Fix issue of missing words due to buffer overflow

manasmagdum · web-flow · commit 4100bdec31b3 · 2023-08-27T16:39:46.000+05:30
diff --git a/llama_cpp/llama.py b/llama_cpp/llama.py
@@ -445,7 +445,7 @@ def detokenize(self, tokens: List[int]) -> bytes:
         """
         assert self.model is not None
         output = b""
-        size = 8
+        size = 16
         buffer = (ctypes.c_char * size)()
         for token in tokens:
             n = llama_cpp.llama_token_to_str_with_model(
