With Langchain moving to V1.0, what changes are needed for using Tavily searches either as a retriever or a tool? Thanks.
Hi there,
No changes are needed — Tavily will continue to work as a retriever or tool with LangChain V1.0.
May
If I want to limit the number of search returns at 10 for TavilySearchAPIRetriever(), do I set k=10 or max_results=10? Or they mean different things? Thanks.
Hi @DavidL
I’d recommend trying out our official integration.
Hi @Shubh this copy of integration guide does not mention Langchain retriever. Does Tavily still support retriever?
Hi @DavidL
We do support the retriever. If you’d like to continue using it, you can set the parameter k (for example: TavilySearchAPIRetriever(k=10)).
However, we recommend using our official integration for the best experience and continued support.
Hi @Shubh
Thanks for getting back to me. I’d like to request that Tavily Search continue to support TavilySearchAPIRetriever(). There are some situations that Retrievers work better than Agents. For example, one can deterministically call Retrievers, without letting LLM decide which tools to use.
Please share my feedback with your product team. Thank you.
David