Competition Rules Data and Evaluation Leaderboard FAQs
Data Description Submission Requirements Evaluation

Data Description

Submission Requirements

For each image in the test set, it is necessary to predict the prompts used to generate the image and convert the prompts into an embedding vector of 384 length. The prediction should be flattened into rows of image (imgId) and embedded (eId) pairs (imgId_eId).

Evaluation

Average cosine similarity between predicted and actual hints embedded vectors

Score to evaluate submission. For precise details on how to calculate the embedding of annotated prompts,

Refer to the relevant reference codes of the organizing committee.