Skip to main content
. 2022 Nov 10;13:6793. doi: 10.1038/s41467-022-34591-0

Fig. 4. Development of benchmark activity in NLP over time.

Fig. 4

Shown are the number of active benchmarks (i.e., benchmarks for which any novel results were reported) vs. number of benchmarks reporting novel state-of-the-art (SOTA) results over time for NLP tasks. A similar plot for computer vision is available in the associated online material and supplementary material.