Correlations of model rankings across benchmark dimensions, instances, and math benchmarks show substantial redundancy, with rank correlations saturating at around 50% of instances in most benchmarks.
It is defined as: R2 = 1 − Pn i=1(yi − ˆyi)2 Pn i=1(yi − ¯y)2 where: • yi are the ground truth values, • ˆyi are the predicted values, • ¯y is the mean of the ground truth values
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Redundancy Principles for MLLMs Benchmarks
Correlations of model rankings across benchmark dimensions, instances, and math benchmarks show substantial redundancy, with rank correlations saturating at around 50% of instances in most benchmarks.