Batch_first = True in RNN. Question specific to a tutorial

I realize there are similar questions particularly this one: Could someone explain batch_first=True in LSTM

However my question pertains to a specific tutorial I found online: https://medium.com/dair-ai/building-rnns-is-fun-with-pytorch-and-google-colab-3903ea9a3a79

I do not understand why they are not using batch_first=true.

I realized that the data being fed to the model is of the form [64, 28, 28].
Because they are working with MNIST data and they specified a batch size of 64
-64 is batch size
-28 (number of rows) is seq_len
-28 (number of cols) is features

Could someone please explain why they are not using batch_first=True?

Thank you very much for your help.
edit: removed a part of question for simplicity

Most likely for performance reasons.
The input will be permuted in the forward method via:

 # transforms X to dimensions: n_steps X batch_size X n_inputs
X = X.permute(1, 0, 2) 

Thank you. I had not seen permute call in the forward method.