Hey

There are ways to manipulate the docs in an index and then persist them. But I didn't find a way to update certain nodes.

For example using a node ID update it's metadata and persist back to the vectorstoreDB

Hola:LlamaIndex:

How can I use huggingface reranker models for the node postprocessing?

Should I create a custom implementation or is there something ready out of the box like the embedding model where you just insert the model name?

Hey 👋
In QuestionsAnsweredExtractor I want to add a custom prompt template and have additional context variables to this template

Under

It calls self.llm.apredict() with the context variables num_questions and context_str.

Is it possible to add more variables dynamically?