Detailed breakdown of 19 questions, provided answers from LLM, three reviewer scores for accuracy, relevance, and readability per question, notes from reviewers (where relevant) explaining rationale for provided score, reviewer name, and annotation (1 = yes, 0 = no) for whether a hallucination was observed with an answer.

This set of information is provided for each of the for LLMs tested, one per worksheet. Also provided in a separate worksheet is the question grouping used to categorize questions in S1 Fig. The last worksheet contains details of answers provided by the RAG model when varying the number of answers (k) the model used to generate a final answer. Additional information on the papers used to generate answers and intermediate answers the model used to generate the final output are also given.
(XLSX)