add comment first basis fuctions

Author: juanitorduz

examples/gaussian_processes/GP-Births.ipynb

@@ -1464,6 +1464,16 @@
     "pm.model_to_graphviz(model=model)"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "id": "6e14cc26",
+   "metadata": {},
+   "source": [
+    "```{attention}\n",
+    "The first two basis vectors for the (periodic) {class}`~pymc.gp.HSGP` sometimes come out to be either all ones or all zeros. This is a problem because it brings an extra intercept in the model and this can hurt sampling. To avoid this, you can use the `drop_first` argument in the {class}`~pymc.gp.HSGP` class.\n",
+    "```"
+   ]
+  },
   {
    "cell_type": "markdown",
    "id": "f1f9d934",

examples/gaussian_processes/GP-Births.myst.md

@@ -429,6 +429,12 @@ with pm.Model(coords=coords) as model:
 pm.model_to_graphviz(model=model)
 ```
 
+```{attention}
+The first two basis vectors for the (periodic) {class}`~pymc.gp.HSGP` sometimes come out to be either all ones or all zeros. This is a problem because it brings an extra intercept in the model and this can hurt sampling. To avoid this, you can use the `drop_first` argument in the {class}`~pymc.gp.HSGP` class.
+```
+
++++
+
 ## Prior Predictive Checks
 
 We run the model with the prior predictive checks to see if the model is able to generate data in a similar scale as the data.