How send extracted features from CNN to an LSTM as an input

Hello guys.
I am new to PyTorch and I am implementing a two-stream action recognition architecture.
For each of them, I have a CNN and at the top an Lstm.
My question is how can I pass the feature extracted from the CNN to the LSTM as input. I am confused a little bit.
Any guidance from you guys will be so helpful for me. Thank you

Or course it depends on your actual use case, but you could treat the height and width of the CNN output as the temporal dimension and the channels as the features and pass it to the RNN.
By default the RNN modules expect an input in the shape [seq_len, batch_size, features] and you could use batch_first=True to pass it as [batch_size, seq_len, features] (the docs give you all expected shapes).