Detailed error message for noise sensitivity metric

[KalyanKS-NLP]

Absence of any of the required inputs (user_input, response, reference or retrieved_contexts) in the test sample results in KeyError. For example, the absence of user_input in the test sample results in the error message "KeyError: user_input".

The error message "KeyError: user_input" is too abstract. I included the following lines of code in the _ascore() function definition which will display a clear error message in the absence of any of the required inputs.

if "reference" not in row or not row["reference"]:
      raise ValueError("reference is missing in the test sample. Please add reference to the test sample.")

 if "user_input" not in row or not row["user_input"]:
      raise ValueError("user_input is missing in the test sample. Please add user_input to the test sample.")

 if "response" not in row or not row["response"]:
       raise ValueError("response is missing in the test sample. Please add response to the test sample.")

 if "retrieved_contexts" not in row or not row["retrieved_contexts"]:
       raise ValueError("retrieved_contexts is missing in the test sample. Please add retrieved_contexts to the test sample.")

[shahules786]

LGTM, thanks bro