In the stochastic volatility example of Getting Started, the model has sigma ~ exp(50).
In cell 16, however:
sigma = pm.Exponential("sigma", 2.0, testval=0.1)
the 50 seems to have changed to 2.0. The pymc3 paper has lambda ~ 1/0.02 = 50. Is this an error in the Getting Started guide?
I think it is changed at some point because Exp(50) is a pretty bad prior that leads to problem during inference
The inconsistency is confusing, so the model specification should be changed. I spent a bit of time searching for alternative parameterizations of the exponential distribution to the one I know.