Ristretto Approximation in smaller bits

[tolulope...@gmail.com]

Hello everyone,

I am new to deep learning and ristretto. I am trying to implement the LeNet Architecture on an hardware using Vivado HLS using 8bit weights. 

I want to know how Ristretto is doing the weight approximation (or quantization) in smaller bits (from 32 bits to 8bits). 

Do I also have access to the quantized (approximated) weights in ristretto at lower bit size. If yes How? 

Thank you.