did I understand correctly that in addition to the full data set, there will be a baseline reranking provided that returns the best 1000 pictures?
If so, how can we use the baseline reranking for testing purposes without submitting to tira?
Thank you so much,
Sorry for the late reply! I am not sure what you refer to. Like last year, we will have a baseline system that strictly uses the Google ranking with “good” and “anti” queries.
You find the relevant information also in the dataset in the rankings.jsonl and you can use that information however you like. We only ask that you are then in your description paper/report open about what information you used and how.