search

ys-zong / VL-ICL

Code for paper: VL-ICL Bench: The Devil in the Details of Benchmarking Multimodal In-Context Learning
https://ys-zong.github.io/VL-ICL/
29 stars 2 forks source link

train_dataset #8

Open ljqnb opened 1 month ago

ljqnb commented 1 month ago

Hello author, I can only find eval dataset, Could you share me the link of train dataset or How to access it?

ys-zong commented 1 month ago

In each folder, query.json is for evaluation, and support.json is the support (train) set for few-shot demonstration.

ljqnb commented 1 month ago

Thank you very much. By the way, where can I find information on Fast Counting & Fast Attribute Matching?

ys-zong commented 1 month ago

These two subsets were added in our newer version. We haven't released them yet but will aim to do so ASAP.

ljqnb commented 1 month ago

Ok, Thank you very much