Wei et al

Reported on 2 benchmarks across 2 tasks

Note: results are matched by exact model name. Different papers may use the same name for different model variants.