Is there a way to save very large Tensor in the format they are printed?

When I print a Tensor , I get the following output.

( 9 ,141,.,.) = 
  8.4626e-01  1.5916e-01  1.2594e+00  ...   1.0844e+00  2.1847e+00  9.3616e-01
 -2.2467e+00 -4.6214e+00 -3.5319e+00  ...  -1.1270e+00 -8.4974e-01  5.3662e-01
 -3.0349e+00 -6.7948e+00 -4.6870e+00  ...  -3.8814e+00 -2.3861e+00  1.9217e-01
                 ...                   ⋱                   ...                
 -1.1074e+00 -1.2833e+00 -1.9130e+00  ...   4.5248e-01 -8.8862e-02  4.6845e-01
 -1.0072e+00 -4.6214e-01 -7.9341e-01  ...   1.8886e+00  7.4593e-01  1.4432e+00
 -9.6281e-01 -3.6548e-01 -6.4941e-01  ...  -7.7183e-01 -8.9020e-01 -4.7354e-01

( 9 ,142,.,.) = 
  2.0206e-01  6.9838e-01 -1.0597e+00  ...  -2.5218e-01  7.8152e-01 -6.4857e-01
 -2.0834e+00 -2.4734e+00 -2.2735e+00  ...  -1.4305e+00 -1.6866e-01 -3.9553e-01
  4.0900e-02 -6.7238e-03  2.6108e+00  ...   4.8603e-01  1.2583e+00 -3.8122e-01
                 ...                   ⋱                   ...                
 -3.5195e-01 -9.8042e-01  7.1058e-02  ...  -5.6329e-01 -5.7361e-01 -1.9677e-01
 -8.0403e-01 -1.2030e+00 -5.3037e-01  ...  -8.2459e-01 -5.3295e-01 -4.8523e-01
 -8.0661e-01 -1.0664e+00 -7.1494e-01  ...  -1.3508e-01 -1.0181e-01 -4.6729e-01

( 9 ,143,.,.) = 
  1.1896e+00 -1.6171e-01 -2.3825e+00  ...  -9.0570e-01 -9.3867e-01 -1.0357e+00
  5.2375e-01 -1.0546e+00 -2.4110e+00  ...  -1.3382e+00 -1.0110e+00 -4.4999e-01
  1.3600e+00 -2.0092e+00 -1.5664e+00  ...  -2.6196e+00 -1.8989e+00 -4.8178e-01
                 ...                   ⋱                   ...                
  1.0739e+00 -4.4255e-01 -6.9291e-02  ...   3.3446e-01 -4.4416e-02  9.7961e-01
  1.2883e+00  4.1974e-01  8.6577e-01  ...  -3.6259e-01  1.1620e-01  3.8721e-01
  6.1010e-01 -3.4533e-01  2.3642e-01  ...  -4.1565e-01 -4.6860e-02  1.8264e-01
[torch.FloatTensor of size 10x144x32x32]

Can I save the tensor in the same format they are printed and without the truncated version ?
I also tried torch.set_printoptions(threshold=550000). But it does not seem to print out un truncated Tensor.
I tried torch.save() and tried to open it with text editor… It has some hex code.

tensor.tolist() then save it whatever way you want as you would save a python nested list.

Is there a work around for Variable?

var.data gives you its tensor. So just var.data.tolist() should work.