List your company

ResearchFormerly WizWand

MIA-Bench

Benchmarks

TaskDatasetSOTA resultTrendResults

Multimodal Instruction Following

MIA-Bench

8.86Score

12

Membership Inference

MIA-Bench 5% Forget Set

69.6Average Performance

12

Multimodal Reasoning

MIA-Bench

4,329.3Length (tokens)

9

MIA-Bench

93.3Accuracy

7

Multi-modal human-preference alignment

MIA-Bench

89.6Score

6

Visual Question Answering

MIA-Bench

68.8Accuracy

4