Graduate VI from experimental

Summary:
- Graduated VI from experimental for the 0.2.0 release => moved it to `ppl/inference`
- Used the reference tool to find all references of the classes/ functions that are being graduated

Differential Revision: D38875109

fbshipit-source-id: 0e4da4038fba9f7652c733483726b6d63f17363f

src/beanmachine/ppl/experimental/tests/vi/vi_gpu_test.py → src/beanmachine/ppl/inference/tests/vi_gpu_test.py

@@ -9,7 +9,7 @@
 import pytest
 import torch
 import torch.distributions as dist
-from beanmachine.ppl.experimental.vi import VariationalInfer
+from beanmachine.ppl.inference.vi import VariationalInfer
 
 cpu_device = torch.device("cpu")
 

src/beanmachine/ppl/experimental/tests/vi/vi_test.py → src/beanmachine/ppl/inference/tests/vi_test.py

@@ -12,11 +12,9 @@
 import scipy.stats
 import torch
 import torch.distributions as dist
-from beanmachine.ppl.experimental.vi import ADVI, MAP, VariationalInfer
-from beanmachine.ppl.experimental.vi.gradient_estimator import (
-    monte_carlo_approximate_sf,
-)
-from beanmachine.ppl.experimental.vi.variational_world import VariationalWorld
+from beanmachine.ppl.inference.vi import ADVI, MAP, VariationalInfer
+from beanmachine.ppl.inference.vi.gradient_estimator import monte_carlo_approximate_sf
+from beanmachine.ppl.inference.vi.variational_world import VariationalWorld
 from beanmachine.ppl.world import init_from_prior, RVDict
 from torch import optim
 from torch.distributions import constraints

src/beanmachine/ppl/experimental/vi/autoguide.py → src/beanmachine/ppl/inference/vi/autoguide.py

@@ -9,8 +9,8 @@
 import torch
 from beanmachine import ppl as bm
 from beanmachine.ppl.distributions.delta import Delta
-from beanmachine.ppl.experimental.vi.variational_infer import VariationalInfer
-from beanmachine.ppl.experimental.vi.variational_world import VariationalWorld
+from beanmachine.ppl.inference.vi.variational_infer import VariationalInfer
+from beanmachine.ppl.inference.vi.variational_world import VariationalWorld
 from beanmachine.ppl.model.rv_identifier import RVIdentifier
 from beanmachine.ppl.world import RVDict
 from torch import distributions as dist

src/beanmachine/ppl/experimental/vi/gradient_estimator.py → src/beanmachine/ppl/inference/vi/gradient_estimator.py

@@ -8,7 +8,7 @@
 from typing import Callable, Mapping
 
 import torch
-from beanmachine.ppl.experimental.vi.variational_world import VariationalWorld
+from beanmachine.ppl.inference.vi.variational_world import VariationalWorld
 from beanmachine.ppl.model.rv_identifier import RVIdentifier
 from beanmachine.ppl.world import RVDict, World
 

src/beanmachine/ppl/experimental/vi/variational_infer.py → src/beanmachine/ppl/inference/vi/variational_infer.py

@@ -10,11 +10,11 @@
 
 import torch
 import torch.optim as optim
-from beanmachine.ppl.experimental.vi.discrepancy import kl_reverse
-from beanmachine.ppl.experimental.vi.gradient_estimator import (
+from beanmachine.ppl.inference.vi.discrepancy import kl_reverse
+from beanmachine.ppl.inference.vi.gradient_estimator import (
     monte_carlo_approximate_reparam,
 )
-from beanmachine.ppl.experimental.vi.variational_world import VariationalWorld
+from beanmachine.ppl.inference.vi.variational_world import VariationalWorld
 from beanmachine.ppl.model.rv_identifier import RVIdentifier
 from beanmachine.ppl.world.world import RVDict
 from tqdm.auto import tqdm

src/beanmachine/ppl/model/statistical_model.py

@@ -8,7 +8,7 @@
 
 import torch
 import torch.distributions as dist
-from beanmachine.ppl.experimental.vi.variational_world import VariationalWorld
+from beanmachine.ppl.inference.vi.variational_world import VariationalWorld
 from beanmachine.ppl.legacy.world import World
 from beanmachine.ppl.model.rv_identifier import RVIdentifier
 from beanmachine.ppl.world import get_world_context