MMRL++: Parameter-Efficient and Interaction-Aware Representation Learning for Vision-Language Models

Benchmark Model Rank Results
prompt-engineering-on-eurosatMMRL++#2Harmonic mean: 91.94
prompt-engineering-on-fgvc-aircraftMMRL++#4Harmonic mean: 42.24
prompt-engineering-on-dtdMMRL++#5Harmonic mean: 74.46
prompt-engineering-on-oxford-102-flowerMMRL++#5Harmonic mean: 87.01
prompt-engineering-on-stanford-carsMMRL++#5Harmonic mean: 78.18
prompt-engineering-on-sun397MMRL++#5Harmonic mean: 81.28
prompt-engineering-on-ucf101MMRL++#5Harmonic mean: 83.81
prompt-engineering-on-caltech-101MMRL++#7Harmonic mean: 96.75
prompt-engineering-on-imagenetMMRL++#7Harmonic mean: 74.44
prompt-engineering-on-food-101MMRL++#9Harmonic mean: 91.1
prompt-engineering-on-oxford-iiit-pet-datasetMMRL++#16Harmonic mean: 96.51