minor revisions

docs/tex/tutorials/automated-transformations.tex

@@ -184,11 +184,11 @@ \subsubsection{Automated Transformations in Inference}
 
 \subsubsection{Acknowledgements \& Remarks}
 
-Automated transformations have largely been popularized by Stan,
-particularly for Hamiltonian Monte Carlo \citep{carpenter2016stan}.
-This design is inspired by Stan's. However, a key distinction is Edward
-provides users flexibility to wield transformations and manipulate
-results in either the original (constrained) or inferred
+Automated transformations have largely been popularized by Stan
+for Hamiltonian Monte Carlo \citep{carpenter2016stan}.
+This design is inspired by Stan's. However, a key distinction is that Edward
+provides users the ability to wield transformations and more flexibly
+manipulate results in both the original (constrained) and inferred
 (unconstrained) space.
 
 Automated transformations are also core to the algorithm automatic

notebooks/automated_transformations.ipynb

@@ -320,11 +320,11 @@
    "source": [
     "## Acknowledgements & Remarks\n",
     "\n",
-    "Automated transformations have largely been popularized by Stan,\n",
-    "particularly for Hamiltonian Monte Carlo (Carpenter et al., 2016).\n",
-    "This design is inspired by Stan's. However, a key distinction is Edward\n",
-    "provides users flexibility to wield transformations and manipulate\n",
-    "results in either the original (constrained) or inferred\n",
+    "Automated transformations have largely been popularized by Stan\n",
+    "for Hamiltonian Monte Carlo (Carpenter et al., 2016).\n",
+    "This design is inspired by Stan's. However, a key distinction is that Edward\n",
+    "provides users the ability to wield transformations and more flexibly\n",
+    "manipulate results in both the original (constrained) and inferred\n",
     "(unconstrained) space.\n",
     "\n",
     "Automated transformations are also core to the algorithm automatic\n",
@@ -333,7 +333,7 @@
     "distributions. However, note the automated transformation from\n",
     "non-negative to reals in Edward is not $\\log$, which is used in Stan;\n",
     "rather, Edward uses $\\textrm{softplus}$ which is more numerically\n",
-    "stable (see also Kucukelbir et al. (2017, Fig. 9).\n",
+    "stable (see also Kucukelbir et al. (2017, Fig. 9)).\n",
     "\n",
     "Finally, note that not all inference algorithms use or even need\n",
     "automated transformations.\n",