search

mlpc-ucsd / BLIVA

(AAAI 2024) BLIVA: A Simple Multimodal LLM for Better Handling of Text-rich Visual Questions
https://arxiv.org/abs/2308.09936
BSD 3-Clause "New" or "Revised" License
260 stars 26 forks source link

how about the ablation study #23

Open jiinhui opened 1 month ago

jiinhui commented 1 month ago

In your paper, from the table of “Results of adding individual techniques of our framework in text-rich VQA benchmarks“，I get the improvement vals of +7.40% ，+31.72%，etc. I want to know how you calculated that. thanks!

gordonhu608 commented 1 month ago

The relative improvement is the absolute value of (our score - previous score) / previous score.

jiinhui commented 1 month ago

The relative improvement is the absolute value of (our score - previous score) / previous score.

thanks, I see.