data processing

[soans1994]

hello author,

I have some doubts. I am new to transformers. I have original data with shape (1000, 50) and convert to time series data as follows
(total_samples=1000, window_or_timestep=10, feature_dim=50).==================> (1000,10,50).
So I have 2 cases. First is original data reshape add dimension (1000, 50, 1). Second is time series converted data (1000, 10, 50)
For RNN/LSTM I use second case.
What should be the input data shape should i use for transformers? is it first (50,1) or second (10, 50)
please give me your feedback.

thank you