Follow up after recent PRs

src/heretic/main.py

@@ -201,7 +201,7 @@ def run():
         p=2,
         dim=1,
     )
-    # we don't need the residuals after computing refusal directions
+    # We don't need the residuals after computing refusal directions.
     del good_residuals, bad_residuals
     empty_cache()
 
@@ -420,8 +420,11 @@ def run():
                             continue
 
                         user = huggingface_hub.whoami(token)
-                        fullname = user.get("fullname", user.get("name", "Unknown"))
-                        email = user.get("email", "Not provided")
+                        fullname = user.get(
+                            "fullname",
+                            user.get("name", "unknown user"),
+                        )
+                        email = user.get("email", "no email found")
                         print(f"Logged in as [bold]{fullname} ({email})[/]")
 
                         repo_id = questionary.text(

src/heretic/utils.py

@@ -48,6 +48,9 @@ def batchify(items: list[T], batch_size: int) -> list[list[T]]:
 
 
 def empty_cache():
+    # Collecting garbage is not an idempotent operation, and to avoid OOM errors,
+    # gc.collect() has to be called both before and after emptying the backend cache.
+    # See https://github.com/p-e-w/heretic/pull/17 for details.
     gc.collect()
 
     if torch.cuda.is_available():