Use python warnings module

abetlen · abetlen · commit 71805353ef3f · 2024-06-04T10:13:15.000-04:00
diff --git a/llama_cpp/llama.py b/llama_cpp/llama.py
@@ -8,6 +8,7 @@
 import ctypes
 import typing
 import fnmatch
+import warnings
 import multiprocessing
 
 from typing import (
@@ -1020,7 +1021,10 @@ def _create_completion(
         model_name: str = model if model is not None else self.model_path
 
         if prompt_tokens[:2] == [self.token_bos()] * 2:
-            print(f'*** WARNING: Detected duplicate leading "{self._model.token_get_text(self.token_bos())}" in prompt, this will likely reduce response quality, consider removing it...', file=sys.stderr)
+            warnings.warn(
+                f'Detected duplicate leading "{self._model.token_get_text(self.token_bos())}" in prompt, this will likely reduce response quality, consider removing it...',
+                RuntimeWarning,
+            )
 
         # NOTE: This likely doesn't work correctly for the first token in the prompt
         # because of the extra space added to the start of the prompt_tokens
