Contributor
My role entailed the annotation of text as well as the creation of RLHF datasets. Responsibilities entailed ranking and scoring the quality of the answers provided, determining the degree of adherence to instructions, checking the facts, measuring safety and coherence, and evaluating overall utility in order to rewrite the output. I have conducted prompt-and-answer evaluation, collected preference data, tagged items based on a classification system, and carried out quality assessment on different domains.