Hello,

node_post processors into my chat engine directly -- not sure what this means?

CondensePlusContextChatEngine with the query engine using the post processor it gives me the following error -- seems like a bug, its not properly passing in a query bundle object

When using something like this

“‘chat_engine = index.as_chat_engine(streaming=True,
                                   use_async=True,
                                   chat_mode = 'condense_plus_context',
                                   llm = llm,
                                   verbose=True,
                                   similarity_top_k = 3,
                                   node_postprocessors = [rerank],
                                   memory = memory,
                                   ) ’”
It does not use the post processors at all. It seems like the kwargs is not taken into account.

Looking at the source code for CondensePlusContext, it should be? But normally you'd want to set the top-k kind of large, and then filter down with reranking to a smaller set of nodes

If the rerank top-n is also 3, its just reordinging them

Yes, when I’m setting a large top k, like 15 it raises a warning about the token limit being too much, I’m setting reranker to 2 there but still getting 3

The thing is that this way, I don’t get the attribute error, so I was thinking maybe the kwargs is not passed through properly 🫣