Anthropic function calling for structured LLM outputs

Anthropic recently added tool use to their API, which is extremely useful for structured outputs. Here we explain how tool use works, how to use it with llm.with_structured_output(...) in LangChain, and how to design fallbacks to catch errors in the validation / parsing of structured output according to a user-defined schema.
Notebook:
github.com/langchain-ai/langc...

[@cemerdem3136]

When can we use Claude 3 models with langchain agents like gpt?

[@jay-dj4ui]

so the key of this clip is the fall_back chain check part? which is nice to let it retry and get the result we need?

[@bharat2277]

Hey guys do you have idea about how to load custom llm model and custom chat model into langchain , where my model are hosted on ec2. I have gone through custom llm chat and model those topic were not much helpfull

[@SaudBako]

I forgot how much computer people like watch blocks and blocks of text.

[@ArcheNovalis]

Looks like it can output improper syntax. A better solution is to use outlines-dev so that it is impossible for the model to select tokens that would make it output improper syntax.

[@ScottzPlaylists]

👎Claude API access is Paid. 👎
Are there any Free API access LLMs, without running them local? ❓

[@thngzys]

Why should someone else pay for your use?

[@ScottzPlaylists]

@@thngzys So much AI stuff is free, I don't know how they all manage to make money. I poor bloke an hope and ask though right❓