I am trying to see if there is a way to run my own node_postprocessor over an HTTP call . The reason for this is the main LlamaIndex ragapp is a lambda and the startup costs of loading a bge re-ranker is high so I was going to have it pre-loaded behind a mini flask app. Are there any guides on how to make a huggingface reranker model like BGE or JINA work behind an API the way like, a cohere or others do? Would i have to write my own postprocessor implementation that wraps this ?
