improve nb

Python/numpyro_pathfinder.ipynb

@@ -496,7 +496,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "Note that the value for `a` is close to the true value of $1.0$. However, the values for `b` and `sigma` are do not match the true values of $2.0$ and $0.5$ respectively. The reason is that we use a `dist.HalfNormal` and `dist.Exponential` prior these parameters respectively so the samples are taken from the unconstrained space."
+    "Note that the value for `a` is close to the true value of `1.0`. However, the values for `b` and `sigma` do not match the true values of `2.0` and `0.5` respectively. The reason is that we are using prior the prior distributions `dist.HalfNormal` and `dist.Exponential` for these parameters respectively. Since these distributions are positive, the sampler transform these parameters to the unconstrained space using a bijective transformation. To compare the results in the original space, we need to apply the inverse transformation."
    ]
   },
   {