How would I go about doing profile guided quantization of an RNN with Glow? All of the user guides I have seen make reference to an image-classifier tool: https://github.com/pytorch/glow/blob/master/docs/Quantization.md
My input is not images but looks something like: [x, h_1, h_2, ..., h_n]
And the output [y, h_1, h_2, ..., h_n]
Where the h_1, h_2,... correspond to the hidden states of the network