rename class

Author: ferrine

pymc3/tests/test_variational_inference.py

@@ -7,7 +7,7 @@
 import pymc3 as pm
 from pymc3 import Model, Normal
 from pymc3.variational import (
-    ADVI, FullRankADVI, SVGD, NF,
+    ADVI, FullRankADVI, SVGD, NFVI,
     Empirical, ASVGD,
     MeanField, FullRank,
     fit, flows
@@ -161,13 +161,13 @@ def simple_model(simple_model_data):
 
 
 @pytest.fixture('module', params=[
-        dict(cls=NF, init=dict(flow='scale-loc')),
+        dict(cls=NFVI, init=dict(flow='scale-loc')),
         dict(cls=ADVI, init=dict()),
         dict(cls=FullRankADVI, init=dict()),
         dict(cls=SVGD, init=dict(n_particles=500, jitter=1)),
         dict(cls=ASVGD, init=dict(temperature=1.)),
     ], ids=[
-        'NF=scale-loc',
+        'NFVI=scale-loc',
         'ADVI',
         'FullRankADVI',
         'SVGD',
@@ -202,11 +202,11 @@ def fit_kwargs(inference, using_minibatch):
             obj_optimizer=pm.adagrad_window(learning_rate=0.01, n_win=50),
             n=12000
         ),
-        (NF, 'full'): dict(
+        (NFVI, 'full'): dict(
             obj_optimizer=pm.adagrad_window(learning_rate=0.01, n_win=50),
             n=12000
         ),
-        (NF, 'mini'): dict(
+        (NFVI, 'mini'): dict(
             obj_optimizer=pm.adagrad_window(learning_rate=0.01, n_win=50),
             n=12000
         ),
@@ -608,7 +608,7 @@ def test_hh_flow():
     cov = pm.floatX([[2, -1], [-1, 3]])
     with pm.Model():
         pm.MvNormal('mvN', mu=pm.floatX([0, 1]), cov=cov, shape=2)
-        nf = NF('scale-hh*2-loc')
+        nf = NFVI('scale-hh*2-loc')
         nf.fit(25000, obj_optimizer=pm.adam(learning_rate=0.001))
         trace = nf.approx.sample(10000)
         cov2 = pm.trace_cov(trace)

pymc3/variational/__init__.py

@@ -25,7 +25,7 @@
     FullRankADVI,
     SVGD,
     ASVGD,
-    NF,
+    NFVI,
     Inference,
     fit
 )

pymc3/variational/inference.py

@@ -691,7 +691,7 @@ def run_profiling(self, n=1000, score=None, obj_n_mc=300, **kwargs):
             n=n, score=score, obj_n_mc=obj_n_mc, **kwargs)
 
 
-class NF(Inference):
+class NFVI(Inference):
     R"""
     Normalizing flow is a series of invertible transformations on initial distribution.
 
@@ -753,7 +753,7 @@ def __init__(self, flow='planar*3',
                  local_rv=None, model=None,
                  scale_cost_to_minibatch=False,
                  random_seed=None, start=None, jitter=.1):
-        super(NF, self).__init__(
+        super(NFVI, self).__init__(
             self.OP, self.APPROX, self.TF,
             flow=flow,
             local_rv=local_rv, model=model,
@@ -772,7 +772,7 @@ def from_flow(cls, flow):
 
         Returns
         -------
-        :class:`NF`
+        :class:`NFVI`
         """
         inference = object.__new__(cls)
         Inference.__init__(inference, KL, flow, None)
@@ -800,10 +800,10 @@ def fit(n=10000, local_rv=None, method='advi', model=None,
         -   'advi->fullrank_advi'  for fitting ADVI first and then FullRankADVI
         -   'svgd'  for Stein Variational Gradient Descent
         -   'asvgd'  for Amortized Stein Variational Gradient Descent
-        -   'nf'  for Normalizing Flow
-        -   'nf=formula'  for Normalizing Flow using formula
+        -   'nfvi'  for Normalizing Flow
+        -   'nfvi=formula'  for Normalizing Flow using formula
 
-    model : :class:`pymc3.Model`
+    model : :class:`Model`
         PyMC3 model for inference
     random_seed : None or int
         leave None to use package global RandomStream or other
@@ -833,7 +833,7 @@ def fit(n=10000, local_rv=None, method='advi', model=None,
         fullrank_advi=FullRankADVI,
         svgd=SVGD,
         asvgd=ASVGD,
-        nf=NF
+        nfvi=NFVI
     )
     if isinstance(method, str):
         method = method.lower()
@@ -853,9 +853,9 @@ def fit(n=10000, local_rv=None, method='advi', model=None,
             inference = FullRankADVI.from_advi(inference)
             logger.info('fitting fullrank advi ...')
             return inference.fit(n2, **kwargs)
-        elif method.startswith('nf='):
+        elif method.startswith('nfvi='):
             formula = method[3:]
-            inference = NF(
+            inference = NFVI(
                 formula,
                 local_rv=local_rv,
                 model=model,