search

deepset-ai / haystack

:mag: AI orchestration framework to build customizable, production-ready LLM applications. Connect components (models, vector DBs, file converters) to pipelines or agents that can interact with your data. With advanced retrieval methods, it's best suited for building RAG, question answering, semantic search or conversational agent chatbots.
https://haystack.deepset.ai
Apache License 2.0
16.67k stars 1.83k forks source link

Output shaper to add the output of the LLM to the document's metadata #4926

Closed ZanSara closed 6 months ago

ZanSara commented 1 year ago

Is your feature request related to a problem? Please describe. Summarizer and document classifiers can currently add their output to the metadata of a document. Right now, it seems like PromptNode can't do the same due to the lack of an output shaper that supports this case.

Describe the solution you'd like Add an output shaper that populates the metadata.

tholor commented 1 year ago

Especially interesting for usage in indexing pipelines. There it would allow us to easily add category tags, sentiment, summaries (...) to docs that can be utilized later at query time (e.g. to filter down the search space efficiently or utilize the metadata for online retrieval/generation steps)

sjrl commented 1 year ago

We have a request for this from one of our clients as well. Right now we are having to use the TransformerSummarizer in the mean time.

anakin87 commented 6 months ago

Closing in favor of #5702 (for 2.x) Won't fix in 1.x