I would like to get the embedding vector as ouput. Is it supported?

#7
by iwonder87 - opened

I would like to get the embedding vector as ouput. Is it supported?
For example, I have a image, I enter a question about the image. I hope to get an embedding vector as an output.

OpenBMB org
edited Aug 5

hi, you should Revise the get_vlllm_embeding function in the modeling_minicpmv.py to get vision_embedding.

hi, you should Revise the get_vlllm_embeding function in the modeling_minicpmv.py to get vision_embedding.

Thanks very much. I manage to use the function to get embeding. One more question, how to set the parameter of data['image_bound'], data['tgt_sizes'] and data['pixel_values']. Is there any demo for it?
Although I mange to run it without error, I'm not sure the result is right or not.

Sign up or log in to comment