I was wondering how we can perform text-to-shape and shape-to-text retrieval evaluation on the test split (i.e. get recall rate, NDCG etc).
From my understanding, the tools/scripts/generate_text_embeddings.sh script evaluates the text-to-text and shape-to-shape retrieval. Is that right?
I was wondering how we can perform text-to-shape and shape-to-text retrieval evaluation on the test split (i.e. get recall rate, NDCG etc).
From my understanding, the
tools/scripts/generate_text_embeddings.shscript evaluates the text-to-text and shape-to-shape retrieval. Is that right?