def test_kl_divergence():
"""Tests kl_divergence"""
# Divergence between a distribution and itself should be 0
dist = tfp.distributions.Normal(0, 1)
assert ops.kl_divergence(dist, dist).numpy() == 0.0
# Divergence between two different distributions should be >0
d1 = tfp.distributions.Normal(0, 1)
d2 = tfp.distributions.Normal(1, 1)
assert ops.kl_divergence(d1, d2).numpy() > 0.0
# Divergence between more different distributions should be larger
d1 = tfp.distributions.Normal(0, 1)
d2 = tfp.distributions.Normal(1, 1)
d3 = tfp.distributions.Normal(2, 1)
assert (ops.kl_divergence(d1, d2).numpy() < ops.kl_divergence(d1,
d3).numpy())
# Should auto-convert probflow distibutions
dist = pf.Normal(0, 1)
assert ops.kl_divergence(dist, dist).numpy() == 0.0
def __call__(self):
return pf.Mixture(pf.Normal(self.m(), self.s()), probs=self.w())
def __call__(self):
return pf.Normal(self.mu(), self.sig())
def __call__(self, x):
return pf.Normal(self.net(x), self.s())
def __call__(self, x):
return pf.Normal(x * self.weight() + self.bias(), self.std())
def __call__(self, x):
return pf.Normal(x @ self.w() + self.b(), self.s())
