Better handling of LLM request failures

strix/llm/__init__.py

@@ -1,12 +1,15 @@
 import litellm
 
 from .config import LLMConfig
-from .llm import LLM
+from .llm import LLM, LLMRequestFailedError
 
 
 __all__ = [
     "LLM",
     "LLMConfig",
+    "LLMRequestFailedError",
 ]
 
+litellm._logging._disable_debugging()
+
 litellm.drop_params = True