0.0282,  0.0160, -0.0450,  0.1699,  0.0707,  0.0674,
          -0.0456, -0.0273,  0.0401, -0.0623, -0.0653,  0.2069, -0.0769,
           0.2017, -0.0514,  0.0674,  0.0370, -0.0931, -0.2761,  0.0760,
          -0.0325, -0.0639,  0.0399, -0.0479, -0.0749, -0.0360,  0.0277,
           0.0862,  0.1340,  0.0353, -0.1419,  0.0599,  0.0825,  0.2321,
           0.1891,  0.1222,  0.0107,  0.0783, -0.0613, -0.0466, -0.0157,
          -0.1413, -0.0466,  0.3371, -0.0088,  0.2287, -0.0819,  0.1114,
           0.1103, -0.1081]]], grad_fn=<SumBackward1>)
 final prediction: tensor([[[-0.0417, -0.0213,  0.0098]]], grad_fn=<ViewBackward0>)
 predicted probabilities: tensor([[[0.3254, 0.3321, 0.3426]]], grad_fn=<SoftmaxBackward0>)