def guide():
d = dist.Categorical(pyro.param("q"))
context1 = pyro.plate("outer", outer_dim, dim=-1)
context2 = pyro.plate("inner", inner_dim, dim=-2)
pyro.sample("w", d, infer={"enumerate": "parallel"})
with context1:
pyro.sample("x", d, infer={"enumerate": "parallel"})
with context2:
pyro.sample("y", d, infer={"enumerate": "parallel"})
with context1, context2:
pyro.sample("z", d, infer={"enumerate": "parallel"})
def model():
d = dist.Categorical(p)
context1 = pyro.plate("outer", outer_dim, dim=-1)
context2 = pyro.plate("inner", inner_dim, dim=-2)
pyro.sample("w", d)
with context1:
pyro.sample("x", d)
with context2:
pyro.sample("y", d)
with context1, context2:
pyro.sample("z", d)
def auto_model():
probs_a = pyro.param("probs_a")
probs_b = pyro.param("probs_b")
probs_c = pyro.param("probs_c")
probs_d = pyro.param("probs_d")
with pyro.plate("a_axis", 2, dim=-1):
a = pyro.sample("a",
dist.Categorical(probs_a),
infer={"enumerate": "parallel"})
pyro.sample("b", dist.Categorical(probs_b[a]), obs=b_data)
with pyro.plate("c_axis", 3, dim=-1):
c = pyro.sample("c",
dist.Categorical(probs_c),
infer={"enumerate": "parallel"})
pyro.sample("d", dist.Categorical(probs_d[c]), obs=d_data)
def auto_guide(data):
probs_a = pyro.param("guide_probs_a")
probs_c = pyro.param("guide_probs_c")
a = pyro.sample("a",
dist.Categorical(probs_a),
infer={"enumerate": "parallel"})
with pyro.plate("data", 2, dim=-1):
pyro.sample("c", dist.Categorical(probs_c[a]))
def model(data):
T, N, D = data.shape # time steps, individuals, features
# Gaussian initial distribution.
init_loc = pyro.param("init_loc", torch.zeros(D))
init_scale = pyro.param("init_scale",
1e-2 * torch.eye(D),
constraint=constraints.lower_cholesky)
# Linear dynamics with Gaussian noise.
trans_const = pyro.param("trans_const", torch.zeros(D))
trans_coeff = pyro.param("trans_coeff", torch.eye(D))
noise = pyro.param("noise",
1e-2 * torch.eye(D),
constraint=constraints.lower_cholesky)
obs_plate = pyro.plate("channel", D, dim=-1)
with pyro.plate("data", N, dim=-2):
state = None
for t in range(T):
# Transition.
if t == 0:
loc = init_loc
scale_tril = init_scale
else:
loc = trans_const + funsor.torch.torch_tensordot(
trans_coeff, state, 1)
scale_tril = noise
state = pyro.sample("state_{}".format(t),
dist.MultivariateNormal(loc, scale_tril),
infer={"exact": exact})
# Factorial probit likelihood model.
with obs_plate:
pyro.sample("obs_{}".format(t),
dist.Bernoulli(logits=state["channel"]),
obs=data[t])
def auto_model(data):
probs_a = pyro.param("model_probs_a")
probs_b = pyro.param("model_probs_b")
probs_c = pyro.param("model_probs_c")
probs_d = pyro.param("model_probs_d")
probs_e = pyro.param("model_probs_e")
a = pyro.sample("a", dist.Categorical(probs_a))
b = pyro.sample("b",
dist.Categorical(probs_b[a]),
infer={"enumerate": "parallel"})
with pyro.plate("data", 2, dim=-1):
c = pyro.sample("c", dist.Categorical(probs_c[a]))
d = pyro.sample("d",
dist.Categorical(Vindex(probs_d)[b, c]),
infer={"enumerate": "parallel"})
pyro.sample("obs", dist.Categorical(probs_e[d]), obs=data)
def model(data):
loc = pyro.sample("loc", dist.Normal(0., 1.))
with pyro.plate("data", len(data), dim=-1):
pyro.sample("obs", dist.Normal(loc, 1.), obs=data)
def model(data=None):
loc = pyro.param("loc", torch.tensor(expected_mean))
scale = pyro.param("scale", torch.tensor(1.0))
with pyro.plate("data", 1000, dim=-1):
x = pyro.sample("x", dist.Normal(loc, scale), obs=data)
return x
def guide():
with pyro.plate("plate", len(data), dim=-1):
p = pyro.param("p", torch.ones(len(data), 3) / 3, event_dim=1)
pyro.sample("x", dist.Categorical(p))
return p
def model():
locs = pyro.param("locs", torch.tensor([-1.0, 0.0, 1.0]))
with pyro.plate("plate", len(data), dim=-1):
x = pyro.sample("x", dist.Categorical(torch.ones(3) / 3))
pyro.sample("obs", dist.Normal(locs[x], 1.0), obs=data)
def guide():
loc = pyro.param("loc", torch.tensor(0.0))
scale = pyro.param("scale", torch.tensor(1.0))
with pyro.plate("plate_outer", data.size(-1), dim=-1):
pyro.sample("x", dist.Normal(loc, scale))
def model():
loc = torch.tensor(3.0)
with pyro.plate("plate_outer", data.size(-1), dim=-1):
x = pyro.sample("x", dist.Normal(loc, 1.0))
with pyro.plate("plate_inner", data.size(-2), dim=-2):
pyro.sample("y", dist.Normal(x, 1.0), obs=data)
def guide():
p = pyro.param("p", torch.tensor([0.5, 0.3, 0.2]))
with pyro.plate("plate", len(data), dim=-1):
pyro.sample("x", dist.Categorical(p))
def model():
locs = pyro.param("locs", torch.tensor([0.2, 0.3, 0.5]))
p = torch.tensor([0.2, 0.3, 0.5])
with pyro.plate("plate", len(data), dim=-1):
x = pyro.sample("x", dist.Categorical(p))
pyro.sample("obs", dist.Normal(locs[x], 1.0), obs=data)
