Good question, I can see how that's confusing. The example is meant to demonstrate using LLMs to check whether *an answer is present within a datasource* before asking for the answer itself. It's an additional check to prevent hallucinations - we can avoid asking for answers when the LLM isn't sufficiently confident that the question can be answered to begin with.
I've updated the codeblocks to say "Article Contains Answer" instead. Hopefully that makes more sense!
I'm a little bit confused about some of the responses to the Ada Lovelace questions:
Question: What nationality was Ada Lovelace?
Answer: True
- True: -3.1281633e-07 (100.0%)
- False: -15.125 (0.0%)
Question: What was an important finding from Lovelace's seventh note?
Answer: True
- True: -5.5122365e-07 (100.0%)
- False: -14.562501 (0.0%)
Why would the answers here be "True"?
Good question, I can see how that's confusing. The example is meant to demonstrate using LLMs to check whether *an answer is present within a datasource* before asking for the answer itself. It's an additional check to prevent hallucinations - we can avoid asking for answers when the LLM isn't sufficiently confident that the question can be answered to begin with.
I've updated the codeblocks to say "Article Contains Answer" instead. Hopefully that makes more sense!
Yup, that's clearer.