Maxim issues

Repositories
Issues
Comments

Results 1 issues of


                                            Maxim

Combining the CLIP embeddings of text and KNNs: conditional_retrieval_encoder

What is the idea behind the unimplemented `conditional_retrieval_encoder` [here](https://github.com/CompVis/retrieval-augmented-diffusion-models/blob/1017f2b51d43ff530f178c72c460bbba83316da6/rdm/models/diffusion/ddpm.py#L811)? Would this be another encoder to combine the CLIP embs of the original text query and KNNs?