The rates of incorrect consensus showed a decreasing trend: 10.8% in 2022 (22/203), 6.3% in 2023 (15/237), and 4.8% in 2024 (11/227), indicating an improvement in the reliability of model agreement over the years.
← all excerpts
Performance evaluation of large language models on Korean medical licensing examination: a three-year comparative analysis.
1
—
—