Disabling missing data imputation

Questions
16

here is an example of what I want to do

import pytensor.tensor as pt
import numpy as np
import pymc as pm
obs_dataset = np.random.rand(5, 6)  
mask = (obs_dataset>1)

#this works 
with pm.Model() as m:

    mu = pm.Normal("mu")
    sigma = pm.HalfNormal("sigma")

    bcast_mu = pt.broadcast_to(mu, obs_dataset.shape)
    bcast_sigma = pt.broadcast_to(sigma, obs_dataset.shape)
    pm.Normal("likelihood", bcast_mu[mask], bcast_sigma[mask], observed=obs_dataset[mask])

m.point_logps()  # {'mu': -0.92, 'sigma': -0.73, 'likelihood': -61.98}

#how to do this ? this doesn't work, how to do this when obs_dataset is mutable ?
with pm.Model() as m:
    obs_dataset = pm.MutableData('obs_dataset', obs_dataset)
    mask = (obs_dataset >1)
    mu = pm.Normal("mu")
    sigma = pm.HalfNormal("sigma")

    bcast_mu = pt.broadcast_to(mu, obs_dataset.shape)
    bcast_sigma = pt.broadcast_to(sigma, obs_dataset.shape)
    pm.Normal("likelihood", bcast_mu[mask], bcast_sigma[mask], observed=obs_dataset[mask])

m.point_logps()

basically the mask is mutable, because the obs_dataset is a MutableData so how can I do the indexing in a mutable way?