adding context window exceeded error to huggingface

litellm/llms/huggingface_restapi.py

@@ -94,7 +94,10 @@ class HuggingfaceRestAPILLM:
                 additional_args={"complete_input_dict": data},
             )
             ## RESPONSE OBJECT
-            completion_response = response.json()
+            try:
+                completion_response = response.json()
+            except:
+                raise HuggingfaceError(message=response.text, status_code=response.status_code)
             print_verbose(f"response: {completion_response}")
             if isinstance(completion_response, dict) and "error" in completion_response:
                 print_verbose(f"completion error: {completion_response['error']}")

litellm/utils.py

@@ -1462,6 +1462,13 @@ def exception_type(model, original_exception, custom_llm_provider):
                         llm_provider="cohere",
                     )
             elif custom_llm_provider == "huggingface":
+                if "length limit exceeded" in error_str:
+                    exception_mapping_worked = True
+                    raise ContextWindowExceededError(
+                        message=error_str,
+                        model=model,
+                        llm_provider="huggingface"
+                    )
                 if hasattr(original_exception, "status_code"):
                     if original_exception.status_code == 401:
                         exception_mapping_worked = True