The community member posted a feature request to create an embeddings class for jina.ai to leverage their "late embeddings" approach via Hugging Face. They provided sample code that still has some custom methods. The comments suggest that this could be implemented as a node parser/text splitter, but one community member is not entirely convinced it's worth the effort to contribute this themselves. Another community member notes that most of the code could be copied into a node-parser integration package implementing the BaseNodeParser class, and that the code uses jina.ai's semantic splitter.
Hi, just a random feature request. Is it possible to create a an embeddings class for jina.ai to leverage their new "late embeddings" approach via HF? Here is the sample code which still has a bunch of custom methods. https://github.com/jina-ai/late-chunking/tree/main/chunked_pooling
