I have three question on “Entropy Calibration - pseudocode”
-
Why should we preserve empty bins after squeeze-expand process on Entropy Calibration[PTQ]? It seems reasonable to preserve empty bins but It will loss some righteousness for simulating quantization with Histogram.
-
How to deal with empty bins to calculate KLDivs?
I’ve deal those zero with epsilon[1e-6]. is it right way?
Thank you for kind answer