Closed MaartenGr closed 2 years ago
Extract the textual representation not through cosine similarity of embeddings but by generating a set of words for each image and running c-TF-IDF over the clusters of words.
Extract the textual representation not through cosine similarity of embeddings but by generating a set of words for each image and running c-TF-IDF over the clusters of words.