labels
tensor([ 21.0000, 23.7500, 26.5000, 29.2500, 32.0000, 34.7500, 37.5000,
40.2500, 43.0000, 45.7500, 48.5000, 51.2500, 54.0000, 56.7500,
59.5000, 62.2500, 65.0000, 67.7500, 70.5000, 73.2500, 76.0000,
78.7500, 81.5000, 84.2500, 87.0000, 89.7500, 92.5000, 95.2500,
70.0000, 72.7500, 75.5000, 78.2500, 81.0000, 83.7500, 86.5000,
89.2500, 92.0000, 94.7500, 97.5000, 100.2500, 103.0000, 105.7500,
108.5000, 111.2500, 114.0000, 116.7500, 119.5000, 122.2500, 125.0000,
127.7500, 130.5000, 133.2500, 136.0000, 138.7500, 141.5000, 144.2500])
logits
tensor([[ -6.8191, -5.2425, -6.6681, ..., -5.7708, -5.2461, -5.2987],
[ -6.4731, -6.5320, -6.7831, ..., -3.5591, -3.8897, -7.2461],
[ -8.7466, -8.0584, -8.0699, ..., -12.5498, -11.9103, -6.9622],
...,
[ -6.4875, -9.2046, -6.9570, ..., -10.9126, -11.3829, -9.2204],
[ -7.5633, -5.9068, -7.0456, ..., -4.6755, -4.4977, -6.5758],
[ -5.3650, -7.0494, -5.5796, ..., -8.1454, -8.3036, -6.3091]],
grad_fn=<LogSoftmaxBackward>