Hi, I'm trying to reproduce the results from your paper. For accuracy on general VLM Benchmarks (mme and pope), I was wondering how you were able to get a 100% accuracy with both llava-phi and llama-vision models? Do you fine-tune the VLMs on these datasets in addition to the FIUBench dataset?
Hi, I'm trying to reproduce the results from your paper. For accuracy on general VLM Benchmarks (mme and pope), I was wondering how you were able to get a 100% accuracy with both llava-phi and llama-vision models? Do you fine-tune the VLMs on these datasets in addition to the FIUBench dataset?