Instruction Generation
Benchmarks
Dataset NameSOTA methodMetricTrendResultsLast Updated
66Correctness
7
Apr 10, 2026
76SR
7
Feb 26, 2026
67.9Better Describes Street View
6
Feb 26, 2026
0.708BLEU-1
5
Feb 26, 2026
0.728BLEU-1
5
Feb 26, 2026
30.6BLEU
3
Feb 26, 2026