cypher_rlhf
Having to rate two model-generated responses separately across various dimensions. To provide preference ranking scores and to write a justification for the response preference ranking and to explain the specifics and logic behind the decision.