diff --git a/relbench/datasets/__init__.py b/relbench/datasets/__init__.py
index 06154cf7..6c9011fe 100644
--- a/relbench/datasets/__init__.py
+++ b/relbench/datasets/__init__.py
@@ -74,7 +74,14 @@ def get_dataset(name: str, download=False) -> Dataset:
     Returns:
         Dataset: The dataset object.
 
-    If `download` is True, the dataset will be downloaded into the cache.
+    If `download` is True, the database comprising the dataset will be
+    downloaded into the cache from the RelBench server. If you use
+    `download=False` the first time, the database will be processed from the
+    raw files of the original source.
+
+    Once the database is cached, either because of download or processing from
+    raw files, the cache will be used. `download=True` will verify that the
+    cached database matches the RelBench version even in this case.
     """
 
     if download:
diff --git a/relbench/tasks/__init__.py b/relbench/tasks/__init__.py
index 1c94167a..3e51a4cf 100644
--- a/relbench/tasks/__init__.py
+++ b/relbench/tasks/__init__.py
@@ -79,7 +79,14 @@ def get_task(dataset_name: str, task_name: str, download=False) -> BaseTask:
     Returns:
         BaseTask: The task object.
 
-    If `download` is True, the task will be downloaded into the cache.
+    If `download` is True, the task tables (train, val, test) comprising the
+    task will be downloaded into the cache from the RelBench server. If you use
+    `download=False` the first time, the task tables will be computed from
+    scratch using the database.
+
+    Once the task tables are cached, either because of download or computing from
+    scratch, the cache will be used. `download=True` will verify that the
+    cached task tables matches the RelBench version even in this case.
     """
 
     if download: