Why not remove describe text embedding in image_embeddings output
kiendn1 opened this issue · 2 comments
kiendn1 commented
I am using the vidore/colqwen2-v0.1 model to embed documents and queries. In the code, I noticed that you removed pixel_values during the query processing, but you didn't remove the described embedding during the image processing. Does this affect the results of late interaction search?
tonywu71 commented
Hi @kiendn1, sorry for the wait 👋🏼
May I ask which line you are referring to? In the main branch, I'm not seeing any reference to pixel_values in the process_queries method.
ManuelFay commented
yeah don't quite get this question - queries have no pixel values, images do...