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

Benchmark Model Rank Results
prompt-engineering-on-caltech-101MMRL++#3Harmonic mean: 96.75
prompt-engineering-on-dtdMMRL++#2Harmonic mean: 74.46
prompt-engineering-on-eurosatMMRL++#1Harmonic mean: 91.94
prompt-engineering-on-fgvc-aircraftMMRL++#2Harmonic mean: 42.24
prompt-engineering-on-food-101MMRL++#8Harmonic mean: 91.1
prompt-engineering-on-imagenetMMRL++#4Harmonic mean: 74.44
prompt-engineering-on-oxford-102-flowerMMRL++#3Harmonic mean: 87.01
prompt-engineering-on-oxford-iiit-pet-datasetMMRL++#7Harmonic mean: 96.51
prompt-engineering-on-stanford-cars-1MMRL++#2Harmonic mean: 78.18
prompt-engineering-on-sun397MMRL++#3Harmonic mean: 81.28
prompt-engineering-on-ucf101MMRL++#4Harmonic mean: 83.81