llm_server/llm/vllm/tokenize.py

@@ -28,7 +28,6 @@ def tokenize(prompt: str, backend_url: str) -> int:
             return j['length']
         except Exception as e:
             print(f'Failed to tokenize using VLLM -', f'{e.__class__.__name__}: {e}')
-            raise Exception
             return len(tokenizer.encode(chunk)) + 10
 
     # Use a ThreadPoolExecutor to send all chunks to the server at once